Psychotherapy Relationships That Work

Psychotherapy Relationships That Work
This page intentionally left blank

Psychotherapy
Relationships
That Work
Evidence-Based
Responsiveness
Second Edition
Edited by
John C. Norcross
1
1
Published in the United States of America by Oxford University Press, Inc.,
198 Madison Avenue, New York, NY, 10016
United States of America
Oxford University Press, Inc., publishes works that further Oxford University’s
objective of excellence in research, scholarship, and education.
Oxford is a registered trade mark of Oxford University Press
in the UK and in certain other countries
© Oxford University Press, Inc. 2011
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, without the prior permission in writing of Oxford
University Press, Inc., or as expressly permitted by law, by licence, or under terms agreed with the
appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope
of the above should be sent to the Rights Department, Oxford University Press, Inc., at the
address above
You must not circulate this work in any other form and you must impose this same condition on
any acquirer
______________________________________________
Library of Congress Cataloging-in-Publication Data
Psychotherapy relationships that work : evidence-based

responsiveness / edited by John C. Norcross. — 2nd ed.
p. ; cm.
Includes bibliographical references.
ISBN 978-0-19-973720-8 (alk. paper)
1. Psychotherapist and patient. 2. Evidence-based psychotherapy.
I. Norcross, John C., 1957-
[DNLM: 1. Professional-Patient Relations—Meta-Analysis.
2. Psychotherapy—methods—Meta-Analysis. 3. Evidence-Based
Practice—Meta-Analysis. WM 420]
RC480.8.P78 2011
616.89’14—dc22
2010037228
______________________________________________
978-0-19-973720-8
1 3 5 7 9 10 8 6 4 2
Typeset in Adobe Garamond Pro
Printed on acid-free pape
Printed in the United States of America
Dedicated to
Emma and Owen

Daily reminders of the healing power of nurturing relationships
Arnold A. Lazarus, Ph.D.

Lifelong champion of adapting psychotherapy to the individual patient
P R E FA C E
A cordial welcome to the second edition of a single professional association (Division

Psychotherapy Relationships That Work. This of Psychotherapy), but this second edition
volume seeks, like its predecessor, to iden- was sponsorship by both the American
tify effective elements of the psychotherapy Psychological Association (APA) Division
relationship and to determine effective of Clinical Psychology and the APA
methods of adapting or tailoring that rela- Division of Psychotherapy. Second, we
tionship to the individual patient. That is, retitled the focus evidence-based psychother-
we summarize the empirical research apy relationships instead of empirically sup-
and clinical practice on what works ported (therapy) relationships to parallel the
in general as well as what works in particu- contemporary movement to the newer ter-
lar. This dual focus has been characterized minology. This title change, in addition,
as “two books in one”: one book on rela- properly emphasizes the confluence of the
tionship elements and one book on adapta- best research, clinical expertise, and patient
tion methods. characteristics in a quality treatment rela-
My hope in this book, as with the first tionship. Third, we expanded the breadth
edition, is to advance a rapprochement of coverage. New reviews were commis-
between the warring factions in the culture sioned on the alliance with children and
wars of psychotherapy and to demonstrate adolescents, the alliance in couple and
that the best available research clearly dem- family therapy, collection of real-time
onstrates the healing qualities of the ther- feedback from clients, patient preferences,
apy relationship. The first edition brought culture, and attachment style. Fourth, we
renewed and corrective attention to the decided to insist on meta-analyses for all
substantial research behind the therapy research reviews. These original meta-anal-
relationship and, in the words of one yses enable direct estimates of the magni-
reviewer (Psychotherapy Research, 2003, tude of association and the ability to search
p. 532), “will convince most psychothera- for moderators. Unfortunately, that also
pists of the rightful place of ESRs (empiri- meant that several relationship elements
cally supported relationships) alongside and adaptation methods in the first edition
ESTs in the treatments they provide.” Note (self-disclosure, transference interpreta-
the desired emphasis on “alongside” treat- tions, anaclitic vs. introjective styles,
ments, not “instead of ” or “better than.” assimilation of problematic experiences)
were excluded due to their insufficient
Changes in the New Edition number of studies. Fifth, we improved the
The aims of this edition of Psychotherapy process for determining whether a particu-
Relationships That Work remain the same as lar relationship element—say, the alliance
its predecessor, but its sponsorship, meth- or empathy—could be classified as
odology, and scope differ. First, the inaugu- demonstrably effective, probably effective,
ral edition of the book was sponsored by or promising but insufficient research
vi i
to judge. We constituted expert panels to evidence-based practices or treatment
establish a consensus on the evidentiary guidelines. We hope our work will inform
strength of the relationship elements and and balance any efforts to focus exclusively
adaptation methods. Experts indepen- on techniques or treatments to the neglect
dently reviewed and rated the meta-analy- of the humans involved in the enterprise.
ses on several objective criteria, thus adding Our third priority was insurance carriers
a modicum of rigor and consensus to the and accreditation organizations, many of
process, which was admittedly less so in the which have unintentionally devalued
first edition of the book. the person of the therapist and the
The net result is a compilation of two centrality of the relationship by virtue of
dozen, cutting-edge meta-analyses devoted reimbursement decisions. Although sup-
to what works in the therapy relationship portive of the recent thrust toward science
and what works in adapting that relation- informing practice, we must remind all
ship to the individual client and his/ parties to the therapy relationship that
her singular situation. This new edition, healing cannot be replaced with treating,
appearing 10 years after the first incarna- caring cannot be supplanted by managing.
tion, presents a slightly slimmer book offer- Finally, Psychotherapy Relationships That
ing more practical, bulleted information Work is intended for psychotherapy
on clinical practice at the end of each researchers seeking a central resource on
chapter. the empirical status of the multiple, inter-
dependent qualities of the therapy
Probable Audiences relationship.
One of our earliest considerations in plan-
ning the first edition of the book concerned Organization of the Book
the intended audiences. Each of psycho- The opening chapter introduces the book
therapy’s stakeholders—patients, practitio- by outlining the purpose and history of the
ners, researchers, trainers, students, interdivisional Task Force and its relation
organizations, insurance companies, and to previous efforts to identify evidence-
policymakers—expressed different prefer- based practices in psychotherapy. This
ences for the content and length of the chapter also presents the key limitations of
volume. our work.
We prepared the book for multiple audi- The heart of the book is composed of
ences but in a definite order of priority. research reviews on the therapist’s relational
First came clinical practitioners and train- contributions and recommended therapeu-
ees of diverse theoretical orientations and tic practices predicated on that research.
professional disciplines. They need to Section II—Effective Elements of the
address urgent pragmatic questions: What Therapy Relationship: What Works in
do we know from the empirical research General—features eleven chapters on rela-
about cultivating and maintaining the tionship elements primarily provided by
therapeutic relationship? What are the the psychotherapist. Chapters 2–5 report
research-supported means of adapting that on broader, more inclusive relationship ele-
relationship to the patient beyond his/her ments. The therapy alliance and group
diagnosis? Our second priority was accorded cohesion are composed, in fact, of multiple
to the mental health disciplines themselves, elements. Chapters 6–9 feature more
specifically those committees, task forces, specific elements of the therapy relation-
and organizations promulgating lists of ship, and Chapters 10–12 review specific
viii pre fac e
therapist behaviors that promote the rela- Report the effect size as weighted r
tionship and favorable treatment results. (in Section II) or d (in Section III).
Section III—Tailoring the Therapy Relation- • Moderators. Present the results of the
ship to the Individual Patient: What Works moderator analyses on the association
in Particular—features eight chapters on between the relationship element and
adaptation methods. They feature probably treatment outcome.
and demonstrably effective means of tailor- If available in the studies, examine the
ing psychotherapy to the entire person possible moderating effects of
beyond diagnosis alone. (1) rater perspective (assessed by therapist,
The final section of the book consists of patient, or external raters),
a single chapter. It presents the Task Force (2) therapist variables, (3) patient factors,
conclusions, including a list of evidence- (4) different measures,
based relationship elements and adaptation (5) time of assessment (when in the course
methods, and our recommendations, of therapy), and (6) type
divided into general, practice, training, of psychotherapy/theoretical orientation.
research, and policy recommendations. • Patient Contribution. The meta-
analyses pertain largely to the
Chapter Guidelines psychotherapist’s contribution to the
With the exception of the bookends relationship; by contrast, this section will
(Chapters 1 and 21), all chapters use the address the patient’s contribution to that
same section headings and adhere to a con- relationship and
sistent structure, as follows: the distinctive perspective he/she brings to
the interaction.
• Introduction (untitled). Introduce the
• Limitations of the Research. Point to
relationship element or the adaptation
the major limitations of both the meta-
method and its historical context.
analysis and the available studies.
• Definitions and Measures. Define in
• Therapeutic Practices. Emphasize what
theoretically neutral language
works. Bullet the practice implications
the relationship element or adaptation
from the foregoing research, primarily in
method. Identify any highly similar
terms of the therapist’s contribution and
or equivalent constructs from
secondarily in terms of the patient’s
diverse theoretical traditions.
perspective.
Review the popular measures used
in the research and included in the These research reviews are based on the
ensuing meta-analysis. results of empirical research linking the
• Clinical Example. Provide several relationship element or adaptation method
concrete examples of the relationship to psychotherapy outcome. Outcome was
behavior being reviewed. Portions of inclusively defined but consisted largely of
psychotherapy transcripts are encouraged. distal posttreatment outcomes. Authors
• Meta-Analytic Review. Compile all were asked to specify the outcome criterion
available empirical studies linking the when a particular study did not employ
relationship behavior to treatment a typical end-of-treatment measure of
outcome in the English language. Use the symptom or functioning. Indeed, the type
Meta-Analysis Reporting Standards of outcome measure was frequently ana-
(MARS) as a general guide for the lyzed as a possible moderator of the overall
information included in the chapter. effect size.
p re fac e ix
Acknowledgments sharing their expertise. Dr. Bruce Wampold
Psychotherapy Relationships That Work would expertly reviewed each meta-analysis and
not have been possible without a decade of provided valuable guidance on the entire
organizational and individual support. On project. Members of the expert consensus
the organizational front, the board of direc- panels critiqued each meta-analysis and
tors of the APA Division of Psychotherapy rated the evidentiary strength of the results;
and the APA Division of Clinical Psychology I appreciate the generosity of Drs. Guillermo
commissioned and supported the Task Bernal, Franz Caspar, Louis Castonguay,
Force. In particular, I am indebted to the Charles Gelso, Mark Hilsenroth, Michael
presidents of the respective divisions: Drs. Lambert, and Bruce Wampold. The Steering
Jeffrey Barnett, Nadine Kaslow, and Jeffrey Committee of the first Task Force assisted
Magnavita of the psychotherapy division, in canvassing the literature, defining the
and Drs. Marsha Linehan, Irving Weiner, parameters of the project, selecting the
and Marvin Goldfried of the clinical divi- contributors, and writing the initial con-
sion. At Oxford University Press, Joan clusions. I am grateful to them all: Steven
Bossert shepherded both books through J. Ackerman, Lorna Smith Benjamin,
the publishing process and recognized early Larry E. Beutler, Charles J. Gelso, Marvin
on that they would compliment Oxford’s R. Goldfried, Clara E. Hill, Michael
landmark Treatments That Work. This J. Lambert, David E. Orlinsky, and Jackson
second edition has been improved by the P. Rainer. Last but never least, my
OUP book team of Sarah Harrington, Jodi immediate family—Nancy, Jonathon, and
Nardi, and Tony Orrantia. Rebecca—tolerated my absences, preoccu-
On the individual front, many people pations, and irritabilities associated with
modeled and manifested the ideal thera- editing this book with a combination of
peutic relationship throughout the course empathy and patience that would do any
of the project. The authors of the respective seasoned psychotherapist proud.
chapters, of course, were indispensible John C. Norcross, PhD
in generating the research reviews and in Clarks Summit, Pennsylvania
x pre fac e
CO N T R I B U TO R S
Jennifer Alonso, B.S. AC Del Re, M.A.

Department of Psychology, Department of Counseling Psychology,
Brigham Young University University of Wisconsin–Madison
Rebecca M. Ametrano, B.A. Gary M. Diamond, Ph.D.
Department of Psychology, Department of Psychology,
University of Massachusetts Amherst Ben-Gurion University of the Negev
Diane B. Arnkoff, Ph.D. Erin M. Doolin, M.Ed.
Department of Psychology, Department of Counseling Psychology,
Catholic University of America University of Wisconsin–Madison
Sara B. Austin, B.S. Robert Elliott, Ph.D.
Department of Psychology, University of School of Psychological Sciences and
Wisconsin–Madison Health, University of Strathclyde
Guillermo Bernal, Ph.D. William D. Ellison, M.S.
Institute for Psychological Research, Department of Psychology,
University of Puerto Rico Pennsylvania State University
Samantha L. Bernecker, B.S. Valentín Escudero, Ph.D.
Department of Psychology, Departamento de Psicología,
Pennsylvania State University Universidad de A Coruña
Larry E. Beutler, Ph.D. Catherine Eubanks-Carter, Ph.D.
Pacific Graduate School of Psychology, Ferkauf Graduate School of Psychology,
Palo Alto University Yeshiva University
Kathy Blau, M.S. Barry A. Farber, Ph.D.
Pacific Graduate School of Psychology, Department of Counseling and Clinical
Palo Alto University Psychology, Teachers College
Arthur C. Bohart, Ph.D. Columbia University
Department of Psychology, California Christoph Flückiger, Ph.D.
State University–Dominguez Hills and Department of Clinical Psychology and
Graduate College of Psychology and Psychotherapy, University of Bern
Humanistic Studies, Saybrook University Myrna L. Friedlander, Ph.D.
Gary M. Burlingame, Ph.D. Department of Educational and
Department of Psychology, Counseling Psychology, University at
Brigham Young University Albany/State University of New York
Jennifer L. Callahan, Ph.D. Charles J. Gelso, Ph.D.
University of North Texas University of Maryland-College Park
Michael J. Constantino, Ph.D. Carol R. Glass, Ph.D.
University of Massachusetts-Amherst Catholic University of America
Don E. Davis, M.S. Leslie S. Greenberg, Ph.D.
Virginia Commonwealth University York University
xi
T. Mark Harwood, Ph.D. Michael A. McDaniel, Ph.D.
Private Practice Department of Management,
Chicago, Illinois Virginia Commonwealth University
Jeffrey A. Hayes, Ph.D. Aaron Michelson, M.S.
Counseling Psychology Program, Pacific Graduate School of Psychology,
Pennsylvania State University Palo Alto University
Laurie Heatherington, Ph.D. J. Christopher Muran, Ph.D.
Department of Psychology, Derner Institute of Advanced
Williams College Psychological Studies,
John Holman, M.S. Adelphi University
Pacific Graduate School of Psychology, John C. Norcross, Ph.D.
Palo Alto University Department of Psychology,
Joshua N. Hook, Ph.D. University of Scranton
Department of Psychology, James O. Prochaska, Ph.D.
University of North Texas Cancer Prevention Research Consortium,
Adam O. Horvath, Ed.D. University of Rhode Island
Faculty of Education & Department of Melanie M. Domenech Rodríguez, Ph.D.
Psychology, Simon Fraser University Department of Psychology,
Ann M. Hummel, M.S. Utah State University
Department of Psychology, Jeremy D. Safran, Ph.D.
University of Maryland-College Park Department of Psychology,
Marc S. Karver, Ph.D. New School for Social Research
Department of Psychology, Lori N. Scott, M.S.
University of South Florida Department of Psychology,
Satoko Kimpara, Ph.D. Pennsylvania State University
Pacific Graduate School of Psychology, Kenichi Shimokawa, Ph.D.
Asian American Community Involvement Family Institute,
(AACI) and Palo Alto University Northwestern University
Marjorie H. Klein, Ph.D. Stephen R. Shirk, Ph.D.
Department of Psychiatry, Department of Psychology,
University of Wisconsin–Madison University of Denver
Gregory G. Kolden, Ph.D. JuliAnna Z. Smith, M.A.
Department of Psychiatry and Center for Research on Families,
Psychology, University of University of Massachusetts Amherst
Wisconsin–Madison Timothy B. Smith, Ph.D.
Paul M. Krebs, Ph.D. Department of Counseling
Department of General Internal Psychology and Special Education,
Medicine, New York University Brigham Young University
Medical Center Xiaoxia Song, Ph.D.
Michael J. Lambert, Ph.D. Department of Psychology,
Department of Psychology, Ohio University
Brigham Young University Joshua K. Swift, Ph.D.
Kenneth N. Levy, Ph.D. Department of Psychology,
Department of Psychology, University of Alaska Anchorage
Pennsylvania State University Dianne Symonds, Ph.D.
Debra Theobald McClendon, Ph.D. Faculty of Community and
Department of Psychology, Health Studies, Kwantlen
Brigham Young University Polytechnic University
xii co n t r i bu to r s
Georgiana Shick Tryon, Ph.D. Chia-Chiang Wang, M.Ed.
Ph.D. Program in Educational Department of Rehabilitation
Psychology, The Graduate Center, Psychology and Special Education,
City University of New York University of Wisconsin–Madison
David Verdirame, M.S. Jeanne C. Watson, Ph.D.
Pacific Graduate School of Department of Adult Education,
Psychology, Palo Alto University University of Toronto
Barbara M. Vollmer, Ph.D. Greta Winograd, Ph.D.
Department of Counseling Psychology Department,
Psychology, University of State University of
Denver New York-New Paltz
Bruce E. Wampold, Ph.D. Everett L. Worthington, Jr., Ph.D.
Department of Counseling Department of Psychology,
Psychology, University of Virginia Commonwealth
Wisconsin–Madison University
co n t r i bu to r s xiii
TA B L E O F C O N T E N T S
Part One • Introduction

1. Evidence-Based Therapy Relationships 3
John C. Norcross and Michael J. Lambert
Part Two • Effective Elements of the Therapy Relationship:

What Works in General
2. Alliance in Individual Psychotherapy 25
Adam O. Horvath, A. C. Del Re, Christopher Flückiger, and Dianne Symonds
3. Alliance in Child and Adolescent Psychotherapy 70
Stephen R. Shirk and Marc S. Karver
4. Alliance in Couple and Family Therapy 92
Myrna L. Friedlander, Valentín Escudero, Laurie Heatherington, and
Gary M. Diamond
5. Cohesion in Group Therapy 110
Gary M. Burlingame, Debra Theobald McClendon, and Jennifer Alonso
6. Empathy 132
Robert Elliott, Arthur C. Bohart, Jeanne C. Watson, and
Leslie S. Greenberg
7. Goal Consensus and Collaboration 153
Georgiana Shick Tryon and Greta Winograd
8. Positive Regard and Affirmation 168
Barry A. Farber and Erin M. Doolin
9. Congruence/Genuineness 187
Gregory G. Kolden, Marjorie H. Klein, Chia-Chiang Wang, and
Sara B. Austin
10. Collecting Client Feedback 203
Michael J. Lambert and Kenichi Shimokawa
11. Repairing Alliance Ruptures 224
Jeremy D. Safran, J. Christopher Muran, and Catherine Eubanks-Carter
12. Managing Countertransference 239
Jeffrey A. Hayes, Charles J. Gelso, and Ann M. Hummel
xv
Part Three • Tailoring the Therapy Relationship to the Individual
Patient: What Works in Particular
13. Reactance/Resistance Level 261
Larry E. Beutler, T. Mark Harwood, Aaron Michelson, Xiaoxia Song,
and John Holman
14. Stages of Change 279
John C. Norcross, Paul M. Krebs, and James O. Prochaska
15. Preferences 301
Joshua K. Swift, Jennifer L. Callahan, and Barbara M. Vollmer
16. Culture 316
Timothy B. Smith, Melanie Domenech Rodríguez, and Guillermo Bernal
17. Coping Style 336
Larry E. Beutler, T. Mark Harwood, Satoko Kimpara, David Verdirame, and
Kathy Blau
18. Expectations 354
Michael J. Constantino, Carol R. Glass, Diane B. Arnkoff,
Rebecca M. Ametrano, and JuliAnna Z. Smith
19. Attachment Style 377
Kenneth N. Levy, William D. Ellison, Lori N. Scott, and
Samantha L. Bernecker
20. Religion and Spirituality 402
Everett L. Worthington, Jr., Joshua N. Hook, Don E. Davis, and
Michael A. McDaniel
Part Four • Conclusions and Guidelines

21. Evidence-Based Therapy Relationships: Research Conclusions and
Clinical Practices 423
John C. Norcross and Bruce E. Wampold
Index 431
xvi ta b l e o f co n t e n ts
PART
1
Introduction
C HA P TER
1 Evidence-Based Therapy Relationships
John C. Norcross and Michael J. Lambert
The culture wars in psychotherapy dramati- controlled/clinical trial), and the scientific-
cally pit the treatment method against medical model? Or do you belong to
the therapy relationship. Do treatments the side of the therapy relationship, the
cure disorders or do relationships heal effectiveness and process-outcome studies,
people? Which is the most accurate vision and the relational-contextual model? Such
for practicing, researching, and teaching polarizations not only impede psychothera-
psychotherapy? pists from working together but also hinder
Like most dichotomies, this one is mis- attempts to provide the most efficacious psy-
leading and unproductive on multiple chological services to our patients.
counts. For starters, the patient’s contribu- We hoped that a balanced perspective
tion to psychotherapy outcome is vastly would be achieved by the adoption of an
greater than that of either the particular inclusive, neutral definition of evidence-
treatment method or the therapy relation- based practice. The American Psychological
ship (Lambert, 1992). The empirical evidence Association (2006, p. 273) did endorse just
should keep us mindful and a bit humble such a definition: “Evidence-based practice
about our collective tendency toward ther- in psychology (EBPP) is the integration of
apist centricity (Bohart & Tallman, 1999). the best available research with clinical
For another, decades of psychotherapy expertise in the context of patient charac-
research consistently attest that the patient, teristics, culture, and preferences.” However,
the therapist, their relationship, the treat- even that definition has been comman-
ment method, and the context all contrib- deered by the rival camps as polarizing
ute to treatment success (and failure). We devices. On the one side, some erroneously
should be looking at all of these determi- equate EBP solely with the best available
nants and their optimal combinations research and particularly the results of
(Norcross, Beutler, & Levant, 2006). RCTs on treatment methods, while on the
But perhaps the most pernicious and other side, some mistakenly exaggerate the
insidious consequence of the false dichot- primacy of clinical or relational expertise
omy of treatment versus relationship has while neglecting research support.
been its polarizing effect on the discipline. Within this polarizing context, in 1999,
Rival camps have developed, and countless the American Psychological Association
critiques have been published on each side (APA) Division of Psychotherapy com-
of the culture war. Are you on the side of the missioned a task force to identify, opera-
treatment method, the RCT (randomized tionalize, and disseminate information on
3
empirically supported therapy relation- the therapy relationship and to determine
ships. That task force summarized its find- effective methods of adapting or tailoring
ings and detailed its recommendations in therapy to the individual patient on the
the first edition of this book (Norcross, basis of his/her (transdiagnostic) character-
2002). In 2009, the Division of Psycho- istics. In other words, we were interested in
therapy along with the Division of Clinical both what works in general and what works
Psychology commissioned a second task for particular patients. This twin focus has
force on evidence-based therapy relation- been characterized as “two books in one”:
ships to update the research base and one book on relationship elements and one
clinical practices on the psychotherapist– book on adaptation methods under the
patient relationship. This second edition, same cover.
appearing 10 years after its predecessor, does For the purposes of our work, we again
just that. adopted Gelso and Carter’s (1985, 1994)
Our hope now, as then, is to advance a operational definition of the therapy rela-
rapprochement between the warring fac- tionship: The relationship is the feelings
tions and to demonstrate that the best and attitudes that therapist and client have
available research clearly supports the heal- toward one another, and the manner in
ing qualities of the therapy relationship and which these are expressed. This definition is
the beneficial value of adapting that rela- quite general, and the phrase “the manner
tionship to patient characteristics beyond in which it is expressed” potentially opens
diagnosis. The bulk of the book summa- the relationship to include everything
rizes the best available research and clinical under the therapeutic sun (see Gelso &
practices on numerous elements of the Hayes, 1998, for an extended discussion).
therapy relationship and on several meth- Nonetheless, it serves as a concise, consen-
ods of treatment adaptation. In doing so, sual, theoretically neutral, and sufficiently
our grander goal is to repair some of the precise definition.
damage incurred by the culture wars in We acknowledge the deep synergy
psychotherapy and to promote integration between treatment methods and the thera-
between science and practice. peutic relationship. They constantly shape
In this chapter, we begin by tracing the and inform each other. Both clinical expe-
purpose and processes of the interdivi- rience and research evidence (e.g., Rector,
sional Task Force. We explicate the need for Zuroff, & Segal, 1999; Barber et al., 2006)
identifying evidence-based elements of the point to a complex, reciprocal interaction
therapy relationship and means of match- between the interpersonal relationship and
ing or adapting treatment to the individual. the instrumental methods. Consider this
In a tentative way, we offer two models to finding from a large collaborative study:
account for psychotherapy outcome as a For patients with a strong therapeutic alli-
function of various therapeutic factors (e.g., ance, adherence to the treatment manual
patient, relationship, technique). The latter was irrelevant for treatment outcome, but
part of the chapter features the limitations for patients with a weak alliance, a moder-
of the Task Force’s work and responds to ate level of therapist adherence was associ-
frequently asked questions. ated with the best outcome (Barber et al.,
2006). The relationship does not exist apart
The Interdivisional Task Force from what the therapist does in terms of
The dual purposes of the interdivisional Task method, and we cannot imagine any treat-
Force were to identify effective elements of ment methods that would not have some
4 i n t ro d u c t i o n
relational impact. Put differently, treat- promulgation of effective psychotherapy. It
ment methods are relational acts (Safran & does so by expanding or enlarging the typi-
Muran, 2000). cal focus of evidence-based practice to ther-
For historical and research convenience, apy relationships. Focusing on one area—
the field has distinguished between rela- in this case, the therapy relationship—may
tionships and techniques. Words like unfortunately convey the impression that
“relating” and “interpersonal behavior” are this is the only area of importance. We
used to describe how therapists and clients review the scientific literature on the
behave toward each other. By contrast, therapy relationship and provide clinical
terms like “technique” or “intervention” are recommendations based on that literature
used to describe what is done by the thera- without, we trust, degrading the simultane-
pist. In research and theory, we often treat ous contributions of the treatments, patients,
the how and the what—the relationship or therapists to outcome. Indeed, we wish
and the intervention, the interpersonal and that more psychotherapists would acknowl-
the instrumental—as separate categories. edge the inseparable context and practical
In reality, of course, what one does and interdependence of the relationship and the
how one does it are complementary and treatment. That can prove a crucial step in
inseparable. To remove the interpersonal reducing the polarizing strife of the culture
from the instrumental may be acceptable wars and in improving the effectiveness of
in research, but it is a fatal flaw when psychotherapy (Lambert, 2010).
the aim is to extrapolate research results An immediate challenge to the Task
to clinical practice (see Orlinsky, 2000; Force was to establish the inclusion and
2005 special issue of Psychotherapy on the exclusion criteria for the elements of the
interplay of techniques and therapeutic therapy relationship. We readily agreed that
relationship). the traditional features of the therapeu-
In other words, the value of a treatment tic relationship—the alliance in individual
method is inextricably bound to the rela- therapy and cohesion in group therapy, for
tional context in which it is applied. Hans example—and the Rogerian facilitative
Strupp, one of our first research mentors, conditions would constitute core elements.
offered an analogy to illustrate the insepa- We further agreed that discrete, relatively
rability of these constituent elements. nonrelational techniques were not part of
Suppose you want your teenager to clean our purview, but that a few relational meth-
his or her room. Two methods for achiev- ods would be included. Therapy methods
ing this are to establish clear standards were considered for inclusion if their con-
and to impose consequences. A reasonable tent, goal, and context were inextricably
approach, but the effectiveness of these interwoven into the emergent therapy rela-
two evidence-based methods will vary on tionship. We settled on three relationship
whether the relationship between you and behaviors (collecting real-time client feed-
the teenager is characterized by warmth back, repairing alliance ruptures, and man-
and mutual respect or by anger and mis- aging countertransference) because these
trust. This is not to say that the methods methods are deeply embedded in the inter-
are useless, merely that how well they work personal character of the relationship itself.
depends upon the context in which they But which relational behaviors to include
are used (Norcross, 2010). and which to exclude under the rubric of
The work of the Task Force applies psy- the therapy relationship bedeviled us, as it
chological science to the identification and has the field.
n o rc ro s s , l a m b e rt 5
How does one divide the indivisible rela- Their respective ratings of demonstrably
tionship? For example, is support similar effective, probably effective, or promising
enough to positive regard or validation to be but insufficient research to judge were then
considered in the same meta-analysis, or is combined to render a consensus. These
it distinct enough to deserve a separate conclusions are presented in the last chap-
research review? We struggled on how finely ter of this book.
to slice the therapy relationship. As David The deliberations of the Steering
Orlinsky opined in one of his e-mail mes- Committee and the expert panels were not
sages, “It’s okay to slice bologna that thin, easy or unanimous. Democracy is messy
but I doubt that it can be meaningfully and inefficient; science is even slower and
done to the relationship.” We agreed, as a painstaking. We debated and, in most
group, to place the research on support in instances, voted on our decisions. We relied
the positive regard chapter, but we under- on expert opinion, professional consensus,
stand that some practitioners may under- and most importantly, reviews of the
standably take exception to collapsing these empirical evidence. But these were all
relationship elements. As a rule, we opted human decisions—open to cavil, conten-
to divide the research reviews into smaller tion, and future revision.
chunks so that the research conclusions
were more specific and the practice impli- Therapy Relationship
cations more concrete. Recent years have witnessed the controver-
In our deliberations, several members of sial compilation of practice guidelines and
the Steering Committee advanced a favor- evidence-based treatments in mental health.
ite analogy: the therapy relationship is like In the United States and other countries,
a diamond, a diamond composed of mul- the introduction of such guidelines has
tiple, interconnected facets. The diamond provoked practice modifications, training
is a complex, reciprocal, and multidimen- refinements, and organizational conflicts.
sional entity. The Task Force endeavored to Insurance carriers and government policy-
separate and examine many of these facets. makers are increasingly turning to such
Once these decisions were finalized, we guidelines to determine which psychother-
commissioned original meta-analyses on apies to approve and fund. Indeed, along
the relationship elements and the adap- with the negative influence of managed
tation methods. The chapters and the care, there is probably issue no more cen-
meta-analyses therein were reviewed and tral to clinicians than the evolution of evi-
subsequently underwent at least one revi- dence-based practice in psychotherapy
sion. Once revised, two consensus panels (Barlow, 2000).
(each composed of five experts) were estab- All of the efforts to promulgate evidence-
lished to review the evidentiary strength based psychotherapies have been noble in
of the relationship element or adaptation intent and timely in distribution. They
method according to the following criteria: are praiseworthy efforts to distill scien-
number of empirical studies, consistency tific research into clinical applications
of empirical results, independence of sup- and to guide practice and training. They
portive studies, magnitude of association wisely demonstrate that, in a climate of
between the relationship element and out- accountability, psychotherapy stands up to
come, evidence for causal link between empirical scrutiny with the best of health
relationship element and outcome, and the care interventions. And within psychol-
ecological or external validity of research. ogy, these have proactively counterbalanced
documents that accord primacy to biomed- Person of the Therapist
ical treatments for mental disorders and Most practice guidelines and evidence-
largely ignore the outcome data for psycho- based practice compilations depict dis-
logical therapies. On many accounts, then, embodied psychotherapists performing
the extant EBP efforts have addressed the procedures on DSM disorders. This stands
realpolitik of the socioeconomic situation in marked contrast to the clinician’s experi-
(Messer, 2001; Nathan, 1998). ence of psychotherapy as an intensely inter-
At the same time, many practitioners personal and deeply emotional experience.
and researchers have found these recent Although efficacy research has gone to
efforts to codify evidence-based treatments considerable lengths to eliminate the indi-
seriously incomplete. While scientifically vidual therapist as a variable that might
laudable in their intent, these efforts have account for patient improvement, the ines-
largely ignored the therapy relationship capable fact is that it’s simply not possible
and the person of the therapist. If one were to mask the person and the contribution of
to read previous efforts literally, disembod- the therapist (Orlinsky & Howard, 1977).
ied therapists apply manualized inter- The curative contribution of the person
ventions to discrete DSM disorders. Not of the therapist is, arguably, as empirically
only is the language offensive on clinical validated as manualized treatments or
grounds to some practitioners, but the psychotherapy methods (Duncan, Miller,
research evidence is weak for validating Wampold, & Hubble, 2010).
treatment methods in isolation from the Multiple and converging sources of
therapy relationship and the individual evidence indicate that the person of the
patient. psychotherapist is inextricably intertwined
Suppose we asked a neutral scientific with the outcome of psychotherapy. A large,
panel from outside the field to review the naturalistic study estimated the outcomes
corpus of psychotherapy research to deter- attributable to 581 psychotherapists treat-
mine what is the most powerful phenome- ing 6,146 patients in a managed care set-
non we should be studying, practicing, and ting. About 5% of the outcome variation
teaching. Henry (1998, p. 128) concludes was due to therapist effects and 0% due
that the panel to specific treatment methods (Wampold
& Brown, 2005). Quantitative reviews of
would find the answer obvious, and therapist effects in psychotherapy outcome
empirically validated. As a general trend studies show consistent and robust effects—
across studies, the largest chunk of probably 5% to 9% of psychotherapy
outcome variance not attributable to outcome (Crits-Christoph et al., 1991). In
preexisting patient characteristics involves reviewing the research, Wampold (2001,
individual therapist differences and the p. 200) concluded that “a preponderance
emergent therapeutic relationship between of evidence indicates that there are large
patient and therapist, regardless of therapist effects . . . and that the effects
technique or school of therapy. This is the greatly exceed treatment effects.”
main thrust of three decades of empirical Two controlled studies examining thera-
research. pist variables in the outcomes of cognitive-
behavioral therapy are instructive (Huppert
What’s missing, in short, are the person et al., 2001; Project MATCH Research
of the therapist and elements of the thera- Group, 1998). In the Multicenter Collab-
peutic relationship. orative Study for the Treatment of Panic
Disorder, considerable care was taken to average at the center was about 8%. In a
standardize the treatment, the therapist, related study of many of the same thera-
and the patients in order to increase the pists (Anderson, Ogles, Patterson, Lambert,
experimental rigor of the study and in order & Vermeersch, 2009), the strongest predic-
to minimize therapist effects. The treat- tor of patient outcome was these therapists’
ment was manualized and structured, the interpersonal skills.
therapists were identically trained and
monitored for adherence, and the patients Relationship Elements
rigorously evaluated and relatively uniform. A second omission from most evidence-
Nonetheless, the therapists significantly based practice guidelines has been the
differed in the magnitude of change among decision to validate only the efficacy
caseloads. Effect sizes for therapist impact of treatments or technical interventions,
on outcome measures ranged from 0% as opposed to the therapy relationship or
to 18%. In the similarly controlled multi- therapist interpersonal skills. This decision
site study on alcohol abuse conducted both reflects and reinforces the ongoing
by Project MATCH, the therapists were movement toward high-quality compara-
carefully selected, trained, supervised, and tive effectiveness research (CER) on brand-
monitored in their respective treatment name psychotherapies. “This trend of putting
approaches. Although there were few out- all of the eggs in the “technique” basket
come differences among the treatments, began in the late 1970s and is now reach-
over 6% of the outcome variance (1%–12% ing the peak of influence” (Bergin, 1997,
range) was due to therapists. Despite impres- p. 83).
sive attempts to experimentally render indi- Both clinical experience and research
vidual practitioners as controlled variables, findings underscore that the therapy rela-
it is simply not possible to mask the person tionship accounts for as much of the
and the contribution of the therapist. outcome variance as particular treatment
Further evidence comes from naturalis- methods (Lambert & Barley, 2002), espe-
tic studies of clinical practice rather than cially after the effects of researcher allegiance
research settings where attempts are made to treatment are accounted for (Luborsky
to reduce individual therapist’s contribu- et al., 1999). An early and influential review
tion to patient outcomes. Okiishi, Lambert, by Bergin and Lambert (1978, p. 180)
Nielsen, and Ogles (2003) examined the anticipated the contemporary research con-
outcomes of clients seen by 56 therapists sensus: “The largest variation in therapy
practicing a variety of treatment methods. outcome is accounted for by pre-existing
Despite the fact that the psychothera- client factors, such as motivation for
pists had similarly disturbed clients, there change, and the like. Therapist personal
were dramatic differences in client outcome factors account for the second largest pro-
as a function of seeing a top-rated thera- portion of change, with technique variables
pist or one at the bottom. On average, cli- coming in a distant third.”
ents seeing a top-rated therapist achieved Even those practice guidelines enjoin-
reliable improvement, while those clients ing practitioners to attend to the therapy
seen by bottom-ranked therapists were relationship do not provide specific, evi-
unchanged or slightly worse off after treat- dence-based means of doing so. The
ment. Client deterioration for the low per- APA Template for Developing Guidelines
formers included one therapist who had (Task Force on Psychological Intervention
21% of his/her clients deteriorate while the Guidelines, 1995, pp. 5–6), for example,
sagely recognizes that factors common to Paul’s (1967) iconic question: “What treat-
all therapies, “such as the clinician’s ability ment, by whom, is most effective for this
to form a therapeutic alliance or to gener- individual with that specific problem, and
ate a mutual framework for change, are under which set of circumstances?” Every
powerful determinants of success across psychotherapist recognizes that what works
interventions” but only vaguely addresses for one person may not work for another;
how research protocols or individual prac- we seek “different strokes for different
titioners should do so. For another exam- folks.”
ple, the scholarly and comprehensive review To many, the means of such matching
on treatment choice from Great Britain was to tailor the psychotherapy to the
(Department of Health, 2001) devotes a patient’s disorder or presenting problem—
single paragraph to the therapeutic rela- that is, to find the best treatment for a
tionship. Its recommended principle is particular disorder. The research suggests
that “Effectiveness of all types of therapy that it is certainly useful for select disorders;
depends on the patient and the therapist some psychotherapies make better mar-
forming a good working relationship” riages with some mental health disorders
(p. 35), but no evidence-based guidance (e.g., Barlow, 2007; Nathan & Gorman,
is offered on which therapist behaviors 2007; Roth & Fonagy, 2004). Indeed, the
contribute to that relationship. Likewise, overwhelming majority of randomized
although most treatment manuals mention clinical trials in psychotherapy compare the
the importance of the therapy relationship, efficacy of specific treatments for specific
few specify which therapist qualities or in- disorders (Lambert, 2011).
session behaviors lead to a curative However, only matching psychotherapy
relationship. to a disorder is incomplete and not always
All of this is to say that extant lists of effective (Wampold, 2001). Particularly
EBPs and best practices in mental health absent from much of the research has been
give short shrift—some would say lip ser- the person of the patient, beyond his/her
vice—to the person of the therapist and disorder. As Sir William Osler, father of
the emergent therapeutic relationship. The modern medicine, said: “It is sometimes
vast majority of current attempts are thus much more important to know what sort
seriously incomplete and potentially mis- of a patient has a disease than what sort of
leading, both on clinical and empirical disease a patient has.”
grounds. Most practice guidelines and evidence-
based compilations unintentionally reduce
Treatment Adaptation our clients to a static diagnosis or prob-
Since the earliest days of modern psycho- lem. The impressive American Psychiatric
therapy, practitioners have realized that Association Practice Guidelines for the
treatment should be tailored to the indi- Treatment of Psychiatric Disorders (2006), to
viduality of the patient and the singularity take one prominent example, is organized
of his/her context. As early as 1919, Freud exclusively around diagnoses. Virtually all
introduced psychoanalytic psychotherapy practice guidelines are directed toward cat-
as an alternative to classical analysis on the egorical disorders. DSM diagnoses have
recognition that the more rarified approach ruled the evidence-based roost to date.
lacked universal applicability (Wolitzky, This choice flies in the face of clini-
2011). The mandate for individualizing cal practice and research findings that a
psychotherapy was embodied in Gordon categorical, nonpsychotic Axis I diagnosis
exercises only a modest impact on treat- Research studies problematically collapse
ment outcome (Beutler, 2000). While the numerous patients under a single diagnosis.
research indicates that certain psychothera- It is a false and, at times, misleading pre-
pies make better marriages for certain supposition in randomized clinical trials
disorders, psychological therapies will be that the patient sample is homogenous.
increasingly matched to people, not simply Perhaps the patients are diagnostically
diagnoses. homogeneous, but nondiagnostic variabil-
The process of creating the optimal ity is the rule, as every clinician also knows.
match in psychotherapy has been accorded It is precisely the unique individual and the
multiple names: adaptation, responsive- singular context that many psychothera-
ness, attunement, matchmaking, custom- pists attempt to treat.
izing, prescriptionism, treatment selection, Moreover, most practice and EBP guide-
specificity factor, differential therapeutics, lines do little for those psychotherapists
tailoring, treatment fit, and individualizing. whose patients and theoretical conceptual-
By whatever name, the goal is to enhance izations do not fall into discrete disorders
treatment effectiveness by tailoring it to the (Messer, 2001). Consider the client who
individual and his/her singular situation. seeks more joy in his/her life, but who
In other words, psychotherapists endeavor does not meet diagnostic criteria for any
to create a new therapy for each patient. disorder, whose psychotherapy stretches
This position can be easily misunder- beyond 20 sessions, and whose treatment
stood as an authority figure therapist pre- objectives are not easily specified in mea-
scribing a specific form of psychotherapy surable, symptom-based outcomes. Current
for a passive client. Far from it: the goal is evidence-based compilations have little to
for an empathic therapist to arrange for contribute to this kind of treatment (see
an optimal relationship collaboratively O’Donohue, Buchanan, & Fisher, 2000,
with an active client on the basis of the cli- for general characteristics of ESTs).
ent’s personality, culture, and preferences. The upshot of these considerations is
If a client frequently resists, for example, that a truly evidence-based psychotherapy
then the therapist considers whether he or will necessarily consider the person of the
she is pushing something that the client psychotherapist, the therapy relationship,
finds incompatible (preferences), or the and means to adapt or tailor that relation-
client is not ready to make those changes ship to the individual patient—in addi-
(stage of change) or is uncomfortable with tion to diagnosis. Otherwise, evidence-based
a directive style (reactance). practice will prove clinically incomplete as
As every clinician knows, different types well as scientifically suspect.
of patients respond more effectively to dif-
ferent types of treatments and relationships. Effect Sizes
Clinicians strive to offer or select a therapy The second edition of this book endeav-
that accords with the patient’s personal char- ors to systemically appraise the empirical
acteristics, proclivities, and worldviews—in research performed on elements of the
addition to diagnosis. Any differential effec- therapy relationship and means of treatment
tiveness of different therapies may well prove adaptation in order to identify what works.
to be a function of cross-diagnostic patient The subsequent chapters feature original
characteristics, such as patient preferences, meta-analyses on the link between the rela-
coping styles, stages of change, personality tionship elements (Section II) and adapta-
dimensions, and culture. tion methods (Section III) to psychotherapy
outcome. Insisting on meta-analyses for mean r of .30. As shown in Table 1.1, this
all these chapters enables direct estimates is a medium effect size. That translates into
of the magnitude of association in the form happier and healthier clients: patients with
of effect sizes. And conducting these meta- empathic therapists tend to progress more
analytic tests with random effects models in treatment and experience a higher prob-
permits generalization to studies outside ability of eventual improvement.
the samples, although the random effects Consider another example, this one
model is slightly less powerful than the involving the effectiveness of tailoring
fixed effect model (Rosenthal, 1995). therapy. The authors of Chapter 16 con-
The meta-analyses in Section II of the ducted a meta-analysis on 65 experimental
book all employed the weighted r. This and quasi-experimental studies, involving
decision improved the consistency among 8,620 patients, which evaluated the impact
the meta-analyses, enhanced their inter- of culturally adapted treatments versus
pretability among the readers (square r for traditional (nonadapted) treatments. The
the amount of variance accounted for), and resultant d of .46 favored those clients
enabled direct comparisons of the meta- receiving a culturally adapted therapy. As
analytic results to one another as well as to seen in Table 1.1, this effect size also repre-
d (the ES typically used when comparing sents a medium, beneficial effect; incor-
the relative effects of two treatments). In all porating clients’ culture into treatment
of these analyses, the larger the magnitude typically enhances the effectiveness of
of r, the higher the probability of patient psychotherapy.
success in psychotherapy. By convention Given the large number of factors contrib-
(Cohen, 1988), an r of .10 in the behav- uting to such success, and the inherent com-
ioral sciences is considered a small effect, plexity of psychotherapy, we do not expect
.30 a medium effect, and .50 a large effect. large, overpowering effects of any one of its
The meta-analyses presented in Section III facets. Instead, we expect to find a number of
of the book, by contrast, employed the helpful facets. And that is exactly what we
weighted d. That is the common indica- find in the following chapters—beneficial,
tor of a difference between two treatments medium-sized effects of several elements of
or conditions: in this case, the difference the complex therapy relationship.
between the conventional or unadapted
therapy and the adapted therapy. In all of Accounting for Psychotherapy
these analyses, the larger the value of d, Outcome
the higher the effectiveness of the specific What, then, accounts for psychotherapy
adaptation or tailoring. By convention success (and failure)? This question repre-
(Cohen, 1988), a d of .30 in the behavioral sents an understandable desire for clarity
sciences is considered a small effect, .50 a and guidance, but we answer with trepi-
medium effect, and .80 a large effect. dation. Our collective ability to answer in
Table 1.1 presents several concrete ways meaningful ways is limited by the huge
to interpret r and d in health care. For variation in methodological designs, theo-
example, the authors of Chapter 6 con- retical orientations, treatment settings, and
ducted a meta-analysis of 57 studies that patient presentations. Of the dozens of vari-
investigated the link between therapist ables that contribute to patient outcome,
empathy and patient success at the end of only a few can be included in any given
treatment. Their meta-analysis, involving a study. How can we divide the indivisible
total of 3,599 clients, found a weighted complexity of psychotherapy outcome?
Table 1.1 Interpretation of Effect Size (ES) Statistics
Cohen’s Type of Percentile of Success rate of Number
d r Benchmark effect treated patientsa treated patientsb needed to treatc
1.00 Beneficial 84 72% 2.2
.90 Beneficial 82 70% 2.4
.80 .50 Large Beneficial 79 69% 2.7
.70 Beneficial 76 66% 3.0
.60 Beneficial 73 64% 3.5
.50 .30 Medium Beneficial 69 62% 4.1
.40 Beneficial 66 60% 5.1
.30 Beneficial 62 57% 6.7
.20 .10 Small Beneficial 58 55% 10.0
.10 No effect 54 52% 20.0
.00 0 No effect 50 50%
−.10 No effect 46 48%
–.20 –.10 Detrimental 42 45%
–.30 Detrimental 38 43%
Sources: Adapted from Cohen (1988); Norcross, Hogan, & Koocher (2008); and Wampold (2001)
a
Each ES can be conceptualized as reflecting a corresponding percentile value: in this case, the percentile standing of the average treated
patient after psychotherapy relative to untreated patients.
b
Each ES can also be translated into a success rate of treated patients relative to untreated patients; a d of .70, for example, would translate into
approximately 66% of patients being treated successfully compared with 50% of untreated patients.
c
Number needed to treat (NNT) refers to the number of patients who need to receive the experimental treatment vis-à-vis the comparison to
achieve one success. An effect size of .70 approximates an NNT of 3: three patients need to receive psychotherapy to achieve a success relative
to untreated patients (Wampold, 2001).
Nonetheless, psychotherapy research has based on decades of research, but not for-
made tremendous strides in clarifying the mally derived from meta-analytic methods
question and addressing the uncertainty. (see Lambert & Barley, 2002, for details).
Thus, we tentatively advance two models The patient’s extratherapeutic change—
that account for psychotherapy outcome, self-change, spontaneous remission, social
averaging across thousands of outcome support, fortuitous events—accounts for
studies and hundreds of meta-analyses, and roughly 40% of success. Common factors,
acknowledging that this matter has been variables found in most therapies regard-
vigorously debated for over six decades. We less of theoretical orientation, probably
implore readers to consider the following account for another 30%. The therapy
percentages as crude estimates, not as exact relationship represents the sine qua non
numbers. of common factors, along with client and
The first model estimates the percentage therapist factors. Technique factors, explain-
of explained psychotherapy outcome vari- ing approximately 15% of the variance, are
ance as a function of therapeutic factors. those treatment methods fairly specific
This comparative importance of each of to the prescribed therapy, such as biofeed-
these factors is summarized in Figure 1.1. back, transference interpretations, desensi-
The percentages presented in Figure 1.1 are tization, or two-chair work. Finally, playing
Individual Other factors
therapist 3%
Expectancy Treatment 7%
(placebo effect) method
15% 8%
Common factors
30%
Unexplained
Therapy variance
Extratherapeutic relationship 40%
change 12%
40%
Techniques
15%
Patient
contribution
30%
Fig. 1.1 % of Improvement in Psychotherapy Patients Fig. 1.2 % of Total Psychotherapy Outcome Variance
as a Function of Therapeutic Factors. Atrributable to Therapeutic Factors.
an important role is expectancy—the pla- How to improve psychotherapy outcome?

cebo effect, the client’s knowledge that he/ Follow the evidence; follow what contrib-
she is being treated and his/her conviction utes to psychotherapy outcome. Begin by
in the treatment rationale and methods. leveraging the patient’s resources and self-
These four broad factors account for the healing capacities; emphasize the therapy
explained outcome variance. relationship and so-called common factors;
The second model begins with the unex- employ research-supported treatment meth-
plained variance in psychotherapy outcome, ods; select interpersonally skilled and clini-
which necessarily decreases the amount of cally motivated practitioners; and adapt all
variance attributable to the therapeutic fac- of them to the patient’s characteristics, per-
tors. As summarized in Figure 1.2, psycho- sonality, and worldviews. This, not simply
therapy research cannot explain all of the matching a treatment method to a particu-
variation in psychotherapy success. To be lar disorder, will maximize success.
sure, some of this is attributable to measure- The differences between the two models
ment error and fallible research methods, help explain the rampant confusion in
but some is also attributable to the complex- the field regarding the relative percentages
ity of human behavior. Thereafter, we esti- accounted for by relationships and tech-
mate that the patient (including severity of niques. The first model (Figure 1.1) presents
disorder) accounts for approximately 30% only the explained variance and separates
of the total variance, the therapy relation- common factors and specific factors, whereas
ship for 12%, the specific treatment method the second model (Figure 1.2) presents the
for 8%, and the therapist for 7% (when not total variance and assigns common factors
confounded with treatment effects). In this to each of the constituent elements. Hence,
model, we assume that common factors are it is essential to inquire whether the percent-
spread across the therapeutic factors—some ages attributable to particular therapeutic
pertain to the patient, some to the therapy factors are based on total or explained vari-
method, some to the treatment method, ance and how common factors are concep-
and some to the therapist him/herself. tualized in a particular model.
Despite the differing percentages, both Force may have failed to make necessary
models converge mightily on several take- distinctions.
home points. One: patients contribute the Another lacuna in the Task Force work is
lion’s share of psychotherapy success (and that we may have neglected, relatively
failure). Simply consider the probable speaking, the productive contribution of
outcome of psychotherapy with an adjust- the client to the therapy relationship. We
ment disorder in a healthy person in the decided not to commission a separate chap-
action stage versus a chronically mentally ter on the client’s contributions; instead, we
ill person presenting in precontemplation/ asked the authors of each chapter to address
denial. Two: the therapeutic relationship them. We encouraged authors to pay atten-
generally accounts for as much psychother- tion to the chain of events among the ther-
apy success as the treatment method. Three: apist’s contributions, the patient processes,
particular treatment methods do matter in and eventual treatment outcomes. This, we
some cases, especially with severe anxiety hoped, would maintain the focus on what
disorders treated via systematic exposure is effective in patient change. Further, all of
(Lambert & Ogles, 2004). Four: Adapting the chapters in Section III examine patient
or customizing therapy to the patient contributions directly in terms of specific
enhances the effectiveness of psychotherapy patient characteristics. Nonetheless, by
probably by innervating multiple path- omitting separate chapters on the client, we
ways—the patient, the relationship, the may be understandably accused of an omis-
method, and the expectancy. Fifth: psycho- sion akin to the error of leaving the rela-
therapists need to consider multiple factors tionship out at the expense of method. This
and their optimal combinations, not only book may be “therapist centric” in minimiz-
one or two of their favorites. ing the client’s relational contribution and
self-healing processes.
Limitations of the Task Force Another prominent limitation across
A single task force can accomplish only so these research reviews is the difficulty of
much work and cover only so much con- establishing causal connections between the
tent. As such, we wish to acknowledge sev- relationship behavior and treatment out-
eral necessary omissions and unfortunate come. The only meta-analysis in Section II
truncations in our work. that contains randomized clinical trials
The products of the Task Force proba- (RCTs) capable of demonstrating a causal
bly suffer from content overlap. We may effect is collecting client feedback. (Note
have cut the “diamond” of the therapy that most of the meta-analyses in Section
relationship too thin at times, leading to III were conducted on RCTs and are capa-
a profusion of highly related and possible of causal conclusions.) Causal infer-
bly redundant constructs. Goal consensus, ences are always difficult to make concerning
for example, correlates highly with parts process variables, such as the therapy rela-
of the therapeutic alliance, but these are tionship. Does the relationship cause
reviewed in separate chapters. Collecting improvement or simply reflect it? The inter-
client feedback and repairing alliance rup- pretation problems of correlational studies
tures, for another example, may represent (third variables, reverse causation) render
different sides of the same therapist behav- such studies less convincing than RCTs.
ior, but these too are covered in separate It is methodologically difficult to meet the
meta-analyses. Thus, to some, the content three conditions to make a causal claim:
may appear swollen; to others, the Task nonspuriousness, covariation between the
process variable and the outcome measure, collaborative, and supportive therapist
and temporal precedence of the process versus a nonempathic, authoritarian, disre-
variable (Feeley, DeRubeis, & Gelfand, spectful, and unsupportive therapist.
1999). We still need to determine whether A final interesting drawback to the pres-
and when the therapeutic relationship is a ent work, and psychotherapy research as a
mediator, moderator, or mechanism of whole, is the paucity of attention paid to
change in psychotherapy (Kazdin, 2007). the disorder-specific and treatment-specific
At the same time as we acknowledge this nature of the therapy relationship. It is pre-
central limitation, let’s remain mindful of mature to aggregate the research on how
several considerations. First, the establish- the patient’s primary disorder or the type
ment of temporal ordering is essential for of treatment impacts the therapy relation-
causal inference, but it is not sufficient. ship, but there are early links. For exam-
In showing that these facets of a therapy ple, in the National Institute on Drug
relationship precede positive treatment Abuse Collaborative Cocaine Treatment
outcome, we can certainly state that the Study, higher levels of the working alliance
therapy relationship is, at a minimum, were associated with increased retention in
an important predictor and antecedent of supportive-expressive therapy, but in cog-
that outcome. Second, within these reality nitive therapy, higher levels of alliance were
constraints, dozens of lagged correlational, associated with decreased retention (Barber
unconfounded regression, structural equa- et al., 2001). In the treatment of severe
tion, and growth curve studies suggest that anxiety disorders, the specific treatments
the therapy relationship probably causally seem to exert a larger effect than the ther-
contributes to outcome (e.g., Barber et al., apy relationship, but in depression, the
2000). For example, using growth curve relationship appears more powerful. The
analyses and controlling for prior improve- therapeutic alliance in the NIMH Treat-
ment and eight prognostically relevant client ment of Depression Collaborative Research
characteristics, Klein and colleagues (2003) Program, in both psychotherapy and phar-
found that the early alliance significantly macotherapy, emerged as the leading force
predicted later improvement in 367 chron- in reducing a patient’s depression (Krupnick
ically depressed clients. Although we need et al., 1996). The therapeutic relationship
to continue to parse out the causal linkages, probably exhibits more impact in some dis-
the therapy relationship has probably been orders and in some therapies than others
shown to exercise a causal association to (Beckner, Vella, Howard, & Mohr, 2007).
outcome. Third, some of the most precious As with research on specific treatments,
behaviors in life are incapable on ethical it may no longer suffice to ask “Does the
grounds of random assignment and experi- relationship work?” but “How does the
mental manipulation. Take parental love as relationship work for this disorder and this
an exemplar. Not a single randomized clini- treatment?”
cal trial has ever been conducted to conclu-
sively determine the causal benefit of a Frequently Asked Questions
parental love on children’s functioning, The interdivisional Task Force on Evidence-
yet virtually all humans aspire to it and Based Therapy Relationships has generated
practice it. Nor can we envision an institu- considerable enthusiasm in the professional
tional review board (IRB) ever approving community, but it has also provoked mis-
a grant proposal to randomize patients in understandings and reservations. Here we
a psychotherapy study to an empathic, address frequently asked questions (FAQs)
about the Task Force’s objectives and convey the impression that it is the only
results. area of importance. This is certainly not
♦ What is the relationship of this task force our intention. Relationship factors are
to the Division 12 Task Force on Research- important, and we need to review the sci-
Supported Treatments (now the standing entific literature and provide clinical and
Committee on Science and Practice)? training recommendations based upon that
Questions abound regarding the con- literature. This can be done without trivial-
nection of the task forces, probably because izing or degrading the effects of specific
they are both associated with the same treatments.
division of the American Psychological ♦ Isn’t your report just warmed-over Carl
Association. Organizationally, the Task Rogers?
Forces are separate creatures. Their respec- No. While Rogers’ (1957) facilitative
tive foci obviously diverge: one looking at conditions are represented in this book,
therapist contributions to the relationship they comprise only about 15% of the
and patient responsiveness, the other look- research critically reviewed. More funda-
ing at treatment methods for specific disor- mentally, we have moved beyond a limited
ders. However, both task forces share the and invariant set of necessary relationship
same book publisher (Oxford University conditions. Monolithic theories of change
Press) and overarching goals (to identify and one-size-fits-all therapy relationships
and promulgate evidence-based practices). are out; adapting the therapy to the unique
♦ Are you saying that treatment methods patient is in.
are immaterial to psychotherapy outcome? ♦ An interpersonal view of psychotherapy
Absolutely not. The empirical research seems at odds with what managed care and
shows that both the therapy relationship bean counters ask of me in my clinical prac-
and the treatment method make frequent tice. How do you reconcile these?
contributions to treatment outcome. It It is true that a dominant image of modern
remains a matter of judgment and method- psychotherapy, among both researchers
ology on how much each contributes, but and reimbursers, is as a mental health treat-
there is virtual unanimity that both the ment. This “treatment” or “medical” model
relationship and the method (insofar as inclines people to define process in terms of
we can separate them) “work.” Looking method, therapists as providers trained in
at either treatment methods or therapy rela- the application of techniques, treatment in
tionships alone is incomplete. We encour- terms of number of contact hours, patients
age practitioners and researchers to look at as embodiments of psychiatric disorders,
multiple determinants of outcome, partic- and outcome as the end result of a treat-
ularly client contributions. ment episode (Orlinsky, 1989).
♦ But are you not exaggerating the effects It is also true that the Task Force mem-
of relationship factors and/or minimizing bers believe this model to be restricted and
the effects of treatments in order to set up the inaccurate. The psychotherapy enterprise is
importance of your work? far more complex and interactive than the
We think not and hope not. With linear “Treatment operates on patients to
the guidance of Task Force members and produce effects” (Bohart & Tallman, 1999).
external consultants, we have tried to avoid We would prefer a broader, integrative
dichotomies and polarizations. Focusing model that incorporates the relational and
on one area—the psychotherapy relation- educational features of psychotherapy, one
ship—in this volume may unfortunately that recognizes both the interpersonal and
instrumental components of psychother- capable of more malleability and “mood
apy, one that appreciates the bidirectional transcendence” than might be expected. In
process of therapy, and one in which the Gurman’s (1973) research, for example,
therapist and patient cocreate an optimal expert therapists appeared to be less handi-
process and outcome. capped by their own “bad moods” than
♦ Won’t these results contribute fur- were their less skilled peers. From the
ther to deprofessionalizing psychotherapy? literature on the cognitive psychology of
Aren’t you unwittingly supporting efforts expertise (Schacht, 1991), experienced
to have any warm, empathic person perform psychotherapists are disciplined improvisa-
psychotherapy? tionalists who have stronger self-regulating
Perhaps some will misuse our conclu- skills and more flexible repertoires than
sions in this way, but that is neither our novices. The research on the therapist’s
intent nor commensurate with our meta- level of experience suggests that experience
analyses. It trivializes psychotherapy to begets heightened attention to the client
characterize it as simply “a good relation- (less self-preoccupation), an innovative
ship with a caring person.” The research perspective, and in general, more endorse-
shows that an effective psychotherapist is ment of an “eclectic” orientation predi-
one who employs specific methods, who cated on client need (Auerbach & Johnson,
offers strong relationships, and who cus- 1977). Indeed, several research studies (see
tomizes both treatment methods and rela- Beutler, Machado, & Neufeldt, 1994) have
tionship stances to the individual person demonstrated that therapists can consis-
and condition. That requires considerable tently use different treatment models in a
training and experience, the antithesis of discriminative fashion.
“anyone can do psychotherapy.” Thus, our clinical experience and the
♦ Are psychotherapists really able to adapt modest amount of research attest that sea-
their relational style to fit the proclivities and soned practitioners can shift back and forth
personalities of their patients? Where is the among different relationship styles for a
evidence we can do this? given case. Whether inexperienced psycho-
Relational flexibility conjures up many therapists can do so is still unanswered.
concerns, but two of particular import in And we caution therapists to be careful that
this question: the limits of human capacity the blending of stances and strategies never
and the possibility of capricious posturing deteriorates into playacting or capricious
(Norcross & Beutler, 1997). Although posturing.
the psychotherapist can, with training ♦ What should we do if we are unable or
and experience, learn to relate in a number unwilling to adapt our therapy to the patient
of different ways, there are limits to our in the manner that research indicates is likely
human capacity to modify relationship to enhance psychotherapy outcome?
stances. It may be difficult to change inter- Five avenues spring to mind. First,
action styles from client to client and ses- address the matter forthrightly with the
sion to session, assuming one is both aware patient as part of the evolving therapeutic
and in control of one’s styles of relating contract and the creation of respective
(Lazarus, 1993) tasks, in much the same way one would
Can psychotherapists authentically differ with patients requesting a form of therapy
from their preferred or habitual style of or a type of medication that research has
relating? There is some research supporting indicated would fit particularly well in their
this assertion. Experienced therapists are case but which is not in your repertoire.
Second, treatment decisions are the result therapy relationship and the particular
of multiple, interacting, and recursive means of adapting it to individual patients.
considerations on the part of the patient, It is premature to proffer the last word, but
the therapist, and the context. A single it is time to codify and disseminate what
evidence-based guideline should be seri- we do know. We look forward to regular
ously considered, but only as one of many updates on our research conclusions and
determinants of treatment itself. Third, practice recommendations.
formal tracking of patient functioning ♦ So, are you saying that the therapy rela-
during the course of psychotherapy pro- tionship (in addition to the treatment method)
vides a systematic way of assessing the is crucial to outcome, that it can be improved
consequences of treatment as it unfolds. by certain therapist contributions, and that it
Determine if, in this particular case, the can be effectively tailored to the individual
treatment is helping. Fourth, an alternative patient?
to the one-therapist-fits-most-patients per- Precisely. And this book, on the basis
spective is practice limits. Without a will- of the empirical research, suggests impor-
ingness and ability to engage in a range of tant directions for practitioners, trainers,
interpersonal stances, the therapist may researchers, and policymakers.
limit his or her practice to clients who fit
that practice. Psychotherapists need not Concluding Reflections
offer all services to all patients. Finally, con- The future of psychotherapy portends the
sider a judicious referral to a colleague who integration of science and service, of the
can offer the relationship stance (or treat- instrumental and the interpersonal, of
ment method or medication) indicated for the technical and the relational in the tradi-
a particular patient. tion of evidence-based practice (Norcross,
♦ Are the Task Force’s conclusions and rec- Freedheim, & VandenBos, 2011). Evidence-
ommendations intended as practice standards? based therapy relationships align with this
No. These are research-based conclu- future and embody a crucial part of
sions that can lead, inform, and guide prac- evidence-based practice, when properly
titioners toward evidence-based therapy conceptualized. We can imagine few prac-
relationships and responsiveness to patient tices in all of psychotherapy that can confi-
needs. They are not legal, ethical, or profes- dently boast that they integrate “the best
sional mandates. available research with clinical expertise in
♦ Well, don’t these represent the offi- the context of patient characteristics, culture,
cial positions of Division 12 (Clinical), and preferences” (American Psychological
Division 29 (Psychotherapy), or the American Association Task Force on Evidence-Based
Psychological Association? Practice, 2006) as well as the relational
No. No. No. behaviors and treatment adaptations pre-
♦ Isn’t it premature to launch a set of sented in this book. We are reminded daily
research-based conclusions on the therapy that research can guide how we create, culti-
relationship and patient matching? vate, and customize that powerful human
Science is not a set of answers; science is relationship.
a series of processes and steps by which we Moreover, we fervently hope this book
arrive closer and closer to elusive answers. will indirectly serve another master: to heal
A vast amount of sophisticated research the damage incurred by the culture wars in
over the past five decades has been con- psychotherapy. If our Task Force is even a
ducted on both the general elements of the little bit successful, then the pervasive gap
between the science and practice commu- Barlow, D. H. (Ed.). (2007). Clinical handbook of
nities will be narrowed, and the insidious psychological disorders: A step-by-step treatment
manual (4th ed.). New York: Guilford.
dichotomy between the therapy relation-
Beckner, V., Vella, L., Howard, I., & Mohr, D. C.
ship and the treatment method will be (2007). Alliance in two telephone-administered
lessened. Phrased more positively, psycho- treatments: Relationship with depression and
therapists from all camps and communities health outcomes. Journal of Consulting and
will increasingly collaborate, and our patients Clinical Psychology, 75, 508–512.
will benefit from the most efficacious treat- Bergin, A. E. (1997). Neglect of the therapist and the
human dimensions of change: A commentary.
ments and relationships available.
Clinical Psychology: Science and Practice, 4, 83–89.
Bergin, A. E., & Lambert, M. J. (1978). The
References evaluation of outcomes in psychotherapy. In
American Psychiatric Association. (2006). Practice S. L. Garfield & A. E. Bergin (Eds.), Handbook
guidelines for the treatment of psychiatric dis- of psychotherapy and behavior change (pp.
orders: Compendium 2006. Washington, DC: 139–189). New York: Wiley.
Author. Beutler, L. E. (2000). David and Goliath: When
American Psychological Association Task Force on empirical and clinical standards of practice meet.
Evidence-Based Practice. (2006). Evidence-based American Psychologist, 55, 997–1007.
practice in psychology. American Psychologist, 61, Beutler, L.E., Machado, P. P. P., & Neufeldt, S.A.
271–285. (1994). Therapist variables. In A. E. Bergin &
Anderson, T., Ogles, B. M., Patterson, C. L., S. L. Garfield (Eds.), Handbook of psychotherapy
Lambert, M. J., & Vermeersch, D. A. (2009). and behavior change (4th ed., pp. 229–269).
Therapist effects: Facilitative interpersonal skills New York: Wiley.
as a predictor of therapist success. Journal of Bohart, A. C., & Tallman, K. (1999). How clients
Clinical Psychology, 65, 755–768. make therapy work: The process of active self-
Auerbach, A. H., & Johnson, M. (1977). In healing. Washington, DC: American Psycho-
A. S. Gurman & A. M. Razin (Eds.), Effective logical Association.
psychotherapy: A handbook of research. New York: Cohen, J. (1988). Statistical power analysis for
Pergamon. the behavioral sciences (2nd ed.). Hillsdale, NJ:
Barber, J. P., Connolly, M. B., Crits-Christoph, P., Erlbaum.
Gladis, L., & Siqueland, L. (2000). Alliance Crits-Christoph, P., Baranackie, K., Kurcias, J. S.,
predicts patients’ outcomes beyond in-treatment Beck, A. T., Carroll, K., Perry, K., et al. (1991).
change in symptoms. Journal of Consulting and Meta-analysis of therapist effects in psycho-
Clinical Psychology, 68, 1027–1032. therapy outcome studies. Psychotherapy Research,
Barber, J. P., Gallop, R., Crits-Christoph, P., Frank, A., 1, 281–91.
Thase, M. E., Weiss, R. D., et al. (2006). The Department of Health. (2001). Treatment choice in
role of therapist adherence, therapist compe- psychological therapies and counseling. London:
tence, and alliance in predicting outcome of Department of Health Publications.
individual drug counseling: Results from the Duncan, B. L., Miller, S. D., Wampold, B. E., &
National Institute on Drug Abuse Collaborative Hubble, M. A. (Eds.). (2010). The heart and soul
Cocaine Treatment Study. Psychotherapy Research, of change (2nd ed.). Washington, DC: American
16, 229–240. Psychological Association.
Barber, J. P., Luborsky, L., Gallop, R., Feeley, M., DeRubeis, R. J., & Gelfand, L. A.
Crits-Christoph, P., Frank, A., Weiss, R. D., (1999). The temporal relation of adherence and
et al. (2001). Therapeutic alliance as a predictor alliance to symptom change in cognitive therapy
of outcome and retention in the National for depression. Journal of Consulting and Clinical
Institute on Drug Abuse Collaborative Cocaine Psychology, 67, 578–582.
Treatment Study. Journal of Consulting & Clinical Gelso, C. J. (Ed.). (2005). The interplay of tech-
Psychology, 69, 119–124. niques and the therapeutic relationship in psych-
Barlow, D. H. (2000). Evidence-based practice: otherapy. Psychotherapy, 42(4), whole.
A world view. Clinical Psychology: Science and Gelso, C. J., & Carter, J. A. (1985). The relationship
Practice, 7, 241–242. in counseling and psychotherapy: Components,
consequences, and theoretical antecedents. The Lambert, M. J., & Barley, D. E. (2002). Research
Counseling Psychologist, 13, 155–243. summary on the therapeutic relationship and
Gelso, C. J., & Carter, J. A. (1994). Components psychotherapy outcome. In J. C. Norcross (Ed.),
of the psychotherapy relationship: Their interac- Psychotherapy relationships that work (pp. 17–32).
tion and unfolding during treatment. Journal of New York: Oxford University Press.
Counseling Psychology, 41, 296–306. Lambert, M. J., & Ogles, B. M. (2004). The
Gelso, C. J., & Hayes, J. A. (1998). The psycho- efficacy and effectiveness of psychotherapy. In
therapy research: Theory, research, and practice. M. J. Lambert (Ed.), Begin and Garfield’s hand-
New York: Wiley. book of psychotherapy and behavior change (5th
Gurman, A. S. (1973). Effects of therapist and ed., pp. 139 – 193). New York: Wiley.
patient mood on the therapeutic functioning of Lazarus, A. A. (1993). Tailoring the therapeutic
high- and low-facilitative therapists. Journal of relationship, or being an authentic chameleon.
Consulting and Clinical Psychology, 40, 48–58. Psychotherapy, 30, 404–407.
Henry, W. P. (1998). Science, politics, and the Luborsky, L., Diguer, L., Seligman, D. A.,
politics of science: The use and misuse of empir- Rosenthal, R., Krause, E.D., Johnson, S., et al.
ically validated treatment research. Psychotherapy (1999). The researcher’s own therapy allegiances:
Research, 8, 126–140. A “wild card” in comparisons of treatment
Huppert, J. D., Bufka, L. F., Barlow, D. H., efficacy. Clinical Psychology: Science and Practice,
Gorman, J. M., Shear, M. K., & Woods, S. W. 6, 95–106.
(2001). Therapists, therapist variables, and cog- Messer, S. B. (2001). Empirically supported treat-
nitive–behavioral therapy outcome in a multi- ments: What’s a nonbehaviorist to do?. In
center trial for panic disorder. Journal of B. D. Slife, R. N. Williams, & S. H. Barlow
Consulting and Clinical Psychology, 69, 747–755. (Eds.), Critical issues in psychotherapy. Thousand
Kazdin, A. E. (2007). Mediators and mechanism Oaks, CA: Sage.
of change in psychotherapy research. In Nathan, P. E. (1998). Practice guidelines: Not yet
S. Nolen-Hoeksema, T. D. Cannon, & ideal. American Psychologist, 53, 290–299.
T. Widiger (Eds.), Annual Review of Clinical Nathan, P. E., & Gorman, J. M. (Eds.). (2007).
Psychology, 3, 1–27. A guide to treatments that work (3rd ed.).
Klein, D. N., Schwartz, J. E., Santiago, N. J., New York: Oxford University Press.
Vivian, D., Vocisano, C., Castonguay, L. G., Norcross, J. C. (Ed.). (2002). Psychotherapy
et al. (2003). Therapeutic alliance in depression relationships that work: Therapist contributions
treatment: Controlling for prior change and and responsiveness to patient needs. New York:
patient characteristics. Journal of Consulting and Oxford University Press.
Clinical Psychology, 71, 997–1006. Norcross, J. C. (2010). The therapeutic relationship.
Krupnick, J. L., Sotsky, S. M., Simmens, S., In B. L. Duncan, S. D. Miller, B. E. Wampold,
Moyer, J., Elkin, I., Watkins, J., et al. (1996). & M. A. Hubble (Eds.), Heart & soul of
The role of the therapeutic alliance in psycho- change in psychotherapy (2nd ed.). Washington,
therapy and pharmacotherapy. Journal of DC: American Psychological Association.
Consulting and Clinical Psychology, 64, 532–539. Norcross, J. C., & Beutler, L. E. (1997). Determining
Lambert, M. J. (1992). Psychotherapy outcome the therapeutic relationship of choice in
research: Implications for integrative and eclectic brief therapy. In J. N. Butcher (Ed.), Personality
therapists. In J. C. Norcross and M. R. Goldfried assessment in managed health care: A practi-
(Eds.), Handbook of psychotherapy integration tioner’s guide. New York: Oxford University
(pp. 94–129). New York: Basic Books. Press.
Lambert, M. J. (2010). Prevention of treatment Norcross, J. C., Beutler, L. E., & Levant, R. F.
failure: The use of measuring, monitoring, & (Eds.). (2006). Evidence-based practices in
feedback in clinical practice. Washington, DC: mental health: Debate and dialogue on the funda-
American Psychological Association. mental questions. Washington, DC: American
Lambert, M. J. (2011). Psychotherapy research Psychological Association.
and its achievements. In J. C. Norcross, Norcross, J. C., Hogan, T. P., & Koocher, G. P.
G. R. VandenBos, & D. K. Freedheim (Eds.), (2008). Clinician’s guide to evidence-based prac-
History of psychotherapy (2nd ed.). Washington, tices: Mental health and the addictions. New York:
DC: American Psychological Association. Oxford University Press.
Norcross, J. C., Freedheim, D. K., & Rector, N. A., Zuroff, D. C., & Segal, Z. V. (1999).
VandenBos, G. R. (2011). Into the future: Cognitive change and the therapeutic alliance:
Retrospect and prospect in psychotherapy. The role of technical and nontechnical factors in
In J. C. Norcross, G. R. VandenBos, & cognitive therapy. Psychotherapy, 36, 320–328.
D. K. Freedheim (Eds.). (2011). History of Rogers, C. R. (1957). The necessary and sufficient
psychotherapy (2nd ed.). Washington, DC: conditions of therapeutic personality change.
American Psychological Association. Journal of Consulting Psychology, 22, 95–103.
O’Donohue, W., Buchanan, J. A., & Fisher, J. E. Rosenthal, R. (1995). Writing meta-analytic
(2000). Characteristics of empirically supported reviews. Psychological Bulletin, 118, 183–192.
treatments. Journal of Psychotherapy Practice and Roth, A., & Fonagy, P. (2004). What works for
Research, 9, 69–74. whom? A critical review of psychotherapy research
Okiishi J., Lambert, M. J., Nielsen, S. L., & (2nd ed.). New York: Guilford.
Ogles, B. M. (2003). In search of supershrink: Safran, J. D., & Muran, J. C. (2000). Negotiating
Using patient outcome to identify effective the therapeutic alliance. New York: Guilford.
and ineffective therapists. Clinical Psychology and Schacht, T. E. (1991). Can psychotherapy educa-
Psychotherapy, 10, 361–373. tion advance psychotherapy integration? A view
Orlinsky, D. E. (1989). Researchers’ images of from the cognitive psychology of expertise.
psychotherapy: Their origins and influence Journal of Psychotherapy Integration, 1, 305–320.
on research. Clinical Psychology Review, 9, Task Force on Psychological Intervention Guidelines.
413–441. (1995). Template for developing guidelines:
Orlinsky, D. E. (2000, August). Therapist interper- Interventions for mental disorders and psychosocial
sonal behaviors that have consistently shown posi- aspects of physical disorders. Washington, DC:
tive correlations with outcome. Paper presented at American Psychological Association.
the 108th annual convention of the American Wampold, B. E. (2001). The great psychother-
Psychological Association, Washington, DC. apy debate: Models, methods, and findings.
Orlinsky, D., & Howard, K. E. (1977). The Mahwah, NJ: Lawrence Erlbaum.
therapist’s experience of psychotherapy. In Wampold, B. E., & Brown, G. S. (2005). Estimating
A. S. Gurman & A. M. Razin (Eds.), Effective variability in outcomes attributable to therapists:
psychotherapy: A handbook of research. New York: A naturalistic study of outcomes in managed
Pergamon. care. Journal of Consulting and Clinical Psychology,
Paul, G. L. (1967). Strategy of outcome research in 73, 914–923.
psychotherapy. Journal of Consulting Psychology, Wolitzky, D. L. (2011). Psychoanalytic theo-
31, 109–118. ries of psychotherapy. In J. C. Norcross,
Project MATCH Research Group. (1998). Therapist G. R. VandenBos, & D. K. Freedheim (Eds.),
effects in three treatments for alcohol problems. History of psychotherapy (2nd ed.). Washington,
Psychotherapy Research, 8, 455–474. DC: American Psychological Association.
PART
2
Effective Elements
of the Therapy
Relationship: What
Works in General
C HA P TER
2 Alliance in Individual Psychotherapy
Adam O. Horvath, AC Del Re, Christoph Flückiger, and Dianne Symonds
Since our last review of the literature was the work of Carl Rogers and his col-
in 2002, research on the alliance in psy- leagues. By applying rigorous empirical
chotherapy has continued to flourish. By methods to the examination of person-
searching the electronic databases at the centered treatment, they not only proved
end of 2000, we located just over 2,000 ref- that the therapy process can be explored
erences using the keywords alliance, helping beyond anecdotal records, but also moved
alliance, working alliance, and therapeutic the concept of the therapeutic relationship
alliance. The same search in early 2010 to the center of the healing process. Rogers
yielded over 7,000 items. The growing attrac- and colleagues generated an important
tion of the alliance concept appears to be body of literature exploring the interper-
the result of a number of related sources: sonal interior of psychotherapy (Rogers,
One reason is the convergence of evi- Gendlin, Kiesler, & Truax, 1967).
dence, staring in the ’70s, that different A third important precursor can be traced
psychotherapies typically produce similar back to the 1930s: A growing curiosity and
beneficial effect for clients (e.g., Luborsky, interest in the integration of diverse theo-
Singer, & Luborsky, 1975; Smith & Glass, ries of psychotherapies (Frank & Frank,
1977; Stiles, Shapiro, & Elliot, 1986). 1991; Rosenzweig, 1936). The desire to
Although the “Dodo bird interpretation” reconcile some conflicting therapeutic
(All have won and all must have their methods and their underlying theories
prizes. . .) of these meta-analyses of psy- eventually led to the founding of the Society
chotherapy effectiveness has proven some- for the Exploration of Psychotherapy
what controversial (Chambless, 2002), Integration (SEPI) in 1983. On the prac-
most therapists and researchers alike have tice side, psychotherapists in North America
accepted the notion that a large part of started to reject the strict boundaries of
what is helpful for clients receiving psycho- classical theories and became increasingly
therapy is shared across diverse treatments. interested in utilizing a variety of effective
The quality of the therapeutic relationship methods irrespective of their “school;” the
in general, and the alliance in particular, field was moving from theoretical monism
are obvious “common factors” shared by to an eclectic pragmatism. The value of
most if not all psychotherapies. aspects of therapist–client relatedness (e.g.,
Another important precursor of the alli- alliance) found ready acceptance among
ance concept, and a pioneering force in the those committed to psychotherapy integra-
development of therapy process research, tion (Goldfried, 1980).
25
But perhaps the most potent force respon- (1910/1913). His basic premise was that
sible for the sustained growth of interest in all relationships were transference based
the alliance was the consistent finding of a (Freud, 1912/1958). Early in his writings,
moderate but robust relationship between he struggled with the question of what keeps
the alliance and treatment outcome across a the analysand in therapy in the face of the
broad spectrum of treatments in a variety of psyche’s unconscious fear and rejection
client/problem contexts (Horvath & Bedi, of exploring repressed material. His first
2002; Horvath & Symonds, 1991; Martin, formulation suggested that he thought that
Ganske, & Davis, 2000). there was an “analyst” within the patient
In this chapter, we reexamine the empir- supporting the healing journey (Freud,
ical evidence linking the alliance to out- 1912/1958). Later he speculated about the
come in individual psychotherapy with reality-based collaboration between therapist
adults. But the relation between alliance and client, a conjoint effort to conquer the
and therapy is only the first level of interest. client’s pain. He also referred to this process
Beyond the strength of the overall alliance– as the unobjectionable or positive transfer-
outcome link, it was our intent to use the ence (Freud, 1913/1940). Both the wisdom
accumulated data to examine the role of of recognizing the client’s attachment to
several potential moderators and mediators the therapist, and his ambiguity about the
that impact this relationship, with particu- status of this attachment (reality based and
lar attention to issues that help us better conscious versus transferential and uncon-
understand the way alliance and treatment scious) has echoed throughout the evolution
results are linked. of the concept.
The term ego alliance was coined by
Definitions and Measures Sterba (1934), who conceptualized it as
The term alliance (also therapeutic alliance, part of the client’s ego-observing process
working alliance, and helping alliance) as it is that alternated with the experiencing (trans-
used in the research literature, can refer to ferential) process. Zetzel (1956) used the
a number of related constructs; at this term therapeutic alliance to refer to the
time we do not have a single, consensually patient’s ability to use the healthy part of
accepted definition of the concept (Horvath her/his ego to link up or join with the
& Luborsky, 1993; Saketopoulou, 1999). analyst to accomplish the therapeutic tasks.
While there are important shared aspects in Greenson (1965, 1967) made a distinction
the way researchers use the construct in the between the working alliance, the client’s
literature (e.g., Bordin, 1980, 1989; Gaston ability to align with the tasks of analysis,
et al., 1995; Hatcher & Barends, 2006; and the therapeutic alliance, which refers to
Horvath & Luborsky, 1993), there are also the capacity of therapist and client to form
nontrivial differences among authors about a personal bond.
the precise meaning of the term (Psycho- During the 1970s efforts were made to
therapy, 43(3), whole). The best way to grasp extrapolate and extend the concept of the
the complexity of the current status of this alliance from its psychodynamic roots to
concept is by briefly reviewing its history. encompass components of the relational
elements of all helping endeavors: Luborsky
Definitions (1976) proposed an extension of Zetzel’s
The concept of the alliance (though not (1956) and Stone’s (1961) concept. He sug-
the term itself ) originated with Freud gested that the alliance between therapist
26 e f f e c t i ve e l e m e n ts o f t he t he r a p y re l at i o n s h i p
and client developed in two phases: The build and ebb in the normal course of
first phase, Type I alliance, involved the events, and that the repair of these stresses
client’s belief in the therapist as a potent in the alliance would constitute the core
source of help, and the therapist providing task of any helping relationship.
a warm, supporting, and caring relation- The most distinguishing feature of the
ship. This level of alliance results in a secure modern pantheoretical reconceptualization
holding relationship within which the work of the alliance is its emphasis on collabora-
of the therapy can begin. The second phase, tion and consensus (Bordin, 1980; Hatcher,
Type II alliance, involved the client’s invest- Barends, Hansel, & Gutfreund, 1995;
ment and faith in the therapeutic process Luborsky, 1976). In contrast to previous
itself, a commitment to the core concepts conceptualizations that emphasized either
undergirding the therapy (e.g., nature of the therapist’s contributions to the relation-
the problem, value of the exploratory pro- ship (i.e., Rogers & Wood, 1974) or the
cess), as well as a willing investment of her unconscious distortions of the relation
or himself, to share the ownership for the between therapist and client, the revised
therapy process. While Luborsky’s (1976, alliance theory emphasized the active col-
1994) assumptions about the therapy pro- laboration between the participants.
cess itself were grounded in psychody- An equally significant consequence of
namic theory, his description of the alliance the way the alliance concept was reintro-
as a therapeutic process was quite general. duced is that there were two different voices
Luborsky and his team also pioneered theorizing about the concept. Each want-
an alliance assessment method for raters, ing to separate the idea from its long his-
using transcripts or audio recordings, to tory within the psychodynamic framework
count signs of in-session events indicative and operationalize the concept in a way
of the presence of either type of alliance. in which it would be compatible with most,
Bordin (1975, 1976, 1989, 1994) pro- if not all, theoretical approaches. But nei-
posed a somewhat different pan-theoretical ther of these theorists (Bordin or Luborsky)
alliance concept he called the working alli- offered a precise definition of how this
ance. His concepts of the alliance were new conceptualization of the alliance
based on Greenson’s’ (1965) ideas as a start- related to (or was different from) other
ing point but departed from the psychody- concepts that are parts of the therapeutic
namic premises even more clearly than relationship. This theoretical ambiguity
Luborsky did. For Bordin, the alliance was created a void which was filled by a number
centrally the achievement of collaborative of alliance measures developed in para-
stance in therapy and was built on three llel between 1978 and 1986. What we
components: agreements on the therapeu- know about the alliance and its relation
tic goals, consensus on the tasks that make to outcome and other therapy variables
up therapy, and a bond between the client has been gleaned from studies that, in prac-
and the therapist. He predicted that differ- tice, define the alliance by the instru-
ent therapies would place different demands ment used to measure it. In this sense, the
on the relationship, thus the “profile” of the instrumentation defines the construct.
ideal working alliance would be different In the following section we review the alli-
across theoretical orientations. Bordin also ance instruments and discuss the differ-
proposed that as therapy progresses, the ences and similarities of their undergirding
strength of the working alliance would conceptualizations.
ho rvat h , re , f lü c k i g e r, s y m o n d s 27
Measures the numbers were similar to those reported
In this chapter we refer to the alliance in for the core measures. In Table 2.1, each
the singular. However, in the database of instrument is identified using the label
201 studies we have assembled for this or identification the authors provided.
meta-analysis, over 30 different alliance However, in the moderator analyses we dis-
measures were used, not counting different cuss later in this chapter, the less often used
versions of the same instrument. Similar to measures (n of use ≤ 3) were merged into one
previous reports, the four core measures: category: “Other.” In this “Other” category
California Psychotherapy Alliance Scale are: some newer alliance measures with rela-
(CALPAS, Gaston & Marmar, 1994), tively few administrations, measures devel-
Helping Alliance Questionnaires (HAq, oped for the specific investigation, and
Alexander & Luborsky, 1987), Vanderbilt instruments originally developed for rela-
Psychotherapy Process Scale (VPPS, tionship constructs other than the alliance.
O’Mally, Suh & Stupp, 1983), and Work- Adding to the diversity of measures is the
ing Alliance Inventory (WAI, Horvath & fact that, over time, the four core instru-
Greenberg, 1986) accounted for approxi- ments have evolved as well and currently
mately two thirds of the data. In examina- exist in a number of different forms (e.g.,
tions of the shared factor structure of the short versions, observer versions, versions
WAI, CALPAS, and HAq, the concept of specific to context and/or application, trans-
“confident collaborative relationship” was lations). The relation of these modified
the central common theme (Hatcher et al., instruments to the original is not always
1995; Hatcher & Barends, 1996). Each of well documented. As we noted, the diver-
these four instruments has been in use for sity in the definition of the alliance
over 20 years and has demonstrated an via the use of a variety of assessment mea-
acceptable levels of internal consistency. sures has become an important issue. The
The methods of reporting reliability of consequences of these differences will be
measures were somewhat inconsistent, but discussed in the section evaluating the inter-
we estimated that clients’ and therapists’ pretation of this corpus of research.
rating of the alliance using these core mea-
sures were in the range of 0.81–0.87
Clinical Examples
(Cronbach’s alpha). Rated (observer) mea-
The alliance represents an emergent quality
sures tended to report interrater reliability
of partnership and mutual collaboration
indexes of similar values. However, the
between therapist and client. As such, it is
shared variance, even among these well-
not the outcome of a particular intervention;
established measures, has been shown to be
its development can take different forms and
less than 50% (Horvath, 2009).
may be achieved almost instantly or nurtured
Fifty four of the research reports in
over a longer period of time depending on
our data set used less-well-validated instru-
the kind of therapy and the stage of treatment
ments or assessment procedures; the rela-
(Bordin, 1994). The following is an excerpt
tion of most of these measures to the core
from an early session that illustrates the
instruments, or to each other, are not well
challenges of negotiating the clients’ whole-
documented, and sometimes nonexistent.
hearted participation in the therapy process:
Relatively little data are available with
respect to their psychometric properties, Client (C): Well aren’t you going to ask
but when this information was provided, me what this reminds me of?
Table 2.1 Research Reports included in the Meta Analysis
Treatment Alliance Outcome
Study Sessions Type Rater Measure Time Measure Rater ES N
Adler (1988) 12 Various C.T WAI, E, L TC, SCL/BSI, RSE, C, T 0.28 44
HAq, CIS IIP, PTQ
Allen et al. ∗ Inpatient T ITAS E, L, A Overall Outcome, T 0.54 37
(1985)a GAS, Outcome
Composite
Allen et al. ∗ Inpatient T ITAS E, L Premature C 0.54 37
(1986)a Termination
Andreoli 6 Crisis T ITAS E Overall Outcome, T 0.57 16
et al. (1993)b intervention Interpersonal
Functioning,
Specific Outcomes
Ankuta 6 Crisis T ITAS E Overall Outcome T 0.02 44
(1993) intervention
Arnow et al. 20 CBT C WAI-S E Premature O 0.10 681
(2003) Termination
Baldwin ∗ Various C WAI M OQ-45 C 0.24 331
(2007)
Barber et al. 20 Various C, T HAq-II, E, M SCL/BSI, Addiction C 0.10 121
(2006) CALPAS Severity Index, BDI
Barber et al. 40 Various C, O HAq-II, E, M SCL/BSI, Addiction C 0.13 83
(1999)a CALPAS Severity Index, BDI
Barber et al. ∗ Dynamic C CALPAS E, M, L BDI C 0.37 83
(2000)
Barber et al. 40 Various C CALPAS E, M Addiction Severity C 0.08 265
(2001)a Index
Barber et al. 36 Dynamic C HAq, E Addiction Severity C 0.10 89
(2008) CALPAS Index
Barkham 12 Interpersonal O CALPAS E Overall Outcome C 0.41 12
et al. (1993)
Bassler et al. 14 w Various C HAq E Overall Outcome C 0.16 237
(1995)
Bethea et al. 8 CBT C, T, HAq II E, M, Drug Use, C, O 0.21 25
(2008) O L, A Functioning
Adherence, Pain
Rating
Bieschke 7 Various C WAI L Change in Distress C 0.38 90
et al. (1995)
Biscoglio ∗ CBT C, T WAI-S E GAS, IIP, SCL/BSI, C, O 0.21 32
(2005) TC
Botella ∗ Various C WAI-S E Premature O 0.16 190
(2008) Termination
(Continued )
29
Table 2.1 Continued
Bredel et al. ∗ Various C NSI E Satisfaction C, O 0.44 78
(2004)
Broome 46 Drug Counseling, C 3-item M Premature C 0.11 167
(1996)b Methodone NSI Termination
Brotman 16 Various C, O WAI, E HRSD O 0.31 51
(2004) HA(r)
Burns et al. 12 w Rehabilitation C WAI-S E Cardiac Depression C 0.12 79
(2007) Scale, Diet Progress,
Exercise and Diet
Self-Efficacy,
General Health
Survey
Busseri et al. ∗ Eclectic C, T WAI E, M SCL/BSI, TC C, T 0.36 54
(2003)
Busseri et al. 8 Eclectic C, T WAI E PTQ, TC, SCL C, T 0.35 50
(2004)
Card (1991) 6 Cognitive- O CALPAS E, M, L STAI, BDI, HRSD, C, O 0.07 55
behavior SCL/BSI
Castonguay 15 Cognitive, O WAI M BDI, HRSD, GAS C, O 0.57 30
et al. (1996) Medication
Chilly 16 Interpersonal C WAI E BDI C 0.52 9
(2004)b
Cislo (1998) 10 Various C HAq A Session Impact C 0.30 47
Clarkin et al. ∗ Inpatient O ITAS A GAS O 0.39 96
(1987) Psychiatric Unit
Cloitre et al. 16 Various C WAI-S E Premature C, O 0.27 30
(2004) Termination, PTSD
Symptoms
Coleman ∗ Eclectic C WAI-S C SCL/BSI, SWLS 0.12 102
(2006)
Connors 12 w Various C, T WAI E DpD, Abstinence C 0.11 579
et al. (1997)a
Constantino 19 CBT, C HAq E, M Purge Frequency 0.29 75
et al. (2005)b Interpersonal
Crits- 54 Dynamic O HAq(cs) E Composite C, T, O 0.39 43
Cristoph Outcome, Residual
et al. (1988) Gain
Davis et al. 26 w CBT O WAI-S A PANSS, WBI 0.43 26
(2007)
de Roten 4 Dynamic C HAq M, E, SCL/BSI, C 0.45 70
et al. (2004) A Evaluation
Questionnaire, SAS
(Continued )
30
Table 2.1 Continued
Dearing 12 CBT C WAI E DpD, Abstinence, C 0.29 208
et al. (2005) Drinking Related
Consequences,
Satisfaction with
Treatment
Deu et al. 10 w Interpersonal C HAq E Depressive O 0.18 17
(2009) Symptoms
Dorsch et al. ∗ Various C HAq II E ACQ, BDI, BSQ, C 0.61 30
(2002) Clinical
Improvement, SCL/
BSI, STAI
Dundon ∗ Various C, T WAI E Abstinence, Sessions O 0.08 194
et al. (2008) Attended
Dunn et al. 18 w CBT C CALPAS E PANSS O -0.11 29
(2006)
Eaton et al. ∗ Various O TARS A Overall Outcome, C, T 0.00 40
(1988) SCL/BSI
Emmerling ∗ Eclectic C WAI-S E GHQ C 0.42 56
et al. (2009)
Fakhoury ∗ Various T HA E Rehospitalization O 0.14 223
et al. (2007)
Feeley 12 Cognitive O HAr A BDI C 0.40 25
(1993)
Ferleger 41 Dynamic O CALPAS E SCL/BSI, TC, C 0.09 40
(1993) Social Adjustment
Florsheim 90–100 Various C, O WAI E, M Drug Use, Teachers C, O 0.22 78
et al. (2000) days (residential and Youths Report
program) Form, Recidivism
Flückiger ∗ CBT C BPSR E, M, L SCL/BSI, IIP, C, T 0.58 47
et al. (2005) Satisfaction,
Improvement, Goal
attainment
Forbes et al. NS Counseling C WAI-S E PTSD symptoms C 0.10 84
(2008)
Forman 6 Rehabilitation C, T WAI M, L Global Outcome C, T 0.48 29
(1990)
Frank et al. 56 Various T ITAS M Premature C, O 0.32 46
(1990) Termination,
Specific Symptoms,
Overall Outcome,
Symptom Severity,
Social Relations
(Continued )
31
Table 2.1 Continued
Freitas ∗ Therapeutic C, T WAI E Lengths of T, O 0.00 80
(2001) Community Treatment,
Neuropsychological
Status
Fries et al. 25 w Various C BPSR A PANSS C, O 0.32 30
(2003)b
Gaiton 24 CBT T, O CALPAS E Composite C, O 0.14 38
(2004) Outcome
Gallop et al. 10 Inpatient Eating C, T WAI E Premature C 0.16 31
(1994) Disorders Unit Termination
Gaston et al. 18 Various C, T CALPAS E, M, L BDI, HRSD C 0.21 18
(1991)a, b
Gaston et al. 18 Dynamic C, T CALPAS A Depression-Anxiety, C 0.15 32
(1994)a, b Interpersonal
Behavior Scale
Gaston et al. 18 Various O CALPAS A BDI, HRSD C 0.34 88
(1998)
Geider ∗ Experiential O CALPAS A Global Outcome C 0.48 10
(1997)
Geiser et al. ∗ Various C HAq II E ACQ, BDI, BSQ, C, T 0.55 231
(2002) GAF
Gerstley 48 Various C, T HAq E Addictive Severity O 0.36 30
et al. (1989) Index
Godfrey 6w CBT O OAS E Chronic Fatique C 0.10 71
et al. (2007)
Gomes- 18 Various O VPPS A Overall Ratings, C, T, O 0.46 35
Swartz MMPI
(1978)a Maladjustment, TC
Greenberg 6 Gestalt C WAI E Scale of Indecision, C, T 0.62 31
et al. (1982) STAI, TC
Greenberg 32 Experiential C WAI A SCL/BSI, IIP, C 0.14 32
et al. (2002) Intrex, TC
Grob et al. 19w Inpatient O ITAS E, M, L Overall T 0.41 60
(1989) Improvement
Gunderson ∗ Various C, T HAq E, M, L SCL/BSI, SAS, C 0.22 28
et al. (1997) GAS
Gunther 15 Various O CALPAS E, L SCL/BSI C 0.25 41
(1991)
Gutfreund 29 Various O CALPAS A SCL/BSI, Dynamic C 0.16 46
(1992) Outcome
(Continued )
32
Table 2.1 Continued
Hansson 4w Inpatient C ITAS E, L SCL/BSI, CPRS, C 0.19 106
et al. (1992) DTES, TC
Hardy et al. 16 w CBT C CALPAS A BDI C 0.71 24
(2001)
Hartley et al. 18 Various O VTAS A Composite Gain C, T, O 0.27 28
(1983) Scale
Hartmann 12 Dynamic O CS E, M, L SCL/BSI, IIP C 0.46 10
(2001)
Hatcher 51 Dynamic C CALPAS Various Improvement to C 0.10 230
et al. (1996) Date
Hawley et al. 16 Various O VTAS A HRSD O 0.27 162
(2006)a
Hayes et al. NS CBT C, O WAI Severity Rating O 0.26 18
(2007)
Hays (1994) 6 Various C, T WAI E Global Outcome, C, T 0.30 29
Personal Growth,
Relations with
Others
∗ Mother–infant O WAI E Growth C 0.35 58
Hervé et al.
(2008) Consultation
Hilliard et al. 25 Dynamic C,T,O SASB M Interject-best/worst, C, T, O 0.21 64
(2000) Intrex SCL/BSI, Global
Outcome
Hopkins 12 Various C, T WAI E SCL/BSI C 0.25 15
(1988)
Hopkins 30 Case T WAI-S C MCAS T 0.24 28
et al. (2006) Management
Horowitz 12 Dynamic O TARS A SCL/BSI, PCS C, O 0.11 52
et al. (1984)
Horvath 10 Various C, T WAI E PTQ C, T 0.49 29
(1981)
Howard 16 Various C WAI E BDI, HRSD, IIP C, O 0.57 47
(2003)
Howard 16 CBT C WAI M BDI C, 0.67 19
et al. (2006)
Huber et al. ∗ Various C, T TRS E BDI, Contentment, C, T 0.28 275
(2003) Premature
Termination
Ilgen et al. ∗ Alcohol and drug C, T WAI E Alcohol Abstincence O 0.11 785
(2006a)a Abstinence Self-Efficacy, DpD
Program
(Continued )
33
Table 2.1 Continued
Ilgen et al. ∗ Alcohol and Drug C, T WAI E Alcohol Abstincence O 0.11 785
(2006b)a Abstinence Self-Efficacy, DpD
Program
Irelan (2004) ∗ Various C WAI E Premature O 0.35 40
Termination
Jacob (2003) 13 Various C WAI E OQ-45, Panic C 0.16 80
Severity
Janecke 38 w Various C HAq E IIP, Satisfaction, C 0.00 50
(2003) Symptom Reduction
Johansson ∗ Various C, T HAq II E SCL/BSI, IIP O 0.23 122
et al. (2006)
Joyce et al. 20 Dynamic C, T NSI A General Symptoms, C, T, O 0.29 64
(1998) Individual
Objectives,
Social-sexual
Adjustment
Joyce et al. 18 Various C, T AAS A Improvement, C, T, O 0.27 144
(2003)a Severity of
Disturbance
Jumes 28 w Inpatient, C WAI E BPRS, GAS O 0.28 121
(1995) Medication
Kabuth et al. ∗ Hospital O HAq E, L Social O 0.41 33
(2005) Development,
Symptom
Reduction
Karver et al. 12 w CBT C, T, WAI-S, E CES-D C 0.12 12
(2008)a O AOCS
Katz (1999) 5 Dynamic C WAI-S E Premature O 0.03 100
Termination
Kech (2008) 16 IPT C NSI A Depression C 0.56 20
Composite
Kelly et al. ∗ Various C, T WAI-S A SCL/BSI C 0.28 83
(2009)
Kivlighan 12 Various C, T WAI E, M, L Interpersonal C 0.17 21
et al. (1995) Problems
Kivlighan 4 Various C WAI E IIP, BIC O 0.55 38
et al. (2000)
Klee et al. 29 Various O TARS E SCL/BSI, Global C 0.23 32
(1990) Outcome
Klein et al. 12 CBT T WAI-S E, M HRSD O 0.31 367
(2003)
(Continued )
34
Table 2.1 Continued
Knaevelsrud 5 w CBT C WAI-S L SCL/BSI, IES C 0.48 41
et al. (2007)
Kokotovic 4 Various C, T WAI E Premature C 0.13 105
et al. (1990) Termination
Kolden 4 Dynamic C TBS E Mental Health C 0.30 60
(1996) Index
Konzag et al. 12 w Various C, T HAq E SCL/BSI C 0.21 225
(2004)
Kramer et al. ∗ Various C HAq A SCL/BSI C 0.25 50
(2008)a
Kramer et al. ∗ Various C HAq A SCL/BSI C 0.80 50
(2009)a
Krupnick 16 Various O VTAS E, A Global Outcome C, O 0.46 206
et al. (1994)a
Krupnick 16 Various O VTAS E, A HRSD, BDI O 0.46 206
et al. (1996)
Kukla et al. ∗ Vocational C WASc A Job Tenure, O -0.18 91
(2009) Program Working Duration
Lansford 12 Dynamic O AWR A Global Outcome C, T, O 0.89 6
(1986)
Lieberman ∗ Acute Inpatient C, T ITGA, EH E Symptom C 0.30 63
et al. (1992) Improvement, GAS,
Premature
Termination,
Defense Style, RSE
Liebler et al. ∗ Various C BPSR M SCL/BSI C 0.07 87
(2004)
Loneck et al. ∗ Intake Interview O VPPS E Referral O 0.23 39
(2002) Appointment
Luborsky 52 Dynamic O HAq(cs), E, L, A Rated Benefits, C, T, O 0.54 20
et al. (1983) HAq(r) Residual Gain,
Success,
Satisfaction,
Improvement
Luborsky 0.79 77
et al. (1985)a
Mallinckrodt 12 Various C, T WAI E Global Outcome C, T 0.63 40
(1993)
Mallinckrodt 15 Brief C WAI E SCL/BSI, Social C 0.54 34
(1996) Interpersonal Support, BDI
(Continued )
35
Table 2.1 Continued
Marmar 18 Various C, T CALPAS E BDI C 0.18 18
et al.
(1989a)b
Marmar 12 Dynamic O CALTARS A Patterns of C 0.39 52
et al. Individual
(1989b) Change
Scores, SCL
Marmarosh ∗ Various C, T WAI-S E SCL/BSI C 0.30 48
et al. (2009)
Marshel 50 Dynamic C HAq, E Premature C -0.06 101
(1986) TARS, Termination
Marziali 12 Dynamic O TARS A Composite C, O 0.35 10
et al. (1981) Outcome
Marziali 20 Dynamic C, T, TARS A Behavioral C, T, O 0.24 42
(1984) O Symptom Index,
SAS, Global
Outcome
Marziali 30 Dynamic C TAS† E, L SAS, Objective C 0.79 17
et al. (1999)b Behavior Index,
SCL/BSI
McNeil 12 Various C, T, AQ A General Symptoms O 0.22 99
(2006) O
McLeod ∗ Various O TPOCS A Trait and Stait C 0.50 22
et al. (2005) Anxiety
Meier et al. ∗ Alcohol and C, T WAI-S E Premature O 0.01 187
(2006a)a Drug Termination
Abstinence
Program
Meier et al. ∗ Alcohol and Drug C, O WAI-S E Premature O 0.01 187
(2006b)b Abstinence Termination
Program
Meyer et al. 16 Various O VTAS E HRSD, BDI C, O 0.49 151
(2002)a
Missirlian 16 Experiential C WAI-S E, M, L SCL/BSI, BDI, IIP, C 0.37 32
et al. (2005) RSE
Mohl et al. ∗ Various C HAq E Premature C 0.30 80
(1991) Termination
Moleiro 20 Alcohol and C STS, A BDI, Composite C, O 0.48 186
(2003)a Drug TPRS Outcome
Abstinence
Program
(Continued )
36
Table 2.1 Continued
Morgan et al. 52 Dynamic O HAr E, A Composite C, T, O 0.59 20
(1982)a Outcome, Rated
Benefits
Moseley 14 Various C WAI E State-Trait Anxiety, C 0.28 25
(1983) Self-Concept, TC,
PTQ
Multon et al. 7 Career C WAI-S E SCL/GSI, C 0.14 42
(2001) Counseling Instability
Muran et al. 20 Cognitive C CALPAS A SCL/BSI, C, T 0.38 37
(1995) Interpersonal
Problems, GAS,
TC, Overall
Outcome
Muran et al. 30 w Various C, T WAI-S E Premature O 0.38 99
(2009) Termination,
Interpersonal
Functioning
O’Malley ∗ Various O VPPS E Overall Outcome, C, T, O 0.55 38
et al. (1983)a TC
Ogrodniczuk 20 Interpretive, C, T NSI A General Symptoms, C, T, O 0.35 67
et al. (2000) Supportive Individual
Objectives,
Social-Sexual
Adjustment
Pantalon 19 w CBT C IVRS A Abstinence, O 0.46 16
et al. (2004) Premature
Termination
Pavio et al. 12 Experiential C WAI E, L SCL/BSI, C 0.24 33
(1998) SASB Introject,
Unfinished
Business
Schale
Piper et al. 19 Dynamic C, T AAS A Composite C, T, O 0.52 64
(1991) Outcome
Piper et al. 19 Dynamic C, T AAS A State-Trait C, T 0.54 30
(1995) Anxiety, BDI,
SCL/BSI, Overall
Usefulness
Piper et al. 20 Dynamic C, T NSI A Composite C, O 0.10 144
(2004)a Outcome
Pos (2007) 18 Experiential C WAI M SCL/BSI, BDI C 0.34 74
Pos et al. 18 Experiential C WAI M SCL/BSI, BDI C 0.34 74
(2009)
(Continued )
37
Table 2.1 Continued
Priebe et al. 20 Case management C BAS E Hospitalization O 0.28 58
(1993) months Index, Work Axis,
Accommodation
Prigatano 6 Neuropsychology T NAS L Productivity O 0.40 35
et al. (1994) months Rehabilitation
Pugh (1991) 12 Various C, T WAI E SCL/BSI, TC C, T 0.18 55
Pyne (1991) 6 Various T, O HAq(r), A Global Outcome, C, T, O 0.34 29
VPPS Premature
Termination
Ramnerö 16 CBT T WAI-S M Outcome O -0.06 59
et al. (2007) Composite
Reiner ∗ Dynamic C TBS E Overall Outcome O 0.40 82
(1987)
Reis et al. 16 Dynamic C WAI E HRSD O 0.07 58
(2004)
Riley (1992) 8 Various C, T WAI, E, L SCL/BSI, TC, GAS C, T, O 0.17 61
CALPAS
Rogers et al. ∗ Case C, T WAI-S E Depressive O 0.27 64
(2008) Management Symptoms
Rounsaville 14 Interpersonal O VPPS E Schedule for O, C 0.25 35
et al. (1987)a Affective
Disorders, SAS,
Patient Self-
Assessment
Rudolf et al. ∗ Dynamic C, T TRS E, L Composite C, T 0.44 238
(1993) Outcome
Safran et al. 20 Cognitive C WAI, E SCL/BSI, C, T 0.53 22
(1991) CALPAS MCMI, BDI,
Global
Success, TC
Sammet ∗ Various C HAq A SCL/BSI, IIP C 0.16 213
et al. (2004)
Samstag 30 Various C, T WAI-S A SCL/BSI, IIP C 0.55 48
et al. (2008)
Santiago 12 CBT C WAI-S E HRSD O 0.22 324
et al. (2005)
Saunders 26 Dynamic C TSR E Mental Health C 0.16 114
(2000)a Index
Saunders 26 Dynamic C TBS E Session Quality, C, O 0.20 113
et al. (1989)a Termination
Outcome
(Continued )
38
Table 2.1 Continued
Schauenburg 11 Dynamic C, T HAq L SCL/BSI, Severity C 0.23 284
et al. (2005) Rating
Schleussner ∗ Dynamic C HAq E Satisfaction C 0.13 57
(2005)
Schönberger 14 Rehabilitation C, T WAI-S E EBIQ C, T 0.14 59
et al.
(2006a)a
Schönberger 14 Rehabilitation C, T WAI-S† E EBIQ C, T 0.31 103
et al.
(2006b)a
Schönberger 14 Rehabilitation C, T WAI-S E Composite C, T 0.14 59
et al. Outcome
(2006c)a
Schönberger 14 Rehabilitation C, T WAI-S E Cognitive C, T 0.14 104
et al. (2007)a Functioning
Sexton 10 Various C WAI-S E BOPS, Beck C, O 0.40 27
(1996) Anxiety Scale,
SAS, GAS,
BSO, Zung,
Global Problem
Rating
Sherer et al. ∗ Rehabilitation C, O, CALPAS, E Premature O 0.18 56
(2007) T NAS Termination,
Productivity,
Functional
Status
Shirk et al. 12 w CBT C, T, AOCS A BDI, Depressive C, T 0.25 50
(2008)a O Symptoms
Solomon 2 years Case C, T WAI L Quality of Life, C, O 0.28 82
et al. (1995) Management Compliance,
Satisfaction with
Treatment, other
Variables
Sonnenberg 11 Inpatient C, T ITAS E SCL/BSI C 0.03 63
(1996)
Spinhoven ∗ Various C, T WAI E Symptom Status O 0.25 70
et al. (2007)
Stevens et al. 30 30 C WAI E, M, L Outcome C, T 0.37 44
(2007) Composite
Stiles et al. 12 Various C ARM A SCL/BSI, BDI, IIP, C 0.25 76
(2004) SAS, RSE
(Continued )
39
Table 2.1 Continued
Strauser et al. ∗ Mental C WAS A Employment C 0.41 97
(2004)b Retardation Prospects, Job
Satisfaction
Strauss ∗ CBT C CALPAS A WISPI C 0.41 25
(2001)
Strauss et al. ∗ CBT C CALPAS A WISPI, SCID-II, C, O 0.45 30
(2006) BDI
Svartberg 20 Dynamic C FAI M SCL/BSI, DAS C 0.38 11
et al. (1994)
Tichenor 16 Various C, T, WAI, A SCL/BSI, Self C, T, O 0.16 8
(1989) O CALPAS, Concept, TC,
HAq(r), HRSD, HRSA
VTAS
Trepka et al. 16 CBT C CALPAS, A BDI C 0.50 30
(2004) ARM
Tryon et al. 19 Various C, T HAq M Premature C 0.20 74
(1990) Termination
Tryon et al. 13 Various C, T WAI-S E Premature C 0.25 86
(1993) Termination
Tryon et al. 10 Various C, T WAI-S E Premature C 0.25 71
(1995) Termination
Tunis et al. 180 Methadone C CALPAS E, M, Premature C 0.34 20
(1995) days Detox. L, A Termination,
Opioid Use, HIV
Risk Behavior
Van et al. 16 Various C HAq M Depressive C 0.24 62
(2008) Symptoms
Vogel et al. 12 CBT C HAq M Y-BOCS O 0.36 37
(2006)
Vronmans 8 Narrative C WAI E, M, L BDI, OQ-45 C 0.48 34
(2007) Therapy
Wettersten 12 Various C WAI A SCL/BSI, C 0.27 32
(2000) Satisfaction
Wettersten 12 Various C WAI A SCL/BSI, C 0.27 32
et al. (2005)b Satisfaction
Wilson et al. 19 Various C HRQ E Frequency of C 0.00 154
(2002) Vomiting
Windholtz 16 Dynamic O VPPS M SCL/BSI, C, T, O 0.20 38
et al. (1988) Overall Change,
TC, GAS
(Continued )
40
Table 2.1 Continued
Yeomans 230 Dynamic O CALPAS E Premature C 0.05 20
et al. (1994) Termination
Zuroff et al. 16 Various O VTAS L DAS, C 0.10 149
(2000) Maladjustment
Composite
Zuroff et al. 16 CBT O BLRI E Maladjustment C, O 0.18 48
(2006) Composite
Notes:
Raters: C = client, T = therapist, O = other/observer
Time E = early, M = middle, L = late, A = averaged alliance
RG = residual gain score
Alliance AAS = Alberta Alliance Scale
Measures: AE = Active Engagement
AOCS = Alliance Observation Coding System
AQ = Alliance Questions
ARM = Agnew Relationship Measure
AWR = Alliance Weakenings and Repairs
BAS = Berlin Alliance Scale
BLRI = Barrett-Lennard Relationship Inventory
BPSR = Bern Post Session Report
CALPAS = California Psychotherapy Alliance Scale
CALTARS = California Therapeutic Alliance Rating Scale
CIS = Client Involvement Scale
CS = Coordination Scale
EH = Patient expectation of helpfulness
FAI = Facilitative Alliance Inventory
HA(r) = Penn Helping Alliance Scale - Rated
HAq = Helping Alliance Questionnaire - Self-Rated
HA(cs) = Helping Alliance Counting Signs
HRQ = Helping Relationship Questionnaire
ITAS = Various Inpatient Therapeutic Alliance Scales
ITGA = Inpatient Task and Goal Agreement
IVRS = Interpersonal Variables Rating Scale
NAS = Neuropsychology Alliance Scale, Prigatano Alliance Scale
NSI = Non Standard Instrument (Measure developed for the specific research project)
OAS = Observer Alliance Scale
SASB = Structural Analysis of Social Behavior
STS TPRS = Systematic Treatment Selection Therapy Process Rating Scale
TARS = Therapeutic Alliance Rating Scale
TBS = Therapeutic Bond Scale
TRS = Therapeutic Relationship Scale
VTAS = Vanderbilt Therapeutic Alliance Rating Scale,
VPPS = Vanderbilt Psychotherapy Process Scale,
WAI = Working Alliance Inventory,
WAI-S = Working Alliance Inventory - Short version
WASu = Working Alliance Survey
WASc = Working Alliance Scale
(Continued )
41
Table 2.1 Continued
Outcome ACQ = Agoraphobic Cognitions Questionnaire
measures: BDI = Beck Depression Inventory
BIC = Battery of Interpersonal Capabilities
BOPS = Brief Outpatient Psychopathology Scale
BPRS = Brief Psychiatric Rating Scale
BSQ = Body Sensation Questionnaire
CES-D = Center of Epidemiologic Studies Depression Scale
CPRS = Comprehensive Psychopathological Rating Scale
DAS = Dysfunctional Attitudes Scale
DpD = Drinking per Day
DTES = Drug Taking Evaluation Scale
EBIQ = European Brain Injury Questionnaire
GAS = Global Assessment Scale
GHQ = General Health Questionnaire
HRSA = Hamilton Rating Scale for Anxiety
HRSD = Hamilton Rating Scale for Depression,
IES = Impact of Event Scale
MCAS = Multnomah Community Ability Scale
PANSS = Positive and Negative Syndrome Scale
PICS = Pattern of Individual Change Scores
PTQ = Post Therapy Questionnaire
RSE = Rosenberg Self-Esteem Index
SCL/BSI = Symptom Checklist 90, Brief Symptom Inventory
SEQ = Session Evaluation Questionnaire
STAI = State-Trait Anxiety Inventory
SWLS = Satisfaction with Life Scale
TC = Target Complaints
WBI = Working Behavior Inventory
WISPI = Wisconsin Personality Disorder Inventory
Y-BOCS = Yale-Brown Obsessive-Compulsive Scale
Zung = Zung’s Self-Rating of Depression
Treatment: ∗
= various length
w = weeks of treatment
†
Partial Scale
a
Studies that share data with other report(s).
b
Studies that contributed multiple independent alliance–outcome relations.
Therapist (T): You think I should? recognize it, and perhaps prevent a
C: You do; always. re-play of the same old grooves . . .
T: Because we agreed that looking at [pause 10 sec]
connection between past relationship T: I said that “we agreed” that this
patterns and how you and [name] are is the way to go, but I get the
getting on is . . . sense that you may not be
C: {voice over} Yes, unfinished convinced that’s so . . . it is
business . . . And all that. such a good idea.
T: It may be that there is a pattern C: Look, I mean . . . you are the
here, which would be useful to therapist and I keep fucking up with
explore and understand better. my “old lady.” So I guess I better
Once we understand it, we can start thinking & talking about these
patterns . . . I wish there was a pill therapy—he drops his previous agenda
or electric shock therapy to . . ., it and demonstrates his commitment to find
would be faster. a way of working collaboratively with his
T: Maybe we better take a step back. patient. Clients frequently have a mixture
I am a therapist, but I can’t give you of hopes and worries about discussing long
a pill or shock you to fix you. And suppressed feelings and memories of deep
looking for these unfinished patterns significance. The therapist’s challenge in
don’t seem to make an awful lot building the alliance is to recognize, legiti-
of sense to you . . . right now. But mize, and work through these issues and
I hear you are willing to be a “good engage the client in a joint exploration of
client” these obstacles.
C: But this not what it is about, about The following excerpt provides another
me being good, I mean, right? brief example of such a process:
T: What would you say if you were not
a “good patient”? Would you rebel? C: “[topic discussed last week]” . . . was
C: I guess I might . . . It’s crazy you interesting . . . But sometimes I can’t
know, before I got married I was a remember what I talked about from
pretty wild dog . . . long hair, one week to the next.
motorcycles, some pretty crazy stuff. T: . . . I think we ended up last talking
T: So, what happened? Where did the about how difficult it is to imagine
“crazy you” go? What did you do how things would be different.
with him? C: <overlap> I sometimes wonder . . .
C: Married, good job, slick house, nice what do therapists do after the
kids, you know . . . session? I mean . . . Do you walk
T: You think I might meet this around the block to forget all this
character? He seems to have been craziness . . .? Do you go home and
shut up but not forgotten . . . He dream about it?
might have something interesting T: Hmm, I . . .
to say . . . C: [overlap] I mean, it is not like having
C: I might be a little afraid of my old a discussion with a friend; though
self . . . But [with different voice]: goodness knows, I sometimes forget
Doc, I’m trash, my old man was about those too. I think to myself,
trash, but he put his money in good does he (T) need to hear all this?
booze; not in psychiatrists’ pockets! How often did I tell you that stuff?
T: He did not have much faith in this I read that Freud sometimes napped
therapy business behind the couch . . . Not, mind
C: Yeah, Of course you should not let you, that I think you nap! But
him write the cheque for the session; sometimes you look tired <Laughs>.
it would for sure bounce . . . [both Oh, don’t mind; this was a useful
laugh] session. <looks at the clock> Are we
done? <Stands up>
In the above excerpt the therapist T: So I guess sometimes you wonder
starts off defending his “modus operandi,” “what is it in it for him (T)”?
but when he becomes aware of the client’s C: I knew you’d say that!
ambivalent feelings about dealing with T: Well . . . I am not “really a friend.”
the past—and possibly about being in It is a strange thing to pour one’s
heart out to someone and then in a way I have not been able to talk
wonder: Did it mean anything to about it before. Last week, I
him? What am I to him? mean . . . But kind of pulled back
C: Yeah, I guess . . . That’s therapy, for and felt mixed up when we
you! <stand up again as to go> started . . .I don’t like risking myself
T: Not sure if you want to talk about much do I? . . . Hmm, I guess I went
this or go? to the right school: “The hit and run
C: Well it is late . . . academy of motherly love” . . . I am
T: Interesting that this came up to-day. so tired of it [pause] . . . I think I am
And . . . kind of left hanging . . . making the connection . . . [pause]
between us. We got someplace today.
C: You mean “Hit & Run” . . .? when
It is important to note that clients, espe-
I don’t . . . get . . . something . . .
cially in the beginning of treatment, may
[I want] I don’t wait for an answer.
appear to be hostile, rejecting, or fearful of
T: There was something you
treatment or the therapist. The therapist’s
wanted . . . from me . . .?
ability to respond with acceptance and an
C: Doesn’t take a rocket scientist to
openness to discuss these challenges is an
figure out . . . When you where
important asset in establishing the alliance.
asking “does it (therapy) work for
There is some research evidence to show
you” {reference to last week’s
that therapists who respond with their own
discussion} I thought here it
negativity to client’s hostile remarks will
comes . . .
likely damage the alliance (Henry, 1994).
T: You mean I’ll quit on you?
The last excerpt offers a brief illustration
C: I know you would not do that.
of what the concept that we psychologists
I know you wouldn’t. But, I mean,
call alliance feels like from the client’s
we are talking about this all this
perspective.
time, and I think . . . I talk about
it to others too {relates an incident C: Yeah, I am more comfortable
of talking about his marriage working with you . . . After finishing
to a colleague} Now I know she with Dr. “K” I was not too sure
{colleague} feels sorry for me, but of about getting into therapy again for
course this doesn’t help either. But two years. My previous therapist—
that’s different. Kind of . . .it’s not I went to him for about a year—he
sympathy I need, but sometimes was great at listening . . . I mean he
feels . . . <voice goes shallow, eyes had a good reputation and I think
moist> he was older than you. He must
T: You want from me . . . how I as have heard of these things before.
person feel . . . about . . . But I thought he was afraid that if
C: <Change of expression; sarcastic> he told me something I’ll do it like
Good fucking time to bring it up! a robot or something. I mean,
T: Does this; like this . . . remind . . . I know these are my decisions and
C: You mean do I do this Hit & Run I got to get my own answers and
with B (wife), yeah. I’ve been sometimes you tell me that I’m
thinking about that. Kind of stupid trying to get around busting my
but interesting; I felt we were own ass by getting you to tell me
really . . . I was telling you something what you think . . .
Th: I . . . account for correlations of the outcome
C: [Chuckle]. It’s OK, you do it quite measures within studies. In addition, we
nicely. But I can tell. [Pause] But extracted some alliance–outcome relations
you respect an honest question and not available previously, and adjusted for
seem to try to work with me the variations in the number of participants
way I want to, not always out of used to calculate alliance–outcome rela-
the book . . . tions within studies. As a result, both the
I mean the other day, last week I mean, ES and sample size (k) associated with some
I was . . . I just could not let go of studies in this report are not identical to the
that anger. I guess I was not very well values reported in previous meta-analyses.
behaved here, as a patient I mean. To locate research with data on the rela-
But it was important for me to hear tion between alliance and outcome from
when you said “you will not let go 2000 to 2009, we searched the PsycINFO
without taking a piece of me.” Then database using the same search parame-
we talked . . . I talked like a normal ters as the Horvath and Bedi (2002) meta-
client. But I needed to get a foot into analysis published in the previous edition
you and hear “ouch” for me to look of this book. In addition, we had access to
at what is happening. I needed your a list of e-mail address for persons with
“ouch” to see into me, and not a whom the first author corresponded on the
finger from on high. subject of the alliance; these individuals
were invited to identify studies meeting the
These brief excerpts were selected to selection criteria.
illustrate how different therapy contexts The criteria for inclusion in this report
draw on diverse therapist resources, and were: (1) the study author referred to the
also the fact that the concept of the alli- therapy process variable as alliance (includ-
ance unites the notions of interventions ing variants of the term), (2) the research
and the development of the relationship in was based on clinical as opposed to analog
therapy. Alliance is built by doing the work data, (3) five or more patients participated
of therapy collaboratively. in the study, and (4) the data reported was
such that we could extract or estimate
Meta-Analytic Review a value indicating the relation between
Sources of Data alliance and outcome.
To locate research published between 1973 In reviewing the retrieved material, we
and 2000, we relied on the three previously discovered that there is a growing literature
published meta-analyses (Horvath & Bedi, linking alliance to the effectiveness of med-
2002; Horvath & Symonds, 1991; Martin, ical interventions as well as a variety of
Garske, & Davis, 2000). However, most social and even legal services. However,
of the previously published effect sizes it was decided that this literature was out-
(ESs) were recomputed for this analysis, side the scope of this report. We chose
using a more detailed coding system to take to focus more narrowly on the relation
account of added features and to better between the quality of the alliance and
identify interdependencies in the underly- outcome in the context of psychological
ing data when more than one research treatments. Alliance research conducted
report shared the same client sample. We on couples and family therapy and alliance
also applied more sensitive statistical analy- research on children were also excluded as
ses to the previously published data to these topics are covered by other chapters
in this volume. However, treatments for appeared overwhelmingly (153) in peer-
substance abuse as well as psychological reviewed journals, with some (5) in book
problems that involve psychoactive medi- chapters, while 43 items came from unpub-
cations are included. lished (mostly dissertations) sources. The
In contrast to previous meta-analyses, later represent a significant increase
we attempted to include research published in the proportion of unpublished research
in languages other than English. Our liter- in the current data compared to previous
ature search was extended to material pub- meta-analyses. In total, the data captures
lished in Italian, German, and French. information based on over 14,000 treat-
A search was conducted of the German ments. (In Table 2.1, we provide ESs
language database (PSYNDEX) using the associated with each manuscript, but the
same inclusion criteria as for the English aggregated effect sizes, and all of the cal-
language searches. One hundred and culations presented below, were adjusted
fifty- two German abstracts were retrieved. for shared (nonindependent) data and
Of these, 17 manuscripts contained usable are based on the 190 independent effects
alliance–outcome data and were included sizes.)
in the analysis. For the French and Italian The number of eligible studies included
literature, we searched in PsycINFO with in this chapter is roughly double the size of
the additional keywords French OR Francais the data that were available in the previous
OR Italian OR Italiano. We accessed the chapter. The growth in the literature over
search platforms EBSCO (USA) and OVID the past decade means not only that there
(Europe). Of the 87 French articles located, are more studies available for analysis, but
73 manuscripts were written in English also that there is a significant increase in
and published in English journals; of the the types of therapies, treatment contexts,
remaining 14 items, 2 had usable alliance– client problems, and research designs cap-
outcome data; these are included in the tured by the current analysis.
analyses. Twenty-six Italian manuscripts Even with an effort to include non-
were located; of these 14 were published in English publications, the geographic distri-
English journals, and none of the Italian- bution of research in our data is strongly
only papers had usable data. In total, 19 biased: 153 manuscripts came from North
research reports unavailable in English were America (134 USA, 19 Canada), 45 from
included in the analysis. Europe (22 from German-speaking coun-
The 201 research reports included in the tries, 10 from Scandinavia, 8 from UK,
meta-analysis are listed in Table 2.1. Thirty- and 8 from other countries in Europe),
nine of these manuscripts were based on a and three research reports came from
shared data; that is, two or more reports pro- Australia. Notwithstanding these limita-
vided alliance–outcome information derived tions, it is reasonable to claim that the data
from a common pool of clients. Thus, some we present closely mirrors the universe of
of these reported effect sizes were not inde- alliance research, since it appears that most
pendent. In addition, 10 research publica- foreign-speaking researchers who do this
tions listed in the table reported multiple kind of work publish in English language
alliance–outcome relations based on two or journals.
more independent samples.
The data on which our analysis is based Methods of Analysis
includes both published (158) and unpub- For our numerical estimates, we used the
lished (53) research. The published research random-effects model. The reasons for this
were twofold. First, given the broad range of correlation (which has a negative skew),
of applications, research designs, and mea- and the Fisher’s z transformation results in
surement approaches within our data, an approximately normal distribution.
we could not assume the existence of an In cases where the primary study reported
underlying, homogeneous, singular, alliance– more than one level of a categorical vari-
outcome index of alliance–outcome rela- able (e.g., both clients’ and therapists’
tions. By using a fixed-effects model, we alliance scores), dependencies at the mod-
would “. . . assume homogeneity of under- erator level were accounted for by randomly
lying treatment effects across studies [and selecting one within-study level per study.
this] may lead to substantial understate- This procedure allowed for a fully indepen-
ment of uncertainty” (National Research dent analysis at the moderator level. This
Council, 1992, p. 187). Second, the random- random selection procedure provided a
effects model, apart from requiring fewer safeguard from violating the assumption of
assumptions, yields a more conservative independence in testing differences among
estimate and hence leads to safer, more levels of moderators; however, using this
trustworthy, conclusions (Cooper, Hedges, procedure also reduced the sample size and
& Valentine, 2009; Hunter & Schmidt, thus the power of the analysis. All proce-
2004). A random-effects model assumes dures for this meta-analysis were conducted
that the studies analyzed are selected from using the MAc (Del Re & Hoyt, 2010) and
a population of studies and thus the results RcmdrPlugin.MAc (Del Re, 2010) meta-
are generalizable to the larger universe of analysis packages for the R statistical
studies. software program (R Development Core
In many studies, there were a number Team, 2009).
of different outcome measures and hence
multiple effect sizes were reported. In order Results
to account for the dependencies among The aggregate effect size, for the 190 inde-
outcome measures, due to multiple within- pendent alliance–outcome relations repre-
study ESs, we employed Hunter & Schmidt’s senting over 14,000 treatments was r =
(2004) aggregation procedures to obtain 0.275. The 95% confidence interval of
one correlation effect size per study. These this aggregated ES ranged from 0.249 to
procedures take into account the correla- 0.301. The aggregated value is adjusted for
tion among within-study outcome mea- sample size, as well as the intercorrela-
sures and thus yield a more precise estimate tion among outcome measures. The magni-
of the population parameter. In cases where tude of the relationship in the current
the primary studies did not provide actual meta-analysis is a little larger but similar to
correlations among outcome measures, the values reported in previous research
the estimate of between-outcome measure (Horvath & Bedi, 2002, r = 0.21, k = 100;
correlation was set to 0.50 (Wampold, Horvath & Symonds, 1991, r = 0.26, k = 26;
1997). Martin, Garske, & Davis, 2000, r = 0.22,
When conducting categorical and con- k = 79). The median effect size of ESs of the
tinuous moderator analyses, all correlations current data set is 0.28 (not adjusted for
were transformed to a Fisher’s z (Fisher, sample size), suggesting that the group of
1924) and then transformed back to r effect sizes we collected are not strongly
for interpretive purposes. The correlation skewed. The overall effect size of 0.275 is
coefficient is known to be nonnormally statistically significant at p < 0.0001 level,
distributed, particularly with high values indicating a moderate but highly reliable
relation between alliance and psychother- hidden in dusty file drawers to generate an
apy outcome. aggregate ES that was no longer statistically
This effect size of 0.275 was esti- significant.
mated based on studies located using elec- Another way to explore the question of
tronic databases. Therefore, this estimate whether there is a sampling bias effecting
is potentially vulnerable to the file drawer the data is by inspecting the funnel plot of
bias (Sutton, 2009): the possibility that the the collection of ES in our set. A funnel
research literature we accessed represents a plot is a diagram of standard error on the
biased sample, as there might be a number vertical axis as a function of effect size on
of studies with smaller or null ESs languish- the horizontal axis. In the presence of bias,
ing in file drawers and unlisted in databases we would expect the plot to show a higher
(possibly rejected by journals because they concentration of studies on one side of the
report nonsignificant results). The conse- mean than the other. Typically, smaller
quence of such a scenario can be evaluated sample size studies (having larger standard
by computing the fail-safe N. This is the errors) are more likely to be published
number of studies with ES = 0 that would if they have larger-than-average effects. In
make the aggregate ES in the database sta- the absence of publication bias, we would
tistically nonsignificant (p > 0.05). We have expect the studies to be distributed rela-
calculated the fail-safe value (Rosenthal, tively symmetrically around the aggregated
1979): there would have to be over one ES. The funnel plot in Figure 2.1 does not
thousand ES = 0 (null) additional studies indicate a strongly biased set of data, but
Publication bias
0.2
0.3
Standard error
0.4
0.5
−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

Fisher’s z
Fig. 2.1 Funnel plot of the ESs in the meta analysis.
neither is it perfectly symmetrical about the relation, we would expect the reported
vertical axes. research results to cluster around a popula-
We investigated two possible sources tion parameter with deviations from
of systemic bias in the distribution of the true value due only to random errors.
ESs: date of publication and study sample We computed the I 2 statistic, which pro-
size. There was a small and statistically vides an estimate of the percentage of
nonsignificant negative time trend observed variance of ESs over and above the amount
( p = 0.082). Over time (1972–2009) research- of variability that can be accounted for by
ers were reporting slightly decreasing ESs. random (chance) variation. The I 2 of 0.56
This makes intuitive sense because recent we obtained indicates that the variance in
studies use more sophisticated methods our data is approximately 56% greater than
for controlling for pre-therapy effects that one would expect if all the studies were
might impact the strength of the alliance. measuring the same relation. This finding,
There are also more studies published in and of itself, is not surprising; research-
recently involving client populations with ers assessed alliance at different points
more severe psychological problems. Both of therapy, in a variety of therapy con-
of these factors would likely exert a down- texts, using therapists, clients, or observers
ward pressure on the correlation between for their evaluations. In addition, outcomes
alliance and outcome. were measured from a variety of perspec-
More surprisingly, we found a signifi- tives, sometimes immediately after treat-
cant relation between sample size and ments, at other times at follow-up points.
ES (r = –.25 p <0.01). The best fitting Heterogeneity in the data encouraged us to
regression line for this puzzling association investigate the possible moderators effect-
is quadratic, nonlinear; the studies with ing the alliance outcome correlations.
sample sizes between 100 to 200 appear
to report lower ESs compared to studies Moderators and Mediators
with both smaller and larger sample size. Alliance Measures
This effect may be an artifact of some sort In our meta-analysis, over 30 different alli-
but will require further investigation. ance instruments were employed, but only
In sum, the overall relation between the four core instruments (CALPAS, HAq,
alliance and outcome in individual psycho- VPPS, WAI) were used in three or more
therapy is robust, is not effected by the file studies. (The HAq family is composed of
drawer problem, and accounts for approxi- two quite different instruments: the original
mately 7.5% of the variance in treatment [1983] versions coded as HAq, and the
outcomes. [1996] revision coded as HAqII). Therefore,
for the current study, we compared the
Variability of Effect Sizes aggregated ES of each of the four core instru-
There is a great deal of variability in the ments plus a residual category called “other.”
alliance–outcome relations across the 190 The box and whisker plot (Figure 2.2) dis-
ESs in the current data set. Similar to what plays the ESs associated with these measures
we found in the previous meta-analysis (range r = 0.23–0.39). The differences
(Horvath & Bedi, 2002), the group of among them were not significant (Q = 1.851,
alliance–outcome relations in this data set df = 5). However, it should be noted that
are not homogenous (Q = 498.42, df = 189, within the four core instruments, only the ES
p < 0.00001). If all the alliance–outcome associated with the CALPAS and the VPPS
research in our data were sampling the same were homogeneous. Effect sizes reported
0.8
0.6
0.4
Effect size
0.2
0.0
CALPAS (k = 24) HAQ (k = 25) HAQ II (k = 5) VPPS (k = 5) WAI (k = 78) OTHER (k = 52)
Alliance measure
Fig. 2.2 Box-and-Whisker plot of Effect Sizes (ES) associated with different alliance measures.
within each of the other measures were more only, if multiple assessments were available
variable than one would expect from chance in a study, we chose the earliest assessment
or random error alone. A likely reason for available for the computations. Figure 2.3
the homogeneity of the the CALPAS and shows the results of this analysis graphi-
the VPPS measures are available in fewer cally. As one would anticipate, the relation
versions than the WAI and the HAq. between alliance and outcome grew in
magnitude as the alliance and outcome
Time-of-Alliance Assessment became closer in time. The omnibus Q sta-
The time-of-alliance assessment was tistic for the overall contrast among these
grouped into four categories: Early: Sessions time categories was highly significant
1–5; Mid: after the fifth, but four or more (Q 17.42, df = 3, p <0.001), but the post-
sessions before end-of-treatment; Late: hoc pairwise multiple comparisons were
within three sessions of end-of-treatment; not statistically significant (p > 0.05), due
and Averaged a combination of assessment to the large within-category heterogeneity.
points. A number of researchers provided
information on multiple assessment times Sources-of-Alliance Assessment
within the same study. In most of our anal- The alliance can be rated from three per-
yses, one ES was randomly selected if mul- spectives: Clients, therapists, and observers.
tiple ESs were available in order to ensure Client and observer ratings are similar
independence of the data. However, for [Clients r = 0.282 (k = 109); Observer
reasons of clinical relevance, in this analysis r = 0.295 (k = 47)], and both of these
0.8
0.6
Effect size
0.4
0.2
0.0
Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

Session alliance rated
Fig. 2.3 Box-and-Whisker plot of Effect Sizes between alliance and outcome measured at different
phases of treatment.
perspectives of the alliance provided better Outcome Measures

prediction of therapy outcome than thera- As was the case with the alliance mea-
pist evaluations (r = 0.196, k = 40). These sures, a broad range of therapy outcome
findings are consistent with previous measures were utilized in the studies in
research (Horvath & Bedi 2002; Horvath our meta-analysis. Of the over 35 out-
& Symonds, 1991). However, the differ- come assessments, only three measures —
ences among these categories were not Beck Depression Inventory (Beck et al.,
statistically significant (Qb = 5.16, df = 2, 1961, BDI), SCL (Derogatis, Lippman, &
p = 0.076). An examination of the distribu- Covi, 1973), and premature termination or
tion of ESs within these categories indi- “dropout”— were utilized in sufficient fre-
cated that the variability of the ESs in the quency (5 ESs or more) to permit analysis.
client and therapist ratings was over 50% As a result, only a subset of 60 ESs that uti-
greater than expected by chance. That is, lized these three outcome measures could
the ESs in the set were heterogeneous, likely be included in this analysis. Table 2.2
due to the variety of measuring instru- displays the results of this analysis. The
ments used. Such a high degree of variabil- Q statistics for this set of moderators is
ity within these categories made it less significant (Q 10.98, df = 2, p = 0.004).
likely that the differences between the Post-hoc analysis indicated a statistically
categories of raters would reach statistical significant difference between studies using
significance. dropout (or premature termination) and
Table 2.2 Relation between the Alliance and Outcome Measures
K ES 95% CI (Lower) 95% CI (Upper) p
Dropout 15 0.164 0.062 0.262 .001

SCL 28 0.276 0.195 0.353 .000
BDI 17 0.409 0.304 0.505 .000
BDI as an outcome measure. In consider- the alliance–outcome relation among them

ing these results, it is helpful to keep in were not statistically significant (Qb = 4.85,
mind that the BDI is most often used as a df = 3, p = 0.183)*. These results support
symptom-specific outcome measure for the claim that the alliance is a pantheoreti-
clients receiving treatment for depression. cal factor in diverse types of treatments.
The relation between alliance and therapy However, it should be noted that only 93
outcome for the treatment for depression out of the total of 190 ES in the data set
tends to be relatively high. Dropout as a could be fitted into the four categories used
treatment outcome was almost exclusively for this analysis.
utilized in studies of clients with substance
abuse problems. While unilateral termina- Raters of Outcome Data
tion represents, in one sense, “hard” out- Similar to the measurement of alliance,
come data, the substance abuse treatments researchers used outcome evaluations
included in our data were highly varied, obtained from clients, therapists, indepen-
and clients in these treatments were often dent observers, or some combination of
volatile and multidisordered. This being these sources. In the data set, 109 ESs
the case, individuals might have termi- were based on clients, 47 ESs on observers,
nated therapy for a variety of reasons, apart 12 ESs on therapists, and 22 ESs were gen-
from lack of progress in treatment. The erated by other sources (e.g., dropouts,
observed differences among these studies days of sobriety, rehospitalization). The
are consistent with this hypothesis; the 15 difference among the alliance–outcome ES
ESs within the dropout group were highly obtained by these disparate raters was statis-
variable (p = 0.37). tically significant (Q = 8.34. df = 3, p < 0.05),
but the post-hoc test of pairwise contrast of
Types of Treatments differences was not statistically significant.
Bordin (1976, 1994) argued that the alli- Again, this result is likely due to the large
ance is a significant factor in all types of variability within these rater categories.
helping relationships. We explored the evi-
dence for this claim by contrasting the Halo Effect
averaged effect sizes associated with CBT, We examined the question of whether
IPT, psychodynamic, and substance abuse ESs were inflated when the alliance and
treatments. The ESs for each type of treat- outcome information came from the
ment, reflecting the strength of the associa- same (e.g., client rates both alliance and
tion between alliance and outcome, were outcome) or different sources (e.g., client
highly significant (p <0.001). But an analy- rated alliance and observer rated outcome).
sis of the contrast between the treatment
categories indicated that the differences in *Qb = between-group statistic
Table 2.3 Alliance outcome correlations disaggregated by raters of assessment.
Alliance rater Outcome rater
Client Therapist Observer Other
Client 0.30 (98) 0.27 (14) 0.21 (40) 0.34 (13)
Therapist 0.18 (31) 0.29 (20) 0.13 (20) 0.30 (12)
Observer 0.24 (28) 0.39 (10) 0.27 (17) 0.58 (7)
Other 0.16 (9) N/A 0.16 (2) 0.22 (8)
Note: Numbers in parentheses (k); Diagonal values represent data generated when the alliance and outcome were rated by the same source.
Data in this table are not independent; some studies provided data from more than one source.
Table 2.3 shows that ESs from studies in is that the categories we coded were not
which the same raters completed both the pure factors, but variables that interact
outcome, and alliance measures were indeed with one another. To better understand
higher on average than those coming from the influence of these variables on the rela-
different rater categories, but the difference tion between alliance and outcome, one
between these values fell slightly below needs to examine the complexities of these
the critical level for statistical significance variables acting together.
(p = 0.079). It is notable that the difference Modeling the full complement of poten-
between same source versus independent tial moderator interactions is statistically
source ESs has increased progressively unmanageable even with a large data set
since this effect was first analyzed in 1991 such as the one we collected. Not all the
using 27 data points (Horvath & Symonds, levels of these categorical variables intersect,
1991). Keeping in mind this apparent and there are computational difficulties
trend, the possibility that same-source eval- because many of the joint values are not
uations might be inflated may be a concern independent (i.e., are based on the same
of clinical significance in the future. data). Taking these limits into consider-
ation, a random effects multipredictor
Publication Sources meta-regression was computed to explore
The lion’s share of the ESs in our data was the joint impact of a subset of the clini-
published in refereed journals (153), fol- cally most interesting categorical modera-
lowed by unpublished studies (43), and 5 tors: The alliance raters (client, therapist,
extracted from books. The effect sizes associ- observer), the alliance measurements (using
ated with these sources were r = 0.287, only the core instruments WAI, HAq, HAq
0.237, and 0.399, respectively. The differ- II, VPPS, and CALPAS), and the three
ences among these ESs were not significant. major outcome indexes (BDI, SCL, and
dropout). The effects of the year of publica-
Interaction among Moderators tion and sample size were controlled in this
We examined a number of categorical analysis.
variables that potentially moderate the Because of the restrictions in the data,
alliance–outcome relation, but within most only 54 ESs could be entered into this
of these moderators, the range of ESs was analysis. However, almost half of the total
quite broad; most levels of moderator variance among these alliance–outcome rela-
categories were themselves heterogeneous. tions were explained by the individual and
One possible reason for such heterogeneity joint effects of these variables (R 2 = 0.46).
It was particularly interesting that the The results of these research studies defies
addition of the joint effect of the alliance easy summary. Multiple studies found some
measure × alliance rater alone contributed support for the existence of quadratic alli-
an R 2 change of 0.23. While the results ance patterns and their relation to posi-
of this analysis cannot be generalized to tive outcome in short-term therapy (e.g.,
the alliance–outcome literature, since only Horvath & Marx, 1991; Kivlighan &
about 25% of the studies could be used for Shaughnessy, 2000; Man, 1973; Miller
these calculations, the findings strongly et al., 1983). But several other studies (e.g.,
suggest that the abundant heterogeneity in Piper et al., 2004; Stevens et al., 2007; Stiles
the research findings is due in large part to et al., 2004) were unable to replicate their
the range of methods used to measure the results and confirm this hypothesis. There
alliance combined with the variety of means is some support for the prediction that
used to assess outcome. linear and increasing levels of the quality
The cost of inclusivity in defining both of alliance over the length of treatment
the outcome and especially the process are associated with better outcome than
(alliance) variable in research praxis is the flat-linear and decreasing-linear patterns
difficulty in arriving at a focused conclu- (de Roten et al., 2004; Kramer et al., 2009;
sion. An exemplar of the how this broad Piper et al., 2004). However, the associa-
conceptualization of variables both enriches tion of linear-increasing pattern with out-
the alliance research literature and at the come is problematic: some researchers
same time creates challenges in generat- analyzed data that was collected close to
ing convergence toward clinically useful the end of therapy; these “late” alliance
conclusion, we will briefly review one of measures are difficult to disentangle from
the most dynamic strands of the current outcome (Horvath & Luborsky, 1993).
research agenda: the investigation of the Kivlighan and Shaughnessy (2000) have
dynamics of the alliance over the span of also identified “U” and reverse or inverted
treatment. “U” patterns in brief therapy. However,
there are number of reports confirming the
Patterns of Alliance over Time existence of these patterns but the number
The patterns of growth and development of published reports unable to confirm the
of the alliance over the course of psycho- pattern (e.g., Kramer et al., 2009; Piper
therapy associated with a good outcome et al., 2004; Stiles et al., 2004; Stevens
have been of continued interest to research- et al., 2007) were about equally divided.
ers. Bordin (1985, 1989) suggested that The diversity among these findings is
the repairs of stresses or tears in the alliance almost certainly related to a number of
will make an important contribution research design and measurment issues. The
toward therapeutic gains. Gelso and Carter different length of therapies is one: research-
(1994) predicted that a rise in early alliance ers have examined treatments ranging
followed by a decline in the quality of the from as few as 6 to over 30 sessions. Some
alliance, followed again by an increase, studies explored the change in the level
would be associated with positive outcomes and the shape of alliance patterns over time
in therapy. There have been a number of independently (Stiles, et al., 2004; Strauss,
research projects aimed at investigating 2001), but others choose not to make those
these and other predictions in order to distinctions (e.g., Kivlighan & Shaughnessy,
document the relation between the various 2000). In addition, a number different
alliance patterns and outcome. alliance measures and a variety of statistical
approaches, have been used by different (i.e., growth or growth-decay-and-growth
researchers, (e.g., regression models, HLM, pattern) may be most beneficial is not yet
and cluster analysis), and there is no consis- resolved.
tency across studies in terms of the length
of time in the course of psychotherapy over Limitations of the Research
which patterns were examined. As well, This chapter is based on a numerical
researchers used different criteria for the synthesis of the research results. While
identification of growth patterns and curve our team has made a sustained effort to
shape. As a result, it is not yet clear if seek out all the available research on
the lack of convergence in this literature alliance–outcome relation, no meta-analysis
signals that the hypothesized good alliance is comprehensive, and this one is no excep-
development patterns are local to specific tion. At the very least, by the time the chap-
contexts (e.g., type of treatment or client ter goes to press, there will be, no doubt, a
problems), and thus cannot be generalized, number of new studies available.
or that the broad diversity of research meth- A more significant challenge is the
ods obscure a yet-to-be discovered general “apples and oranges” problem (Hunter &
pattern of good alliance development. Schmidt, 1990, p. 521). In including all
There are a couple of hypotheses, how- research in which the authors refer to the
ever, with promise of convergence: The process variable as alliance, we might have
most consistent finding appears to be the collected and summarized a number of
proposition that some fluctuation, that is, different kinds of things. This is a serious
change over time, particularly in the mid- concern, especially in light of the fact that
phase of therapy, is associated with posi- the ESs in this data set are quite diverse.
tive outcome when contrasted with a linear A practical response to this conceptual
or stable alliance pattern. It is also safe to problem is to assert that this chapter reports
conclude that no single pattern of alliance the results of alliance–outcome relation as
development or growth has been consis- it is researched.
tently documented as better or more pre- There are also some technical constraints
dictive of good outcome than alternative to the analyses we reported. We chose to
shapes, across different kinds of therapies use independent data whenever possible.
in different lengths of treatments. There To achieve this, on many occasions we
is mounting evidence that in treatments needed to randomly discard some data (ES)
where the quality of alliance is steadily in order to make sure that only one result
declining over time, the outcome is usually from a particular data pool was used in each
poor (Stiles, 2004). There is also conver- analysis. As a result, we lost power to detect
gence that some variability in the quality differences in a number of analyses. In the
of the alliance is likely an indication of long run, the use of independent data is
superior outcome compared to a situation statistically well justified, but the resulting
where the alliance is level and stable, so constraints put on the computations are
long as the overall quality of the alliance also important to consider.
does not decline over time (Safran, 1993;
Safran, Crocker, McMain, & Murray, Therapeutic Practices
1990; Safran & Muran, 1996, 2000; Strauss The positive relation between the quality of
et al., 2006). But, the amount of varia- the alliance and treatment outcomes for
tion that is optimal for outcome, or indeed many different types of psychotherapies is
the period over which the fluctuation confirmed in this meta-analysis. The question
of whether alliance contributes to outcome through some media (e.g., Internet,
beyond early therapy gains (e.g., Feeley, telephone). Different forms of therapy
DeRubeis, & Gelfand, 1999) has also call on diverse relational resources and
been largely resolved: a number of studies different levels of intimacy and intensity.
that controlled for this factor found that The therapist and client must find the
alliance is predictive of outcome above level of collaboration suited to achieve
and beyond early gains (e.g., Barber, et al., the work of therapy—even if they do not
2001; Brotman, 2004; Constantino, have face-to-face contact.
Arnow, Blasey & Agras, 2005; Gaston, • In the early phases of therapy,
et al., 1991; Klein, et al., 2003; Strauss, modulating the methods of therapy (tasks)
et al., 2006). While the overall ES of to suit the specific client’s needs,
r = 0.275 accounts for a relatively modest expectations, and capacities is important
proportion of the total variance in treat- in building the alliance. Clients are often
ment outcome, the magnitude of this cor- naïve in their expectation of what therapy
relation makes it one of the strongest and entails, how they have to participate in the
most robust predictors of treatment success process, and unaware of the links between
that research has been able to document what is happening moment to moment
(Wampold, 2001). In the following sec- during the session and the changes they
tion, we condense some of the most salient desire. Bridging the client’s expectations
points for practicing therapists. and what the therapist believes to be the
most appropriate interventions is an
• The alliance is not the same as the important and delicate task. Alliance
therapeutic relationship. The relationship emerges, in part, as a result of the smooth
is made of several interlocking elements coordination of these elements.
(empathy, responsiveness, creating a safe • Therapists need to closely monitor
secure environment, etc.) The alliance is the client’s perspective on the alliance
one way of conceptualizing what has been throughout treatment. It is frequently the
achieved by the appropriate use of these case that therapists’ and clients’
elements. perceptions of the alliance, particularly
• The fostering of the alliance is not early in treatment, do not converge.
separate from the interventions therapists Misjudging the client’s experience of the
implement to help their clients; it is alliance (i.e., believing that it is in good
influenced by, and is an essential and shape when the client does not share this
inseparable part of, everything that perception) could render therapeutic
happens in therapy. In this sense, the interventions less effective.
therapist does not “build alliance” but • The strength of the alliance, within
rather does the work of treatment in such or between sessions, often fluctuates in
a way that the process forges an alliance response to a variety of in-therapy factors,
with the client. such as the therapist challenging clients
• The development of a “good enough” to grapple with difficult conflicts,
alliance early in therapy is vital for therapy misunderstandings, and transference.
success. The sense of collaboration creates These “normal” variations —as long as
a working space, room to introduce new they are attended to and resolved— are
ways of addressing the clients concerns. associated with good treatment outcomes.
• The alliance matters in all forms of • Therapists’ non-defensive response to
therapy, including treatments mediated client negativity or hostility is critical for
∗
maintaining a good alliance. Therapists Barber, J. P., Connolly, M. B., Crits-Christoph, P.,
ought to neither internalize nor to ignore Gladis, L., & Siqueland, L. (2000). Alliance
predicts patients’ outcome beyond in-treatment
client’s negative responses.
change in symptoms. Journal of Consulting and
• Clients presenting with high Clinical Psychology, 68(6), 1027–1032.
interpersonal anxiety or with personality ∗
Barber, J. P., Gallop, R., Crits-Christoph, P.,
disorders tend to be particularly Barrett, M. S., Klostermann, S., McCarthy, K. S.,
challenging in terms of alliance et al. (2008). The role of the alliance and tech-
development and maintenance. niques in predicting outcome of supportive-
expressive dynamic therapy for cocaine
dependence. Psychoanalytic Psychology, 25(3),
461–482.
References ∗
Barber, J. P., Gallop, R., Crits-Christoph, P.,
Asterisk indicates works used in meta- Frank, A., Thase, M. E., Weiss, R. D., et al.
analysis. (2006). The role of therapist adherence, therapist
∗
Adler, J. V. (1988). A study of the working alliance in competence, and alliance in predicting outcome
psychotherapy. Unpublished doctoral dissertation, of individual drug counseling: Results from the
University of British Columbia. National Institute Drug Abuse Collaborative
Alexander, L. B., & Luborsky, L. (1987). The Penn Cocaine Treatment Study. Psychotherapy Research,
Helping Alliance Scales. In L. S. Greenberg 16(2), 229–240.
∗
& W. M. Pinsoff (Eds.), The psychotherapeutic Barber, J. P., Luborsky, L., Crits-Cristoph, P.,
process: A research handbook (pp. 325–356). Thase, M. E., Weiss, R., Frank, A., et al. (1999).
New-York: Guilford. Therapeutic alliance as predictor of outcome in
∗
Allen, J. G., Tarnoff, G., & Coyne, L. (1985). treatment of cocaine dependence. Psychotherapy
Therapeutic alliance and long-term hospital Research, 9, 54–73.
∗
treatment outcome. Comprehensive Psychiatry, Barber, J. P., Luborsky, L., Gallop, R.,
26, 187–194. Crits-Christoph, P., Frank, A., Weiss, R. D.,
∗
Allen, J. G., Tarnoff, G., Coyne, L., Spohn, H. E., et al. (2001). Therapeutic alliance as a predic-
Buskirk, J. R., & Keller, M. W. (1986). An inno- tor of outcome and retention in the National
vative approach to assessing outcome of long- Institute on Drug Abuse Collaborative Cocaine
term psychiatric hospitalization. Hospital and Treatment Study. Journal of Consulting and
Community Psychiatry, 37, 376–380. Clinical Psychology, 69(1), 119–124.
∗ ∗
Andreoli, A., Frances, A., Gex-Mabry, M., Barkham, M., Agnew, R. M., & Culverwell, A.
Aapro, N., Gerin, P., & Dazord, A. (1993). Crisis (1993). The California Psychotherapy Alliance
intervention in depressed patients with and Scales: A pilot study of dimensions and ele-
without DSM-II-R personality disorders. Journal ments. British Journal of Medical Psychology, 66,
of Nervous and Mental Disease, 181, 732–737. 157–165.
∗ ∗
Ankuta, G. Y. (1993). Initial and later therapeutic Bassler, M., Potratz, B., & Krauthauser, H. (1995).
alliance and psychotherapy outcome. Unpublished Der “Helping Alliance Questionnaire” (HAQ) von
doctoral dissertation, Michigan State University, Luborsky [The Helping Alliance Questionnaire
East Lansing. (HAQ) by Luborsky]. Psychotherapeut, 40(1),
∗
Arnow, B. A., Manber, R., Blasey, C., Blalock, J. A., 23–32.
Rothbaum, B. O., Thase, M. E., et al. (2003). Beck, A. T., Ward, C. H., Mendelson, M., Mock, J.,
Therapeutic reactance as a predictor of out- & Erbaugh, J. (1961). An inventory for measur-
come in the treatment of chronic depression. ing depression. Archives of General Psychiatry, 4,
Journal of Consulting and Clinical Psychology, 71, 561–571.
∗
1025–1035. Bethea, A. R., Acosta, M. C., & Haller, D. L.
∗
Baldwin, S. A., Wampold, B. E., & Imel, Z. E. (2008). Patient versus therapist alliance: whose
(2007). Untangling the alliance-outcome corre- perception matters? Journal of Substance Abuse
lation: Exploring the relative importance of Treatment, 35(2), 174–183.
∗
therapist and patient variability in the alliance. Bieschke, K. J., Bowman, G. D., Hopkins, M.,
Journal of Consulting and Clinical Psychology, Levine, H., & McFadden. (1995). Improve-
75(6), 842–852. ment and satisfaction in short-term therapy at
a university counseling center. Journal of College ∗
Busseri, M. A., & Tyler, J. D. (2003).
Student Development, 36, 553–559. Interchangeability of the Working Alliance
∗
Biscoglio, R. L. (2005). Patient and therapist per- Inventory and Working Alliance Inventory,
sonality, therapeutic alliance, and overall outcome Short Form. Psychological Assessment, 15(2),
in Brief Relational Therapy. Unpublished doctoral 193–197.
dissertation, City University, New York, NY. ∗
Busseri, M. A., & Tyler, J. D. (2004). Client-
Bordin, E. S. (1975, September). The working alli- therapist agreement on target problems, working
ance: Basis for a general theory of psychotherapy. alliance, and counselling outcome. Psychotherapy
Paper presented at the Society for Psychotherapy Research, 14(1), 77–88.
Research, Washington, DC. ∗
Card, C. A. (1991). Therapeutic alliance, comple-
Bordin, E. S. (1976). The generalizability of the mentarity, and distress during stress prevention
psychoanalytic concept of the working alliance. training after HIV testing. Unpublished doctoral
Psychotherapy: Theory, Research, and Practice, 16, dissertation, New School for Social Research,
252–260. New York.
Bordin, E. S. (1980, June). Of human bonds that ∗
Castonguay, L. G., Goldfried, M. R., Wiser, S.,
bind or free. Paper presented at the Society for Raue, P. J., & Hayes, A. M. (1996). Predicting
Psychotherapy Research, Pacific Grove, CA. the effect of cognitive therapy for depression:
Bordin, E. S. (1989, June). Building therapeutic A study of unique and common factors. Journal
alliances: The base for integration. Paper presented of Clinical and Consulting Psychology, 64,
at the Society for Psychotherapy Research, 497–504.
Berkley, CA. Chambless, D. L. (2002). Beware the dodo bird:
Bordin, E. S. (1994). Theory and research on the The dangers of overgeneralization. Clinical
therapeutic working alliance: New directions. In Psychology: Science and Practice, 9(1), 13–16.
A. O. Horvath & L. S. Greenberg (Eds.), The ∗
Chilly, C. (2004). The effects of the interpersonal
working alliance: Theory, research, and practice. framework of intervention on therapeutic compli-
(pp. 13–37). New York: Wiley. ance, alliance, and outcome. Unpublished doc-
∗
Botella, L., Corbella, S., Belles, L., Pacheco, M., toral dissertation, York University, Toronto,
Maria Gomez, A., Herrero, O., et al. (2008). Canada.
Predictors of therapeutic outcome and process. ∗
Cislo, D. A. (1988). Client perceptions of the thera-
Psychotherapy Research, 18(5), 535–542. peutic alliance and session outcome. Unpublished
∗
Bredel, S., Brunner, R., Haffner, J., & Resch, F. doctoral dissertation, University of Toledo,
(2004) Behandlungserfolg, Behandlungserleben Toledo, OH.
und Behandlungszufriedenheit aus der Sicht von ∗
Clarkin, J. F., & Crilly, J. L. (1987). Therapeutic
Patienten, Eltern und Therapeuten - Ergebnisse alliance and hospital treatment outcome. Hospital
einer evaluativen Studie aus der stationären and Community Psychiatry, 38, 871–875.
Kinder- und Jugendpsychiatrie [Treatment suc- ∗
Cloitre, M., Chase Stovall-McClough, K.,
cess, experience of treatment, and treatment sat- Miranda, R., & Chemtob, C. M. (2004).
isfaction from the viewpoint of patients, parents, Therapeutic alliance, negative mood regulation,
and therapists]. Praxis der Kinderpsychologie und and treatment outcome in child abuse-related
Kinderpsychiatrie. 53(4), 256–276. posttraumatic stress disorder. Journal of Consulting
∗
Broome, K.M. (1996). Antisocial personality and and Clinical Psychology, 72(3), 411–416.
drug abuse treatment process. Unpublished doc- ∗
Coleman, D. (2006). Client personality, working
toral dissertation. Texas Christian University, alliance and outcome: A pilot study. Social Work
Fort Worth. in Mental Health, 4(4), 83–98.
∗ ∗
Brotman, M. A. (2004). Therapeutic alliance and Connors, G. J., Carroll, K. M., DiClemente, C. C.,
adherence in cognitive therapy for depression. Longabaugh, R., & Donovan, D. M. (1997).
Unpublished doctoral dissertation, University of The therapeutic alliance and its relationship to
Pennsylvania, Philadelphia. alcoholism treatment participation and outcome.
∗
Burns, J. W., & Evon, D. (2007). Common and Journal of Consulting and Clinical Psychology,
specific process factors in cardiac rehabilitation: 65(4), 588–598.
Independent and interactive effects of the work- ∗
Constantino, M. J., Arnow, B. A., Blasey, C., &
ing alliance and self-efficacy. Health Psychology, Agras, W. S. (2005). The association between
26(6), 684–692. patient characteristics and the therapeutic alliance
in cognitive-behavioral and interpersonal ther- alcohol dependence. Drug & Alcohol Dependence,
apy for bulimia nervosa. Journal of Consulting 95(3), 230–236.
∗
and Clinical Psychology, 73(2), 203–211. Dunn, H., Morrison, A. P., & Bentall, R. P. (2006).
Cooper, H., Hedges, L.V., & Valentine, J. C. The relationship between patient suitability,
(2009). The handbook of research synthesis and therapeutic alliance, homework compliance
meta-analysis (2nd edition). New York, Russell and outcome in cognitive therapy for psycho-
Sage Foundation. sis. Clinical Psychology & Psychotherapy, 13(3),
∗
Crits-Cristoph, P., Cooper, A., & Luborsky, L. 145–152.
∗
(1988). The accuracy of therapists’ interpreta- Eaton, T. T., Abeles, N., & Gutfreund, M. J.
tions and the outcome of dynamic psycho- (1988). Therapeutic alliance and outcome:
therapy. Archives of General Psychiatry, 56, Impact of treatment length and pretreatment
490–495. symptomology. Psychotherapy: Theory, Research,
∗
Davis, L. W., & Lysaker, P. H. (2007). Therapeutic Practice and Training, 25, 536–542.
∗
alliance and improvements in work performance Emmerling, M. E., & Whelton, W. J. (2009).
over time in patients with schizophrenia. The Stages of change and the working alliance in
Journal of Nervous and Mental Disease, 195(4), psychotherapy. Psychotherapy Research, 19(6),
353–357. 687–698.
∗ ∗
Dearing, R. L., Barrick, C., Dermen, K. H., & Fakhoury, W. K. H., White, I., Priebe, S., & PLAO
Walitzer, K. S. (2005). Indicators of client Study Group (2007). Be good to your patient:
engagement: Influences on alcohol treatment How the therapeutic relationship in the treat-
satisfaction and outcomes. Psychology of Addictive ment of patients admitted to assertive outreach
Behaviors, 19(1), 71–78. affects rehospitalization. Journal of Nervous and
Del Re., A. C. (2010). RcmdrPlugin.MAc: Meta- Mental Disease, 195, 789–791.
∗
Analysis with Correlations (MAc) Rcmdr Plug-in. Feeley, W. M. (1993). Treatment components of
R package version 1.0.5. http://CRAN.Rproject. cognitive therapy for major depression: The good,
org/package=RcmdrPlugin.MAc the bad, and the inert. Unpublished doctoral
Del Re., A. C., & Hoyt, W. T. (2010). MAc: Meta- dissertation. University of Pennsylvania,
Analysis with Correlations. R package version Philadelphia.
1.0.5. http://CRAN.R-project.org/package=MAc Feeley, M., DeRubeis, R. J., & Gelfand, L. A.
∗
de Roten, Y., Fischer, M., Drapeau, M., Beretta, V., (1999). The temporal relation of adherence and
Kramer, U., Favre, N., et al. (2004). Is one assess- alliance to symptom change in cognitive therapy
ment enough? Patterns of helping alliance devel- for depression. Journal of Consulting and
opment and outcome. Clinical Psychology & Clinical Psychology, 67(4), 578–582.
∗
Psychotherapy, 11(5), 324–331. Ferleger, N. A. (1993). The relationship of depen-
∗
Deu, A., Heimbrock, F., & Röhrle B. (2009). Der dency and self-criticism to alliance, complementarity
Zusammenhang zwischen Merkmalen Sozialer and outcome in short-term dynamic psychotherapy.
Netzwerke und der therapeutischen Beziehung. Unpublished doctoral dissertation, Fordham
Eine kombinierte Prozess-Outcome-Studie [The University, Bronx, NY.
relation between social network attributes Fisher, R. A. (1924) On a distribution yielding
and the therapeutic relationship - A combined the error functions of several well known statis-
process-outcome study]. In Röhrle, B., & tics. Proceedings of the International Congress of
Laireiter, A. R. (Eds.). Soziale Unterstützung Mathematics, Toronto, 2, 805–813.
∗
und Psychotherapie (pp. 227–253). Tübingen, Florsheim, P., Shotorbani, S., Guest-Warnick, G.,
Germany: dgvt Verlag. Barratt, T., & Hwang, W. (2000). Role of the
∗
Dorsch, B., & Joraschky, P. (2002). Katamnestische working alliance in treatment of delinquent
Untersuchung an Patienten mit einer Angststörung boys in community-based programs. Journal of
nach stationärer Therapie [Follow-up study of Clinical Child Psychology, 29, 94–107.
∗
patients with anxiety disorders after inpatient Flückiger, C., Regli, D., & Grawe, K. (2005).
therapy]. Lage, Germany: Verlag Hans Jacobs. Allgemeine Psychotherapie im Gruppensetting:
∗
Dundon, W. D., Pettinati, H. M., Lynch, K. G., Das Zusammenspiel von gruppen- und inter-
Xie, H., Varillo, K. M., Makadon, C., et al. ventionsspezifischen Wirkfaktoren [General psy-
(2008). The therapeutic alliance in medical- chotherapy in the group setting: Interaction
based interventions impacts outcome in treating between group- and intervention-specific factors].
Gruppenpsychotherapie und Gruppendynamik, International. Section B: Physical Sciences and
41(3), 306–322. Engineering, 65(09), 4828.
∗ ∗
Forbes, D., Parslow, R., Creamer, M., Allen, N., Gallop, R., Kennedy, S. H., & Stern, S. D.
McHugh, T., & Hopwood, M. (2008). (1994). Therapeutic alliance on an inpatient
Mechanisms of anger and treatment outcome in unit for eating disorders. International Journal
combat veterans with posttraumatic stress disor- of Eating Disorders, 16, 405–410.
der. Journal of Traumatic Stress, 21(2), 142–149. Gaston, L., Goldfried, M. R., Greenberg, L. S.,
∗
Forman, N.W. (1990). The nature of trait empathy Horvath, A. O., Raue, P. J., & Watson, J.
in clients with chronic pain and their counselors (1995). The therapeutic alliance in psychody-
and its impact on the development of the working namic, cognitive-behavioral and experiential
alliance and outcome. Unpublished doctoral dis- therapies. Journal of Psychotherapy Integration,
sertation, Ohio State University, Columbus. 15, 1–26.
∗
Frank, A. F., & Gunderson, J. G. (1990). The role Gaston, L., & Marmar, C. (1994). The California
of the therapeutic alliance in the treatment of Psychotherapy Alliance Scales. In A. O. Horvath
schizophrenia. Archives of General Psychiatry, 47, & L. S. Greenberg (Eds.), The working alliance:
228–236. Theory, research and practice . New York: Wiley.
∗
Frank, J. D., & Frank, J. B. (1991). Persuasion and Gaston, L., Marmar, C. R., Thompson, L. W., &
healing: A comparative study of psychotherapy (3rd Gallager, D. (1991). Alliance prediction of out-
ed.). Baltimore: Johns Hopkins University Press. come beyond in-treatment symptomatic change as
∗
Freitas, T. T. (2001). The relationship between psychotherapy progresses. Psychotherapy Research,
neurocognitive impairment, working alliance, and 1, 104–112.
∗
length of stay in a therapeutic community. Gaston, L., Piper, W. E., Debbane, E. G.,
Unpublished doctoral dissertation, Norfolk State Bienvenu, J. P., & Garant, J. (1994). Alliance
University, Virginia. and technique for predicting outcome in
Freud, S. (1912/1958). The dynamics of trans- short- and long-term analytic psychotherapy.
ference [Zur Dynamik der Übertragung]. In Psychotherapy Research, 4, 121–135.
∗
J. Strachey (Ed. & Trans.), The standard edition of Gaston, L., Thompson, L., Gallager, D.,
the complete psychological works of Sigmund Freud Cournoyer, L. G., & Gagnon, R. (1998).
(Vol. 12, pp. 99–108). London: Hogarth Press. Alliance, technique and their interactions in
Freud, S. (1913). On the beginning of treatment: predicting outcome of behavioral, cognitive and
Further recommendations on the technique of brief dynamic therapy. Psychotherapy Research, 8,
psychoanalysis [Zur Einleitung der Behandlung– 190–209.
∗
Weitere Ratschläge zur Technik der Psycho- Geider, F. J. (1997). Die Messung therapeutischer
analyse]. In J. Strachey (Ed.), Standard edition of Allianz: Dimensionierung und Konstruktva-
the complete psychological works of Sigmund Freud lidierung einer deutschen Version der California
(Vol. 12, pp. 122–144). London: Hogarth. Psychotherapeutic Alliance Scales (CALPAS) [The
Freud, S. (1913/1940). The technique of psycho- measurement of therapeutic alliance: Dimension-
analysis. In J. Strachey (Ed.), Standard edition of ality and construct validation of a German ver-
the complete psychological works of Sigmund Freud sion of the California Psychotherapeutic Alliance
(Vol. 23, pp. 172–182). London: Hogarth. Scales (CALPAS)]. Unpublished doctoral disser-
∗
Fries, A., Pfammatter, M., Andres, K., & tation, University of Ruprecht-Karls-University,
Brenner, H. D. (2003). Wirksamkeit und Heidelberg, Germany.
∗
Prozessmerkmale einer psychoedukativen und Geiser, F., Bassler, M., Bents, H., Carls, W.,
bewältigungsorientierten Gruppentherapie für Joraschky, P., Michelitsch, B., et al. (2002).
schizophren und schizoaffektiv Erkrankte Bewertung des Therapieerfolgs durch Patienten mit
[Outcome and process characteristics of a psy- Angststörungen nach stationärer Psychotherapie
choeducational and coping-oriented group [Patient’s subjective assessment of therapy out-
therapy for patients with schizophrenia and come after inpatient psychotherapy for anxiety
schizoaffective disorder]. Verhaltenstherapie, disorders.] Der Nervenarzt, 73(1), 59–64.
13(4), 237–243. Gelso, C. J., & Carter, J. A. (1994). Components
∗
Gaiton, L. R. (2004). Investigation of therapeutic of the psychotherapy relationship: Their interac-
alliance in a treatment study with substance- tion and unfolding during treatment. Journal of
abusing women with Ptsd. Dissertation Abstracts Counseling Psychology, 41, 296–306.
∗ in short-term inpatient psychiatric care.
Gerstley, L., McLellan, A.T., Alterman, A.I.,
Woody, G.E., Luborsky, L., & Prout, M. (1989). Scandinavian Journal of Social Medicine, 20,
Ability to form an alliance with the therapist: A 45–50.
∗
possible marker of prognosis for patients with Hardy, G. E., Cahill, J., Shapiro, D. A., Barkham,
antisocial personality disorder. American Journal M., Rees, A., & Macaskill, N. (2001). Client
of Psychiatry, 146, 508–512. interpersonal and cognitive styles as predictors of
∗ response to time-limited cognitive therapy for
Godfrey, E., Chalder, T., Ridsdale, L., Seed, P.,
& Ogden, J. (2007). Investigating the ‘active depression. Journal of Consulting and Clinical
ingredients’ of cognitive behaviour therapy and Psychology, 69(5), 841–845.
∗
counselling for patients with chronic fatigue in Hartley, D. E., & Strupp, H. H. (1983). The
primary care: Developing a new process mea- therapeutic alliance: Its relationship to out-
sure to assess treatment fidelity and predict come in brief psychotherapy. In J. Masling
outcome. British Journal of Clinical Psychology, (Ed.), Empirical studies in analytic theories. (Vol. 1,
46(3), 253–272. pp. 1–37). Hillside, NJ: Erlbaum.
∗
Goldfried, M. R. (1980). Toward the delinea- Hartmann, H. A. (2001). Clients’ contribution
tion of therapeutic change principles. American to the alliance and the role it plays in outcome.
Psychologist, 35, 991–999. Unpublished doctoral dissertation, Adelphi
∗ University. New York, NY.
Gomes-Schwartz, B. (1978). Effective ingredients
∗
in psychotherapy: Prediction of outcome from Hatcher, R. L., & Barends, A. W. (1996). Patient’s
process variables. Journal of Consulting and view of the alliance in psychotherapy: Exploratory
Clinical Psychology, 46, 1023–1035. factor analysis of three alliance measures. Journal
∗
Greenberg, L. S., & Malcolm, W. (2002). Resolving of Clinical and Consulting Psychology, 64,
unfinished business: Relating process to outcome. 1326–1336.
Journal of Consulting and Clinical Psychology, Hatcher, R. L., & Barends, A. W. (2006). How a
70(2), 406–416. return to theory could help alliance research.
∗ Psychotherapy: Theory, Research, Practice, Training,
Greenberg, L. S., & Webster, M. C. (1982).
Resolving decisional conflict by Gestalt two- 43(3), 292–299.
chair dialogue: Relating process to outcome. Hatcher, R. L., Barends, A., Hansell, J., &
Journal of Counseling Psychology, 29, 468–477. Gutfreund, M. J. (1995). Patient’s and therapist’s
Greenson, R. R. (1965). The working alliance shared and unique views of the therapeutic
and the transference neuroses. Psychoanalysis alliance: An investigation using confirmatory
Quarterly, 34, 155–181. factor analysis in a nested design. Psychoanalysis
Greenson, R. R. (1967). Technique and practice of Quarterly, 63, 636–643.
∗
psychoanalysis. New York: International University Hawley, L. L., & Moon-Ho Ringo, H. (2006). The
Press. relationship of perfectionism, depression, and
∗ therapeutic alliance during treatment for
Grob, M. C., & Eisen, S. V. (1989). Most likely to
succeed: Correlates of good versus poor hospital depression: Latent difference score analysis.
outcome in young adult inpatients. Psychiatric Journal of Consulting & Clinical Psychology,
Hospital, 20, 23–30. 74(5), 930–942.
∗ ∗
Gunderson, J. G., Najavits, L. M., Leonhard, C., Hayes, S. A., Hope, D. A., VanDyke, M. M., &
Sullivan, C. N., & Sabo, A. N. (1997). Ontogeny Heimberg, R. G. (2007). Working alliance for cli-
of the therapeutic alliance in borderline patients. ents with social anxiety disorder: Relationship with
Psychotherapy Research, 7, 301–309. session helpfulness and within-session habituation.
∗ Cognitive Behaviour Therapy, 36(1), 34–42.
Gunther, G. J. (1992). Therapeutic alliance, patient
∗
object relations and outcome in psychotherapy. Hays, V.L. (1994). The effects of the therapeutic
Unpublished doctoral dissertation, Michigan alliance and social support on therapy outcome and
State University, East Lansing. mental health of women. Unpublished doctoral
∗ dissertation, University of Wisconsin, Madison.
Gutfreund, M. J. (1992). Therapist interventions:
Their relation to therapeutic alliance and outcome in Henry, W. P., & Strupp, H. H. (1994). The
dynamic psychotherapy. Unpublished doctoral dis- therapeutic alliance as interpersonal process.
sertation, Michigan State University, East Lansing. In A. O. Horvath & L. S. Greenberg (Eds.), The
∗ working alliance: Theory, research and practice.
Hansson, L., & Berglund (1992). Stability of ther-
apeutic alliance and its relationship to outcome New York: Wiley.
∗ Horvath, A. O., & Luborsky, L. (1993). The role
Hervé, M.-J., White-Koning, M., Paradis-
Guennou, M., Guiraud, A., Picot, M.-C., of the therapeutic alliance in psychotherapy.
Grandjean, H., et al. (2008). Adaptation d’une Journal of Consulting and Clinical Psychology, 61,
echelle d’alliance therapeutique au contexte des 561–573.
consultations mere-nourrisson. Etude prelimi- Horvath, A. O., & Marx, R. W. (1991). The devel-
naire [Adaptation of a therapeutic alliance scale opment and decay of the working alliance during
in mother-infant consultations. A preliminary time-limited counselling. Canadian Journal of
study]. Devenir, 20(1), 65–85. Counselling, 24, 240–259.
∗ Horvath, A. O., & Symonds, B. D. (1991).
Hilliard, R. B., Henry, W. P., & Strupp, H. H.
(2000). An interpersonal model of psychother- Relation between working alliance and out-
apy: Linking patient and therapist developmental come in psychotherapy: A meta-analysis. Journal
history, therapeutic process, and types of of Counseling Psychology, 38, 139–149.
∗
outcome. Journal of Consulting and Clinical Howard, I. (2003). Interpersonal factors in relation-
Psychology, 68(1), 125–133. ship to therapy outcome and therapeutic alli-
∗ ance in a cohort of multiple sclerosis patients.
Hopkins, M., & Ramsundar, N. (2006). Which fac-
tors predict case management services and how do Unpublished doctoral dissertation, Alliant
these services relate to client outcomes? Psychiatric International University, San Francisco.
∗
Rehabilitation Journal, 29(3), 219–222. Howard, I., Turner, R., Olkin, R., & Mohr, D. C.
∗ (2006). Therapeutic alliance mediates the rela-
Hopkins, W. E. (1988). The effects of conceptual
level matching on the working alliance and out- tionship between interpersonal problems and
come in time-limited counseling. Unpublished depression outcome in a cohort of multiple scle-
doctoral dissertation, University of Maryland, rosis patients. Journal of Clinical Psychology,
College Park. 62(9), 1197–1204.
∗ ∗
Horowitz, M. J., Marmar, C., Weiss, D. S., Huber, D., Brandl, T., Henrich, G., & von Rad, M.
DeWitt, K. N., & Rosenbaum, R. (1984). Brief (2003). Interaktive Beziehungsmuster und ihre
psychotherapy of bereavement reactions: The Bedeutung für psychotherapeutische Verän-
relationship of process to outcome. Archives of derungen [Satisfaction and compliance after a
General Psychiatry, 41, 438–448. psychosomatic-psychotherapeutic consultation
∗ session]. Zeitschrift für Klinische Psychologie und
Horvath, A. O. (1981). An exploratory study of the
working alliance: Its measurement and relationship Psychotherapie, 32(2), 85–93.
to outcome. Unpublished doctoral dissertation, Hunter, J. E., & Smith, F. L. (1990). Methods of
Vancouver, Canada. Meta-analysis: Correcting error and bias in research
Horvath, A. O. (2009). Conceptual and methodolog- findings. London: Sage.
ical challenges in alliance research: Is it time for a Hunter, J. E., & Schmidt, F. L. (2004). Methods of
change. Paper presented at the European Society meta-analysis (2nd edition). Thousand Oaks,
for Psychotherapy Research Conference. Bolzano, CA: Sage.
∗
Italy. Ilgen, M. A., McKellar, J., Moos, R., & Finney, J. W.
Horvath, A. O., & Bedi, R. P. (2002). The alliance. (2006a). Therapeutic alliance and the relationship
In J. C. Norcross (Ed.), Psychotherapy relation- between motivation and treatment outcomes in
ships that work: Therapist contributions responsive- patients with alcohol use disorder. Journal of
ness to patients (pp. 37–70). New York: Oxford Substance Abuse Treatment, 31(2), 157–162.
∗
University Press. Ilgen, M., Tiet, Q., Finney, J., & Moos, R. H.
Horvath, A. O., Gaston, L., & Luborsky, L. (2006b). Self-efficacy, therapeutic alliance, and
(1993). The role of alliance in psychotherapy. alcohol-use disorder treatment outcomes. Journal
In N. E. Miller, L. Luborsky, J. P. Barber, & of Studies on Alcohol, 67(3), 465.
∗
J. P. Docherty (Eds.), Psychodynamic treatment Irelan, C. S. (2004). The therapeutic alliance, resolu-
research: A handbook for clinical practice. tion of alliance ruptures and termination outcome.
(pp. 247–274). New York: Basic Books. Central Michigan University, Michigan.
∗
Horvath, A. O., & Greenberg, L. S. (1986). Jacob, K. L. (2003). Toward understanding change
Development of the working alliance Inventory. in psychotherapy: A closer look at the relation-
In L. S. Greenberg & W. M. Pinsof (Eds.), The ship between client-therapist match and treat-
psychotherapeutic process: A research handbook ment outcome. Unpublished doctoral dissertation,
(pp. 529–556). New York: Guilford. Clark University, Worchester, MA.
∗ outpatient therapy. Psychotherapy: Theory,
Janecke, N. (2003). Bedeutung der Motivation und
therapeutischen Beziehung in der Psychotherapie Research, Practice, Training, 46(2), 193–202.
∗
[The importance of motivation and therapeutic Kivilighan, D. M., & Shaugnessy, P. (1995).
alliance in psychotherapy]. Unpublished doc- Analysis of the development of the working alli-
toral dissertation, University of Innsbruck, ance using hierarchical linear modeling. Journal
Innsbruck, Austria. of Counseling Psychology, 42, 338–349.
∗ ∗
Johansson, H. K., & Eklund, M. (2006). Helping Kivlighan, D. M, Jr., & Shaughnessy, P. (2000).
alliance and early dropout from psychiatric out- Patterns of working alliance development:
patient care: The influence of patient factors. A typology of client’s working alliance ratings.
Social Psychiatry and Psychiatric Epidemiology, Journal of Counseling Psychology, 47(3), 362–371.
∗
41(2), 140–147. Klee, M. R., Abeles, N., & Mullere, R. T. (1990).
∗ Therapeutic alliance: Early indicators, course,
Joyce, A. S., & Piper, W. E. (1998). Expectancy,
the therapeutic alliance, and treatment out- and outcome. Psychotherapy, 27, 166–174.
∗
come in short-term individual psychotherapy. Klein, D. N., Schwartz, J. E., Santiago, N. J.,
Journal of Psychotherapy Practice and Research, 7, Vivian, D., Vocisano, C., Castonguay, L. G.,
236–248. et al. (2003). Therapeutic alliance in depres-
∗ sion treatment: Controlling for prior change and
Joyce, A. S., Ogrodniczuk, J. S., Piper, W. E., &
McCallum, M. (2003). The alliance as mediator patient characteristics. Journal of Consulting and
of expectancy effects in short-term individual Clinical Psychology, 71(6), 997–1006.
∗
therapy. Journal of Consulting and Clinical Knaevelsrud, C., & Maercker, A. (2007). Internet-
Psychology, 71(4), 672–679. based treatment for PTSD reduces distress and
∗ facilitates the development of a strong thera-
Jumes, M. T. (1995). The developmental course of
the inpatient working alliance in state psychiatric peutic alliance: A randomized controlled clinical
hospital clients. Unpublished doctoral disserta- trial. BMC Psychiatry, 7, 13–10.
∗
tion, University of Texas Southwestern Medical Kokotovic, A. M., & Tracey, T. J. (1990). Working
Center, Dallas. alliance in the early phase of counseling. Journal of
∗ Counseling Psychology, 37, 16–21.
Kabuth, B., De Tychey, C., & Vidailhet, C. (2005).
∗
Alliance therapeutique avec les mères et evolu- Kolden, G. G. (1996). Change in early sessions
tion clinique des enfants d’un hôpital de of dynamic therapy: Universal processes and
jour [Mothers’ therapeutic alliance and children the generic model of psychotherapy. Journal
outcome in a day hospital]. Annales Médico of Clinical and Consulting Psychology, 64,
Psychologiques, 163(6), 486–492. 489–496.
∗ ∗
Karver, M., Shirk, S., Handelsman, J. B., Fields, S., Konzag, T., Bademer-Greulich, U., Bahreke, U.,
Crisp, H., Gudmundsen, G., et al. (2008). & Fikentscher, E. (2004). Therapeutische
Relationship processes in youth psychotherapy: Beziehung und Therapieerfolg bei der station-
Measuring alliance, alliance-building behaviors, ären Psychotherapie von Persönlichkeitsstörungen
and client involvement. Journal of Emotional & [Therapeutic relationship and outcome in psy-
Behavioral Disorders, 16(1), 15–28. chotherapy for personality disorders]. Zeitschrift
∗ für Psychosomatische Medizin und Psychotherapie,
Katz, J. (1999). Self-handicapping, the working alli-
ance, interpersonal tendencies and the prediction 50(4), 394–405.
∗
of drop-out. Unpublished doctoral dissertation, Kramer, U., de Roten, Y., Beretta, V., Michel, L.,
New School for Social Research, New York. & Despland, J.-N. (2009). Alliance patterns
∗ over the course of short-term dynamic psycho-
Kech, S. (2008). Einflussfaktoren auf den
Behandlungserfolg der Interpersonellen Psychother- therapy: The shape of productive relationships.
apie bei stationären Depressionspatienten: Analyse Psychotherapy Research, 19(6), 699–706.
∗
der Wirkmechanismen [Determinats of success Kramer, U., de Roten, Y., Beretta, V., Michel, L.,
of interpersonal psychotherapy with depressive & Despland, J.-N. (2008). Patient’s and
inpatients: Analysis of effective factors.] Published therapist’s views of early alliance building in
doctoral dissertation, University of Freiburg, dynamic psychotherapy: Patterns and relation
Freiburg, Germany. http://www.freidok.uni- to outcome. Journal of Counseling Psychology,
freiburg.de/volltexte/5318/ 55(1), 89–95.
∗ Kramer, U., de Roten, Y., Beretta, V., Michel, L., &
Kelly, A. E., & Yuan, K.-H. (2009). Clients’
secret keeping and the working alliance in adult Despland, J.-N. (2009). Alliance patterns over
the course of short-term dynamic psychother- alliance methods for predicting outcomes of psy-
apy: The shape of productive relationships. chotherapy: A counting signs vs. a global rating
Psychotherapy Research, 19(6), 669–706. method. Journal of Nervous and Mental Disease
∗
Krupnick, J. L., Elkin, I., Collins, J., Simmens, S., 171, 480–491.
∗
Sotsky, S. M., Pilkonis, A., & Watkings, J. T. Luborsky, L., McLellan, A. T., Woody, G. E.,
(1994). Therapeutic alliance and clinical out- O’Brien, C. P., & Auerbach, A. (1985). Therapist
come in the NIMH Treatment of Depression success and its determinants. Archives of General
Collaborattive Research Program: Preliminary Psychiatry, 42, 602–611.
findings. Psychotherapy, 31, 28–35. Luborsky, L., Singer, B., & Luborsky, L. (1975).
∗
Krupnick, J. L., Sotsky, S. M., Simmens, A., Comparative studies of psychotherapies; “Is
Moyer, J., Elkin, I., Watkins, J., et al. (1996). it true that everybody has won and all must
The role of the alliance in psychotherapy and have prizes”? Archives of General Psychiatry, 32,
pharmacotherapy outcome: Findings in the 995–1008.
∗
National Institute of Mental Health treatment Mallinckrodt, B. (1993). Session impact, work-
of depression collaborative research program. ing alliance, and treatment outcome in brief
Journal of Clinical and Consulting Psychology, 64, counseling. Journal of Counseling Psychology, 40,
532–539. 25–32.
∗ ∗
Kukla, M., & Bond, G. R. (2009). The working Mallinckrodt, B. (1996). Change in working
alliance and employment outcomes for people alliance, social support and psychological symp-
with severe mental illness enrolled in voca- toms in brief therapy. Journal of Counseling
tional programs. Rehabilitation Psychology, 54(2), Psychology, 43, 448–455.
∗
157–163. Marmar, C. R., Gaston, L., Gallager, D., &
∗
Lansford, E. (1986). Weakenings and repairs of the Thompson, L. W. (1989b). Therapeutic alli-
working alliance in short-term psychotherapy. ance and outcome in behavioral, cognitive, and
Professional Psychology: Research and Practice, 17, brief dynamic psychotherapy in late-life depres-
364–366. sion. Journal of Nervous and Mental Disease 177,
∗
Lieberman, P. B., von Rehn, S., Dickie, E., & 464–472.
∗
Elliott, B. (1992). Therapeutic effects of brief Marmar, C., Weiss, D. S., & Gaston, L. (1989a).
hospitalization: The role of a therapeutic alli- Toward the validation of the California
ance. Journal of Psychotherapy Practice & Research, Therapeutic Alliance Rating System. Psychological
1(1), 56–63. Assessment: A Journal of Consulting and Clinical
∗
Liebler, A., Biella, M., Schulz, W., & Hartmann, U. Psychology, 1, 46–52.
∗
(2004). Differenzielle Therapieverläufe und Marmarosh, C. L., Gelso, C. J., Markin, R. D.,
Behandlungserfolg in der stationären Grup- Majors, R., Mallery, C., & Choi, J. (2009). The
penpsychotherapie [Differential courses of treat- real relationship in psychotherapy: Relationships
ment and their relation to outcome in inpatient to adult attachments, working alliance, transfer-
group psychotherapy]. Gruppenpsychotherapie und ence, and therapy outcome. Journal of Counseling
Gruppendynamik. 40(1), 1–21. Psychology, 56(3), 337–350.
∗ ∗
Loncek, B., Banks, S., Way, B., & Bonaparte, E. Marshel, R. L. (1986). An exploration of the
(2002). An empirical model of therapeutic pro- relationship between the therapeutic alliance
cess for psychiatric emergency room clients with and premature termination. Unpublished doc-
dual disorders. Social Work Research, 26(3), 132. toral dissertation, Northwestern University,
Luborsky, L. (1976). Helping alliances in psycho- Evanston, IL.
therapy. In J. L. Cleghhorn (Ed.), Successful psy- Martin, D. J., Garske, J. P., & Davis, K. M. (2000).
chotherapy (pp. 92–116). New York: Brunner/ Relation of the therapeutic alliance with out-
Mazel. come and other variables: A meta analytic review.
Luborsky, L. (1994). Therapeutic alliance measures Journal of Clinical and Consulting Psychology, 68,
as predictors of future benefits of psychotherapy. 438–450.
∗
In A. O. Horvath & L. S. Greenberg (Eds.), The Marziali, E. (1984). Three viewpoints on the
working alliance: Theory, research and practice. therapeutic alliance scales: Similarities, differ-
New York: Wiley. ences and associations with psychotherapy out-
∗
Luborsky, L., Crits-Cristoph, P., Alexander, L., come. Journal of Nervous and Mental Disease
Margolis, M., & Cohen, M. (1983). Two helping 172, 417–423.
∗ ∗
Marziali, E., Marmar, C., & Krupnick, J. (1981). Moleiro, C. M. D. M. (2003). Change in therapy:
Therapeutic alliance scales: Development and A model of the effects of systematic treatment
relationship to psychotherapy outcome. American selection variables and quality of alliance on indi-
Journal of Psychiatry, 138, 361–364. vidual growth. Unpublished doctoral disserta-
∗
Marziali, E., Munroe-Blum, H., & McClery, L. tion, University of California, Santa Barbara.
∗
(1999). The effects of the therapeutic alliance Morgan, R., Luborsky, L., Crits-Cristoph, P.,
on the outcomes of individual and group psy- Curtis, H., & Solomon, J. (1982). Predicting
chotherapy with borderline personality disorder. the outcomes of psychotherapy by the Penn
Psychotherapy Research, 9, 452–467. Helping Alliance Rating Method. Archives of
∗
McLeod, B. D., & Weisz, J. R. (2005). The Therapy General Psychiatry, 39, 397–402.
∗
Process Observational Coding System-Alliance Moseley, D. (1983). The therapeutic relationship and
Scale: Measure characteristics and prediction its association with outcome. Unpublished thesis,
of outcome in usual clinical practice. Journal University of British Columbia, Vancouver,
of Consulting and Clinical Psychology, 73(2), Canada.
∗
323–333. Multon, K. D., Heppner, M. J., Gsybers, N. C.,
∗
McNeil, D. C., (2006). Patient, therapist, and Zook, C., & Ellis-Kalton, C. A. (2001).
observer perspectives on cohesion and alliance Client psychological distress: An important
and their relationship to outcome in psychody- factor in career counseling. The Career
namic group psychotherapy for persons experienc- Development Quarterly, 49, 324–335.
∗
ing complicated grief. Unpublished doctoral Muran, J. C., Gorman, B. S., Safran, J. D.,
dissertation, University of Alberta, Edmonton, Twining, L, Samstag, I. W., & Winston, A.
Canada. (1995). Linking in-session change to overall
∗
Meier, P. S., & Donmall, M. C. (2006). Differences outcome in short term cognitive therapy.
in client and therapist views of the working Journal of Consulting and Clinical Psychology,
alliance in drug treatment. Journal of Substance 63, 651–657.
∗
Use, 11(1), 73–80. Muran, J. C., Safran, J. D., Gorman, B. S.,
∗
Meier, P. S., Donmall, M. C., McElduff, P., Samstag, L. W., Eubanks-Carter, C., &
Barrowclough, C., & Heller, R. F. (2006). The Winston, A. (2009). The relationship of early
role of the early therapeutic alliance in predict- alliance ruptures and their resolution to pro-
ing drug treatment dropout. Drug & Alcohol cess and outcome in three time-limited psycho-
Dependence, 83(1), 57–64. therapies for personality disorders. Psychotherapy:
∗
Meyer, B., Pilkonis, P. A., Krupnick, J. L., Theory, Research, Practice, Training, 46(2),
Egan, M. K., Simmens, S. J., & Sotsky, S. M. 233–248.
(2002). Treatment expectancies, patient alli- National Research Council (1992). Combining
ance and outcome: Further analyses from the information: Statistical issues and opportunities
National Institute of Mental Health Treatment for research. Washington DC: National Academy
of Depression Collaborative Research Program. Press.
∗
Journal of Consulting and Clinical Psychology, Ogrodniczuk, J. S., Piper, W. E., Joyce, A. S., &
70(4), 1051–1055. McCallum, M. (2000). Different perspectives of
Miller, N. E., Luborsky, L., Barber, J. P., & the therapeutic alliance and therapist technique in
Docherty, J. P. (1993). Psychodynamic treat- 2 forms of dynamically oriented psychotherapy.
ment research: A handbook for clinical practice. Canadian Journal of Psychiatry, 45, 452–458.
∗
New York, NY US: Basic Books. O’Malley, S. S., Suh, C. S., & Strupp, H. H.
∗
Missirlian, T. M., Toukmanian, S. G., Warwar, S. H., (1983). The Vanderbilt Psychotherapy Process
& Greenberg, L. S. (2005). Emotional arousal, Scale: A report on the scale development and a
client perceptual processing, and the working process-outcome study. Journal of Clinical and
alliance in experiential psychotherapy for depres- Consulting Psychology, 51, 581–586.
∗
sion. Journal of Consulting and Clinical Psychology, Pantalon, M. V., Chawarski, M. C., Falcioni, J.,
73(5), 861–871. Pakes, J., & Schottenfeld, R. S. (2004). Linking
∗
Mohl, P. C., Martinez, D., Ticknor, C., Huang, M., process and outcome in the community rein-
& Cordell, L. (1991). Early dropouts from forcement approach for treating cocaine depen-
psychotherapy. Journal of Nervous and Mental dence: A preliminary report. American Journal of
Disease, 179(8), 478–481. Drug & Alcohol Abuse, 30(2), 353–367.
∗ ∗
Pavio, S. C., & Bahr, L. B. (1998). Interpersonal Ramnerö, J., & Öst, L.-G. (2007). Therapists’ and
problems, working alliance, and outcome in clients’ perception of each other and working
short-term experiential therapy. Psychotherapy alliance in the behavioral treatment of panic dis-
Research, 8, 392–406. order and agoraphobia. Psychotherapy Research,
∗
Piper, W. E., Azim, H. F. A., Joyce, A. S., 17(3), 328–337.
∗
MacCallum, M., Nixon, G. W. H., & Segal, P. S. Reiner, P. A. (1987). The development of the thera-
(1991). Quality of object relations vs. interper- peutic alliance. Unpublished doctoral disserta-
sonal functioning as predictor of therapeutic alli- tion, University of North Carolina, Chapel Hill.
∗
ance and psychotherapy outcome. Journal of Reis, S., & Grenyer, B. F. S. (2004). Fearful attach-
Nervous and Mental Disease 179, 432–438. ment, working alliance and treatment response
∗
Piper, W. E., Boroto, D. R., Joyce, A. S., for individuals with major depression. Clinical
McCallum, M., & Azim, H. F. A. (1995). Pattern Psychology & Psychotherapy, 11(6), 414–424.
∗
of alliance and outcome in short-term individual Riley, J. D. (1992). The impact of pre-therapy and
psychotherapy. Psychotherapy, 32, 639–647. extra-therapy role preparation on the development
∗
Piper, W. E., Ogrodniczuk, J. S., & Joyce, A. S. of therapeutic alliance and outcome. Unpublished
(2004). Quality of object relations as a modera- doctoral dissertation, Pacific Graduate School of
tor of the relationship between pattern of Psychology, Palo Alto, CA.
alliance and outcome in short-term individual Rogers, C. R., Gendlin, G. T., Kiesler, D. V., &
psychotherapy. Journal of Personality Assessment, Truax, L. B. (1967). The therapeutic relationship
83(3), 345–356. and its impact: A study of psychotherapy with schizo-
∗
Pos, A. E., & B. (2007). Experiential treatment phrenics. Madison: University of Wisconsin Press.
for depression: A test of the experiential theory of Rogers, C. R., & Wood, J. K. (1974). Client-
change, differential effectiveness, and predictors of centered theory: Carl R. Rogers. . In A. Burton
maintenance of gains. Unpublished dissertation, (Ed.), Operational theories of personality
York University. Toronto, Canada. (pp. 211–258.). New York: Bruner/Mazel.
∗ ∗
Pos, A. E., Greenberg, L. S., & Warwar, S. H. Rogers, N., Lubman, D. I., & Allen, N. B. (2008).
(2009). Testing a model of change in the Therapeutic alliance and change in psychiatric
experiential treatment of depression. Journal symptoms in adolescents and young adults
of Consulting and Clinical Psychology, 77(6), receiving drug treatment. Journal of Substance
1055–1066. Use, 13(5), 325–339.
∗
Priebe, S., & Gruyters, T. (1993). The role of Rosenthal, R. (1979). The file drawer problem
helping alliance in psychiatric community care: and tolerance for null results. Psychological
A prospective study. Journal of Nervous and Bulletin, 86, 638–41.
Mental Disease, 181, 552–557. Rosenzweig, S. (1936). Some implicit common
∗
Prigatano, G. P., Klonoff, P. S., O’Brien, K. P., factors in diverse methods of psychotherapy.
Altman, I. M., Amin, K., Chiapello, D., American Journal of Orthopsychiatry, 6, 412–415.
∗
Shepherd, J., Cunningham, M., & Mora, M. Rounsaville, B. J., Chevron, E. S., Prusoff, B. A.,
(1994). Productivity after neuropsychologically Elkin, I., Imber, S., Sotsky, S., (1987). The
oriented milieu rehabilitation. Journal of Head relation between specific and general dimensions
Trauma Rehabilitation, 9, 91–102. of the psychotherapy process in interpersonal
∗
Pugh, M. A. (1991). The working alliance, prether- psychotherapy of depression. Journal of Clinical
apy interpersonal relationships and therapeutic out- and Consulting Psychology, 55, 379–384.
∗
come. Unpublished doctoral dissertation, Virginia Rudolf, G., & Manz, R. (1993). Zur prognos-
Commonwealth University, Richmond. tischen Bedeutung der therapeutischen Arbeits-
∗
Pyne, S. C. (1991). Therapist’s use of countertransfer- beziehung aus der Perspektive von Patienten
ence in short-term psychotherapy: A focus on rescu- und Therapeuten [The prognostic relevance of
ing and counterhostility. Unpublished doctoral a therapeutic working alliance as seen by patients
dissertation, University of Colorado, Boulder. and therapists]. Psychotherapie, Psychosomatik,
R Development Core Team (2009). R: A language Medizinische Psychologie. 43(6), 193–199.
and environment for statistical computing. Safran, J. D. (1993). The therapeutic alliance
R Foundation for Statistical Computing, rupture as a transtheoretical phenomenon:
Vienna, Austria. ISBN 3-900051-07-0, http:// Definitional and conceptual issues. Journal of
www.R-project.org. Psychotherapy Integration, 3(1), 33–49.
Safran, J. D., Crocker, P., McMain, S., & Murray, P. Langzeitstudie zur Psychotherapieforschung
(1990). The therapeutic alliance rapture as a [Effectiveness of individual psychological psy-
therapy event for emperical investigation. Psycho- choanalytic psychotherapy - A longitudinal study
therapy: Theory, Research, Practice and Training, on psychotherapy research]. Zeitschrift für indi-
27, 154–165. vidualpsychologie, 30(1), 51–77.
Safran, J. D., & Muran, C. (2000). Negotiating the Schönberger, M., Humle, F., & Teasdale, T. W.
therapeutic alliance: A relational treatment guide. (2006a). Subjective outcome of brain injury
New York: Guilford. rehabilitation in relation to the therapeutic work-
∗
Safran, J. D., & Wallner, L. K. (1991). The relative ing alliance, client compliance and awareness.
predictive validity of two therapeutic alliance Brain Injury, 20(12), 1271–1282.
∗
measures in cognitive therapy. Psychological Schönberger, M., Humle, F., & Teasdale, T. W.
Assessment, 3, 188–195. (2006b). The development of the therapeutic
Saketopoulou, A. (1999). The psychotherapeu- working alliance, patients’ awareness and their
tic alliance in psychodynamic psychotherapy: compliance during the process of brain injury
Theoretical conceptualizations and research rehabilitation. Brain Injury, 20(4), 445–454.
∗
findings. Psychotherapy, 36, 329–342. Schönberger, M., Humle, F., & Teasdale, T. W.
∗
Sammet, V. I., Staats, H., & Schauenburg, H. (2007). The relationship between clients’ cogni-
(2004). Beziehungserleben und Therapieergebnis tive functioning and the therapeutic working
in stationärer Psychotherapie [Relationship expe- alliance in post-acute brain injury rehabilitation.
riences and therapeutic outcome in inpatient Brain Injury, 21(8), 825–836.
∗
psychotherapy]. Zeitschrift für Psychosomatische Schönberger, M., Humle, F., Zeeman, P., &
Medizin und Psychotherapie, 50(4), 376–393. Teasdale, T. (2006c). Working alliance and
∗
Samstag, L. W., Muran, J. C., Wachtel, P. L., patient compliance in brain injury rehabilitation
Slade, A., Safran, J. D., & Winston, A. (2008). and their relation to psychosocial outcome.
Evaluating negative process: A comparison of Neuropsychological Rehabilitation, 16(3), 298–314.
∗
working alliance, interpersonal behavior, and Sexton, H. (1996). Process, life events, and symp-
narrative coherency among three psychother- tomatic change in brief eclectic psychotherapy.
apy outcome conditions. American Journal of Journal of Consulting and Clinical Psychology, 64,
Psychotherapy, 62(2), 165–194. 1358–1365.
∗ ∗
Santiago, N. J., Klein, D. N., Vivian, D., Sherer, M., Evans, C. C., Leverenz, J., Stouter, J.,
Arnow, B. A., Blalock, J. A., Kocis, J. H., et al. Irby, J. W., Jae Eun, L., et al. (2007). Therapeutic
(2005). The therapeutic alliance and CBASP- alliance in post-acute brain injury rehabilitation:
specific skill acquisition in the treatment of Predictors of strength of alliance and impact
chronic depression. Cognitive Therapy and of alliance on outcome. Brain Injury, 21(7),
Research, 29, 803–817. 663–672.
∗ ∗
Saunders, S. M. (2000). Examining the relationship Shirk, S. R., Gudmundsen, G., Kaplinski, H. C.,
between the therapeutic bond and the phases & McMakin, D. L. (2008). Alliance and out-
of treatment outcome. Psychotherapy: Theory, come in cognitive-behavioral therapy for adoles-
Research, Practice, Training, 37(3), 206–218. cent depression. Journal of Clinical Child &
∗
Saunders, S. M., Howard, K. I., & Orlinsky, D. E. Adolescent Psychology, 37(3), 631–639.
(1989). The therapeutic bond scales: Psychometric Smith, M. L., & Glass, G. V. (1977). Meta-analysis
characteristics and relationship to treatment effec- of psychotherapy outcome studies. American
tiveness. Psychological Assessment, 1, 323–330. Psychologist, 32, 752–760.
∗ ∗
Schauenburg, H., Dinger, U., & Starck, M. (2005). Solomon, P., Draine, J., & Delaney, M. A. (1995).
Zur Bedeutung der Einzeltherapeuten für das The working alliance and consumer case man-
Therapieergebnis in der stationären Psycho- agement. Journal of Mental Health Adminstration,
therapie - eine Pilotstudie [The significance of 22, 126–134.
∗
individual psychotherapists for psychothera- Sonnenberg, R. T. (1996). An examination of
peutic outcomes in inpatient psychotherapy. the outcome of a brief psychiatric hospitalization
A pilot study]. Psychotherapie, Psychosomatik, with particular reference to the therapeutic alliance
Medizinische Psychologie, 55(7), 339–346. and select patient characteristics, Unpublished
∗
Schleussner, D. (2005). Wirksamkeit individualpsy- doctoral dissertation, University of Nebraska,
chologisch-psychoanalytischer Psychotherapie - Eine Lincoln.
∗ ∗
Spinhoven, P., Giesen-Bloo, J., van Dyck, R., Tichenor, V. (1989). Working alliance: A measure
Kooiman, K., & Arntz, A. (2007). The therapeu- comparison. Unpublished doctoral dissertation,
tic alliance in schema-focused therapy and University of Maryland, Chapel Hill, MD.
transference- focused psychotherapy for border- Tichenor, V., & Hill, C. E. (1989). A comparison of
line personality disorder. Journal of Consulting & six measures of working alliance. Psychotherapy:
Clinical Psychology, 75(1), 104–115. Theory, Research, and Practice, 26, 195–199.
∗
Sterba, R. F. (1934). The fate of the ego in analytic Trepka, C., Rees, A., Shapiro, D. A., Hardy, G. E.,
therapy. International Journal of Psychoanalysis, & Barkham, M. (2004). Therapist competence
115, 117–126. and outcome of cognitive therapy for depression.
∗
Stevens, C. L., Muran, J. C., Safran, J. D., Cognitive Therapy & Research, 28(2), 143–157.
∗
Gorman, B. S., & Winston, A. (2007). Levels Tryon, G., & Kane, A. S. (1990). The helping alli-
and patterns of the therapeutic alliance in brief ance and premature termination. Canadian
psychotherapy. American Journal of Psychotherapy, Psychologist, 3, 223–238.
∗
61(2), 109–129. Tryon, G. S., & Kane, A. S. (1993). Relationship
∗
Stiles, W. B., Glick, M. J., Osatuke, K., Hardy, G. E., of working alliance to mutual and unilateral
Shapiro, D. A., Agnew-Davies, R., et al. (2004). termination. Journal of Counseling Psychology,
Patterns of alliance development and the 40, 33–36.
∗
rupture-repair hypothesis: Are productive rela- Tryon, G. S., & Kane, A. S. (1995). Client involve-
tionships U-shaped or V-shaped? Journal of ment, working alliance and type of psycho-
Counseling Psychology, 51(1), 81–92. therapy termination. Psychotherapy Research, 5,
Stiles, W. B., Shapiro, D., & Elliot, R. (1986). 189–198.
∗
Are all psychotherapies equivalent? American Tunis, S. L., Deluchi, K. L., Schwartz, K.,
Psychologist, 41, 165–180. Banys, P., & Sees, K. L. (1995). The relationship
Stone, L. (1961). The Psychoanalytic Situation: of counselor and peer alliance to drug use and
An Examination of Its Development and Essential HIV risk behaviors in a six-month methadone
nature. New York: International Universities detoxification program. Addictive Behaviors, 20,
Press. 395–405.
∗ ∗
Strauser, D. R., Lustig, D. C., & Donnell, C. Van, H. L., Hendriksen, M., Schoevers, R. A.,
(2004). The relationship between working Peen, J., Abraham, R. A., & Dekker, J. (2008).
alliance and therapeutic outcomes for individu- Predictive value of object relations for therapeu-
als with mild mental retardation. Rehabilitation tic alliance and outcome in psychotherapy for
Counseling Bulletin, 47(4), 215–223. depression: An exploratory study. Journal of
∗
Strauss, J. L. (2001). Patterns and impact of the Nervous & Mental Disease, 196(9), 655–662.
∗
alliance and affective arousal in psychotherapy: Vogel, P. A., Hansen, B., Stiles, T. C., &
An application to cognitive therapy for avoidant Gotestam, K. G. (2006). Treatment motivation,
and obsessive-compulsive personality disorders. treatment expectancy, and helping alliance as
Unpublished doctoral dissertation, University of predictors of outcome in cognitive behavioral
Miami, Coral Gables, FL. treatment of OCD. Journal of Behavior Therapy
∗
Strauss, J. L., Hayes, A. M., Johnson, S. L., and Experimental Psychiatry, 37(3), 247–255.
∗
Newman, C. F., Brown, G. K., Barber, J. P., et al. Vronmans, L. P. (2007). Process and outcome of
(2006). Early alliance, alliance ruptures, and symp- narrative therapy for major depressive disorder in
tom change in a nonrandomized trial of cognitive adults: Narrative reflexivity, working alliance and
therapy for avoidant and obsessive-compulsive improved symptom and interpersonal outcomes.
personality disorders. Journal of Consulting and Unpublished doctoral dissertation, Queensland
Clinical Psychology, 74(2), 337–345. University of Technology, Brisbane, Australia.
Sutton, A. J. (2009). Publication bias. In H. Cooper, Wampold, B. E. (2001). The great psychotherapy
L. V. Hedges, & J. C. Valentine (Eds.), The hand- debate: Models, methods, and findings. Mahwah,
book of research synthesis and meta-analysis (2nd NJ: Erlbaum.
ed.). New York: Russell Sage Foundation. Wampold, B. E., Mondin, G. W., Moody, M.,
∗
Svartberg, M., & Stiles, T. C. (1994). Therapeutic Stich, F., Benson, K., & Ahn, H. (1997). A meta-
alliance, therapist competence and client change analysis of outcome studies comparing bona fide
in short-term anxiety-provoking psychotherapy. psychotherapies: Empiricially, ‘all must have
Psychotherapy Research, 4, 20–33. prizes.’ Psychological Bulletin, 122(3), 203–215.
∗ ∗
Wettersten, K. B. (2000). Solution-focused brief Yeomans, F. E., Gutfreund, J., Selzer, M. A.,
therapy, the working alliance, and outcome: A com- Clarkin, J., Hull, J. W., & Smith, T. E. (1994).
parative analysis. Unpublished doctoral disserta- Factors related to drop-outs by borderline
tion, University of Kansas, Lawrence. patients: Treatment contract and therapeutic
∗
Wettersten, K. B., Lichtenberg, J. W., & alliance. Journal of Psychotherapy Practice and
Mallinckrodt, B. (2005). Associations between Research, 3, 16–24.
working alliance and outcome in solution- Zetzel, E. R. (1956). Current concepts of transfer-
focused brief therapy and brief interper- ence. International Journal of Psychoanalysis, 37,
sonal therapy. Psychotherapy Research, 15(1–2), 369–376.
∗
35–43. Zuroff, D. C., & Blatt, S. J. (2006). The therapeutic
∗
Wilson, G. T., Fairburn, C. C., Agras, W. S., relationship in the brief treatment of depres-
Walsh, B. T., & Kraemer, H. (2002). Cognitive- sion: Contributions to clinical improvement and
behavioral therapy for bulimia nervosa: Time enhanced adaptive capacities. Journal of Consult-
course and mechanisms of change. Journal ing and Clinical Psychology, 74(1), 130–140.
∗
of Consulting and Clinical Psychology, 70(2), Zuroff, D. C., Blatt, S. J., Sotsky, S. M.,
267–274. Krupnick, J. L., Martin, D. J., Sanislow, C. A.,
∗
Windholtz, M. J., & Silbershatz, G. (1988). et al. (2000). Relation of therapeutic alliance and
Vanderbilt psychotherapy process scale: A repli- perfectionism to outcome in brief outpatient
cation with adult outpatients. Journal of Clinical treatment for depression. Journal of Clinical &
& Consulting Psychology, 56, 56–60. Consulting Psychology, 68, 114–124.
C HA P TER
Alliance in Child and
3 Adolescent Psychotherapy
Stephen R. Shirk and Marc S. Karver
The therapeutic alliance has a long history in engaging children and adolescents in a
the child and adolescent psychotherapy liter- working relationship is a major challenge
ature dating to the work of Anna Freud for those who treat young clients.
(1946). In contrast, research on the alliance in Relationship processes, including the alli-
youth treatment is relatively new. In their ance, have been neglected in the child and
2003 meta-analysis of relationship predictors adolescent literature for other reasons. In
of child and adolescent treatment outcomes, contrast to the adult literature, where treat-
Shirk and Karver (2003) identified only one ment equivalence has prompted the search
study that met the inclusion criteria used in for common factors, the development of spe-
adult alliance meta-analyses. Fortunately, the cific treatment methods has remained the
last decade has produced a substantial increase focal point of youth therapy research. This
in research on the alliance in child and ado- difference in focus is not just another exam-
lescent treatment, but the total number of ple of child research lagging behind its adult
studies still pales in comparison to the adult counterpart but, rather, reflects the absence
literature (see Chapter 2). of a “Dodo Bird verdict” for treatment equiv-
The discrepancy between the adult and alence in youth outcomes. Broad-band meta-
youth alliance research is not surprising. In analyses of youth treatment outcomes
general, research on child and adolescent indicate that behavioral treatments tend to
therapy has lagged behind its adult coun- produce significantly better results than non-
terpart in many areas. Yet, developmen- behavioral therapies across many childhood
tal differences between children and adults disorders (Weisz, Weiss, Alicke, & Klotz,
make the alliance especially important in 1987; Weisz, Weiss, Han, & Granger, 1995),
youth treatment. Children and adolescents a finding that holds up even after controlling
rarely refer themselves for treatment (Shirk for differences in methodological quality
& Saiz, 1992). Often young clients fail to (Weiss & Weisz, 1995). Although this per-
acknowledge the existence of psychologi- spective has its critics (Miller, Wampold, &
cal problems, and when they do, they attri- Varhely, 2008; Spielmans, Gatlin, & McFall,
bute their cause to environmental factors. 2010), the focus on specific treatment proce-
Children, like persons with severe mental dures has dislocated research on common
illness, lack both awareness of their prob- factors in the child and adolescent literature.
lems and interest in self-exploration that Despite this trend, research on the alliance in
facilitate involvement in therapy (Wright, youth therapy has expanded in recent years
Everett, & Roisman, 1986). Consequently, (Shirk & Karver, 2006; Zack, Castonguay,
70
& Boswell, 2007), partially in response to parenting strategies (Kazdin & Wassell,
growing recognition of within-treatment 2000). Although the content of therapeutic
variability in outcomes. work varies across treatments, the association
The aims of this chapter are threefold. between alliance and outcome is presumed
First, we review definitions, measures, and to be mediated through involvement in treat-
clinical examples of the alliance in the ment tasks.
child and adolescent literature. Of particu- In contrast to this perspective, play ther-
lar importance are developmental issues apists have long emphasized the curative
that distinguish youth and adult alliances. nature of the therapy relationship (Axline,
Second, we provide a meta-analytic review 1947). In this tradition, the child’s experi-
of alliance–outcome associations in child ence of the therapist as supportive, attuned,
and adolescent therapy. Third, we summa- and nonjudgmental was essential for thera-
rize the research on client factors and peutic change (Shirk & Russell, 1996;
therapist strategies that facilitate alliance Wright, Everett, & Roisman, 1986). Drawing
formation with children and adolescents. on the work of Rogers (1957), therapy was
not conceptualized as treatment, as some-
Definitions and Measures thing you do to the child, but rather as
Two views of the therapeutic relationship an opportunity for growth. The relational
were prominent in the early history of child conditions of empathy, genuineness, and
therapy. Anna Freud (1946) observed that an positive regard are posited as the active
“affectionate attachment” between child and ingredients of therapy. The development
therapist is a “prerequisite for all later work” and maintenance of an emotional bond
in child therapy (p. 31). In this early state- facilitates emotional and behavioral change.
ment, we find an enduring distinction in the Associations between bond and outcome
alliance literature, the distinction between in this tradition are direct rather than medi-
bond and work, between the emotional rela- ated through therapeutic work.
tionship and the collaborative relationship Common to the foregoing perspectives
(Estrada & Russell, 1999; Shirk & Saiz, is an emphasis on an emotional connec-
1992). Of equal importance, the link between tion between child and therapist. Emotional
bond and collaboration is framed function- bond, then, appears to be a core compo-
ally; the emotional bond enables the child to nent of alliance with children. This view has
work purposefully on the tasks of therapy. taken root in recent approaches to assessing
The bond itself is not posited as curative, but the alliance in child and adolescent ther-
rather as a catalyst for promoting therapeutic apy (e.g., Shirk, Gudmundsen, Kaplinski,
work. Interestingly, this view is revived in & McMakin, 2008; Shirk & Russell, 1996;
later cognitive-behavioral formulations of Shirk & Saiz, 1992).
the therapy relationship. The alliance serves In contrast, some have criticized this
specific technical procedures and can facili- perspective for failing to acknowledge the
tate child involvement in tasks ranging from social contractual features of the therapeutic
exposure to homework completion (Kendall, alliance. “Traditional theories of child and
Comer, Marker, Creed, & Puliafico, 2009; adolescent psychotherapy appear to have
Shirk & Karver, 2006). In the area of parent overly focused on the bond as necessary and
management training, where parents rather sufficient for change” (DiGiuseppe, Linscott,
than children are the focus of treatment, the & Jilton, 1996, p. 87). From this per-
alliance is hypothesized to improve parent spective, the central component of alli-
attendance and adherence to adaptive ance, especially with older children and
s h i rk , k a rve r 71
adolescents, consists of agreements regard- adults, the relationship with a therapist
ing treatment goals and the methods for fills a need for such connection. Indeed,
accomplishing them. The fact that youth this type of relationship might be quite
are typically referred by others makes the therapeutic for relationship-deprived chil-
establishment of agreements both difficult dren. However, other children may anchor
and essential for treatment collaboration. their positive feelings for their therapists
Given the press toward autonomy in ado- on features not typically regarded as thera-
lescence, this issue takes on added impor- peutic, for example, how fun, stimulating,
tance. At present, however, it is not clear or rewarding the therapist might be. In
if goal agreement is equally relevant for such cases it is unclear if the “bond” reflects
younger child clients as with adolescents. an experience of the therapist as an “ally,”
As these clinical perspectives suggest, there or as a valued playmate. In fact, A. Freud
are important parallels between adult and (1946) distinguished this type of relation-
youth models of alliance. Consistent with ship from the alliance. In the latter, the
Bordin’s (1979) pantheoretical model, three therapeutic bond is based on experiencing
facets of alliance—emotional bond, task the therapist as someone who can be counted
collaboration (work), and agreements (goal on for help with emotional or behavioral
consensus)—are prominent in the youth lit- problems. This is a rather tall order for many
erature. Although it is tempting to view this children and adolescents, and possibly for
convergence as evidence for configural invari- some adults. But it draws attention to the
ance in the alliance construct across age potential developmental differences and the
groups, at least two studies have failed to multiplicity of meanings in the emotional
fully support the three-factor model with bond in child versus adult therapy.
youth. These studies produced a single-factor A second developmental issue concerns
solution, thus suggesting that features of the the task dimension of the alliance. In the
alliance may be less differentiated at younger adult literature, tasks are framed in terms of
ages (DiGiuseppe et al., 1996; Faw, Hogue, agreements about the content and methods
Johnson, Diamond, & Liddle, 2005). of therapy; in essence, whether there is
Of equal importance, a number of devel- consensus between client and therapist on
opmental issues contribute to differences the substance of therapy (Bordin, 1979).
across youth and adults in the nature of Such judgments may exceed the cognitive
bond, task, and goal dimensions of alliance. capacities of many child and adolescent
Consider first the therapy bond. Anna clients. For example, studies of children’s
Freud (1946) noted some time ago that understanding of therapy have shown
a child’s relationship with the therapist important developmental progressions in
could arise from a number of sources, not their recognition of therapy processes and
all of them developmentally equivalent. parameters (e.g., Shirk & Russell, 1998, for
For many children, the relationship with a a review). For example, children’s causal rea-
therapist is an opportunity to obtain grati- soning may limit their ability to understand
fications not available in other contexts. As links between specific therapy tasks and
A. Freud observed, “if no one at home plays subsequent therapy goals (Shirk, 1988).
games with the child, for example, he might Perhaps it should not be surprising that
like to come to treatment because there research finds little agreement between child
a grown-up pays attention to him” (Sandler, and therapist ratings of task collaboration,
Kennedy, & Tyson, 1980, p.47). For chil- but greater convergence for therapy bond
dren who lack sustaining relationships with (Shirk & Saiz, 1992). Such developmental
concerns have prompted some investigators therapists are faced with establishing and
to suggest that task collaboration with chil- maintaining an alliance with the youth and
dren is best assessed through observation his or her parent(s). Most research on alli-
(Karver et al., 2008; Shirk & Karver, 2006). ance–outcome relations with children
In essence, the task dimension of alliance in and adolescents has focused on the youth–
youth therapy, especially with children, therapist relationship. A notable exception
should be operationalized as observed is in the area of parent management train-
participation in therapy tasks and not as ing where parents are the primary focus of
agreements about such tasks. child therapy. It is possible that alliances
Developmental issues complicate the goal with parents and youth may relate to differ-
dimension of the alliance as well. An impor- ent sets of outcomes. For example, Hawley
tant difference between adult and youth and Weisz (2005) found that parent, but
therapy is the involvement of other family not youth, alliance predicted better therapy
members aside from the identified client. participation. Youth alliance, but not parent
Minimally, parents or guardians are involved alliance, predicted symptom change. These
in transportation to and payment of therapy. findings suggest that a strong alliance with
Quite often, however, parents are more parents is important for treatment continu-
actively involved as informants about client ation, whereas the youth alliance may be
functioning, collateral participants, or even more critical for treatment outcomes.
as therapeutic collaborators who help with At present, a unified definition of alliance
treatment implementation outside sessions. has yet to emerge in the youth literature. In
Consequently, the therapist is faced with fact, our review of the research literature
multiple sets of goals, and often the goals of reveals 10 different alliance measures for
parents and youth diverge. Agreement on children and adolescents. No study has
goals, then, is complicated by whose goals examined the concurrent validity of the most
are considered. A study of clinic-referred frequently used measures. Thus, it is not
children (Hawley & Weisz, 2003) examined clear if different measures with similar names
therapist, child, and parent agreement about are assessing the same facets of the alliance.
the most important problems to be addressed The two most frequently used patient
in therapy. Amazingly, more than 75% of and therapist report instruments in youth
child, parent, and therapist triads began research are the Working Alliance Inventory
treatment without agreement on even one (WAI; Horvath & Greenberg, 1989) and
target problem. Nearly half failed to agree the Therapeutic Alliance Scale for Children
on one broad problem domain such as (TASC; Shirk & Saiz, 1992). The WAI has
aggression versus depression. It is interesting been used primarily with adolescents
to note that therapists agreed with parents and the TASC with children and young
more often than with children. Such evi- adolescents. Although the WAI, originally
dence suggests that agreement on goals may developed for adult therapy, has been
mean something quite different in youth modified for use with adolescents (Linscott,
therapy than in adult therapy At present, it DiGuiseppe, & Jilton, 1993), the original
is not clear if agreement between parent and or short version has been employed
therapist or child and therapist is a better most frequently. The WAI measures the
predictor of treatment outcome. quality of the therapeutic relationship across
A related issue involves the presence of three subscales: bonds, tasks, and goals. The
multiple alliances in youth treatment even final item pool for the measure was gener-
when it is child focused. Unlike with adults, ated on the basis of content analysis of
Bordin’s (1979) model of working alliance. or task categories and consistently sorted
Expert raters evaluated items for goodness items were retained. The resulting coding
of fit with the working alliance construct. system includes eight bond items and six
The TASC was developed specifically task collaboration items. Interrater reliabil-
for child therapy and also was based on ity has been shown to be good across items.
Bordin’s model (1979). Two dimensions Bond and task dimensions are highly
are assessed: bond between child and thera- correlated, consistent with what has been
pist, and level of task collaboration. Unlike found with youth self-reports of alliance
the WAI, task collaboration does not refer dimensions, suggesting that alliance may
to agreements on tasks, but to ratings be a unitary construct in youth therapy.
of actual collaboration on tasks such as
“talking about feelings” and “trying to solve Clinical Examples
problems.” The therapist version of the The following verbal interactions derived
TASC involves ratings of the child’s bond from a composite of cases reflect different
and task involvement rather than the thera- features of the therapeutic alliance with
pist’s own. Although items on the bond young clients. The first example illustrates a
subscale remain constant, items on the task strong emotional bond between a young
collaboration scale vary with type of treat- adolescent and her therapist:
ment in order to be consistent with CBT Therapist: So, what is it like when
(cognitive-behavioral therapy) or psycho- you’re feeling really down?
dynamic tasks. The subscales show good Client: I get like I don’t want to talk to
internal consistency and relatively high anyone. I’m like get away, leave me
levels of stability over a 4- to 7-week period alone. My dad asks me how I’m doing
(Shirk et al., 2008). Although therapist and I just say nothing or walk away.
and child agreement on bond ratings are Therapist: You just want some space.
medium to strong, agreement is substan- You don’t want to be pushed.
tially lower for task collaboration. Client: Exactly.
A number of observational measures Therapist: In here, I’m going to ask you
have appeared in the youth literature but a lot about how you are feeling. If you
none have become the “gold standard.” feel like I’m pushing you, is it possible
One measure that was developed specifi- you will not want to talk with me.
cally for child and adolescent therapy and Client: I don’t think that’ll happen
that has been used in more than one study because you’re not in my face.
is the Therapy Process Observation Coding Talking gets my stress out. When
System–Alliance Scale (McLeod & Weisz, I’m in a bad mood on the day of
2005). This observation scale took as its our meetings, I look forward to our
starting point the distinction between talking . . . it helps keep me going
bond and task collaboration found in factor because I know you get me.
analyses of child and therapist reports of
alliance (Shirk & Saiz, 1992) and factor In the next example, the goal dimension
analyses of process codes (Estrada & Russell, of the alliance is prominent. Here the
1999). Items from a broad range of mea- therapist explores the adolescent’s goals
sures that mapped onto the bond and task for therapy.
dimensions were initially included in the Therapist: I know what your parents are
item pool, and redundant items reduced. hoping for from our therapy, but
Expert raters then sorted items into bond what are your goals?
Client: I want to stop worrying that I’ll Therapist: So using what we’ve worked
say something wrong, so I won’t just on might have a payoff?
feel all stuck. Client: Yeah, like what we practiced in
Therapist: It sounds like you’d like to here.
shift your focus away from all the
It should be noted that other features
things that make you worry and feel
of interactions can reflect the condition of
trapped.
the alliance. For example, a child who is
Client: Yeah, when I think about what
unresponsive to therapist questions or who
could happen, I become so nervous
is only willing to talk about topics unrelated
I just avoid everyone. I want to go
to problems or issues that prompted ther-
walk up to the ins (popular girls) and
apy demonstrates behaviorally low levels of
just be right there talking and not all
collaboration. Similarly, many children will
what if.
actively participate in games and unstruc-
Therapist: So, if we could change how
tured play but will avoid talking about con-
much you worry and think about all
cerns or practicing relevant skills in session.
the negatives that would be a good
Though such children appear to like their
result?
therapist, it is not evident that the therapist
Client: Definitely, I’m tired of worrying
is viewed as someone who could help with
all the time.
emotional or behavioral problems.
In the final example, an older child talks
with his therapist about dealing with anger.
Prior Reviews
The client’s statements reflect the collabora-
In their 2003 meta-analysis of associations
tive aspect of the alliance:
between relationship variables and treatment
Client: I feel better since we last talked. outcomes, Shirk and Karver found only 23
That stuff we worked on was pretty studies with quantifiable relationships pub-
helpful. lished over the previous 27 years. The major-
Therapist: That’s cool. Great. What ity of these studies did not assess alliance per
did you do? se but evaluated various dimensions of rela-
Client: Like . . . I forgot what it is tionship quality such as therapist warmth,
called . . . like . . . I controlled my therapeutic climate, or treatment participa-
temper . . . when I got angry . . . tion. Only ninestudies examined the alli-
I was like OK like take a deep ance, and of these nine only one evaluated
breath . . . then I walked away. the alliance prospectively in individual ther-
Therapist: Great. It helped bring your apy. Overall, Shirk and Karver (2003) found
anger down. that relationship variables are related to
Client: mmhmm. youth treatment outcomes with a weighted
Therapist: You made a good decision. mean correlation of 0.20, slightly smaller
Some people get angry and are than, but similar to, estimates from the adult
like, hey, I’m right, I’m not backing literature (see Chapter 2).
down. Although Shirk and Karver (2003) did
Client: If I get up in their face when not report results for alliance studies alone,
I’m mad, I end up losing anyway. a reanalysis of the data indicates that the
Therapist: Losing anyway? weighted mean alliance–outcome association
Client: Yeah, I pay for it later. Get in in the sample of nine studies was r = 0.25.
trouble and stuff. It should be noted, however, that this
estimate includes both prospective and child and adolescent literature. First, consis-
concurrent measurement of alliance and tent with earlier meta-analyses, we expected
outcome, and the assessment of alliance in alliance to predict outcomes and hypothe-
individual and family therapy. sized that this association might be moder-
In a subsequent meta-analysis, Karver, ated by timing of alliance assessment. Prior
Handelsman, Fields, & Bickman (2006) research on relationship processes in child
specifically examined alliance–outcome rela- and adolescent treatment has been criticized
tions for child and adolescent clients. These for the inclusion of studies involving the
estimates did not include parent and family concurrent measurement of alliance and
alliance data. Karver et al. (2006) identified outcome (Shirk & Karver, 2003). It has been
10 studies that assessed youth alliance in suggested that concurrent measurement
relation to outcome. Correlations varied inflates estimates of association. In order to
widely across studies and ranged from 0.05 address this possibility, we examined associa-
to 0.49 with a weighted mean correlation of tions by timing of alliance measurement.
0.21. Like the original meta-analysis by Shirk Second, some evidence suggests that parent
and Karver (2003), these results indicate a alliance may be more strongly associated
moderate association between alliance and with treatment continuation and attendance
outcome in child and adolescent therapy. than treatment outcome. (Hawley & Weisz,
2005), thus we examined this possibility by
Meta-Analytic Review comparing correlations by type of alliance
In recent years, there has been significant (parent vs. youth). Third, consistent with
growth in the number of studies evaluating prior results (Shirk & Karver, 2003), we did
alliance–outcome relations in child and not expect a difference in the strength of
adolescent therapy. In order to provide an alliance–outcome relations by treatment
estimate of this association based on a larger type; instead, we expected comparable asso-
sample of studies than previously reported, ciations across behavioral and nonbehavioral
we conducted a meta-analysis of current therapies. We did, however, predict differ-
evidence on alliance–outcome relation- ences in alliance–outcome relations as a
ships in child and adolescent therapy. Our function of type of problem treated, as pre-
meta-analysis is restricted to child- and vious research (Shirk & Karver, 2003) found
adolescent-focused treatments and does somewhat stronger alliance–outcome rela-
not include studies of family therapy (which tions for youth with externalizing (disrup-
are reviewed in Chapter 4). Consistent with tive) problems than with internalizing
prior youth meta-analyses, we included (emotional) problems. Therefore, we exam-
both prospective and concurrent assess- ined outcomes by type of problem. Finally,
ments of alliance and outcome because we did not expect age-related differences in
of the limited number of studies in the alliance–outcome relations. Comparisons
literature. However, we provide separate were made between studies with child sam-
estimates of association for each design. ples and those with adolescent samples.
Because parents are often included in child
and adolescent therapy, and in fact, they Alliance–Outcome Studies
may be the focus of behavioral treatments, To identify applicable studies that mea-
we also provide an estimate of association sured the relationship between alliance and
between parent alliance and outcome. outcome, a three- pronged approach was
Our analyses were guided by a number of used. First, prior reviews of the alliance-
hypotheses based on prior findings in the to-outcome relationship were examined for
qualifying manuscripts (Karver et al., 2006; Calculation of Effect Sizes
Shirk & Karver, 2003). Citations of these Because most studies reported results (alli-
articles were then examined as a means to ance-to-outcome relationships) as corre-
identify additional manuscripts. Second, lations, the product-moment correlation
the PsycINFO database was searched from coefficient r was used as our effect size
2004 forward to identify articles that have estimate. All results (typically product-
been published since the last major meta- moment correlation coefficient rs) in each
analytic review of the therapeutic alliance study were converted to Fisher’s Z in order
in child and adolescent therapy. Finally, to normalize the r distribution (Hedges &
Google Scholar was used to search for Olkin, 1985). For all studies it was possi-
studies that may have been missed and ble to compute effect sizes, thus no effects
for unpublished manuscripts. For both were imputed as zero. In most studies, more
searches, child and adolescent were used in than one alliance–outcome relationship
conjunction with the terms alliance or rela- was reported. In order to correct for bias
tionship and therapeutic or therapy. due to correlated effects within studies and
To be included in the current meta- an unequal number of associations reported
analysis, studies had to meet the following in different studies, we averaged (simple
criteria: (1) the study had to include a spe- mean) the Fisher’s Z’s for each study. In
cific measure explicitly described in the order to calculate a more precise estimate
manuscript as an alliance measure; (2) the of the overall relationship between alliance
alliance had to be related to some indicator and outcome, we weighted the average
or measure of posttreatment outcome and effect size (Z) for each study by the number
not another process variable; (3) the study of participants in the study. We weighted
had to be of individual or group mental each effect size so that the final estimate of
health treatment delivered to a youth under the alliance-to-outcome relationship prop-
age 18 or a parent; (4) the study could not erly accounts for the fact that more precise
be an analog study; (5) the study needed to estimates should be given more weight in
be available in English; (6) the study must the aggregate. The weighted effect sizes for
have included at least 10 participants; and each study were aggregated, and then this
(7) if the study did not directly report a sum was divided by the sum of the weights
correlation between alliance and outcome, (number of participants per study minus 3)
enough information had to be available in for each study, resulting in an estimate of
the manuscript to calculate the effect size. the overall alliance-to-outcome relationship.
The resulting sample consisted of 29 stud- This weighted effect size Z was then con-
ies with 2,202 youth clients and 892 parents. verted back to the product-moment correla-
Studies were coded for type of alliance (youth tion coefficient r. We then analyzed effect
vs. parent), timing of alliance measurement size estimates by the type of alliance, timing
(early- or middle- during first two thirds of of alliance measurement, type of therapy,
therapy sessions, late- during the last third of type of problem treated, and child age.
therapy sessions, at termination, or posttreat-
ment, or combined), type of treatment (cog- Results of the Meta-Analysis
nitive-behavioral/behavioral, nonbehavioral, The 29 studies that met inclusion criteria are
or mixed), type of problem (internalizing, displayed in Table 3.1. The sample includes
externalizing, mixed, or substance abuse), 26 published studies and 3 doctoral disser-
and mean age of youth (child = less than 12; tations. Twenty-eight studies evaluated the
adolescent = ages 12–18). child or adolescent alliance, and 10 studies
Table 3.1 Reviewed Studies, Alliance Measures, Classifications, and Effect Sizes
Study N Alliance measure Classifications Wt. mean r
Adler (1998) 92 Parent Evaluation Questionnaire Both ages 0.24
Mixed problems
Nonbehavioral
Auerbach et al. (2008) 39 Working Alliance Inventory - Short Adolescent 0.12
Substance Abuse
Nonbehavioral
Champion (1998) 19 Child Behavior in Therapy Scale Child 0.18
Mixed problems
Nonbehavioral
Chiu et al. (2009) 34 Therapy Process Observation Child 0.21
System - Alliance Internalizing
Behavioral
Colson et al. (1991) 69 Therapeutic Alliance Difficulty Scale Adolescent 0.28
Mixed problems
Nonbehavioral
Creed & Kendall (2005) 68 Therapeutic Alliance Scale for Children Adolescent 0.30
Internalizing
Behavioral
Darchuck (2007) 40 Working Alliance Inventory - Short Adolescent 0.25
Substance Abuse
Nonbehavioral
Diamond et al. (2006) 353 Working Alliance Inventory - Short Adolescent 0.20
Substance abuse
Behavioral
Eltz et al. (1995) 38 Penn Helping Alliance Questionnaire Adolescent 0.32
Mixed problems
Nonbehavioral
Florsheim et al. (2000) 78 Working Alliance Inventory Adolescent 0.12
Externalizing
Nonbehavioral
Gavin et al. (1999) 60 Treatment Alliance Scale Adolescent 0.03
Mixed problems
Nonbehavioral
Green (1996) 25 Family Engagement Scale Child 0.58
Empathy and Understanding Scale Mixed problems
Nonbehavioral
Green 20 Family Engagement Scale Adolescent −0.04
Empathy and Understanding Scale Mixed problems
Nonbehavioral
Handwerk et al. (2008) 71 Working Relationship Scale Adolescent 0.25
Mixed problems
Behavioral
Hawley & Garland (2008) 78 Working Alliance Inventory - Short Adolescent 0.29
Mixed problems
Nonbehavioral
(Continued )
78
Table 3.1. Continued
Study N Alliance measure Classifications Wt. mean r
Hawley & Weisz (2005) 81 Therapeutic Alliance Scale for Children Both 0.13
Mixed problems
Nonbehavioral
Hintikka (2006) 45 Working Alliance Inventory Adolescent 0.07
Mixed problems
Nonbehavioral
Hogue et al. (2006) 56 Vanderbilt Therapeutic Alliance Scale Adolescent −0.02
Substance abuse
Behavioral
Holmqvist et al. (2007) 59 Penn Helping Alliance Questionnaire Adolescent 0.13
Externalizing
Combined
Karver et al. (2008) 23 Alliance Observation Coding System Adolescent 0.08
Working Alliance Inventory Internalizing
Combined
Kaufman et al. (2005) 45 Working Alliance Inventory - Short Adolescent 0.00
Mixed problems
Behavioral
Kazdin et al. (2006) 310 Working Alliance Inventory Child 0.29
Therapeutic Alliance Scale for Children Externalizing
Behavioral
Kazdin et al. (2005) 185 Working Alliance Inventory Child 0.21
Externalizing
Behavioral
Kazdin & Whitley (2006) 218 Working Alliance Inventory Child 0.30
Therapeutic Alliance Scale for Children Externalizing
Behavioral
McLeod & Weisz (2005) 22 Therapy Process Observation Child 0.25
System - Alliance Internalizing
Nonbehavioral
Shirk et al. (2008) 50 Therapeutic Alliance Scale Adolescent 0.26
for Adolescents Internalizing
Behavioral
Smith et al. (2008) 55 Penn Helping Alliance Questionnaire Adolescents 0.36
Mixed problems
Nonbehavioral
Tetzlaff et al. (2005) 434 Working Alliance Inventory - Short Adolescent 0.24
Substance abuse
Combined
Zaitsoff et al. (8) 36 Penn Helping Alliance Questionnaire Adolescent 0.48
Eating Disorders
Nonbehavioral
79
examined the parent alliance. In terms of an rw = 0.38 (CI = +/− 0.24). As predicted,
timing of alliance measurement, 15 studies studies that included a measure of alliance
assessed the alliance early or in the middle of later in treatment, either as part of an aver-
treatment, 8 studies measured alliance late or age, change score, or slope, or simply as a
posttreatment, and 8 studies assessed alliance late measure of alliance, yielded larger effects
over time as an average and 2 assessed alliance ( p < 0.05) than studies that measured alli-
as a slope or change score. In terms of types of ance early in treatment (See Figure 3.1).
treatment, 20 studies involved nonbehavioral With regard to client characteristics, age
therapy, 14 involved behavioral or cognitive- and type of problem were examined as
behavioral therapy, and 4 involved mixed potential moderators. Studies with child
therapies, usually the inclusion of family ther- samples (under age 12) yielded significantly
apy with individual therapy. Eleven studies larger weighted mean correlations (r = 0.27;
included child samples, 12 included only CI = +/− 0.08) than studies with adolescent
adolescents, and 5 studies included both chil- samples (r = 0.17; CI = +/− 0.05). A number
dren and adolescents. One study could not of studies included both children and ado-
be classified. Finally, 9 studies focused on lescents and produced a mean correlation of
internalizing problems, 7 on externalizing 0.12 (CI= +/− 0.10), also significantly dif-
problems, 5 on substance abuse, 6 on mixed ferent from the child estimate ( p < 0.05).
problems, and 1 on eating disorders. As Previous research (e.g, Shirk & Karver,
shown in Table 3.1, weighted correlations 2003) did not find age-related differences in
averaged 0.19 with a confidence interval of alliance–outcome associations, and this find-
+/− 0.04 (range = −.09 to 0.59). This esti- ing is somewhat surprising given the clinical
mate is very similar to earlier results for the focus on alliance difficulties with adolescents
association between relationship variables (Castro-Blanco & Karver, 2010).
and outcomes (rw = 0.20), but slightly lower Effect sizes did not differ across types
than the estimate based on studies that mea- of treatment. The weighted mean correla-
sured alliance and outcome (rw = 0.25). The tions for behavioral/cognitive behavioral,
current meta-analysis includes over three nonbehavioral, and mixed therapies were
times as many alliance studies and excluded 0.18 (CI = +/− 0.05), 0.19 (CI = +/− 0.06),
studies of family therapy. and 0.20 (CI = +/− 0.08), respectively.
These results are consistent with earlier
Mediators and Moderators findings reported by Shirk and Karver
A number of variables were expected to (2003). Nor was there a difference in the
moderate the strength of association between weighted mean correlation for therapist–
alliance and outcome. One methodological parent and therapist–youth alliance. Both
variable was timing of alliance assessment. It associations averaged rw=19.
was hypothesized that concurrent measure- Consistent with earlier findings (Shirk &
ments would show stronger associations Karver, 2003), strength of alliance–outcome
than prospective designs. Studies that relations varied as a function of type of treated
assessed alliance early produced an rw = 0.15 problem. The weighted mean correlation for
(CI = +/− 0.06), those that measured it late externalizing problems was 0.24 (CI = +/−
or posttreatment resulted in an rw = 0.24 0.07). In contrast, internalizing problems,
(CI = +/− 0.09), those that used an average substance abuse, and mixed problems showed
resulted in an rw = 0.21 (CI = +/− 0.07), weighted mean correlations of 0.17 (CI =
and the two that used a slope or a change +/− 0.11), 0.14 (CI = +/− 0.07), and 0.20
score across early and late alliances yielded (CI = +/− .07), respectively. One study
evaluated youth with eating disorders and outcome in child and adolescent therapy.
and produced a mean correlation of 0.53 Estimates of alliance–outcome relations in the
(CI = +/− 0.34). The strength of alliance– child and adolescent literature may be slightly
outcome relations differed between studies inflated by the inclusion of studies that only
with externalizing samples and those with measure alliance late in treatment. However,
substance abuse samples ( p < 0.05). Other studies that assessed alliance with methods
differences were not reliable. The difference that included early measures of alliance—-
in alliance–outcome effect size for youth with through averaged scores, slopes, or change
externalizing versus internalizing problems scores—produced effects that were slightly
was in the same direction as previous research, larger than the overall estimate for all studies.
but it did not attain statistical significance. Thus, results support the view that the thera-
Despite theoretical models that posit a peutic alliance is an important predictor of
mediated relation between alliance and outcome in youth therapy.
outcome, no study examined mediation The strength of alliance–outcome rela-
through involvement in specific treatment tions did not vary with type of treatment.
tasks. Only one study (Karver et al., 2008) The alliance thus appears to be important
demonstrated a strong link between early for therapies that vary widely in terms
alliance and later task involvement. of specific treatment procedures, including
therapies that focus on teaching contingency
Summary of Meta-Analytic Findings management to parents. The alliance is an
Results from our meta-analysis indicate a important component of broad classes of
small-to-medium association between alliance child and adolescent therapy, and it appears
0.8
0.6
Effect size
0.4
0.2
0.0
Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

Session alliance rated
Fig. 3.1 Alliance effect sizes and confidence intervals by time of measurement
to contribute to outcomes in treatments as therapy outcomes? One way of thinking
different as manual-guided CBT and non- about this issue is to use a common metric
directive, play therapy. It is noteworthy and to benchmark effects from comparative
that alliance also is predictive of outcome outcome studies with the results of the cur-
in parent-focused therapies. rent meta-analysis. The mean correlation
The strength of association between alli- obtained in this study, rw = 0.19, converts
ance and outcome varied as a function of to an effect size, d, of approximately 0.39.
several variables. Our results showed a slightly How does this compare with effect sizes
stronger association between alliance and obtained by comparing two or more youth
outcome among children compared with treatments involving different procedures?
adolescents. A prior meta-analysis did not Two recent estimates are relevant. In a
find such a difference (Shirk & Karver, comparison of evidence-based treatments
2003). It is possible that age-related dif- (EBTs) with usual-care (UC) therapy, Weisz,
ferences in alliance–outcome associations Jensen-Dose, and Hawley (2006) found
are confounded with the typical problems EBTs to be superior to UC with an effect size
treated in these two age groups. Specifically, of 0.25 for direct comparisons of psychother-
child studies often include samples of disrup- apies. In an effort to estimate the impact of
tive children and adolescent studies include specific treatment methods on outcome,
substance abuse samples. Given that correla- Miller et al. (2008) conducted a meta-
tions were stronger for externalizing samples analysis of youth comparative outcome stud-
than substance abuse samples, the age differ- ies in which two or more bona fide treatments
ences in alliance–outcome associations could were evaluated. Although they found some
be due to age differences in problem types. evidence for method effects, the total out-
It is not clear why stronger associations come variance explained by treatment
between alliance and outcome are found method was 0.037, or when converted to a
with externalizing youth, as they were in correlation, an r of 0.19, identical to the
our earlier meta-analysis (Shirk & Karver, mean correlation obtained in our current
2003). We have speculated that the chal- meta-analysis. Thus, when benchmarked to
lenge of engaging oppositional and disrup- findings from comparisons of treatment
tive youth increases variability in alliance, methods, the contribution of alliance to out-
and thus increases the possibility of larger come appears similar to the contribution of
correlations than those obtained in samples specific methods. Of course, it is possible that
with a more attenuated range of alliance differences in specific methods are more criti-
scores. It is also possible that alliance plays cal for some disorders than others, for exam-
an especially critical role in the treatment ple, obsessive-compulsive disorder versus
of disruptive problems, possibly by facili- depression, just as alliance appears to be more
tating the internalization of skills or an strongly related to outcome for externalizing
empathic attitude toward others. compared to internalizing disorders.
The question of how important the A critical question, then, is the relation
alliance is to youth outcomes deserves between the alliance and specific treatment
some comment. When viewed in terms of factors in child and adolescent therapy.
explained variance, the estimate of less than Are these factors complementary and do
4% of total outcome variance seems rather they contribute to outcome in an additive
small. However, an important question is manner, or might alliance actually account
how does this compare with the contribu- for differences in outcome currently
tion of specific treatment methods to youth attributed to specific methods? Only one
published study in the child and adolescent One exception is motivation for change.
literature has attempted to evaluate the latter Consistent with the adult literature (see
possibility. Kaufman and colleagues (2005) Chapter 2), results generally show that
examined the alliance as a potential media- youth with greater motivation, more prob-
tor of treatment effects obtained in a com- lem recognition, and more stated reasons
parison of group CBT and a Life Skills for changing their behavior form more
group for adjudicated, depressed adoles- positive alliances (Christensen & Skogstad,
cents. These investigators found more posi- 2009; Estrada & Russell, 1999; Fitzpatrick
tive alliances in the group CBT condition & Irannejad, 2008), though one study
than in the Life Skills condition, but this failed to find these associations (Garner,
difference in alliance did not account Godley, & Funk, 2008).
for significant variance in depression out- Within the youth clinical literature, ado-
comes. Unfortunately, few comparative lescence is viewed as one of the most difficult
outcome studies have assessed the alliance periods for alliance formation (Castro-
in the youth literature; consequently, the Blanco & Karver, 2010; Meeks, 1971). The
relative contribution of alliance and specific developmental press toward autonomy, the
factors cannot be readily addressed. increasing centrality of peer relationships,
and growing doubts about adults’ capacity
Client and Therapist Contributions for understanding youth experiences can
A small but growing number of studies contribute to alliance difficulties. One might
have examined predictors of the therapeu- expect, then, relatively clear evidence for
tic alliance in child and adolescent therapy. more positive alliances among children than
Emerging evidence points to a number of adolescents. Yet, the limited evidence is
client characteristics and therapist behav- mixed with two studies showing more posi-
iors associated with alliance formation. tive alliances among older youth (Garner,
Godley, & Funk, 2008; Gavin, Wamboldt,
Pretreatment Predictors of Child Sorokin, Levy, & Wamboldt, 1999), and
and Adolescent Alliance two studies reporting younger children
Numerous clinical accounts have high- forming better alliances (Creed & Kendall,
lighted some of the potential challenges to 2005; DeVet, Kim, Charlot-Swilly, & Ireys,
engaging children and adolescents in a 2003). It is certainly possible that adoles-
working alliance (Castro-Blanco & Karver, cents’ greater capacity for understanding
2010; A. Freud, 1946; Meeks, 1971). therapy rationale and tasks might contribute
Among the obstacles to alliance formation to better alliances among older than younger
are the limited problem recognition and youth. However, the paucity of research evi-
acknowledgement, the tendency to attri- dence on this issue pales in comparison with
bute problems to extermal sources, low practice-based observations of alliance diffi-
motivation for change, absence of self-re- culties with adolescents. Given that most
ferral (or the presence of coaxed or coerced treatment studies focus on either children or
referral), and a lack of understanding of the adolescents, direct comparisons of alliance
therapy processes. Although some of these processes are likely to remain limited.
factors could be indicative of psychopathol- Two other pretreatment factors have
ogy, most are simply a consequence of attracted some attention in the research
developmental level. Surprisingly few stud- literature: type and severity of psychopathol-
ies have examined direct links between these ogy and interpersonal functioning. Research
developmental factors and youth alliance. on symptom severity and alliance formation
has produced mixed results. Three studies Some studies have taken a more indirect
found no relationship between initial symp- approach to looking at the contribution of
tom severity and subsequent alliance pretreatment relationship factors to alliance
(Bickman et al., 2004; DeVet et al., 2003; formation. The idea behind the indirect
Eltz et al., 1995), one found a negative rela- approach is that if youth have poor family
tionship between overall level of maladaptive relationships or low levels of social support,
functioning and alliance (Green, Kroll, Imrie, these factors might indicate poor inter-
Frances, Begum et al., 2001), and two actu- personal skills or poor prior experiences of
ally showed a positive relationship between healthy relationships. Consistent with this
initial severity and alliance (Christensen & perspective, a number of studies have found
Skogstad, 2009; Shirk et al., 2008). It is likely social support—both youth and parent
that greater clarity will be attained by exam- reported—to be related to youth and/or
ining specific symptoms (e.g., depressive parent alliances (DeVet et al., 2003; Fields
symptoms) and alliance rather than overall et al., 2010; Garner, Godley, & Funk,
symptom severity. For example, clinical 2008; Hawley & Garland, 2008; Kazdin
accounts have highlighted the unique chal- & Whitley, 2006). As these results suggest,
lenges of engaging youth with high levels of alliance formation is a social process, and
oppositional and rule-breaking behavior youth relational experiences and compe-
(Gallager, Kurtz, & Blackwell, 2010). High tencies appear to impact the alliance.
levels of defiance, distrust of adult authority, Finally, it should be noted that gender and
and externalization of problems have been race have been considered in a small number
found to make alliance formation especially of studies. The evidence is split on the role of
difficult with this group. (Bickman et al., gender; two studies showed no association
2004; Garcia & Weisz, 2002) between gender and alliance (Creed &
In contrast to symptom severity, inter- Kendall, 2005; Fitzpatrick & Irannejad,
personal variables appear associated with 2008), but three showed females to rate the
youth alliance. A recurrent finding in the alliance more positively than male youth
adult literature is that quality of past (Christensen & Skogstad, 2009; Eltz, Shirk,
and present relationships predict alliance & Sarlin, 1995; Wintersteen, Mensinger, &
quality (e.g., Hersoug, Monsen, Havik, Diamond, 2005). One study showed the
& Hoglend, 2002; Mallinckrodt, Coble, & opposite with males rating the alliance more
Gantt, 1995). One study (Eltz, Shirk, & positively than females (Hawke, Hennen &
Sarlin, 1995) found that youth with more Gallione, 2005). Given that the majority of
interpersonal problems, but not greater child and adolescent therapists are female, it
overall problem severity, had more alliance would be useful to know if gender matching
difficulties, and another (Fields et al., 2010) has an impact on alliance, and if so, for whom.
found social competence to be related to The evidence with regard to race effects is
alliance. In the former study, a history extremely limited. Hawke and colleagues
of child maltreatment predicted early alli- (2005) found Hispanic and African-American
ance difficulties even after controlling for youth to have stronger youth–therapist
problem severity. Perhaps it should not be alliances than European-American youth,
surprising that youth with interpersonal but Wintersteen et al. (2005) found no
trauma histories, especially abuse in the race effects. The impact of therapist–youth
context of the family, would be cautious matching on race has not been examined.
forming a close relationship with an adult The reality is that we know very little about
caregiver (therapist). whether gender, race, or matching on these
variables is related to the alliance in youth patronizing, predicted weaker therapist
therapy. reported alliance at Session 7.
Two findings seem especially important.
Therapist Strategies for Strengthening First, therapist collaborative behavior,
the Youth Alliance including the establishment of treatment
Despite the importance of the alliance in goals with the child, may be critical for alli-
youth therapy, research on therapist behav- ance formation. Recognition and valida-
iors that contribute to alliance formation tion of the child’s perspective on this critical
has only recently emerged. With the excep- issue may help differentiate the therapist
tion of research on family therapy (Diamond, from parents, and help the child view the
Liddle, Hogue, & Dakof, 1999), all of therapist as an ally. Second, therapists need
these studies have been conducted with to balance their approach between setting a
children and adolescents in individual CBT. collaborative tone without being either too
Although generalizations to other forms of formal or overly familiar with the child.
child therapy would be premature, some Similarly, therapists need to balance their
patterns have emerged across these initial focus on problems with the maintenance of
studies and deserve consideration. rapport. Too little focus on problems can
Two studies address therapist strate- amount to supporting avoidance, and too
gies with children. Creed and Kendall (2005) great a focus can undermine alliance.
examined a set of therapist behaviors hypoth- Therapists must be mindful of the child’s
esized to promote or interfere with alliance level of tolerance and work to gradually
formation during the first three sessions of facilitate the child’s ability to talk about anxi-
CBT with anxious children. Child and ther- ety. Therapist flexibility about the pace of
apist reports were used to assess the alliance treatment may be critical for alliance
at Sessions 3 and 7. Child-reported alliance formation.
at Session 3 was positively associated with Chu and Kendall (2009) evaluated thera-
therapist collaboration strategies, including pist flexibility as a strategy for promot-
presenting therapy as a team effort, building ing client involvement in CBT for child
a sense of togetherness by using words like anxiety disorders. In this context, flexibility
“we,” “us,” and “let’s,” and by helping the is conceptualized as treatment adapta-
child set goals for therapy. In contrast, two tion occurring within the parameters of
therapist verbal behaviors predicted a weaker treatment fidelity: this is a way of indivi-
alliance at Session 3. Not surprisingly, “push- dualizing manual-guided therapy. In fact,
ing the child to talk” about anxiety and anx- therapist flexibility was significantly associ-
ious situations was negatively associated with ated with increases in child involvement
early child alliance. Similarly, therapist efforts over the course of therapy. Later child
to “emphasize common ground,” that is, involvement predicted treatment improve-
therapists’ comments like “Me, too!” in ment and lower levels of impairment. Results
response to children’s statements about inter- from this study, in conjunction with find-
ests and activities were predictive of weaker ings reported by Creed and Kendall (2005),
alliances. For therapist-reported alliance, suggest that therapists who provide CBT in
none of the therapist behaviors predicted a flexible manner, who are less didactic, less
alliance scores at Session 3, but collaborative pressing, and better able to integrate client
strategies predicted better alliances at needs within the treatment protocol are
Session 7. Talking to the child in an overly more likely to facilitate better alliances and
formal manner, that is, being too didactic or greater treatment involvement.
In addition to the foregoing research on focus in the first session. After initially social-
child anxiety, Shirk and colleagues have izing the adolescent to treatment, therapists
published a series of studies on engagement who reported more positive alliances later
of depressed adolescents. Their work has turned their attention to responding to ado-
focused on the evaluation of three clusters lescent concerns while providing some hope
of therapist engagement strategies—motiva- for change. In a complementary manner,
tional strategies that focus on goal setting therapists who opened and closed the first
and mobilizing clients’ intention to change, session with a focus on treatment socializa-
socialization strategies that focus on clarify- tion were more likely to have adolescents
ing roles and tasks in therapy, and experien- report positive alliances than therapists who
tial strategies that focus on eliciting the did not follow this pattern.
client’s experience and the provision of Finally, Jungbluth and Shirk (2009)
support. In addition, alliance-impeding uncovered an interesting pattern in their
behaviors (therapist lapses) were examined, analysis of first-session engagement behav-
including therapists’ failure to acknowledge ior and subsequent adolescent involvement
client emotion, therapists misunderstanding in CBT for depression. Therapists who
client’s statements, and therapist criticism. provided less structure in the first session
In their first study, Karver et al. (2008) were more likely to have adolescents who
examined engagement strategies in relation were highly involved in CBT tasks in later
to early alliance in a small sample of adoles- sessions than therapists who initiated ther-
cents who had attempted suicide and were apy with high structure. Low structure is
treated with either problem-solving therapy not equivalent to therapist inactivity but,
or nondirective, supportive therapy. Results rather, indicates greater exploration of ado-
showed that therapist lapses were the most lescents’ experiences and motives as well as
robust predictor of subsequent alliance greater provision of support.
across both conditions. Failure to respond Although the foregoing results are prom-
to expressed emotion was one of the most ising, they also must be considered prelimi-
characteristic problems. nary. At present only five studies have been
In a second study, Russell, Shirk, and published that address the contribution
Jungbluth (2008) examined the same set of of therapist behaviors and strategies to alli-
engagement strategies in a school-based ance formation and therapy involvement.
trial of CBT for adolescent depression. All of these studies have been conducted in
A unique feature of this study was the pre- the context of CBT, and only two child-
diction of subsequent alliance from tempo- hood disorders, anxiety and depression,
ral patterns in therapist behavior rather than have been considered. Nevertheless, some
from overall frequency of behavior. Reliable initial patterns can be discerned.
temporal patterns in therapist engagement Results with children suggest that “push-
behavior were identified. Therapists who ing” child clients, especially in a more
steadily increased their focus on being formal, didactic manner, is counterproduc-
responsive and remoralizing the adolescent tive for alliance formation. With adoles-
after the first 10 minutes of their sessions, cents there is evidence that greater attention
but who then dampened the rate of increase to the teen’s experience, especially in a
over the rest of the session, reported way that acknowledges the adolescent’s
more positive alliances two sessions later. perspective and expressions of emotion, is
These results suggest that more positive alli- associated with more positive alliance and
ances are associated with a shift in therapist involvement. Taken together, these results
suggest that a less directive and less task- First, too many studies still rely on out-
focused approach to therapy is critical at come assessments from a single source. As a
the start of treatment. Subsequent alliance result, alliance–outcome relations can be
and involvement appear to benefit from inflated by shared source variance. Second,
therapists taking the time to attend to their studies that include multiple outcome mea-
client’s experiences and life stories. Efforts to sures from multiple sources often fail to
engage the client by pushing or praising are distinguish among primary and secondary
contraindicated. This pattern is somewhat outcomes. Although it makes sense at this
inconsistent with the emphasis on psycho- early stage of research to explore a wide
education as a method of treatment social- range of outcomes, too few associations are
ization and suggests that client-centered based on specific hypotheses. It is possible
strategies at the start of therapy may be that the alliance as an interpersonal con-
more effective for alliance formation. struct will predict interpersonal outcomes
like changes in relational schema, social
Limitations of the Research engagement, or support-seeking behaviors.
As indicated in our review of definitions Third, research must evaluate the tempo-
and measures, the field has yet to reach ral relationship between alliance and out-
consensus about the crucial features of the comes. Research has revealed early gains in
therapeutic alliance with youth. We were child and adolescent therapy, yet no study
not surprised to find a large number of alli- has examined early alliance in relation to
ance measures in a relatively small number early symptom changes. This last issue is
of studies. One way to advance the field particularly important, as no study in the
would be to conduct a study in which mul- youth literature has ruled out the possibility
tiple measures of alliance are administered that alliance is actually predicted from early
in order to derive core underlying dimen- improvement. Designs that account for pos-
sions. Such an empirical approach could sible symptom changes prior to the measure
anchor future development of the alliance of alliance are clearly needed. And finally,
construct with youth. no study in the youth literature has evalu-
Our meta-analysis revealed some impor- ated the contribution of alliance to outcome
tant progress in research on alliance– while controlling for client adherence to
outcome relations with children and specific treatment tasks. Alliance may very
adolescents. First, the number of studies well predict involvement in specific therapy
has more than tripled since the first meta- components (Karver et al., 2008), but too
analysis of relationship predictors of child few studies assess alliance in relation to other
outcomes. Many new studies do not mea- important process variables.
sure alliance and outcome concurrently, as In conclusion, the alliance has a long
was found in the earlier meta-analysis, but history in the child and adolescent litera-
examine associations prospectively. And ture. Recent research progress on alliance–
there is clearly a trend toward assessing alli- outcome relations indicates that this
ance at multiple points in treatment and long-standing interest is clearly justified.
evaluating alliance trajectories in relation Alliance is a predictor of youth therapy
to outcomes. Finally, a growing number outcomes and may very well be an essential
of studies assess alliance from multiple ingredient that makes diverse child and
perspectives—client, therapist, parent, and adolescent therapies work. Future research
observer. Despite this progress, areas for on the relative contribution of alliance and
improvement remain. specific factors to youth outcomes, as well
as the contribution of alliance to child and • Although it can be tempting to try to
adolescent utilization of specific treatment connect with young clients by finding
procedures, will surely clarify its clinical shared interests and activities, initial
importance. evidence suggests that such efforts may be
counterproductive. Alliance formation
Therapeutic Practices appears to be better served by emphasizing
• Alliances with both youth and their the collaborative nature of therapy.
parents are predictive of treatment • Attending to client’s experiences and
outcomes. Consequently, psychotherapists acknowledging their expressed emotion are
need to attend to the development of crucial for alliance formation, especially
multiple alliances, not just to the alliance with adolescents. Providing an opportunity
with the youth. A solid alliance with the for client-directed interaction at the start
parent may be particularly important for of psychotherapy appears to set the stage
treatment continuation. for subsequent treatment involvement.
• Parents and youth often have divergent
views about treatment goals. Formation of
References
a therapeutic alliance with both youth and
parent requires the therapist to attend to An asterisk (∗) indicates studies included in
multiple perspectives and to develop a the meta-analysis.
∗
Adler, A. G. (1998). The alliance between child
treatment plan that accommodates both therapist and parent: How it predicts treatment
youth and parent perspectives. outcome and reflects parent characteristics.
• The maintenance of a positive alliance Unpublished dissertation.
∗
over time predicts successful outcomes with Auerbach, S., May, J., Stevens, M., & Kiesler, D.
youth. Therapists are advised to monitor (2008). The interactive role of working alli-
ance and counselor-client interpersonal behav-
alliance over the course of treatment. iors in adolescent substance abuse treatment.
Alliance formation is not simply an early International Journal of Clinical and Health
treatment task, it is a recurrent task. Psychology, 8(3), 617–629.
• Youth are likely to have a limited Axline, V. (1947). Play therapy. Boston: Houghton
understanding of therapy. Initial results Mifflin.
suggest that early alliance formation with Bickman, L., Andrade, A., Lambert, E., Doucette, A.,
Sapyta, J., Boyd A., et al. (2004). Youth thera-
youth requires the therapist to balance peutic alliance intensive treatment settings. The
active listening to the youth with Journal of Behavioral Health Services and Research.
providing an explicit framework for 31(2), 134–148.
understanding therapy processes (roles, Bordin, E. S. (1979). The generalizability of the
tasks, relevance). Overemphasizing the psychoanalytic concept of the working alliance.
latter to the exclusion of the former Psychotherapy: Theory, Research, and Practice, 16,
252–260.
appears to interfere with alliance Castro-Blanco, D., & Karver, M. (2010). The elusive
formation, at least with adolescents. alliance: Treatment engagement strategies with with
• “Pushing” youth to talk about high-risk adolescents. Washington, DC: APA Books.
∗
emotionally sensitive material appears to Champion, K. M. P. (1998). The change process
undermine alliance, at least among in therapeutic programs for children. (ProQuest
younger clients. Therapists can “scaffold” Information & Learning). Dissertation Abstracts
International: Section B: The Sciences and
young clients’ emotional disclosures by Engineering, 59, 90.
initially focusing on emotional experiences ∗
Chiu, A. W., McLeod, B. D., Har, K., & Wood, J. J.
of low intensity and gradually attending (2009). Child-therapist alliance and clinical out-
to emotions of higher intensity. comes in cognitive behavioral therapy for child
anxiety disorders. Journal of Child Psychology and Estrada, A., & Russell, R. (1999). The development
Psychiatry, 50, 751–758. of the child psychotherapy process scales (CPPS).
Chu, B., & Kendall, P. (2009). Therapist respon- Psychotherapy Research, 9(2), 154–166.
siveness to child engagement: Flexibility within Faw, L., Hogue, A., Johnson, S., Diamond, G. M.,
manual-based CBT for anxious youth. Journal & Liddle, H. A. (2005). The Adolescent
of Clinical Psychology, 65(7), 736–754. Therapeutic Alliance Scale (ATAS): Initial psy-
Christensen, M., & Skogstad, R. (2009). What chometrics and prediction of outcome in family-
predicts quality of the therapeutic alliance in a based substance abuse prevention counseling.
cognitive behavioral treatment for children with Psychotherapy Research, 15(1–2), 141–154.
anxiety disorders? Therapeutic alliance measured Fields, S., Handelsman, J., Karver, M. S., &
from the patient, therapist and observer perspec- Bickman, L. (2010). Youth and parent predictors
tive. Unpublished manuscript. Det Psykologiske of the therapeutic alliance. Manuscript submitted.
Facultet. Fitzpatrick, M., & Irannejad, S. (2008). Adolescent
∗
Colson, D.B., Cornsweet, C., Murphy, T., readiness for change and the working alliance in
O’Malley, F., Hyland, P.S., McParland, M., counseling. Journal of Counseling & Development.
Coyne, L. (1991). Perceived treatment difficulty 86(4), 438–445.
∗
and the therapeutic alliance in an adolescent Florsheim, P., Shotorbani, S., Guest-Warnick, G.,
psychiatric hospital, American Journal of Ortho- Barratt, T., & Hwang, W. (2000). Role of the
psychiatry, 61, 221–229. working alliance in the treatment of delinquent
∗
Creed, T., & Kendall, P. (2005). Therapist alliance- boys in community-based programs. Journal of
building behavior within a cognitive-behavioral Clinical Child Psychology, 29, 94–10.
treatment for anxiety in youth. Journal of Consult- Freud, A. (1946). The psychoanalytic treatment of chil-
ing and Clinical Psychology, 73(3), 498–505. dren. New York: International Universities Press.
∗
Darchuk, A. J. (2007). The role of the therapeutic Garcia, J., & Weisz, J. (2002). When youth mental
alliance and its relationship to treatment out- health care stops: Therapeutic relationship prob-
come and client motivation in an adolescent lems and other reasons for ending youth outpa-
substance abuse treatment setting. (ProQuest tient treatment. Journal of Consulting and Clinical
Information & Learning). Dissertation Abstracts Psychology, 70(2), 439–443.
International: Section B: The Sciences and Garner, B., Godley, S., & Funk, R. (2008).
Engineering, 68, 5, 195. Predictors of early therapeutic alliance among
DeVet, K., Kim, Y., Charlot-Swilley, D., & Ireys H. adolescents in substance abuse treatment. Journal
(2003). The therapeutic relationship in child of Psychoactive Drugs, 40(1) 55–65.
∗
therapy: Perspectives of children and mothers. Gavin, L. A., Wamboldt, M. Z., Sorokin, N.,
The Journal of Clinical Child & Adolescent Levy, S. Y., & Wamboldt, F. S. (1999). Treatment
Psychology, 32 (2), 277–283. alliance and its association with family func-
Diamond, G., Liddle, H., Hogue, A., & Dakof, G. tioning, adherence, and medical outcome in
(1999). Alliance-building interventions with adolescents with severe, chronic asthma. Journal
adolescents in family therapy: A process study. of Pediatric Psychology, 24(4), 355–365.
∗
Psychotherapy: Theory, Research, Practice, Training, Green, J.M. (1996) Engagement and empathy:
36(4), 355–368. A pilot study of the therapeutic alliance in out-
∗
Diamond, G.S., Liddle, H.A., Wintersteen, M.B., patient child psychiatry, Child Psychology and
Dennis, M.L., Godley, S.H., & Tims, F. (2006). Psychiatry Review, 1, 130–138.
∗
Early therapeutic alliance as a predictor of treat- Green, J., Kroll, L., Imrie, D., Frances, F.,
ment outcome for adolescent cannabis users in Begum, K., Harrison, L., et al. (2001). Health gain
outpstient treatment. The American Journal of and outcome predictors during inpatient and
Addictions, 15, 26–33. related day treatment in child and adolescent
DiGiuseppe, R., Linscott, J., & Jilton, R. (1996), psychiatry. Journal of the American Academy of
Developing the therapeutic alliance in child- Child & Adolescent Psychiatry, 40(3), 325–332.
∗
adolescent psychotherapy. Applied and Preventive Handwerk, M., Smith, G., Thompson, R.,
Psychology, 5, 85–100. Spellman, D., & Daly, D. (2008). Psychotropic
∗
Eltz, M., Shirk, S., & Sarlin, N. (1995). Alliance medication utilization at a group-home residen-
formation and treatment outcome among mal- tial facility for children and adolescents. Journal
treated adolescents. Child Abuse & Neglect, 19(4), of Child and Adolescent Psychopharmacology,
419–431. 18(5), 517–525.
Hawke, J., Hennen, J., & Gallione, P. (2005). Karver, M., Handelsman, J., Fields, S., &
Correlates of therapeutic involvement among Bickman, L. (2006). Meta-analysis of relation-
adolescents in residential drug treatment. The ship variables in youth and family therapy:
American Journal of Drug and Alcohol Abuse, Evidence for different relationship variables in
31(1), 163–177. the child and adolescent treatment literature.
∗
Hawley, K., & Garland, A. (2008). Working Clinical Psychology Review, 26, 50–65.
∗
alliance in adolescent outpatient therapy: Youth, Karver, M., Shirk, S., Handelsman, J. B., Fields, S.,
parent and therapist reports and associations Crisp, H., Gudmundsen, G., et al. (2008).
with therapy outcomes. Child and Youth Care Relationship processes in youth psychotherapy:
Forum, 37(2), 59–74. Measuring alliance, alliance-building behaviors,
Hawley, K., & Weisz, J. (2003). Child, parent and and client involvement. Journal of Emotional and
therapist (dis)agreement on target problems in Behavioral Disorders, 16(1), 15–28.
∗
outpatient therapy: The therapist’s dilemma and Kaufman, N., Rohde, P., Seeley, J., Clarke, G.,
its implications. Journal of Consulting and & Stice, E. (2005). Potential mediators of
Clinical Psychology, 71(1), 62–70. cognitive-behavioral therapy for adolescents
∗
Hawley, K. M., & Weisz, J. R. (2005). Youth versus with comorbid major depression and conduct
parent working alliance in usual clinical care: disorder. Journal of Consulting and Clinical
Distinctive association with retention, satisfac- Psychology, 73(1), 38–46.
∗
tion, and treatment. Journal of Clinical Child and Kazdin, A. E., Marciano, P. L., & Whitley, M. K.
Adolescent Psychology, 34, 117–128. (2005). Therapeutic alliance in cognitive-
Hedges, L. V., & Olkin, I. (1985). Statistical behavioral treatment of children referred for
methods for meta-analysis. New York: Academic oppositional, aggressive, and antisocial behavior.
Press. Journal of Consulting and Clinical Psychology, 73,
Hersoug, A., Monsen, J., Havik, O., & Hoglen, P. 726–730.
(2002). Quality of early working alliance in psy- Kazdin, A., & Wassell, G. (2000). Predictors of
chotherapy: Diagnoses, relationship and intra- barriers to treatment and therapeutic change in
psychic variables as predictors. Psychotherapy and outpatient therapy for antisocial children and
Psychosomatics, 71(1), 18–27. their families. Mental Health Services Research, 2,
∗
Hintikka, U., Laukkanen, E., Marttunen, M., & 27–40.
∗
Lehtonen, J. (2006). Good working alliance Kazdin, A. E., & Whitley, M. K. (2006).
and psychotherapy are associated with positive Pretreatment social relations, therapeutic alli-
changes in cognitive performance among ado- ance, and improvements in parenting prac-
lescent psychiatric inpatients. Bulletin of the tices in parent management training. Journal
Menninger Clinic, 70, 316–335. of Consulting and Clinical Psychology, 74(2),
∗
Hogue, A., Dauber, S., Stambaugh, L., Cecero, C., 346–355.
∗
& Liddle, H. (2006). Early therapeutic alli- Kazdin, A. E., Whitley, M., & Marciano, P. L.
ance and treatment outcome in individual and (2006). Child-therapist and parent-therapist
family therapy for adolescent behavior problems. alliance and therapeutic change in the treatment
Journal of Consulting and Clinical Psychology, of children referred for oppositional, aggressive,
74(1) 121–129. and antisocial behavior. Journal of Child
∗
Holmqvist, R., Hill, T., & Lang, A. (2007). Psychology and Psychiatry, 47, 436–445.
Treatment alliance in residential treatment of Kendall, P., Comer, J., Marker, C., Creed, T., Puliafico,
criminal adolescents. Child and Youth Care A., Hughes, A., et al. (2009). In-session exposure
Forum, 36(4), 163–178. tasks and therapeutic alliance across the treatment
Horvath, A. O., & Greenberg, L. S. (1989). of childhood anxiety disorders. Journal of Consulting
Development and validation of the Working and Clinical Psychology, 77(3), 517–525.
Alliance Inventory. Journal of Counseling Linscott, J., DiGiuseppe, R., & Jilton, R. (1993).
Psychology, 36(2), 223–233. A measure of TA in adolescent psychotherapy. Poster
Jungbluth, N., & Shirk, S. (2009). Therapist presented at the 101st Annual Convention of the
strategies for building involvement in cognitive– American Psychological Association, Toronto,
behavioral therapy for adolescent depression. Canada.
Journal of Consulting and Clinical Psychology, Mallinckrodt, B., Coble, H., & Gantt, D. (1995).
77(6), 1179–1184. Working alliance, attachment memories, and
social competencies of women in brief therapy. ∗
Smith, B., Duffee, D., Steinke, C., Haung, Y., &
Journal of Counseling Psychology, 42(1), 79–84. Larkin, H. (2008). Outcomes in residential
∗
McLeod, B., & Weisz, J. (2005). The therapy pro- treatment for youth: The role of early engage-
cess observational coding system-alliance scale: ment. Children and Youth Services Review, 30(12),
Measure characteristics and prediction of outcome 1425–1436.
in usual clinical practice. Journal of Consulting and Spielman, G., Gatlin, E., & McFall, J. (2010). The
Clinical Psychology, 73(3), 323–333. efficacy of evidence-based psychotherapy versus
Meeks, J. (1971). The fragile alliance: An orientation usual care for youths: Controlling confounds
to the outpatient psychotherapy of the adolescent. in a meta-analysis. Psychotherapy Research, 20,
Oxford England: Williams & Wilkins. 234–246.
Miller, S., Wampold, B., & Varhely, K. (2008). ∗
Tetzlaff, B. T., Kahn, J. H., Godley, S. H.,
Direct comparisons of treatment modalities for Godley, M. D., Diamond, G. S., & Funk, R. R.
youth disorders: A meta-analysis. Psychotherapy (2005). Working alliance, treatment satisfaction,
Research, 18(1), 5–14. and patterns of post-treatment use among ado-
Rogers, C. (1957). The necessary and sufficient lescent substance users. Psychology of Addictive
conditions for therapeutic personality change. Behavior, 19, 199–207.
Journal of Consulting Psychology, 21, 95–103. Weiss, B., & Weisz, J. (1995). Relative effective-
Russell, R., Shirk, S., & Jungbluth, N. (2008). ness of behavioral versus nonbehavioral child
First-session pathways to the working alliance psychotherapy. Journal of Consulting and Clinical
in cognitive-behavioral therapy for adolescent Psychology, 63(2), 317–320.
depression. Psychotherapy Research, 18(1), 15–27. Weisz, J., Jensen-Doss, A., & Hawley, K. (2006).
Sandler, J., Kennedy, H., & Tyson, R. (1980). The Evidence-based youth psychotherapy versus
techniques of child psychoanalysis: Discussions with usual clinical care: A meta-analysis of direct com-
Anna Freud. Cambridge, MA: Harvard University parisons. American Psychologist, 61, 671–689.
Press. Weisz, J., Weiss, B., Han, S., Granger, D., &
Shirk, S. (Ed.). (1988). Cognitive development and Morton, T. (1995). Effects of psychotherapy
child psychotherapy. New York: Plenum. with children and adolescents revisited: A meta-
∗
Shirk, S., Gudmundsen, G., Kaplinski, H., & analysis of treatment outcome studies. Psycho-
McMakin, D. (2008). Alliance and outcome logical Bulletin, 117(3), 450–468.
in cognitive-behavioral therapy for adolescent Weisz, J. R., Weiss, B., Alicke, M. D., & Klotz, M. L.
depression. Journal of Clinical Child and (1987). Effectiveness of psychotherapy with
Adolescent Psychology, 37(3), 631–639. children and adolescents: A meta-analysis for
Shirk, S., & Karver, M. (2003). Prediction of clinicians. Journal of Consulting and Clinical
treatment outcome from relationship variables Psychology, 55, 542–549.
in child and adolescent therapy: A meta-analytic Wintersteen, M., Mensinger, J., & Diamond, G.
review. Journal of Consulting and Clinical (2005). Do gender and racial differences between
Psychology, 71, 462–471. patient and therapist affect therapeutic alli-
Shirk, S., & Karver, M. (2006). Process issues ance and treatment retention in adolescents?
in cognitive-behavioral therapy for youth. In Professional Psychology: Research and Practice,
P. C. Kendall (Ed.), Child and adolescent therapy: 36(4), 400–408.
Cognitive-behavioral procedures (pp. 465–491). Wright, L., Everett, F., & Roisman, L. (1986).
New York: Guilford Press. Experiential psychotherapy with children. Baltimore:
Shirk, S., & Russell, R. L. (1996). Change processes Johns Hopkins University Press.
in child psychotherapy: Revitalizing treatment and Zack, S., Castonguay, L., & Boswell, J. (2007).
research. New York: Guilford Press. Youth working alliance: A core clinical construct
Shirk, S., & Russell, R. L. (1998). Process issues in in need of empirical maturity. Harvard Review of
child psychotherapy. In A. Bellack & M. Hersen Psychiatry, 15, 278–285.
(Eds.), Comprehensive clinical psychology. London: ∗
Zaitsoff, S., Doyle, A., Hoste, R., & le Grange, D.
Pergamon. (2008). How do adolescents with bulimia
Shirk, S., & Saiz, C. (1992). Clinical, empirical, nervosa rate the acceptability and therapeu-
and developmental perspectives on the thera- tic relationship in family-based treatment?
peutic relationship in child psychotherapy. International Journal of Eating Disorders, 41(5),
Development and Psychopathology, 4, 713–728. 390–398.
C HA P TER
4 Alliance in Couple and Family Therapy
Myrna L. Friedlander, Valentín Escudero, Laurie Heatherington, and Gary M. Diamond
Although the salience of the working of our original meta-analysis of the CFT
alliance in couple and family therapy (CFT) alliance outcome studies published through
was recognized over 20 years ago, it has 2008. We summarize the literature on
received far less theoretical and empirical moderators, mediators, and client contri-
attention than has the alliance in individual butions to CFT alliances, discuss the limi-
psychotherapy. In their seminal work on tations of the research, and conclude with
CFT alliances, Pinsof and Catherall (1986; recommended clinical practices on the basis
Catherall, 1984) took Bordin’s (1979) of the meta-analysis.
conceptualization of the alliance as a point
of departure and applied the goal, task, and Definitions and Measures
bond constructs to three interpersonal Creating and sustaining CFT alliances are
facets of the alliance in family treatment complicated by the fact that family members
(self-with-therapist, other-with-therapist, often seek psychotherapy as a last resort,
and group-with-therapist). The rationale when the conflicts among them seem irre-
was that not only do family members vary concilable. Moreover, it is common for
in the degree to which they form a personal family members to have different motives
bond and agree with the therapist about and motivational levels for treatment, and
treatment goals and tasks, but also each sometimes disagree about whether there
person observes, can report on, and is influ- even exists a problem that requires profes-
enced by how others in the family feel sional attention. Even when a problem is
about the therapy and by how the couple jointly acknowledged, for example, “We
or family unit as a whole is responding to fight all the time,” therapy may not be seen
what is taking place in treatment (Pinsof & as the solution, or individuals’ goals may
Catherall, 1986). Thus, from its first differ (“You need to stop drinking” versus
introduction into the literature, the CFT “We need to be a couple—it’s like we’re
alliance was described as unique, complex, living parallel lives”) (Friedlander, Escudero,
and multilayered. It is no longer in question & Heatherington, 2006; Lambert, Skinner,
that, as a treatment format, CFT demands & Friedlander, in press). Consequently,
a unique conceptualization of the alliance. family members’ willingness to engage in
In this chapter we define CFT alliances, various therapy tasks may also differ (“Why
summarize the major observational and should we argue about my drinking if you
self-report measures, and offer an extended don’t even want to stay married to me?”).
clinical example. We then report the results Indeed, it is not uncommon for one partner
92
or family member to feel like a therapy (Friedlander, 2000). Breaches of safety can
hostage (“Come to therapy with me or severely undermine a client’s trust in the
else …”) or expect the therapist to take therapist and the therapeutic process.
sides, particularly if the problem is defined Moreover, the degree of safety felt by family
in zero-sum terms (e.g., to divorce or not, members can change as new problems are
to have a baby or not, to relocate or not) revealed and explored and as different
(Friedlander, Escudero, & Heatherington, family members join or leave treatment
2006). (Beck, Friedlander, & Escudero, 2006).
CFT alliances develop simultaneously on What feels safe to the children when only
an individual level (self-with-therapist) and their father is there, for example, might
a group level (group-with-therapist). Just as feel quite unsafe when their stepmother is
in individual therapy, alliances in CFT present. Likewise, it may seem safer in
involve the creation of a strong emotional couples therapy to discuss conflicts over
bond as well as negotiation of goals and parenting than to explore expectations
tasks with the therapist. A unique charac- about intimacy or sexuality. In CFT, the
teristic of CFT, though, is that at any point conjoint nature of the treatment and ever-
in treatment there are multiple alliances changing composition of sessions makes
that interact systemically. For example, the creating a safe environment both compli-
degree to which a mother likes the psycho- cated and critical.
therapist and is engaged in the treatment A related construct is the group aspect of
may have a facilitating (or hindering) effect the alliance, which has alternately been
on her son’s willingness to trust the thera- conceptualized as allegiance (Symonds &
pist. The son’s involvement also depends on Horvath, 2004), within-family alliance
the mother–son bond and whether he agrees (Pinsof, 1994), and shared sense of purpose
with his mother about the nature of the (Friedlander, Escudero, & Heatherington,
problems, goals, or need for treatment. 2006). A complex part of the conjoint
Moreover, the family members’ degree of therapy process, this “we-ness” refers not
comfort with one another affects each only to a willingness to collaborate in treat-
person’s willingness to negotiate goals with ment but also to a strong emotional bond
the others and with the therapist. In other between and among family members. In
words, every individual simultaneously other words, the within-family alliance has
creates a personal alliance with the thera- more to do with family members’ thoughts,
pist, and each person’s alliance with the feelings, and behavior toward one another
therapist can negatively or positively affect than it does with any one person’s alliance
the others’. with the therapist considered in isolation.
An important aspect of CFT alliances is Moreover, the within-family alliance
the degree to which family members feel develops simultaneously and in interaction
safe and comfortable with each other in with all of the individual alliances. Indeed,
the therapeutic context. The revelation of research shows that family members often
secrets and in-session exploration of conflicts see their personal relationships with the
are not easily left behind in the consulting therapist differently from their allegiance
room at the end of the session. Family with each other (e.g., Beck et al., 2006;
members go home together, and therapy Friedlander, Lambert, Escudero, & Cragun,
can only progress if they feel that the 2008; Lambert et al., in press). For this
material discussed in-session is not used reason, a complete picture of the alliance
against them during the course of the week requires some accounting of how well the
f r i e d l a n d e r, e s c u d e ro , h e at h e r i n g to n , d i a m o n d 93
family works together in therapy as well (Pinsof, Zinbarg, & Knobloch-Fedders,
as how similarly individuals feel about the 2008) only has 12 items in 3 subscales: Self/
therapist. Group, Other, and Within. Self and Group
When alliances are “split” (Heatherington were combined because these subscales
& Friedlander, 1990; Pinsof & Catherall, were indistinguishable both statistically and
1986) or “unbalanced” (Robbins, Turner, experientially, and factor analyses did not
Alexander, & Perez, 2003), at least one support the independence of the original
family member has a stronger bond with the Goals, Tasks, and Bonds subscales.
therapist than do other family members. Only one measure of the alliance
There is ample evidence that, in both cou- includes the element of safety. In the System
ples and family therapy, split alliances occur for Observing Family Therapy Alliances
frequently and vary in severity (Heatherington (SOFTA; Friedlander, Escudero, &
& Friedlander, 1990; Mamodhoussen, Heatherington, 2006; Friedlander, Escudero,
Wright, Tremblay, & Poitras-Wright, 2005; Horvath et al., 2006) or Sistema de la
Muñiz de la Peña, Friedlander, & Escudero, Observación de la Alianza en Terapia Familiar
2009). In family therapy, although we might (SOATIF; Escudero & Friedlander, 2003),
expect parents to feel a greater connection Safety is one of four alliance dimensions.
with the therapist than do their adolescents, In brief, Safety within the Therapeutic
and indeed such a split alliance pattern has System reflects each client’s degree of comfort
been found in several studies, in at least one taking risks, being vulnerable, and exploring
study several of the adolescents felt closer to conflicts with a therapist and other family
the therapist than did their parents (Muñiz members, Engagement in the Therapeutic
de la Peña et al., 2009). Although severely Process reflects Bordin’s (1979) agreement
split alliances often lead to premature termi- with the therapist on tasks and goals,
nation, this is not invariably the case (Muñiz Emotional Connection with the Therapist is
de la Peña et al., 2009), and splits may occur similar to Bordin’s concept of client–therapist
in reverse as new topics or are explored or bond, and Shared Sense of Purpose within the
secrets are revealed. Family refers to productive family collabora-
To assess CFT alliances, the most widely tion (the within-family alliance).
used self-report measures are the Couple The pantheoretical SOFTA contains
Therapy Alliance Scale (CTAS; Pinsof & observational (SOFTA-o; Friedlander,
Catherall, 1986) and the Family Therapy Escudero, & Heatherington, 2006;
Alliance Scale (FTAS), which were revised Friedlander, Escudero, Horvath, et al., 2006;
to include the Within Alliance, “My part- Friedlander, Lambert, & Muñiz de la Peña,
ner and I….” (couple) and “Some of the 2008) and 16-item self-report (SOFTA-s;
other members of my family and I …” Friedlander, Escudero, & Heatherington,
(family). Like the couple version of the 2006; Friedlander, Lambert et al., 2008;
Working Alliance Inventory (WAI-Co; Lambert & Friedlander, 2008) measures
Symonds, 1999), the CTAS and FTAS from both client and therapist perspectives.
reflect Bordin’s (1979) concept of goals, Whereas the client version reflects the
tasks, and bonds. However, unlike the strength of the alliance, the therapist version
63-item WAI-Co, which also has a thera- assesses alliance-related interventions. Using
pist version and yields a total alliance score the SOFTA-o, trained raters observe a vid-
as well as 9 subscale scores (Goals, Tasks, eotaped or live session, tallying the frequency
and Bonds crossed with Self, Partner, and of specific positive behaviors, for example,
Group), the shortened CTAS-r and FTAS-r “Client introduces a problem for discussion”
(Engagement), and negative behaviors, for Because it was obvious that conjoint ses-
example, “Family members try to align with sions with the four family members would
the therapist against each other” (Shared not be productive at this stage of treatment,
Purpose), and using these tallies to make a the therapist proposed holding the next
global rating for each alliance dimension. To two sessions with the teens and the parents
date there has only been one published study separately. Indeed, this seemed to be the
with the therapist version (Friedlander, only arrangement that could provide even a
Lambert et al., 2008), a comparison of good modest amount of safety. In the children’s
and poor outcome cases with the same session, the boy willingly expressed concern
therapist. about his sister, but he cavalierly dismissed
CFT alliances have also been studied his own problems. For her part, the girl
using the Vanderbilt Therapeutic Alliance denied being anxious or having eating
Scale (Hartley & Strupp, 1983), an obser- problems and, instead, complained about
vational measure developed for individual how her brother constantly annoyed her.
psychotherapy and subsequently revised for As siblings, they collaborated minimally,
CFT (VTAS-R; Diamond, Liddle, Dakof, each one only willing to talk about the oth-
& Hogue, 1996). Recently, the 26 Patient er’s problems. When asked about relations
Contribution items in the VTAS-R were with their parents, both teens remained
pared down through factor and item analy- silent. Finally, the son asked if the therapist
ses to a 5-item scale (Shelef & Diamond, thought he could help their parents, but
2008) that includes three client behaviors the boy refused to clarify the basis for this
reflecting bonds and tasks and two client + request. Notably, the alliance seemed split:
therapist behaviors reflecting goals and The boy was visibly more involved and con-
tasks. The VTAS-R requires raters to nected with the (male) therapist than was
provide global judgments of each client’s his sister, whose sense of safety appeared to
overall behavior and therapist–client inter- be quite fragile.
actions across an entire session, for example, In their conjoint session, mother and
“To what extent did the patient indicate father demonstrated even greater unease
that she experiences the therapist as under- with each other. Not only did they not
standing and supporting her?” (Shelef & make eye contact or confer with one
Diamond, 2008, p. 439). another, but they also sat on either end of
the couch, their bodies positioned in oppos-
Case Example ing directions. Although both parents
A middle-aged couple brought their cooperatively described the children’s prob-
reluctant 15-year-old son and 13-year-old lems, they refused to discuss their own
daughter to psychotherapy. The girl, who relationship. Finally, the husband haltingly
exhibited anxiety and an eating disorder explained that “after something happened,”
(only at home), refused to speak in the he and his wife had agreed that the marriage
session, as did the boy, who had vandalized was finished. This “emotional divorce” was
a neighbor’s car and was failing in school. unknown to the children, however. Because
While the parents barely glanced at each neither parent was willing to leave the
other, both adamantly insisted that their home, they planned to continue living
children were in desperate need of help. together until both children grew up and
Thus ended the first session, which clearly moved out. After that, they would separate.
evidenced a lack of safety all around and an Neither client was willing to consider
exceedingly poor within-family alliance. couples work, as they were in agreement
that their marriage was a “lost cause.” They to keep the family together.” For the first
were, however, willing to come for sessions time, the spouses looked at and spoke
if it would help their children. directly to one another. When the husband
Interestingly, within each subsystem made a joke that his wife smiled at, the ther-
there was a shared sense of purpose, at least apist commented that they both seemed to
about why they would continue coming to be experiencing “deep hurts” that they were
therapy: The children agreed to be seen so afraid to express. He said that he wanted it
that the therapist would help their parents, to “see if there was another way for [their
and the parents agreed to come in order to marital] relationship to improve, if only to
help their children. keep on helping the children.”
Given this curious arrangement and The mother, who seemed to trust the
everyone’s clear fear of taking emotional therapist a great deal, admitted thinking
risks, the therapist continued to see each that the children were “reacting” to the
subsystem weekly. He found it relatively emotional divorce. Moving closer to her,
easier to develop a personal bond with the the therapist softly commented that, “As a
son and the mother. By working hard to parent myself, I know it’s extremely hard to
enhance his connection with the daughter realize that something I’ve done has hurt
and father, the therapist gradually became a my children.” The mother responded with
trusted figure, and slowly everyone began tears, and at the end of the session admitted
to engage more freely in the therapeutic that the children “deserved to be told” about
work. Because the adolescents adamantly the status of the marriage. The husband
refused to acknowledge their own problems agreed, albeit reluctantly.
and were clearly protecting their parents— The parents chose to reveal the secret at
never criticizing them or even acknow- home rather than in a family session. In their
ledging their parents’ overt hostility—the next session (alone) with the therapist, the
therapist focused solely on improving the teens no longer felt the need to protect
sibling bond. In one homework assign- their parents. They responded positively to
ment, for example, the girl was asked to the therapist’s empathic response to their
choose a set of digital family photos that expressions of resentment and sadness. When
held happy memories for her, and her the daughter burst out, “But we’re not a real
brother was asked to arrange these pictures family!,” the therapist replied by proposing a
into a slideshow that he would set to new shared sense of purpose, in other words,
music. a common goal: “I disagree. Both of your
As brother and sister began fighting less parents care for you and want the best for
at home and cautiously started to enjoy you, and both of you feel the same for your
each other’s company, both parents began parents. I’m sure you can learn to work
to trust the therapist more. However, they together so that everyone has a happier life.”
rarely looked at one another in session, and Over the next month, the teens pushed
the chasm between them remained as deep their parents into committing to couples
as ever. therapy, with the goal of either working out
Alone with the parents in Week 5, the their differences or deciding to separate.
therapist made some strategic moves. Although the son’s grades in school did not
Focusing first on the within-system alliance, improve substantially, he had no further
he praised their mutual dedication to their delinquent offenses. The daughter remained
children, pointing out how they were both highly stressed but ate normally and began
willing to “sacrifice [their] personal happiness spending more time with friends.
This case illustrates how an alliance- Seven of the 24 were studies of couples
empowering approach can potentially therapy (2 of which were conducted in
repair seriously broken within-family groups), and the remaining 17 were family
attachments. By strategically focusing on studies in which at least a portion of the
different alliances and different aspects of treatment was conducted conjointly. The
each alliance, this therapist moved a stalled total number of clients who participated in
treatment forward. He began by separating these 24 studies is 1,461. Studies examined
the parents and children to enhance safety both treatment as usual as well as specifically
and negotiate different problem definitions defined approaches, including cognitive-
and goals within each subsystem. With the behavioral therapy, functional family
adolescents, the therapist relied heavily on therapy, family-based therapy, systemic
five interventions that have been shown to and ecosystemic-oriented therapy, emotion-
improve poor alliances with teens (Diamond focused therapy, integrative problem-
et al., 1999): He emphasized trust, honesty centered therapy, multidimensional family
and confidentiality; he explained the impor- therapy, brief strategic family therapy,
tance of collaborating in therapy; he defined parent management training, and psycho-
personally meaningful goals for each child; educative family therapy. Most of the treat-
and, most importantly, he presented him- ments were no more than 20 sessions in
self as an ally in the one thing the children length, and the majority (65%) described
agreed on—helping their parents. Then, by the therapy as manualized treatment,
encouraging engagement in therapy tasks although only a few studies provided infor-
through his personal bond with each family mation about treatment integrity.
member, the therapist eventually redefined The problems targeted in these treat-
the family’s problem and the treatment ments ranged from parent–adolescent
goals in a way that was both respectful and communication difficulties to substance
challenging. The success of this process abuse, child abuse or neglect, and schizo-
goal, to create a within-family shared sense phrenia. Some studies were highly specific
of purpose, seemed largely due to the thera- in defining the presenting problems or
pist’s attending to and emphasizing the disorders, but many other studies identi-
strong parent–child bonds. fied the clients’ problems in a general way.
The samples in the more naturalistic studies
Meta-Analysis tended to have a variety of presenting
Table 4.1 summarizes 24 studies in which issues.
CFT alliances, self-reported and observed, The instruments and methods used to
were used to predict treatment retention, evaluate outcome reflect the variability in
improvement midtreatment, and/or final problems treated. In terms of measures,
outcomes. To obtain as comprehensive a roughly 50% of the studies used an obser-
sample as possible, we cross-referenced vational methodology. Most evaluated the
articles known to us and searched electron- alliance early in the therapy, and only a few
ically (PsycINFO, PubMed, Social Sciences studies assessed the alliance at different
Citation Index) for additional alliance- stages of treatment (early, middle, and
related studies in CFT. Unpublished late). The observational instruments were
dissertations were excluded, as were analog primarily the SOFTA-o and the VTAS; the
studies. Only articles published in English WAI and CTAS/FTAS were most often
with validated measures of the alliance were used to measure self-reported alliance.
included. Five of the 24 studies did not measure
Table 4.1 Summary of Alliance Outcome Studies in CFT
Study Treatment Alliance Outcome Overall
effect size
Therapy Format Rater Measure Time Measure Wt. Ave.
model Ave. r N
Bourgeois et al. (1990) CSP Couples C, T CTAS E Dyadic Adjustment 0.43 63
group Scale, Marital Happiness
Scale, Potential Problem
Checklist
Brown & O’Leary (2000) CBT, PE Couples O WAI E Psychological 0.53 70
group Maltreatment of Women
Scale, Modified Conflict
Tactics Scale
Escudero et al. (2008) BFT Family O SOFTA E, L Perceived therapeutic 0.22 68
improvement
Flicker et al. (2008) FFT Family O VTAS-R E Completion vs. dropout 0.25 86
Friedlander et al. (2008) FS Family O SOFTA E Perceived improvement- 0.35 33
so-far
Greenberg et al. (2010) EFT Couple C CTAS E Enright Forgiveness 0.35 40
Inventory
Hawley & Garland (2008) FS Family C, T WAI E, L Youth Symptom 0.33 36.7
Self-Report
Hogue et al. (2006) MDFT Family O VTAS-R E Child Behavior 0.05 44
Checklist (internalizing,
externalizing), Timeline
Follow-Back interview
for substance use
Johnson et al. (2006) EcoS Family, C FTAS L Outcome 0.46 32
home (tasks) Questionnaire,
based Inventory of Parent
and Peer Attachment
Johnson & Ketring (2006) EcoS Family, C FTAS L Outcome Questionnaire 0.10 430
home (bond) 45.2 -Symptom Distress
based subscale, Conflict Tactics
Scale - Physical Aggression
Subscale
Johnson & Talitman EFT Couple C CTAS E Dyadic Adjustment 0.54 23
(1997) Scale, Miller Social
Intimacy Scale
Kazdin et al. (2005) PE Family T, C WAI, E, L Treatment Improvement 0.32 49
CTAS Scale, Marital Satisfaction
Scale
Knobloch et al. (2007) IPCT Couple C CTAS E, L Marital Satisfaction 0.39 37
Inventory Revised
Pereira et al. (2006) FBT Family O WAI E, L Eating Disorders 0.32 31.4
Examination
(Continued )
98
Table 4.1 Continued
Study Treatment Alliance Outcome Overall
effect size
Therapy Format Rater Measure Time Measure Wt. Ave.
model Ave. r N
Pinsof et al. (2008) IPCT Couple C CTAS-R E, L COMPASS Treatment 0.31 80
Assessment System,
Marital Satisfaction
Inventory Revised
Quinn et al. (1997) FS Family C FTAS E Goal achievement 0.53 19
and expectation of
maintenance
Raytek et al. (1999) CBT Marital O VTAS-R E Attrition status 0.37 66
(completers, partial
completers, vs. early
dropouts)
Robbins et al. (2003) FFT Family O VTAS-R E Completion vs. dropout 0.29 34
Robbins et al. (2006) MDFT Family O VTAS-R E Completion vs. dropout 0.35 30
Robbins et al. (2008) BSFT Family O VTAS-R E Completion vs. dropout 0.36 31
Shelef et al. (2005) MDFT Family C, O WAI E Global Appraisal 0.24 59
of Individual Needs,
Substance Problem
Index
Shelef & Diamond MDFT Family O VTAS-R E, L Completion vs. dropout, 0.41 45
(2008) days of cannabis use
Smerud & Rosenfarb PE Family O SOFTA L Brief Psychiatric Rating 0.54 28
(2008) Scale, Social Adjustment
Scale-II, days until first
rehospitalization, days
until first use of rescue
medication, Patient
Rejection Scale
Symonds & Horvath NS Couple C WAI-Co E Marital Satisfaction 0.37 22.4
(2004) Scale (female)
Note: N refers to the average sample size for the various correlations used in the within-study meta-analysis, i.e., not the N in the entire study.
CSP = Couple Survival Program; CBT = cognitive-behavioral therapy; PE = psycho-educative; BFT = brief family therapy; FFT = functional
family therapy; FS = family systems; EFT = emotion-focused therapy; MDFT = multidimensional family therapy; EcoS = ecosystemic therapy;
FBT = family-based therapy; IPCT= integrative problem-centered therapy; BMT = behavioral marital therapy; BSFT = brief strategic family
therapy; NS = not specified. C = client self-report; T = therapist self-report; O= external observer. CTAS = Couple Therapy Alliance Scale
(Pinsof & Catherall, 1986); CTAS-r = Couple Therapy Alliance Scale-Revised (Pinsof et al., 2008); FTAS = Family Therapy Alliance Scale
(Pinsof & Catherall, 1986); SOFTA-o: System for Observing Family Therapy Alliances - observer (Friedlander, Escudero, & Heatherington,
2006); VTAS-R = Vanderbilt Therapeutic Alliance Scale - Revised (Diamond et al., 1996); WAI = Working Alliance Inventory (Horvath &
Greenberg, 1986); WAI-Co = Working Alliance Inventory - couple (Symonds & Horvath, 2004). E = early; L = late.
client outcome but, rather, only explored and Weinberger (2009), which are based on
associations between alliance and treatment Hunter and Schmidt’s (1990) random-
retention. effects approach. For studies that reported
For the meta-analysis of correlation coef- statistics other than correlation coefficients
ficients, we used the recommendations and (e.g., a t test to compare the alliance in
computation program of Diener, Hilsenroth, families that completed therapy versus those
that dropped out), we calculated the corre- differences in the designs, measures, treat-
sponding conversions to r (using Diener ment formats, and problems treated in the
et al.’s (2009) computation program). group of 24 studies made it essential to test
Because of the complex structure of for variables that may moderate the average
alliance in family therapy (multiple partici- effect size. We performed a chi-square test
pants generating multiple levels of analysis), to examine the homogeneity of various
most of the 24 studies reported more than subsamples.
a single alliance–outcome correlation. In The resulting χ2 = 43.52, p < 0.005,
these cases, we calculated a meta-analytic indicated that the null hypothesis of homo-
statistic within each study in order to geneity was rejected. In other words, there
maintain statistical assumption of indepen- was unaccounted-for variability among the
dence. These calculations were conducted effects produced by the various studies.
in the same way as the meta-analytic calcu- Consequently, we explored various moder-
lations for the entire group of 24 studies. ators that might explain the heterogeneity.
That is, the effect sizes listed in the far right
column of Table 4.1 were calculated from Moderators and Mediators
aggregated correlations within each study We computed a series of moderator analy-
(Diener et al., 2009; Hunter & Schmidt, ses with various subsamples. First, analysis
1990). We also calculated the weighted of the 17 family studies showed a similar
average effect size of the 24 studies, mini- average weighted effect size (r = 0.24;
mizing sampling error by weighting each z = 6.55, p < 0.005), and the null hypothesis
study by its sample size. of homogeneity was rejected (χ2 = 27.77,
The resulting weighted average effect size p < 0.05). In other words, the group of
for the 24 studies was r = 0.26. The standard- studies with a family therapy format had
ized normal test for determining whether an significant unexplained variability in the
aggregate r is statistically significant yielded relation between alliance and outcome.
z = 8.13, which is sufficiently large to reject (The subsample of seven couple studies was
the null hypothesis (p < 0.005). The upper too small to test for either effect size or
and lower limits of the weighted average homogeneity.)
effect size calculated for a 95% confidence Second, we analyzed the 13 studies that
interval were 0.33 and 0.20, respectively. used observational (rather than self-report)
These results indicate that the association measures of the alliance. This result showed
between alliance and outcome in couple a somewhat higher global effect size
and family therapy was statistically signifi- (weighted average r = 0.33; z = 8.85,
cant. According to conventional bench- p < 0.001), and the variability in these stud-
marks, an r of 0.26 (d = 0.53), which is a ies was not large enough to reject the null
medium effect size in the behavioral hypothesis of homogeneity (χ2 = 16.48, ns).
sciences, is quite similar to the r = 0.275 By contrast, the 11 studies that used self-
reported in Chapter 2 of this book for reported alliances had a slightly lower effect
the alliance–outcome relation in individual size (weighted average r = 0.22; z = 4.46,
therapy. p < 0.01). In this analysis, the homoge-
In addition to statistical significance, neity hypothesis was rejected (χ2 = 22.39,
the meta-analytic results demonstrated that p < 0.02), possibly due to the diversity of
alliance accounted for a substantial propor- perspectives (parents’, children’s, partners’
tion of variance in CFT retention and/or and therapists’) in self-report studies of
outcome. At the same time, evident alliance.
Taken together, these results mean that permutations (mother–child, father–
in every subsample tested, the alliance in mother, etc.). For example, with adoles-
CFT accounted for a substantial proportion cents who have externalizing problems
of variance in treatment retention and/or (Shelef & Diamond, 2008) or anorexia
outcome. However, the most homogeneous (Pereira, Lock, & Oggins, 2006), research
studies were the observational ones, likely has found that the parents’ (but not the
due to the fact that trained external observ- youths’) alliances predicted treatment
ers use well-defined behavioral criteria to completion. Further, in other studies with
assess the alliance. By contrast, in the family externalizing adolescents (Robbins et al.,
studies and in the studies that relied on self- 2006, 2008), both the children’s and the
report to assess alliance, the alliance–outcome parents’ alliances with the therapist discrim-
relation was more variable. inated dropout from completer families.
These divergences in effect sizes across Unbalanced alliances tend to be negatively
the outcome studies, as well as complex related to retention, and this relation is also
findings within these studies, raise ques- moderated in complex ways by family role.
tions about the circumstances under which Unbalanced father-adolescent alliances
the alliance figures more or less strongly in (where the first person in the pair is the one
outcomes. Because very few direct tests of with the higher alliance) discriminated
moderators and mediators have been dropout from completer families in
conducted, a meta-analysis was not possible. functional family therapy (FFT; Robbins,
In this section, we summarize what is Turner, Alexander, & Perez, 2003), and
known and what has been suggested about increasingly unbalanced mother–adolescent
moderators and mediators of the alliance– alliances and unbalanced mother–father
outcome relation in CFT. alliances characterized families who
dropped out of brief strategic family therapy
Alliance and Treatment Retention (Robbins et al., 2008). Further, unbalanced
Good outcomes depend on treatment parent–adolescent alliances in FFT discrim-
attendance, and retention in family therapy inated dropout from completer Hispanic
is particularly challenging, as any one families but not Anglo families (Flicker,
person’s strong negative feelings can lead to Turner et al., 2008).
premature termination for the entire family. Therapist experience has not been system-
For this reason there has been significant atically manipulated in any study, although
work in CFT on strategies for engaging and experience did differ across studies, prompt-
retaining families, especially families with ing some thoughtful speculation (Flicker &
drug-using adolescents (cf. Liddle et al., Turner, 2008) about how it might account
1992; Szapoznick et al., 1988). for differing results. Therapist experience
Regarding retention, the only clear was positively associated with the alliance
moderator is family role—parent, spouse, (as measured by observer ratings of thera-
child. First, we note that the composite pist behavior and errors in technique) in
index of CFT alliance, that is, an average conjoint alcoholism treatment for couples
of all family members’ alliances, is not (Raytek, McCready, Epstein, & Hirsch,
predictive of retention (or outcome). 1999). A qualitative analysis revealed that
Rather, more nuanced indices of alliance experienced therapists were relatively more
matter: (a) the interplay of each individual active, more responsive to topics initiated
family member’s alliance with the therapist, by clients, more flexible in following man-
and (b) unbalanced alliances, in various ualized treatment guidelines, and better
at managing the couples’ negativity. The parent measures did not. Moreover, adoles-
authors suggested that such responsiveness cent alliance predicted outcome only in
and flexibility strengthened the emotional cases in which the parent–therapist alliance
bond with the therapist and the clients’ was moderate to strong. In a study of
involvement in treatment, which in turn outpatient psychotherapy “as usual” that
facilitated retention in therapy. combined individual and conjoint parent–
teen sessions, youths’ alliance predicted
Couples Therapy Outcomes outcomes (youth symptom improvement,
In general, with respect to gender, the man’s family functioning) as reported by all family
alliance tends to be more strongly associ- members, whereas parents’ alliance predicted
ated with outcome in both group marital fewer outcomes and only their own (i.e.,
therapy (Bourgeois, Sabourin, Wright, not their childrens’) ratings of treatment
1990; Brown & O’Leary, 2000) and success (Hawley & Garland 2008).
couples therapy (Symonds & Horvath, Interestingly, there is no evidence that
2004). Less frequently, the woman’s therapist gender, race/ethnicity, or thera-
alliance is the stronger predictor of out- pist–family ethnic match are significant
come (Knobloch-Fedders, Pinsof, & Mann, factors in the strength of alliance. Nor have
2007). Explanations for the gender differ- they been found to moderate the CFT
ence focus on the documented greater alliance–outcome relation.
reluctance of men to engage in treatment, Type of treatment may moderate the
as well as their relative power in some alliance–outcome relation, given the differ-
couples (especially where there is abuse), ences in findings across studies that
and women’s relatively higher commitment employed different kinds of treatment. In a
and “ability to work toward positive study of behavioral family management
outcomes regardless of the relative strength treatment for schizophrenia (Smerud &
of their relationship with the therapist” Rosenfarb, 2008), only the relatives’
(Symonds & Horvath, 2004, p. 453). observed alliances predicted the patient’s
reoccurrence of symptoms, a finding that
Family Therapy Outcomes underscores the importance of family
In outcome studies of conjoint family environment in preventing relapse in major
therapy, family role emerged as the most mental illness. Interestingly, patients’
consistent (albeit complex) potential mod- alliance predicted less rejection by relatives
erator; its effects vary depending on the and less care burden, suggesting that
measures used and the treatment adminis- alliances in one subsystem may have posi-
tered. A study of family treatment for tive effects on others. Another study (Hogue
anorexia nervosa (Pereira et al., 2006), et al., 2006) that compared cognitive-
for example, found that adolescents’ (but behavior therapy (CBT) and multidimen-
not parents’) observed alliance with the sional family therapy (MDFT) for adolescent
therapist predicted early weight gain, whereas substance abuse found that in CBT, the
parents’ alliance later in therapy was associ- adolescents’ alliance was not associated with
ated with teens’ overall weight gain. outcome, whereas in MDFT both the youths’
Similarly, in a study of family treatment and the parents’ alliances were associated
for adolescent substance abuse (Shelef et al., with outcomes, albeit in different ways.
2005), observer measures (but not self- Finally, in a study of home-based family
report measures) of adolescents’ alliance therapy, the youths’ attachment, as measured
predicted posttreatment outcomes, whereas by trust in each parent, in tandem with the
“tasks” dimension of the alliance, predicted the alliance, just as some authors have
symptom reduction (Johnson, Ketring, claimed that all therapeutic interventions
Rohacs, & Brewer, 2006). The quality of are indistinguishable from alliance building
parents’ attachment relationships with their and maintenance (e.g., Hatcher & Barends,
children was not a significant moderator, 2006). To some extent, clients’ collabora-
however. tion in therapy depends on the therapist’s
We located only one study that specifi- theoretical approach. Couples in behavioral
cally tested a mediating model (Friedlander therapy, for example, spend less time access-
et al., 2008). In this study, the within- ing primary emotions than do couples in
family alliance (shared sense of purpose emotion-focused therapy. Yet, on a differ-
within the family) mediated the relation- ent process level, clients’ alliance-related
ship between the parents’ observed sense of behaviors cut across therapy approaches
safety in Session 1 and their ratings of and formats. That is, like successful clients
improvement-so-far in Session 3. In other in individual therapy, successful family
words, parents who felt comfortable in the members form a close, trusting bond with
first session were more likely to exhibit a their therapists and negotiate (and renego-
strong within-family alliance that, in turn, tiate) treatment goals and tasks. Regardless
predicted their perceptions of improvement of the kinds of in-session or out-of-session
after the third session. In another study tasks, clients who have a shared sense of
(Escudero et al., 2008), within-family purpose listen respectfully to one another,
alliance was the only observed alliance validate each other’s perspective (even when
dimension to increase over time; this they disagree), offer to compromise, and
dimension predicted therapists’ perceptions avoid excessive cross-blaming, hostility, and
of the alliance and ratings of improvement- sarcasm. Family members who feel safe
so-far after Session 6. Although within- and comfortable in therapy are emotion-
family alliance was not tested as a mediator, ally expressive, ask each other for feed-
the latter findings suggest that a stronger back, encourage one another to open up
shared sense of purpose may indeed be an and speak frankly, and share thoughts
important step along the way to treatment and feelings, even painful ones, that have
success. never been expressed at home (Friedlander,
Escudero, & Heatherington, 2006).
Client Contributions
By its very definition, the alliance construct Couples Therapy
implies interaction and collaboration. For While scant, the literature offers some
this reason, isolation of family members’ evidence about the personal characteristics
contributions to the alliance is somewhat and in-session behaviors of clients who
artificial. On the other hand, considering develop strong working relationships in cou-
client contributions is essential as, during ples therapy. Research suggests that whereas
the session, therapists must gauge clients’ psychiatric symptoms are not associated
receptivity and reactions to therapeutic with alliance formation (Knobloch-Fedders,
change attempts. Pinsof, & Mann, 2004; Mamodhoussen
The clinical CFT literature focuses almost et al., 2005), greater trust in the couple rela-
exclusively on therapist behavior, with far tionship (Johnson & Talitman, 1997) and
less emphasis on client participation in treat- less marital distress (Johnson & Talitman,
ment. One could well argue that all client 1997; Knobloch-Fedders et al., 2004) are
behavior contributes to (or detracts from) predictive of more favorable alliances. In one
study (Knobloch-Fedders et al., 2004), a study of family-based therapy for anorexia
alliance development differed for men nervosa found that teens with relatively
and women. For men, recalling positive more weight and eating concerns found it
experiences in the family of origin was particularly difficult to establish an alliance
most critical for early alliance development, with the therapist (Pereira et al., 2006). On
whereas marital distress had a negative the other hand, the nature of adolescents’
impact on the alliance later on. For women, emotional problems played no role in
sexual dissatisfaction was negatively associ- alliance development in a study of MDFT
ated with the alliance throughout therapy, for drug-using adolescents (Shelef &
and women’s family-of-origin distress Diamond, 2008). That study showed no
contributed to a split alliance early in the variability in teens’ alliance-related behav-
process. ior based on pretreatment externalizing or
With respect to in-session behavior, the internalizing behaviors.
findings from one study (Thomas et al., Not surprisingly, alliances are stronger
2005) reflect the complexity of CFT when family members respond favorably to
alliances. Results showed that men were therapists’ alliance-building interventions.
less likely to agree with the therapist on the In a comparison of two families treated by
goals for treatment when their partners the same experienced therapist (Friedlander,
made negative statements about them, Lambert et al., 2008), clients in the poor-
whereas women tended to feel more nega- outcome case were less likely than those in
tive about therapy tasks when they were the good-outcome case to respond posi-
challenged by their partners. Both men and tively to the therapist’s alliance-related
women had a stronger bond with the thera- behaviors. Another small sample study,
pist when their partners self-disclosed, and although not directly assessing the alliance,
they felt more distant from the therapist has implications for client contributions
when their partners challenged or made to a shared sense of purpose. In this
negative comments about them. study, family members moved from disen-
gagement with each other to productive
Family Therapy in-session collaboration when, with the
In community-based family therapy, therapist’s help, they were willing and able
parental differentiation of self, assessed prior to explore the underlying basis for their
to the beginning of treatment, predicted disengagement and recognize some moti-
stronger perceived alliances after Session 3 vation for breaking through the impasse
(Lambert & Friedlander, 2008). Well- (Friedlander, Heatherington, Johnson, &
differentiated individuals are able to balance Skowron, 1994).
thinking and feeling, autonomy and togeth-
erness (Bowen, 1978). The most closely Limitations of the Research
associated aspects of these two constructs The body of research covered in the meta-
were emotional reactivity and safety. That analysis is small but solid. Diverse client
is, parents who reported being generally populations and therapy approaches have
less emotionally reactive tended to feel safer been sampled, and many of the treatments
and more comfortable in conjoint family studied have strong empirical support
therapy. and/or have been delivered by experienced
There is some evidence to suggest that therapists. Under these conditions, the
diagnosis or presenting problem maybe finding that alliances predict treatment
associated with the alliance. For example, retention and outcome over and above
specific therapy methods strengthens importance of attachment for a couple’s
the case for the unique contribution of level of intimacy and for a family’s level of
relationship variables in CFT. Nonetheless, cohesion, attachment may well mediate or
there are limitations in this body of moderate the relationship between alliance
work that require caution in interpreting and treatment outcomes.
the findings and applying them to Although alliances develop and change
practice. over the course of psychotherapy, as several
Considerable variation across studies in studies have shown, we have little knowl-
alliance instruments and in timing and edge about how multiple alliances develop
rating source of the measurement make it over time and interact with each other.
challenging to interpret results. Under- Furthermore, we have little information
standably but unfortunately, sample sizes about how therapists behave in order to
in many of the studies are small, rendering best nurture and sustain working alliances
it difficult to test more than one or two with multiple clients over time. Finally, we
moderators with confidence. Thus, much still have not fully answered the question of
of what we know about moderators is spec- whether early symptom improvement in
ulative, based on results across different CFT actually prompts alliance develop-
studies. ment or is the consequence of a strong
The bulk of the research to date focuses alliance.
on drug-abusing, externalizing adolescents.
The alliance–outcome associations found,
as well as the kinds of therapist behaviors
Therapeutic Practices
shown to strengthen the alliance, may • The first, perhaps most important
largely be specific to these kinds of families. practice implication is that the therapeutic
For this reason, it is unwise to generalize alliance is a critical factor in the process
from these results to families with younger and outcome of CFT. Our meta-analysis
children or to families whose children have underscores the necessity to be aware of
internalizing problems (depression, anxiety, what is going within the system itself as
eating disorders). The effects of unbalanced well as to monitor the personal bond and
alliances, for example, may be weaker or agreement on goals and tasks with each
nonexistent in families in which, by virtue individual family member.
of the child’s age, symptoms, or psycho- • Knowing the importance of
logical dynamics, the children are less the alliance in CFT, practitioners are
inclined to resist treatment. Moreover, encouraged to systematically evaluate
there are few studies of the alliance in family their alliances with clients. One option
treatments for adults with major mental is periodically to ask clients to complete
disorders, such as family psychoeducation brief self-report measures of the alliance.
for bipolar disorder and schizophrenia, Doing so would not only provide the
despite the demonstrated efficacy of these therapist with crucial information
treatments. Finally, little research to date regarding each client’s private experience
has examined individual psychodynamics but also would provide the impetus for
(e.g., attachment styles) as moderators. directly addressing the quality of the
Attachment, in particular, has been a relationship and the therapy process
fruitful area of inquiry in the alliance and, if necessary, for focusing specifically
research on individual therapy (see Chapter on improving safety or repairing a
19 by Levy et al., this volume). Given the seriously split alliance.
• Our findings suggest that evaluating to give me more space”) that compete with
the alliance based on observation is a one another. In this manner, therapists can
skill that can be taught. Observationally facilitate a shared sense
measured alliances were as predictive of of purpose between family members
outcome and more homogeneous in regarding the goals of treatment and how
their effect size than alliances measured they can productively collaborate to
using self-report. These findings suggest achieve these goals.
that therapists may be taught, or • Indeed, shared sense of purpose, a
alternatively may train themselves, to whole system aspect of the alliance, seems
validly assess the strength of their alliance to be a particularly important dimension
with different family members by of the alliance. Creating a safe space,
reviewing videotaped sessions (as was which is critical early on in therapy, is
shown to be effective in the exploratory important for all therapy participants.
training study of Carpenter, Escudero, & A therapist who allies too strongly with an
Rivett, 2008). adolescent may unwittingly damage his or
• The meta-analytic findings also her alliance with the parents, particularly
underscore the need to develop alliances when the latter are expecting the child to
with all family members. Therapists change but are not expecting to be
frequently identify more easily with, personally challenged by the therapist.
or feel a greater affinity to, one family • In short, each person’s alliance
member than another. However, the matters and alliances are not
studies we reviewed indicate that each interchangeable. Thus, clinicians should
and every alliance exerts both a direct build and maintain strong alliances with
and an interactive effect on the course each party and be cognizant of the ways in
of treatment. For example, whereas a which, depending on the couple or family
woman may feel connected to and dynamics, the whole alliance is not equal
aligned with the therapist from the to the sum of its parts.
outset of treatment, it may be the degree
to which her disenfranchised partner
develops a connection with the therapist References
that determines outcome. Findings from Beck, M., Friedlander, M. L., & Escudero, V.
studies examining unbalanced alliances (2006). Three perspectives on clients’ experiences
underscore this point, in that couples of the therapeutic alliance: A discovery-oriented
and families that experienced disparate investigation. Journal of Marital and Family
alliances with the therapist tended to Therapy, 32, 355–68.
fare more poorly in treatment. Bordin, E. S. (1979). The generalizability of the
psychoanalytic concept of the working alliance.
• One approach to developing multiple Psychotherapy, 16, 252–60.
alliances is to establish overarching Bourgeois, L., Sabourin, S., & Wright, J. (1990).
systemic goals (e.g., “It sounds like what Predictive validity of therapeutic alliance in
the two of you want is a relationship in group marital therapy. Journal of Consulting and
which you feel both connected and that Clinical Psychology, 58, 608–613.
you can sometimes do your own thing”) Bowen, M. (1978). Family therapy in clinical
practice. NY: Jason Aronson.
that resonate for all family members rather Brown, P. D., & O’Leary, K. D. (2000). Therapeutic
than focusing on first-order, individual alliance: Predicting continuance and success in
goals (i.e., “I want him to stop watching group treatment for spouse abuse. Journal of
sports on TV every Sunday” or “I want her Consulting and Clinical Psychology, 68, 340–45.
Carpenter, J., Escudero, V., & Rivett, M. (2008). Friedlander, M. L., Bernardi, S., & Lee, H. (2010).
Training family therapy students in conceptual Better versus worse family therapy sessions as
and observation skills relating to the therapeutic reflected in clients’ alliance-related behavior.
alliance: An evaluation. Journal of Family Therapy, Journal of Counseling Psychology, 57, 198–204.
30, 409–22. Friedlander, M. L., Escudero, V., & Heatherington, L.
Catherall, D. (1984). The therapeutic alliance in (2006). Therapeutic alliances with couples and
individual, couple, and family therapy. Unpub- families: An empirically-informed guide to practice.
lished doctoral dissertation, Northwestern Washington, DC: American Psychological
University, Chicago, IL. Association.
Diamond, G. M., Liddle, H. A., Dakof, G. A., & Friedlander, M. L., Escudero, V., Horvath, A.,
Hogue, A. (1996). Revised version of the Vanderbilt Heatherington, L., Cabero, A., & Martens, M.
Therapeutic Alliance Scale. Unpublished manu- (2006). System for Observing Family Therapy
script, Temple University, Philadelphia, PA. Alliances: A tool for research and practice. Journal
Diamond, G. M., Liddle, H. A., Hogue, A., & of Counseling Psychology, 53, 214–25.
Dakof, G. A. (1999). Alliance-building inter- Friedlander, M. L., Heatherington, L., Johnson, B.,
ventions with adolescents in family therapy: & Skowron, B. (1994). “Commitment to
A process study. Psychotherapy, 36, 355–68. engage:” A change event in family therapy.
Diener, M. J., Hilsenroth, M. J., & Weinberger, J. Journal of Counseling Psychology, 41, 1–11.
(2009). A primer on meta-analysis of correlation Friedlander, M. L., Lambert, J. E., Escudero, V.,
coefficients: The relationship between patient- & Cragun, C. (2008). How do therapists
reported therapeutic alliance and adult attach- enhance family alliances? Sequential analyses of
ment style as an illustration. Psychotherapy therapist–client behavior in two contrasting
Research, 19, 519–26. cases. Psychotherapy: Theory, Research, Practice,
Escudero, V., & Friedlander, M. L. (2003). Training, 45, 75–87.
El sistema de observación de la alianza terapeutica Friedlander, M. L., Lambert, J. E., & Muñiz de la
en intervención familiar (SOATIF): Desarrollo Peña, C. (2008). A step toward disentangling the
trans-cultural, fiabilidad, y aplicaciones del alliance/improvement cycle in family therapy.
instrumento. Mosaico (Journal of the Spanish Journal of Counseling Psychology, 55, 118–24.
Federation of Family Therapy Associations), Greenberg, L. S., Warwar, S., & Malcolm, W.
25, 32–36. (2010). Emotion-focused couples therapy and
Escudero, V., Friedlander, M. L., Varela, N., & facilitation of forgiveness. Journal of Marital and
Abascal, A. (2008). Observing the therapeutic Family Therapy, 36, 28–42.
alliance in family therapy: Associations with Hartley, D., & Strupp, H. H. (1983). The
participants’ perceptions and therapeutic therapeutic alliance: Its relationship to outcome
outcomes. Journal of Family Therapy, 30, in brief psychotherapy. In J. Masling (Ed.).,
194–214. Empirical studies of psychoanalytic theories (Vol. 1,
Feeley, M., DeRubeis, R. J., & Gelfand, L. A. pp. 1–37). Hillsdale, NJ: Erlbaum.
(1999). The temporal relation of adherence and Hatcher, R. L., & Barends, A. W. (2006). How a
alliance to symptom change in cognitive therapy return to theory could help alliance research.
for depression. Journal of Consulting and Clinical Psychotherapy: Theory, Research, Practice, &
Psychology, 67, 578–82. Training, 43, 292–99.
Flicker, S. M., Turner, C. W., Waldron, H. B., Hawley, K. M., & Garland, A. F. (2008). Working
Ozechowski, T. J., & Brody, J. L. (2008). Ethnic alliance in adolescent outpatient therapy: Youth,
background, therapeutic alliance, and treatment parent, & therapist reports and association with
retention in functional family therapy with ado- therapy outcomes. Child and Youth Care Forum,
lescents who abuse substances. Journal of Family 27, 59–74.
Psychology, 22, 167–70. Heatherington, L., & Friedlander, M. L. (1990).
Friedlander, M. L. (2000). Observational coding Couple and family therapy alliance scales:
of family therapy processes: State of the art. In Empirical considerations. Journal of Marital and
A. P. Beck & C. M. Lewis (Eds.), The process of Family Therapy, 16, 299–306.
group psychotherapy: Systems for analyzing change. Hogue, A., Dauber, S., Stambaugh, L. F., Cecero, J. J.,
(pp. 67–84). Washington, DC: American & Liddle, H. A. (2006). Early therapeutic
Psychological Association. alliance and treatment outcome in individual
and family therapy for adolescent behavior in couples undergoing couple therapy. Journal of
problems. Journal of Consulting and Clinical Marital and Family Therapy, 31, 159–69.
Psychology, 74, 121–29. Muñiz de la Peña, C., Friedlander, M. L., &
Horvath, A. O., & Greenberg, L. S. (1986). The Escudero, V. (2009). Frequency, severity, and
development of the Working Alliance Inventory. evolution of split family alliances: How observ-
In L. S. Greenberg & W. M. Pinsof (Eds.), The able are they? Psychotherapy Research, 19, 133–42.
psychotherapeutic process: A research handbook. Pandyaa, K., & Herlhyb, J. (2009). An exploratory
(pp. 529–56). New York: Guilford Press. study into how a sample of a British South Asian
Hunter, J. E., & Schmidt, F. L. (1990). Methods of population perceive the therapeutic alliances in
meta-analysis: Correcting error and bias in research family therapy. Journal of Family Therapy, 31,
findings. Newbury Park, CA: Sage. 384–404.
Johnson, L. N., & Ketring, S. A. (2006). The Pereira, T., Lock, J., & Oggins, J. (2006). Role of
therapy alliance: A moderator in therapy therapeutic alliance in family therapy for adoles-
outcome for families dealing with child abuse cent anorexia nervosa. International Journal of
and neglect. Journal of Marital and Family Eating Disorders, 39, 677–84.
Therapy, 32, 345–54. Pinsof, W. B. (1994). An integrative systems
Johnson, L. N., Ketring, S. A., Rohacs, J., & perspective on the therapeutic alliance:
Brewer, A. L. (2006). Attachment and the Theoretical, clinical, and research implications.
therapeutic alliance in family therapy. American In A. O. Horvath & L. S. Greenberg (Eds.), The
Journal of Family Therapy, 34, 205–218. working alliance: Theory, research, and practice
Johnson, S. M., & Talitman, E. (1997). Predictors (pp. 173–95). New York: Wiley & Sons.
of success in emotionally focused marital Pinsof, W. B., & Catherall, D. (1986). The integra-
therapy. Journal of Marital and Family Therapy, tive psychotherapy alliance: Family, couple, and
23, 135–53. individual therapy scales. Journal of Marital and
Kazdin, A. E., Marciano, P. L., & Whitley, M. K. Family Therapy, 12, 137–51.
(2005). The therapeutic alliance in cognitive- Pinsof, W. B., Zinbarg, R., &
behavioral treatment of children referred for Knobloch-Fedders, L. M. (2008). Factorial and
oppositional, aggressive, and antisocial behavior. construct validity of the revised short form
Journal of Consulting and Clinical Psychology, Integrative Psychotherapy Alliance Scales for
73, 726–30. family, couple, and individual therapy. Family
Knobloch-Fedders, L. M., Pinsof, W. M., & Process, 47, 281–301.
Mann, B. J. (2007). Therapeutic alliance and Quinn, W. H., Dotson, D., & Jordan, K. (1997).
treatment progress in couple psychotherapy. Dimensions of therapeutic alliance and their
Journal of Marital & Family Therapy, 33, 245–57. associations with outcome in family therapy.
Lambert, J. E., & Friedlander, M. L. (2008). Psychotherapy Research, 7, 429–38.
Relationship of differentiation of self to adult Rait, D. S. (1998). Perspectives on the therapeutic
clients’ perceptions of the alliance in brief family alliance in brief couples and family therapy.
therapy. Psychotherapy Research, 43, 160–66. In J. D. Safran & J. C. Muran (Eds.), The thera-
Lambert, J., Skinner, A., & Friedlander, M. L. peutic alliance in brief psychotherapy (pp. 171–91).
(in press). Problematic within-family alliances Washington, DC: American Psychological
in conjoint family therapy: A close look at five Association.
cases. Journal of Marital and Family Therapy. Raytek, H. S., McCready, B. S., Epstein, E. E., &
Liddle, H. A., Dakof, G., Diamond, G., Holt, M., Hirsch, L. S. (1999). Therapeutic alliance and
Aroyo, J., & Watson, M. (1992). The adolescent the retention of couples in conjoint alcoholism
module in multidimensional family therapy. treatment. Addictive Behaviors, 24, 317–30.
Adolescent substance abuse: Etiology, treatment, Robbins, M. S., Liddle, H. A., Turner, C. W.,
and prevention. In G. W. Lawson & A. W. Dakof, G. A., Alexander, J. F., & Kogan, S. M.
Lawson (Eds.), Adolescent substance abuse: (2006). Adolescent and parent therapeutic alli-
Etiology, treatment, and prevention (pp. 165–86). ances as predictors of dropout in multidimen-
Gaithersburg, MD: Aspen Publishers. sional family therapy. Journal of Family Psychology,
Mamodhoussen, S., Wright, J., Tremblay, N., & 20, 108–116.
Poitras-Wright, H. (2005). Impact of marital Robbins, M. S., Mayorga, C. C., Mitrani, V. B., Turner,
and psychological distress on therapeutic alliance C. W., Alexander, J. F., & Szapocznik, J. (2008).
Adolescent and parent alliances with therapists prospective change process study. Journal
in brief strategic family therapy with drug- of Consulting and Clinical Psychology, 76,
using Hispanic adolescents. Journal of Marital & 505–510.
Family Therapy, 34, 316–28. Symonds, B. D. (1999). The measurement of alliance
Robbins, M. S., Turner, C. W., Alexander, J. F., & in short term couples therapy. Unpublished
Perez, G.A. (2003). Alliance and dropout in doctoral dissertation, Simon Fraser University,
family therapy for adolescents with behavior Burnaby, British Columbia, Canada.
problems: Individual and systemic effects. Symonds, B. D., & Horvath, A. O. (2004).
Journal of Family Psychology, 17, 534–44. Optimizing the alliance in couple therapy. Family
Shelef, K., & Diamond, G. M. (2008). Short Process, 43, 443–55.
form of the Vanderbilt Therapeutic Alliance Szapocznik, J., Perez-Vidal, A., Brickman, A.,
Scale: Development, reliability, and validity. Foote, F. H., Santisteban, D. A., Hervis, O.,
Psychotherapy Research, 18, 433–43. et al. (1988). Engaging adolescent drug users
Shelef, K., Diamond, G. M., Diamond, G. S., & and their families into treatment: A strategic
Liddle, H. A. (2005). Adolescent and parent structural approach. Journal of Consulting and
alliance and treatment outcome in multidimen- Clinical Psychology, 56, 552–57.
sional family therapy. Journal of Consulting and Thomas, S. E. G., Werner-Wilson, R. J., &
Clinical Psychology, 73, 689–98. Murphy, M. J. (2005). Influence of therapist and
Smerud, P. E., & Rosenfarb, I. S. (2008). The client behaviors on therapy alliance. Contemporary
therapeutic alliance and family psychoeducation Family Therapy: An International Journal, 27,
in the treatment of schizophrenia: An exploratory 19–35.
C HA P TER
5 Cohesion in Group Therapy
Gary M. Burlingame, Debra Theobald McClendon, and Jennifer Alonso
Cohesion is the most popular of several by providing a tabular summary of

relationship constructs (e.g., alliance, group therapeutic practices that have been linked
climate, group atmosphere) in the clinical to increased cohesion. Our intent in this
and empirical literature on groups. Over chapter is to illuminate the coherence in
time it has become synonymous with the cohesion literature, present the meta-
the therapeutic relationship in group analytic conclusions, and offer group leaders
psychotherapy (Burlingame, Fuhriman, & the measures and practices to improve treat-
Johnson, 2002). From the perspective of a ment outcomes.
group member, relationships are comprised
of three structural components: member– Definitions and Measures
member, member–group, and member– Definitions of cohesion have traveled a
leader. From the perspective of the therapist, serpentine trail (Bednar & Kaul, 1994;
relationships include the same three Crouch, Bloch, & Wanlass, 1994; Kivlighan,
components as well as two additional ones: Coleman, & Anderson, 2000) ranging
leader–group and, in the case of a cother- from broad and diffuse (e.g., forces that
apy, leader–leader. The complexity of these cause members to remain in the group,
multilevel structural definitions coupled “sticking-togetherness”) to focused (e.g.,
with the dynamic interplay among them attractiveness, alliance) and structurally
has created an array of competing cohesion coherent (e.g., tripartite relationship; Yalom
instruments and an absence of a consensual & Leszcz, 2005). At different times, review-
definition. ers have pleaded for definitional clarity with
In this chapter, we review the multiple two noting that “there is little cohesion in
measures of group cohesion and then discuss the cohesion research” (Bednar & Kaul,
a new measure that elucidates group rela- 1978, p. 800). Indeed, instruments tapping
tionships by suggesting two latent factors group acceptance, emotional well-being,
that explain common variance among these self-disclosure, interpersonal liking, and
group therapy relationship instruments— tolerance for personal space have been used
quality and structure. We provide clinical as measures of cohesion (Burlingame et al.,
examples to illustrate the multiple facets of 2002). Behavioral definitions have included
cohesion in group work. We then present attendance, verbal content, early termina-
an original meta-analytic review of cohesion’s tion, physical seating distance, amount
relation with treatment outcome and of eye contact, and the length of time
discuss potential moderators. We conclude group members engaged in a group hug
110
(Hornsey, Dwyer, & Oei, 2007). The defi- vertical and horizontal cohesion (Dion,
nitional challenges of cohesion are reflected 2000). Vertical cohesion represents a
by one team’s observation that “just about member–leader relationship and refers to a
anything that has a positive valence [with group member’s perception of the group
outcome] has been interpreted at some leader’s competence, genuineness, and
point as an index of cohesion” (Hornsey, warmth. Horizontal cohesion describes a
Dwyer, Oei, & Dingle, 2009, p. 272). group member’s relationship with other
Empirical investigations examining the group members and with the group as a
multidimensional structure of cohesion whole. The second dimension contrasts task
have reported as few as two and as many as cohesion (task performance) or the work of
five dimensions (Braaten, 1991; Cattell & the group with affective or emotional
Wispe, 1948; Griffith, 1988; Selvin & cohesion (interpersonal/emotional support;
Hagstrom, 1963) with common factors Griffith, 1988). In task cohesion, members
including vertical and horizontal cohesion are drawn to the group to accomplish a
as well as task and social/affective cohesion. given task, while in affective cohesion mem-
After reviewing the literature, we believe bers feel connected because of the emotional
there is ample evidence to support two support the group experience affords.
fundamental definitional dimensions of The measures of cohesion that have been
cohesion. The first dimension relates to the most frequently studied are summarized in
structure of the therapeutic relationship in Table 5.1. All but the Harvard measure
groups and is most often referred to as (Budman et al., 1989) are self-report,
Table 5.1 Cohesion measures that appear two or more times in the literature
Cohesion Measure Description of Measure Scales Dimensions:
Direction &
Function
Group Climate • Engaged measures the degree of self-disclosure, cohesion, and • Horizontal
Questionnaire (GCQ; work orientation in the group. • Affective
MacKenzie, 1981, • Avoiding examines the degree to which individuals rely on the
1983) other group members or leaders, avoiding responsibility for
their own change process.
• Conflict examines interpersonal conflict and distrust.
Group Atmosphere • Group Cohesion includes: Autonomy, Affiliation, Involvement, • Horizontal & Vertical
Scale (GAS; Silbergeld Insight, Spontaneity, Support, and Clarity. • Affective
et al., 1975) • Submission examines group conformity.
• Aggression, Order, Practicality, and Variety contribute to other
aspects of perceived environment. Authors did not define
these scales.
Feelings about group • Modified from Schutz (1957) Cohesiveness Questionnaire; • Horizontal & Vertical
(Lieberman, Yalom & 13-item Likert scale • Affective
Miles, 1973) • No subscales; items ask members to reflect on group
participation, liking of group, inclusion in the group and
feelings about leader
• Designed to measure attractiveness of a group for its members
and perceived belonging or acceptance by other members in
the group.
(Continued)
bu r l i n g a me , mcc l e n d o n , a lo n s o 111
Table 5.1 Continued
Direction &
Function
Gross (1957) Cohesion • No subscales reported • Horizontal
Scale Revised • Questions examine: group fit, perceived inclusion, attraction to • Affective & Task
(Lieberman et al., group activities, likability of members, how well the group
1973) works together, and the like.
Group Cohesion Member–member: • Horizontal & Vertical
(Piper et al., 1983) • Affective & Task
• Positive qualities examines likability, trust, and ease of
communication.
• Personal compatibility examines attraction, similarity, and
desire for personal friendship.
• Significance as a group member examines personal
importance.
Member–leader:
• Positive qualities examines likability, trust, attraction, and
ease of communication.
• Dissatisfaction with leader’s role examines discontent with style,
communication, and level of personal disclosure.
• Personal compatibility examines similarity and desire for
friendship.
Member–group:
• Mutual stimulation and effect examines engagement, inclusion,
and influence.
• Commitment to group examines attending the group and desire
for the group to continue.
• Compatibility of the group examines fit and attractiveness to
the group.
Group Environment • Relationship examines cohesion, leader support, and the • Horizontal & Vertical
Scale (GES; Moos, amount of freedom of action and expression of feelings • Affective & task
1986; Moos & encouraged in the group.
Humphrey, 1974) • Personal Growth examines independent action and expression
among members, the degree of the group’s task orientation,
the group’s encouragement of discussion of personal problems,
and anger and disagreement.
• System Maintenance and Change measures the degree of
organization, structure and rules in the group, role of the leader
in making decisions and enforcing rules, and how much the
group promotes diversity and change in its own process.
Group Attitude • 20-items no subscales reported, measures attraction to • Horizontal
Scale (Evans & Jarvis, group. • Affective
1986) • Illustrative items: “I feel involved in what is happening
in my group”, “If I were told my group would not meet today
I would feel bad” & “I feel it would make a difference to the
group if I were not here”.
• Initially validated against the Moos and Humphrey 1974
version of GES.
(Continued)
112
Table 5.1 Continued
Direction &
Function
Group cohesion • 23-items based on combining selected items from the Group • Horizontal
questionnaire Attitude Questionnaire and Stokes (1983) 3-factor • Affective
(GCQ23: van Andel questionnaire; no subscales reported.
et al., 2003; Trijsburg • Illustrative items: “The group is honest and straightforward”,
et al., 2004). “I feel involved in what is happening in my group” & “There
are people in the group I would enjoy spending time with
outside the group session”.
Harvard Group • Behavioral process scale rated by clinicians from videotapes • Horizontal
Cohesiveness Scale • Five subscales and a global score: (1) withdrawal and self- • Affective & Task
(Budman et al., 1987, absorption vs. interest and involvement, (2) mistrust vs. trust,
1989) (3) disruption vs. cooperation, (4) abusiveness vs. expressed
caring and (5) unfocused vs. focused
• Global scale called fragmentation vs. global cohesiveness
and several were developed by modifying four commonly used relationship measures
previous measures. We have classified each by having 662 members from 111 different
measure in Table 5.1 by its use of the groups complete a copy of each. This study
structural and affective/task definitions of found that a two-dimensional model
cohesion. All assess horizontal cohesion (quality & structure) explained a majority
between members and their group, while of the common variance across the four
fewer than half focus on a member’s rela- measures. Specifically, positive bond, positive
tionship with the leader (vertical). Similarly, work, and negative relationship factors
affective bond is universally assessed by all explained how members perceived the
measures while the task cohesion is assessed quality of the relationship in both non-
by a third of the measures. clinical and clinical groups. Positive bond
A different relation with outcome might described the affective relationship members
result from using different cohesion mea- felt with their leader (vertical cohesion)
sures. In fact, we have proposed a two-factor and in member-to-member relationships
definition of the therapeutic relationship in (horizontal cohesion). Positive work equally
groups to potentially clarify mixed results captured the tasks and goals of the group
in the literature: (1) belonging and accep- while negative relationship captured empathic
tance factors—cohesion and member– failure with the leader and conflict in
leader alliance; and (2) interpersonal work the group. Interestingly, members were
factors—group working alliance, individ- unable to distinguish member-to-group
ual working alliance, and group climate from member-to-member relationships
(Burlingame, Mackenzie, & Strauss, 2004). yielding a two-factor structural dimension:
We undertook a series of studies to evaluate member–leader/–member.
a toolkit containing several therapeutic The three relationship qualities (positive
relationship measures (Strauss, Burlingame, bond, positive work, & negative relationship)
& Bormann, 2008). The first study and two structural factors (member–leader
(Johnson, Burlingame, Davies, & Gleave, & member–member) stimulated subsequent
2005) estimated the empirical overlap of studies that attempted to replicate these
findings across clinical settings and coun- nonclinical process groups replicating the
tries. One study (Bormann & Strauss, 2007) first study (Johnson et al., 2005). Another
collected data from 67 inpatient psychody- population was added (seriously mentally
namic groups drawn from 15 hospitals in ill inpatients; SMI) to determine if the
Germany and Switzerland. Both the rela- model could be used with seriously ill
tionship quality and structure dimensions members; groups were primarily psycho-
emerged, but unlike the first study, these educational (Krogel, 2009; Krogel &
authors found support for three structural Burlingame, 2009). Using a sample of
components (member–member, member– 485 group members, they found the same
leader, & member–group). The next study two-dimensional relationship quality and
(Lorentzen, Høglend, & Ruud, 2008) tested structure model based on a 30-item
the same four measures and reported a simi- solution.
lar two-dimensional model that varied by These studies improve our consensual
stage of treatment (Bakali, Baldwin, & definition of cohesion. Table 5.2a weds the
Lorentzen, 2009). More specifically, early past definition of affective and task cohe-
sessions produced a strong member– sion dimensions from Table 5.1 with the
leader positive bond while later sessions relationship quality and structure model.
(10–11 & 17–18 sessions) included positive As can be seen, affective cohesion is split by
bonds with both other members and the the emotional valence of the item loading
group. either on the positive bond or negative
Findings from these studies led to an relationship dimensions. Table 5.2b depicts
item reduction process to determine if a how horizontal and vertical cohesion load
subset of “practice friendly” items could on the relationship structure dimension.
be identified from the original measures Specifically, horizontal cohesion captures
that contained over 80 items. A four-person the quality of member-to-member relation-
team with 75 collective years of clinical ships and vertical cohesion captures the
experience used both empirical (statistical quality of member-to-leader relationships.
fit with two-dimensional model) and Interestingly, the member–group dimen-
clinical criteria (does it provide actionable sion has been the most elusive theoretical
clinical information?) to produce a 40-item construct to empirically detect with
instrument called the Group Question- mixed findings from preceding studies.
naire that measured the three quality The positive and negative valence items
and structural factors (Burlingame, 2010). that consistently load on the member–
After item consensus was achieved, data group factor across the five studies were
was collected from counseling centers and principally drawn from MacKenzie’s
Table 5.2a Modified Framework for Understanding Cohesion Using Relationship Quality
Relationship Structure
Relationship Quality Member–member Member–leader Member–group
Positive Bond Affective cohesion—positive feelings & belonging
Positive Work Task cohesion
Negative Affective cohesion—empathic failure & conflict
Table 5.2b Modified Framework for Understanding Cohesion Using Relationship Structure
Relationship Structure
Relationship Quality Member–member Member–leader Member–group
Positive Bond
Horizontal Vertical Group-as-a-whole
Positive Work Cohesion Cohesion Cohesion
(Climate)a
Negative
a
cf. McClendon & Burlingame (in press)
(1983) Group Climate Questionnaire going after the last session (smiles at
(see Table 5.1). leader). Member–group positive work
Pete to group: I’ve had this rotten
Clinical Examples headache all day … it would have
The multidimensional complexity of group been real easy to stay home from
cohesion makes it impossible to provide a almost anything … but not from our
single, concrete example. However, the group. Member–group positive bond
relationship quality and structure model Mary to group: Yeah, today as I thought
(Tables 5.2a, 5.2b) provide a practice- about coming to group I knew that
friendly framework to recognize group Steve was going to make me laugh,
behavior that might facilitate cohesion. We everybody else in the group is so good
selected a transcript from Session 14 of a to give me their advice and support,
15-session group therapy (Burlingame & and I enjoy everybody so much.
Barlow, 1996). The segment begins with a Member–member/group positive bond
leader acknowledging that the next session Leader to group: That’s great. I really
will be the last and then probes regarding think every single person needs this
the work achieved over the course of the kind of a positive association, maybe
group. Group members don’t respond to not in a formal setting like this, but
the work probe but instead focus on the somehow or another like this, we
positive bond. The quality and structure need it. We really do. Every human
categories from Table 5.2 are identified by being needs it. Leader–member/group
italics. positive bond
Leader to group: This is our next-to-last The relationship quality/structure model
session. In thinking about our group, also accounts for the multiplicity of relation-
I wondered if anyone would care to ships in the group. This allows a leader to
speak to how they met their goals consider multiple aspects of the therapeutic
over the past 14 weeks? Leader–group relationship as they plan interventions. The
positive work probe following dialogue includes all three relation-
Mary to leader/group: Well … I think ship structures (member–member, member–
we’ve all had fun, I know I have. leader, and member–group) and begins with
In fact, we talked after you left last a leader probe regarding a conflictual event
night. We’re gonna keep our group that happened at the end of the last group.
Leader to Steve: Steve, you OK? You Steve to Susan: Well, I’m glad you’re
seemed upset at the end of our last here … because I’ve been worried
group meeting. Leader–member, about you [Steve goes on to inquire
negative relationship probe about Susan’s situation and tell her
Steve: I need to apologize to you all his thoughts about it. This goes on
because I was a little bit abrupt with for quite a while.] member–member
you last week and I … thought that positive bond
was kinda tacky, uh … even though I Susan to Steve: Thank you. The reason
said it was none of you damned I tore out of work so fast to get here
business. is because I knew I’d get the reception
[Group laughs] … But uh, what I I just got. [Susan starts to cry and
meant was I’m not handling it well group laughs lightly, leader pats
and, therefore, I can’t share anything Susan on the shoulder and Susan
with you. I have nothing to give pats Mary on the knee] member–
[laughs] because I … uh, I’m not group/member positive bond
handling it well. Member–leader Steve to Susan: I apologize for being
negative relationship abrupt with you last week. That was
Leader: You’ve done a lot of good work tactless. I’m sorry. Member–member
over the past few months, but right negative relationship
now you feel like you’ve got nothing Susan to Steve: It didn’t bother me, but
to give—that you’re no longer I accept your apology. It means a lot
handling it well. Leader–member to me that you’d check in with me on
positive work that. Member–member positive bond
Steve: I also feel badly that Susan is
In this dialogue we see Steve interacting
not here today, I miss her. Member–
with a notable level of interpersonal risk
member positive bond [later interaction
with the group leader. When Susan arrives
will reveal an underlying member–
we see multiple levels of positive bond,
member negative relationship valence]
which undoubtedly supports Steve’s ability
Steve: I’ve been thinking about her and
to handle a second negative relationship
her crisis a great deal, and I almost
concern from the last session with Susan.
called you [leader] up to get her
The next segment reflects the end of the
phone number. I know we’re not
group session with continued evidence of
supposed to interact outside …
member–group cohesion as another group
[Steve goes on to tell the group what
member who reinitiates discussion about
he has been thinking about Susan’s
continuing to meet after the group has for-
situation. As they are talking Susan
mally ended:
comes into the group and the whole
group cheers when she enters]. Mary to group: I would like to
Group-member positive bond see us work more on what we
toward Susan discussed last week, and that’s
Leader to Susan: We wanted you to continue it all until it finishes.
to be here so bad, some of us were I really am very interested in that …
thinking that you had a crisis and for your information … sort of a,
we were worried. Leader–member you know, forming after the
positive bond [Susan explains why group … Member–group
she is late]. positive work/bond
Susan to Mary, leaning on her terms: group psychotherapy, group therapy,
shoulder: I don’t know if we can live support groups, group counseling, cohesion,
without each other [dramatically]. group cohesion, cohesiveness, and group cli-
[Group laughs.] Member–member mate. Each abstract was reviewed for fit
positive bond with the above inclusion criteria and, if
Mary to group: Uh, yeah . . . uh once deemed promising, the article was retrieved
a month or something like that or and again reviewed for fit. A total of 24
whatever … I’m easy … but just to articles were included using this method.
get together and see how we’re doing Next, the reference sections of obtained
and talk it over and support each articles were reviewed, and 42 unduplicated
other. Member–group positive bond studies were identified and reviewed result-
ing in 6 studies being included. Finally, six
of the most frequently used cohesion mea-
Meta-Analytic Review
sures (Group Environment Scale, Piper’s
Before undertaking this meta-analysis,
Cohesion Questionnaire Scale, Group
we reviewed the literature for similar or
Climate Questionnaire, Group Atmosphere
related meta-analyses. Three cohesion meta-
Scale, Shulz’s Cohesion Questionnaire,
analyses were located (Evans & Dion, 1991;
Gross Cohesion Scale; cf. Table 5.1) across
Gully, Devine, & Whitney, 1995; Mullen
the 30 identified studies were searched using
& Copper, 1994), but none focused on
Google Scholar, yielding 1,027 abstracts.
cohesion in group psychotherapy; all
Ten additional studies were added, yielding
examined cohesion’s relationship to task
a final data set of 40 studies.
performance in nontherapeutic settings.
Thus, we relied upon five published group
Coding and Analysis
therapy meta-analyses to develop inclusion
We selected and coded 19 variables, many
criteria herein (Burlingame, Fuhriman, &
of which had been found to moderate
Mosier, 2003; Hoag & Burlingame, 1997;
outcome in previous group therapy meta-
Kosters, Burlingame, Nachitgall, & Strauss,
analyses. Five variables that assessed study
2006; Lipsey & Wilson, 2001; McRoberts,
characteristics were coded: year of publica-
Burlingame, & Hoag, 1998). To be included
tion, type of cohesion, outcome measure,
in our meta-analysis, studies must have
and time when cohesion was assessed. Three
included: (a) a group that was comprised of
leader (experience, orientation, single leader
at least three members, (b) groups meeting
vs. co-led groups) and three member
for the purpose of counseling, psycho-
characteristics (gender, diagnosis, treatment
therapy, or personal growth, (c) at least one
setting) were examined. The largest numbers
quantitative measure of both cohesion and
of variables were associated with the group
outcome, (d) data that allowed the calcula-
itself. Specifically, we were interested in
tion of effect sizes as weighted correlations,
the degree of structure associated with the
and (e) English text.
group treatment given the recent emphasis
on manual-based treatments. We coded for
Search Strategy specific practices in the studies that were
Articles were obtained by searching used to increase cohesion. We also coded
PsycINFO, MedLine, and Google Scholar for groups that allowed greater interaction
for publications between January 1969 among members, believing this might lead
and May 2009. A total of 1,506 abstracts to higher levels of cohesion than those that
were retrieved using the following search are more problem focused.
Group treatments varied from psycho- If heterogeneity is found, variability among
educational through psychotherapy/ the study’s effect size mean would be higher
counseling to personal growth groups, so than what would be expected from sam-
we also coded this variable. In an earlier pling error. Moderator results are ultimately
meta-analysis (McRoberts et al., 1998), interpreted with more confidence when
homogeneous groups were associated with heterogeneity exists.
greater improvements. Thus, we coded for A random-effects model was used to
identical or similar diagnoses and present- determine whether differences in the cohe-
ing problems (homogeneous) contrasted sion–outcome relationship existed across
with heterogeneous member composition the 19 variables. Random effects assume
to determine if the correlation between that studies are selected from a population
cohesion and outcome might be greater for of studies and that variability between stud-
homogeneous groups. ies is the result of sampling error. This ana-
The task group literature has sug- lytic model is recommended as a more
gested that group size may moderate the conservative test (Hedges & Vevea, 1998;
cohesion–outcome link, so we coded for Lipsey & Wilson, 2001).
small, medium, and large groups. Finally,
since dose of therapy has been shown to Results
moderate overall outcome, we coded for A summary of study characteristics is pro-
both session length and number of group vided in Table 5.3. As one might expect,
sessions. most of the groups (80%) had a therapy
Eight raters (one graduate student and focus. The majority (58%) of studies were
seven undergraduate students) were trained published after 2000, although over a
on a codebook to rate articles unrelated to fourth were published prior to 1990, cap-
the studies herein using an 85% criterion turing several classic papers (e.g., Braaten,
level of agreement with interrater reliability 1989; Budman et al., 1989; Roether &
being high (kappa = 0.73). After achieving Peters, 1972; Yalom, Houts, Zimerberg, &
this criterion, raters were paired and inde- Rand, 1967). We elected to include a few
pendently coded the same article contained personal growth group studies that met our
in our meta-analysis. Complete agreement criteria and used a task format (e.g., Flowers,
was required with discrepancies resolved by Booraem, & Hartman 1981; Hurley, 1989;
the graduate student. In a few instances, Kivlighan & Lilly, 1997) since all are
the first and second authors met with the frequently cited in cohesion–outcome
graduate student (third author) to clarify literature.
discrepancies. Past reviewers have concluded that
A number of studies used several outcome cohesion has shown a positive relation with
and cohesion measures, thus creating mul- patient improvement in nearly every
tiple cohesion–outcome correlations from published report (Tschuschke & Dies,
a single study. When this occurred, we aver- 1994). Our own previous, narrative review
aged the several values (weighted by n) so concluded that approximately 80% of
that only one correlation per study was published studies demonstrated a positive
included in subsequent analyses. Following association between group cohesion and
calculation of the aggregate correlation, treatment outcome (Burlingame et al.,
we examined the degree of heterogeneity 2002).
in the results across studies using the The results from our meta-analysis with
Q-statistic (Berkeljon & Baldwin, 2009). each study depicted in Figure 5.1 show
Table 5.3 Study Characteristics
Variable % N
Number of Studies 40
Year of publication (median) 1,997.7
Overall number of clients 3,323
Average age of clients 36.4
Average number of sessions 23.5
Theoretical orientation of group
Cognitive/behavioral 33 13
Psychodynamic/Existential 25 10
Humanistic/Interpersonal/Supportive 20 8
Eclectic 8 3
1
Unknown 20 8
Primary Diagnosis
Informal 35 14
Anxiety Disorder 13 5
Mood Disorder 18 7
Substance Disorder 3 1
Eating Disorder 5 2
Personality Disorder 13 5
Medical Condition (not Somatic disorder) 5 2
Unknown 18 7
Country
North America 50 20
Europe 23 9
Canada 18 7
Australia 5 2
Role of Group
Only group/group as primary treatment 10 4
Part of milieu of treatment (e.g. medication, individual therapy) 23 9
Unknown 68 27
Setting
Inpatient 15 6
Outpatient 68 27
Unknown 18 7
(Continued)
119
Table 5.3 Continued
Variable % N
Location
University Counseling Center 3 1
Clinic or Private Practice 13 5
Hospital 45 18
Community Mental Health Center 05 2
Community Location 3 1
Classroom Setting 13 5
Unknown 20 8
Type of Outcome Measure
General psychological distress 38 15
Depression 30 12
Anxiety 15 6
Quality of Life/General well being 20 8
Interpersonal Problems / Relationships 23 9
Self Esteem 13 5
Other 45 18
Unknown1 8 3
Number of Cohesion Measure administrations
Once 10 4
Twice 20 8
Three times 20 8
Four times 0 0
Five or more times 48 19
Unknown 3 1
Note: 1. Values don’t add up to 40 because some studies used multiples
a less glowing picture than past reviews, overall conclusion from 40 studies published
including our own. Only 43% of the studies across a 4-decade span is a positive relation
posted a statistically significant correlation between cohesion and outcome.
between cohesion and patient improvement.
Nonetheless, the weighted aggregate correla- Moderators and Mediators
tion for the 40 studies was a statistically sig- Until recently, there have been few empirical
nificant r = 0.25 with a 95% confidence studies examining moderator or mediator
interval of .17 to .32 (SE = .04) which is con- variables for the cohesion–outcome link
sidered to be a moderate effect. Thus, the (Hornsey et al., 2007). Mediators have
Study name Statistics for each study
Effect Lower Upper Z- p- Correlation and 95% CI

size limit limit value value
Andel et al (2003) 0.19 -0.14 0.48 1.15 0.25
Antonuccio et al (1987) 0.00 -0.19 0.19 0.00 1.00
Beutal et al (2006) 0.23 0.06 0.39 2.68 0.01
Braaten (1989) 0.21 0.02 0.38 2.20 0.03
Budman et al (1989) 0.63 0.48 0.74 6.91 0.00
Crowe & Grenyer (2008) 0.30 -0.06 0.60 1.63 0.10
Falloon 1981 0.16 -0.12 0.42 1.12 0.26
Flowers et al (1981) 0.56 0.09 0.83 2.28 0.02
Gillaspy et al (2002) 0.19 -0.10 0.45 1.30 0.19
Grabhorn et al (2002) 0.18 -0.11 0.44 1.22 0.22
Hilbert et al (2007) 0.24 0.08 0.39 2.84 0.00
Hoberman et al (1988) 0.38 0.09 0.61 2.50 0.01
Hurley (1997) 0.35 0.28 0.41 9.49 0.00
Hurley (1989) 0.70 0.64 0.75 16.7 0.00
Joyce et al (2007) 0.09 -0.10 0.28 0.92 0.36
Kipnes et al (2002) 0.00 -0.57 0.57 0.00 1.00
Kivlinghan & Lilly (1997) 0.36 0.00 0.64 1.96 0.05
Levenson & Macgowan (2004) 0.33 0.09 0.54 2.61 0.01
Lipman et al (2007) 0.15 -0.18 0.45 0.89 0.37
Lorentzen et al (2004) 0.30 -0.33 0.75 0.93 0.35
Mackenzie & Tschuschke (1993) 0.46 -0.05 0.78 1.79 0.07
Marmarosh et al (2005) 0.54 0.39 0.67 6.01 0.00
Marziali et al (1997) 0.19 -0.32 0.62 0.72 0.47
May et al (2008) 0.18 0.01 0.34 2.06 0.04
Norton et al (2008) 0.30 0.04 0.53 2.21 0.03
Oei & Brown (2006) -0.04 -0.19 0.12 -0.51 0.61
Ogrodniczuk & Piper (2003) 0.22 0.03 0.39 2.28 0.02
Ogrodniczuk et al (2005) 0.22 -0.01 0.43 1.90 0.06
Ogrodniczuk (2006) 0.42 0.12 0.65 2.69 0.01
Ratto & Hurley (1995) 0.23 -0.12 0.53 1.28 0.20
Rice (2001) 0.00 -0.26 0.26 0.00 1.00
Roether & Peters (1972) -0.18 -0.43 0.10 -1.26 0.21
Rugel & Barry (1990) 0.10 -0.28 0.46 0.50 0.62
Ryum et al (2009) 0.15 -0.24 0.50 0.74 0.46
Taft et al (2003) 0.18 -0.01 0.36 1.86 0.06
Taube-Schiff et al (2007) 0.43 0.11 0.67 2.56 0.01
Tschuschke & Dies (1994) 0.72 0.35 0.90 3.27 0.00
Woody & Adesky (2002) 0.17 -0.12 0.43 1.15 0.25
Wright & Duncan (1986) 0.13 -0.26 0.49 0.64 0.52
Yalom et al (1967) 0.11 -0.30 0.48 0.52 0.60
−1 −0.5 0 0.5 1 1.5
Favors negative relationship Favors positive relationship
Fig. 5.1 Weighted effect size for cohesion-outcome relationship
been proposed (e.g., member acceptance, from the Group Climate Questionnaire
support, self-disclosure, and feedback), but (MacKenzie, 1983), which posted the
there has been little progress due to the second highest weighted correlation (0.35).
varied definitions and confounds with Interestingly, one of the oldest cohesion
group cohesion (Hornsey et al., 2007). measures (Group Environment Scale)
Our analysis of the studies found only a posted the lowest r value.
handful of moderators among the 19 coded The cohesion–outcome relationship was
variables. explored to see if it varied by outcome
measure. As with the cohesion instrument
Study Characteristics analysis, most measures were used too
None of the study characteristics (publica- infrequently to test for reliable differences.
tion year, outcome or cohesion measure, or However, two measures (SCL-90, BDI)
time of assessment) explained variability that assess general psychiatric and depres-
among the studies. However, there were a sive symptoms, respectively, were each used
number of cohesion measures that were in a dozen studies posting reliable values
used two or more times, and their weighted that were at or above the meta-analytic
averages are depicted in Figure 5.2. The average. Higher weighted averages were
most frequently used was the engaged scale found on both interpersonal and self-esteem
Wtd. Ave. Lower Upper Times
Cor. Cor. limit limit used
Cohesion
Group Atmosphere Scale (GAS-C; Silbergeld et al, 1975) 0.25 0.21 0.01 0.45 5
Group Cohesion (GC ; Piper et al 1983) 0.15 0.17 0.00 0.43 5
Group Environment Scale (GES; Moos, 1986) 0.05 0.07 -0.08 0.19 2
Stuttgarter Bogen (Czogalik & Koeltzow, 1987) 0.35 0.52 0.09 0.72 3
Harvard Group Cohesiveness Scale (GCS; Budman, et al, 1987) 0.58 0.38 0.00 0.70 2
Group Climate Questionnaire (GCQ; MacKenzie, 1981, 1983) 0.35 0.29 0.00 0.70 12
Gross Cohesion (Gross, 1957; Lieberman et. al, 1973) 0.23 0.20 0.00 0.58 6
Outcome
Outcome Questionnaire (OQ; Lambert, et al, 1997)) 0.26 0.27 0.24 0.31 2
Inventory of Interpersonal Problems (IIP; Horowitz et al, 1988) 0.40 0.36 0.25 0.42 3
Therapy Project List (Braaten, 1989) 0.21 0.21 0.18 0.23 2
Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965) 0.54 0.51 0.45 0.58 2
Profile of Mood States (POMS; McNair, et al, 1981) 0.32 0.32 0.24 0.40 2
Beck Anxiety Inventory (BAI; Beck, et al, 1988) 0.13 0.11 -0.08 0.27 3 −0.2 0 0.2 0.4 0.6 0.8
Beck Depression Inventory(BDI; Beck, et al, 1996) 0.23 0.20 0.00 0.59 11
Symptom Checklist (SCL-90; Derogatis, 1977) 0.35 0.27 0.00 0.66 12
Fig. 5.2 Weighted Correlations and range by cohesion and outcome measure
measures (IIP, RSES), but these results are demonstrated across all three major diagnos-
heavily influenced by the two student tic classifications; Axis I (r = 0.17), Axis II
growth group studies (Hurley, 1989; (r = 0.41), and V-code (r = 0.26).
Kivlighan & Lilly, 1997). Thus, it remains
unclear how these outcome measures might Leader Variables
operate in clinical populations. A conclusion No evidence was found to support that
that seems warranted is that the cohesion– leader experience or single versus co-led
outcome relationship is well supported groups explained differences in the
when outcome is defined by general psychi- cohesion–outcome correlations. However,
atric and depressive symptoms. Moreover, there was a difference in the cohesion–
since the SCL-90 and BDI are also two of outcome relation when one considered the
the most frequently used instruments to differences across the theoretical orienta-
evaluate the effectiveness of group psycho- tion of the group leader; Q = 23.56, df = 9,
therapy, the generalizability to outcome p < 0.05. Leaders espousing an interper-
appears sound. sonal orientation posted the highest
cohesion–outcome relation (r = 0.58), with
Member Variables psychodynamic (r = 0.25) and cognitive-
Of the member variables, only one explained behavioral (r = 0.18) orientations posting
differences in the cohesion–outcome link. the lowest values. The remaining orienta-
The average age of participants was nega- tions posted either statistical trends
tively associated with effect size magnitude (humanistic, r = 0.21) or no reliable rela-
within studies (r = −.63; p < 0.0001). Studies tionship (behavioral, eclectic). This argues
with relatively younger group members for cohesion being considered as an evi-
tended to yield effect sizes of higher magni- dence-based relationship factor for groups
tude than studies with relatively older group using a cognitive-behavioral, psychody-
members; Q = 14.92, df = 1, p < 0.05. This namic, and interpersonal orientation.
finding was not explained by client symp-
tom severity; cohesion was reliably related to Group Variables
outcome in both inpatient and outpatient Four group variables proved useful in
settings (r = 0.29 and 0.24, respectively). explaining differences in the cohesion–
Furthermore, the positive association of outcome association. In the past, we
group cohesion and client outcome was (Burlingame et al., 2004; Fuhriman &
Burlingame, 1994) have suggested that the higher cohesion–outcome correlations than
best test of the cohesion–outcome relation did groups lasting 12 or fewer sessions
would be to examine studies that empha- (r = 0.17). Interestingly, there was no statisti-
sized the importance of cohesion as a thera- cally significant difference between groups
peutic strategy. If group cohesion is lasting 13–19 sessions (r = 0.36) and for 20
undervalued or neglected by a group leader, or more sessions (r = 0.31).
its presence would likely be diminished and
perhaps attenuate its relationship with out- Other Potential Moderators
come. Two previous studies which included Two studies that fell outside our meta-
cognitive pre-training to enhance cohesion analytic review parameters—because they
showed higher cohesion for the pre-trained were recently published or the text is in
groups (Santarsiero, Baker & McGee, 1995; German—suggest a potential patient mod-
Palmer, Baker & McGee, 1997). The one erator variable. In the first study, members’
study herein that described procedures for interpersonal style moderated the relation-
enhancing cohesion in their methods section ship between cohesion and outcome
(Kivlighan & Lilly 1997) pre-trained par- with 73 depressed German inpatients
ticipants using videotapes and produced a (Schauenburg, Sammet, Rabung & Strack,
cohesion–outcome correlation (r = .36) that 2001). Specifically, patients with interper-
was slightly higher yet not statistically sonal problems described as “too friendly”
different from the weighted average. improved more when their cohesion
Another significant group variable decreased during therapy, whereas patients
(group focus) found a difference between with cold or hostile interpersonal problems
problem specific and interactive groups; improved most when their experience of
Q = 4.75, df = 1, p < 0.05. Problem specific cohesion increased during the group.
groups were comprised of members with A recent replication of that study involved
similar diagnoses, and group time appeared 327 mixed-diagnosis adults treated on a psy-
to be principally focused on this common chodynamically oriented inpatient psycho-
theme. Interactive groups had members therapy unit in Germany (Dinger &
who were more interactive, and group time Schauenburg, 2010). Higher levels of cohe-
appeared to be less structured. Interactive sion as well as an increase in cohesion over
groups posted a higher cohesion–outcome the life of the group were associated with
correlation than problem-specific groups greater symptom improvement, replicating
(r = 0.38 and r = 0.21, respectively). the findings of our meta-analysis. Once
The cohesion–outcome relationship again patients who described themselves as
proved to be statistically significant when too cold and who reported increased cohe-
examining the size of the group, Q = 4.54, sion posted the greatest improvement; the
df = 1, p < 0.05. Groups comprised of opposite was evident for those who described
5–9 members in each session posted the themselves as too friendly. The value of this
strongest cohesion–outcome relationship study is that it offers one theory-driven
(r = 0.35) whereas groups of any other size explanation of both the positive and neutral
(fewer than 5 members present or more than relationship findings in this meta-analysis.
9 members) were much weaker (r = 0.16). Could a member’s interpersonal style explain
Finally, there were differences in the past mixed cohesion–outcome findings?
cohesion–outcome correlations by number Unfortunately, the “jury is out” on this
of sessions; Q = 6.87, df = 2, p < 0.05. question since the primary measure assess-
Groups lasting more than 12 sessions posted ing cohesion in both studies falls short on
psychometric support, thus attenuating our due to the scant number of studies that focus
confidence in its conclusions. on the leader (Burlingame et al., 2004).
Even if one were to uncover a handful of
Limitations of the Research studies testing the same member or leader
One of the clear challenges in understand- moderator, it is highly likely that different
ing and utilizing cohesion as an evidence- measures of cohesion would be used.
based principle has been the variability in We believe the biggest limitation to our
definition and measurement. The two- findings and the research in general is the
dimensional model (structure and quality) heterogeneity of study characteristics that
offers a promising, parsimonious, and is often hidden in meta-analyses. Even
empirically based definition of the latent though we used 40 studies, when one con-
structure inherent in measures of group siders the possible interactions among dif-
relationship. It suggests that leaders pay ferent diagnoses, settings, orientations, and
attention to the “who” (member, group, & type of groups, considerable caution must
leader), “what” (are we getting work done?), be invoked in interpreting our results.
and “how” (positive and negative emotional There are simply too few studies to ade-
valence) of group relationships. quately test for potential interactions
A second challenge in the literature has between the characteristics tested herein.
been the mixed findings regarding cohesion’s For example, the larger relation between
relation with treatment outcome. Some cohesion and outcome makes sense for
studies support its relationship with out- groups that last longer than 12 sessions.
come; others show no association. The meta- However, what we don’t know is how this
analysis clarifies this confusion by pointing plays out for different theoretical orienta-
out differences between measures, theoreti- tions and clinical settings.
cal orientations, group length, and group Finally, the question of causality cannot be
focus. For instance, cognitive-behavioral addressed in these correlational studies.
groups make up 80% of the recent group Perhaps the strongest evidence to support a
literature (Burlingame & Baldwin, in press), causal relationship was the finding from
and it’s clear that a cohesion–outcome link studies that intentionally used interventions
exists for these groups. Similarly, longer and to enhance cohesion that resulted in a stron-
more interactive groups produce larger ger cohesion–outcome relation. However, as
cohesion–outcome correlations, but even pointed out above, these studies were not
short groups (fewer than 12 sessions) still based on groups comprised of members
show a cohesion–outcome link. having a formal psychiatric diagnosis.
There are several specific limitations to
the findings herein. First, it is virtually
impossible to assess potential member or
We see the following therapeutic practices
leader moderators because of the absence of
supported by our meta-analysis:
research on this topic in the group therapy
literature. For instance, low psychological • Cohesion is reliably associated
mindedness and more severe symptoms have (r = 0.25) with group outcome when
been linked to early dropout (Burlingame outcome is defined as reduction in
et al., 2004), but to our knowledge, these symptom distress or improvement in
have never been formally tested as modera- interpersonal functioning. All group
tors of cohesion–outcome. Additionally, leaders should foster cohesion in its
there is no consensus on leader moderators multiple manifestations.
• Cohesion is certainly involved with Table 5.4. These behaviors track onto a
patient improvement in groups using a behavioral rating scale (Group Psychotherapy
cognitive-behavioral, psychodynamic, or Interventions Rating Scale; GPIRS) devel-
interpersonal orientation. oped from interventions suggested in our
• Group leaders emphasizing member first cohesion chapter (Burlingame et al.,
interaction, irrespective of theoretical 2002). In two studies (Sternberg & Trijsburg,
orientation, post higher cohesion– 2005; Snijders, Trijsburg, De Groot, &
outcome links than problem-focused Duivenvoorden, 2005), group structure,
groups. Thus, it is important to encourage verbal interaction, and emotional climate
member interaction. interventions were positively correlated with
• Cohesion explains outcome member-reported cohesion. A recent North
regardless of the length of the group, but American study (Chapman, Baker, Porter,
is strongest when a group lasts more than Thayer & Burlingame, 2010) translated the
12 sessions and is comprised of 5–9 Dutch GPIRS into English and replicated
members. Group cohesion obviously these findings. Small-to-moderate correla-
requires the correct balance of member tions were found with each GPIRS subscale,
interaction and time to build. suggesting that leader interventions intended
• Younger group members experience to affect emotional climate, manage verbal
the largest outcome gains when cohesion interaction, and maintain group structure were
is present within their groups. Fostering moderately correlated with member-reported
cohesion may be particularly useful for levels of cohesion. Leader interventions facil-
those working with young people (e.g., itating structure, emotional climate, and
counseling centers and adolescents). managing verbal interaction were positively
• Cohesion contributes to group related to cohesion and negatively related
outcome across different settings to interpersonal distrust and conflict, repli-
(inpatient and outpatient) and diagnostic cating the Dutch findings on a different
classifications. Thus, all leaders should measure of cohesion.
actively engage in interventions that foster Cohesion is integrally related to the
and maintain cohesion. success of group therapy, and the research
has identified specific behaviors that
In this regard, we would point to therapist enhance cohesion. For these reasons, we
behaviors that can enhance group cohesion. recommend the behaviors in Table 5.4 to
These specific interventions are depicted in group practitioners.
Table 5.4 Group Psychotherapy Intervention Rating Scale (GPIRS)

Group Structuring
Setting treatment expectation Set group agendas (such as discussion topics or group activities)
Described rationale underlying treatment
Establishing group procedures Discussed group rules (such as time, attendance, absences, tardiness,
confidentiality, participation)
Identified and discussed fears/concerns regarding self disclosure
Structured exercises that focus on emotional expression and exchange
(Continued)
Table 5.4 Continued
Role preparation Discussed member roles and responsibility
Discussed leader roles and responsibility
Verbal Interaction
Verbal style Modeled giving personal information in the “here and now”
and interaction Modeled appropriate member-member behavior
Modeled appropriate self disclosure
Modeled appropriate feeling disclosure
Maintained moderate control
Facilitated appropriate member-member interaction
Self disclosure Encouraged self disclosure without “forcing it”
Encouraged self disclosure relevant to the current group agenda
Helped members understand that disclosed issues achieve more resolution than
undisclosed issues
Encouraged here-and-now vs. story-telling disclosure
Interrupted ill-timed or excessive member disclosure
Elicited member-member feeling disclosure (versus informational disclosures)
Leader shared relevant personal experience from outside of therapy (without
being judgmental or overly-intellectual)
Feedback Reframed injurious feedback (interrupting, if necessary)
Restated corrective feedback by member
Used consensus to reinforce feedback (toward therapist or group member)
Balanced positive and corrective leader-to-member feedback
Encouraged positive feedback
Gave structured feedback exercise
Helped balance positive and corrective member-to-member feedback
Therapist helped members apply in-group feedback to out-of-group situations
Creating and Maintaining a Therapeutic Emotional Climate
Leader contribution Maintained balance in expressions of emotional support and confrontation
Showed understanding of the members and their concerns
Refrained from conveying personal feelings of hostility and anger in response to
negative member behavior
Leader was not defensive when interventions failed
Leader was not defensive when confronted by a member
Maintained an active engagement with the group and its work
Used nonjudgmental language with members
Modeled expressions of open and genuine warmth
Encouraged active emotional engagement between group members
Fostered a climate of both support and challenge
Responded at an emotional level
Developed and/or facilitated relationships with and among group members
Helped members recognize why they feel a certain way (identifying underlying
concerns or motives)
Member contribution Prevented or stopped attacking and judgmental expressions between members
Assisted members in describing their emotions
Recognized and responded to the meaning of groups members’ comments
Prevented situations in which members felt discounted, misunderstood,
attacked, or disconnected
Involved members in describing and resolving conflict (instead of avoiding
conflict)
Elicited verbal expressions of support among group members
Encouraged members to respond to other members’ emotional expression (such
as acceptance, belonging, empathy)
126
Braaten, L. (1991). Group cohesion: A new multi-
References
dimensional model. Group, 15, 39–55.
An asterisk (∗) indicates studies included in ∗
Braaten, L. J. (1989). Predicting positive goal
the meta-analysis. attainment and symptom reduction from early
∗
Antonuccio, D. O., Davis, C., Lewinsohn, P. M., group climate dimensions. International Journal
& Breckenridge, J. S. (1987). Therapist variables of Group Psychotherapy, 39(3), 377–87.
related to cohesiveness in a group treatment for Broker, M., Rohricht, F., & Priebe, F. (1995). Initial
depression. Small Group Research, 18, 557–64. assessment of hospital treatment by patients with
Bakali, J., Baldwin, S., & Lorentzen, S. (2009). paranoid schizophrenia: A predictor of outcome.
Modeling group process constructs at three Psychiatry Research, 58(1), 77–81.
stages in group psychotherapy. Psychotherapy Budman, S. H., Demby, A., Feldstein, M.,
Research, 19, 332–43. Redondo, J., Scherz, B., Bennett, M. J., et al.
Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1987). Preliminary findings on a new instru-
(1988). An inventory for measuring clinical ment to measure cohesion in group psychother-
anxiety: Psychometric properties. Journal of apy. International Journal of Group Psychotherapy,
Consulting and Clinical Psychology, 56, 893–97. 37, 75–94.
∗
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Budman, S. H., Soldz, S., Demby, A., Feldstein, M.,
Manual for the Beck Depression Inventory-II. San Springer, T., & Davis, M. S. (1989). Cohesion,
Antonio, TX: Psychological Corporation. alliance and outcome in group psychotherapy.
Bednar, R. L., & Kaul, T. J. (1978). Experiential Psychiatry, 52(3), 339–50.
group research: Current Perspectives. In Burlingame, G. (2010). Small group treatments:
S. L. Garfield & A. E. Bergin (Eds.), Handbook Introduction to special section. Psychotherapy
of psychotherapy and behavior change (2th ed., Research, 20(1), 1–7.
pp. 769–815). New York: Wiley. Burlingame, G., & Baldwin, S. (2010). History of
Bednar, R. L., & Kaul, T. J. (1994). Experiential group psychotherapy. In History of Psychotherapy
group research: Can the canon fire? In (2nd ed.), J. Norcross, G. VandenBos, &
A. E. Bergin & S. L. Garfield (Eds.), Handbook D. Freedheim. (Eds). (pp. 505–515), Washington
of psychotherapy and behavior change (4th ed., DC: American Psychological Association.
pp. 631–63). New York: Wiley. Burlingame, G. M., & Barlow, S. (1996). Outcome
Berkeljon, A., & Baldwin, S. A. (2009). An and process differences between professional and
introduction to meta-analysis for psychotherapy nonprofessional therapists in time-limited group
outcome research. Psychotherapy Research, 19(4), psychotherapy. International Journal of Group
511–518. Psychotherapy, 46, 455–78.
∗
Beutal, M. E., Knickenberg, R. J., Krug, B., Burlingame, G. M., MacKenzie, K. R., & Strauß, B.
Mund, S., Schattenburg, L., & Zwerenz, R. (2004). Evidence-based small group treatments.
(2006). Psychodynamic focal group treatment In M. Lambert, A. E. Bergin, & S. L. Garfield
for psychosomatic inpatients—with an emphasis (Eds.), Handbook of psychotherapy and behavior
on work-related conflicts. International Journal change (5th ed.). New York: Wiley & Sons.
of Group Psychotherapy, 56(3), 285–305. Burlingame, G. M., & Fuhriman, A. (1994).
Bormann, B., & Strauß, B. (2007). Gruppenklima, Epilogue. In A. Fuhriman & G. M. Burlingame
Kohäsion, Allianz und Empathie als Kompo- (Eds.), Handbook of group psychotherapy: An
nenten der therapeutischen Beziehung in empirical and clinical synthesis (pp. 559–62).
Gruppenpsychotherapien–Überprüfung eines New York: Wiley & Sons.
Mehrebenen-Modells. [Group climate, cohesion, Burlingame, G., Fuhriman, A., & Johnson, J.
alliance, and empathy as components of the thera- (2002). Cohesion in group psychotherapy. In
peutic relationship within group psychotherapy–Test J. C. Norcross (Ed). Psychotherapy Relationships
of a multilevel model]. Gruppenpsychotherapie und That Work (pp. 71–87). New York: Oxford
Gruppendynamik, 43, 1–20. University Press.
Bormann, B., & Strauß, B. (2009, June). The Burlingame, G. M., Fuhriman, A. F., & Mosier, J.
German Group Questionnaire: A multi-site valida- (2003). The differentiated effectiveness of group
tion study. Paper presented at the annual meeting psychotherapy: A meta-analytic review. Group
of the Society for Psychotherapy Research, Dynamics: Theory, Research, and Practice,
Santiago, Chile. 7(1), 3–12.
Cattell, R. B., & Wispe, L. G. (1948). The dimen- abuse treatment outcomes. Psychotherapy
sions of syntality in small groups. Journal of Research, 12(2), 213–29.
∗
Social Psychology, 28, 57–78. Grabhorn, R., Kaufhold, J., & Overbeck, G.
Chapman, C., Baker, E., Porter, G., Thayer, S., & (2002). The role of differentiated group experi-
Burlingame, G. (2010). Rating group therapist ence in the course of inpatient psychotherapy. In
interventions: The validation of the Group Serge P. Shohov (Ed.), Advance in psychology
Psychotherapy Rating Scale. Group Dynamics: research, Volume 12, (pp. 141–54). Hauppauge,
Theory, Research, & Practice, 14(1), 15–31. NY: Nova Science Publishers.
Crouch, E. C., Bloch, S., & Wanlass, J. (1994). Griffith, J. (1988). Measurement of group cohesion
Therapeutic factors: Interpersonal and intraper- in U.S. Army units. Basic and Applied Social
sonal mechanisms. In A. Fuhriman & Psychology, 9, 149–71.
G. M. Burlingame (Eds.), Handbook of group Gross, E. F. (1957). An empirical study of the concepts
psychotherapy (pp. 269–315). New York: Wiley. of cohesiveness and compatibility. Honors thesis,
∗
Crowe, T. P., & Grenyer, B. F. S. (2008). Is therapist Harvard University, Cambridge, MA.
alliance or whole group cohesion more influen- Gully, S. M., Devine, D. J., & Whitney, D. J.
tial in group psychotherapy outcomes? Clinical (1995). A meta-analysis of cohesion and
Psychology and Psychotherapy, 15, 239–46. performance: Effects of level of analysis and
Czogalik, D., & Koltzow, R. (1987). Zur Normierung task interdependence. Small Group Research,
des Stuttgarter Bogens. Gruppenpsychotherapie 26, 497–520.
und Gruppendynamik, 23, 36–45. Hedges, L. V., & Vevea, J. L. (1998). Fixed- and
Derogatis, L. R. (1977). SCL-90: Administration, random-effects models in meta-analysis.
scoring, and procedures manual-R (revised). Psychological Methods, 3, 486–504.
∗
Baltimore: John Hopkins University School of Hilbert, A., Saelens, B. E., Stein, R. I., Mockus, D. S.,
Medicine, Clinical Psychometrics Research Unit. Welch, R. R., Matt, G. E., et al. (2007).
Dinger, U., & Schauenburg, H. (2010). Effects of Pretreatment and process predictors of outcome in
individual cohesion and patient interpersonal interpersonal and cognitive behavioral psychother-
style on outcome in psychodynamically oriented apy for binge eating disorder. Journal of Consulting
inpatient group psychotherapy. Psychotherapy and Clinical Psychology, 75(4), 645–51.
Research, 20(1), 22–29. Hoag, M. J., & Burlingame, G. M. (1997).
Dion, K. L. (2000). Group cohesion: From “field of Evaluating the effectiveness of child and adoles-
forces” to multidimensional construct. Group cent group treatment. Journal of Clinical Child
Dynamics, 4, 7–26. Psychology, 26, 234–46.
∗
Evans, C. R., & Dion, K. L. (1991). Group cohesion Hoberman, H. M., Lewinsohn, P. M., & Tilson, M.
and performance: A meta-analysis. Small Group (1988). Group treatment of depression: Individual
Research, 22, 175–86. predictors of outcome. Journal of Consulting and
Evans, N. J., & Jarvis, P. A. (1986). The Group Clinical Psychology, 56(3), 393–98.
Attitude Scale: A measure of attraction to group. Hornsey, M., Dwyer, L., & Oei, T. (2007). Beyond
Small Group Behavior, 17, 203–216. cohesiveness: Reconceptualizing the link between
∗
Falloon, I. R. H. (1981). Interpersonal variables group processes and outcomes in group psycho-
in behavioural group therapy. British Journal of therapy. Small Group Research, 38, 567–92.
Medical Psychology, 54, 133–41. Hornsey, M., Dwyer, L., Oei, T., & Dingle, G. A.
∗
Flowers, J. V., Booraem, C. D., & Hartman, K. A. (2009). Group processes and outcomes in group
(1981). Client improvement on higher and lower therapy: Is it time to let go of cohesiveness?
intensity problems as a function of group International Journal of Group Psychotherapy,
cohesiveness. Psychotherapy: Theory, Research, and 59(2), 267–78.
Practice, 18, 246–251. Horowitz, L., Rosenberg, S., Baer, B., Ureno, F., &
Fuhriman, A., & Burlingame, G., (Eds.). (1994). Villasenor, V. (1988). Inventory of interpersonal
Handbook of group psychotherapy: An empirical problems: Psychometric properties and clinical
and clinical synthesis. New York: John Wiley applications. Journal of Consulting and Clinical
& Sons. Psychology, 56(6), 885–92.
∗ ∗
Gillaspy, J. A., Jr., Wright, A. R., Campbell, C., Hurley, J. R. (1989). Affiliativeness and outcome
Stokes, S., & Adinoff, B. (2002). Group alliance in interpersonal groups: Member and leader
and cohesion as predictors of drug and alcohol perspectives. Psychotherapy, 26(4), 520–23.
∗ Lieberman, M., Yalom, I., & Miles, M. (1973).
Hurley, J. R. (1997). Interpersonal theory and
measures of outcome and emotional climate Encounter groups: First facts. New York: Basic
in 111 personal development groups. Group Books.
∗
Dynamics: Theory, Research, and Practice, 1(1), Lipman, E. L., Waymouth, M., Gammon, T.,
86–97. Carter, P., Secord, M., Leung, O., Mills, B., &
Johnson, J. E., Burlingame, G. M., Olsen, J. A., Hicks, F. (2007). Influence of group cohesion on
Davies, D. R., & Gleave, R. L. (2005). Group maternal well-being among participants in a
climate, cohesion, alliance, and empathy in support/education group program for single
group psychotherapy: Multilevel structural mothers. American Journal of Orthopsychiatry,
equation models. Journal of Counseling Psychology, 77(4), 543–49.
52, 310–21. Lipsey, M. W., & Wilson, D. B. (2001). Practical
∗ meta-analysis. Applied Social Research Methods
Joyce, A. S., Piper, W. E., & Ogrodniczuk, J. S.
(2007). Therapeutic alliance and cohesion vari- Series, Vol. 49). Thousand Oaks, CA: Sage.
ables as predictors of outcome in short-term Lorentzen, S., Høglend, P., & Ruud, T. (2008,
group therapy. International Journal of Group June). The efficacy of short- and long-term
Psychotherapy, 57(3), 269–96. psychodynamic group psychotherapy: A Norwegian
∗ multi-center study. Panel presentation at the
Kipnes, D. R., Piper, W. E., & Joyce, A. S. (2002).
Cohesion and outcome in short-term psychody- annual meeting of the Society for Psychotherapy
namic groups for complicated grief. International Research, Barcelona, Spain.
∗
Journal of Group Psychotherapy, 52(4), 483–509. Lorentzen, S., Sexton, H. C., & Hoglend, P.
Kivlighan, D. M., Coleman, M. N., & (2004). Therapeutic alliance, cohesion and
Anderson, D. C. (2000). Process, outcome, and outcome in a long-term analytic group. A prelim-
methodology in group counseling research. In inary study. Nordic Journal of Psychiatry,
S. D. Brown & R. W. Lent (Eds.), Handbook 58, 33–40.
of counseling psychology (3rd ed., pp. 767–96). MacKenzie, K. R. (1981). Measurement of
New York: Wiley. group climate. International Journal of Group
∗ Psychotherapy, 31, 287–95.
Kivlighan, D. M., & Lilly, R. L. (1997).
Developmental changes in group climate as MacKenzie, K. R. (1983). The clinical application
they relate to therapeutic gain. Group Dynamics: of a group climate measure. In R. R. Dies &
Theory, Research, and Practice, 1, 208–21. K. R. MacKenzie (Eds), Advances in group
Kosters, M., Burlingame, G., Nachtigall, C., & psychotherapy: Integrating research and practice
Strauss, B. (2006). A meta-analytic review of the (pp. 159–70). New York: International
effectiveness of inpatient group psychotherapy. Universities Press.
∗
Group Dynamics: Theory, Research, & Practice. Mackenzie, R. K., & Tschuschke, V. (1993).
10(2), 143–63. Relatedness, group work, and outcome in long-
Krogel, J. (2009). The Group Questionnaire: A new term inpatient psychotherapy groups. Journal of
measure of the group relationship. Provo, UT: Psychotherapy Practice and Research, 2, 147–56.
∗
Brigham Young University. Marmarosh, C., Holtz, A., & Schottenbauer, M.
Krogel, J., & Burlingame, G. (2009). The group (2005). Group cohesiveness, group-derived
questionnaire: A new measure of the therapeutic collective self-esteem, group-derived hope, and
group relationship. Paper presented at the annual the well-being of group therapy members. Group
meeting of the Society for Psychotherapy Research. Dynamics: Theory, Research, and Practice, 9(1),
Santiago, Chile. 32–44.
∗
Lambert, M. J., Hansen, N. B., Umpress, V., Marziali, E., Munroe-Blum, H., & McCleary, L.
Lunnen, K., Okiiski, J., & Burlingame, G. M. (1997). The contribution of group cohesion and
(1997). Administration and scoring manual for group alliance to the outcome of group psycho-
the OQ-45.2. Stevenson, MD: American therapy. International Journal of Group
Professional Credentialing Services. Psychotherapy, 47(4), 475–97.
∗ ∗
Levenson, J. S., & Macgowan, M. J. (2004). May, A. M., Duivenvoorden, H. J., Korstjens, I.,
Engagement, denial, and treatment progress Weert, E. V., Hoekstra-Weebers, J. E.,
among sex offenders in group therapy. Sexual Borne, B. V. D., et al. (2008). The effect of group
Abuse: A Journal of Research and Treatment, cohesion on rehabilitation outcome in cancer
16(1), 49–63. survivors. Psycho-Oncology, 17, 917–25.
McClendon-Theobald, D., & Burlingame, G. (in Piper, W. E., Marrache, M., Lacroix, R.,
press). Group climate: Construct in search of Richardsen, A. M., & Jones, B. D. (1983).
clarity. In R. Conyne (Ed.), Oxford handbook of Cohesion as a basic bond in groups. Human
group counseling. New York: Oxford University Relations, 36, 93–108.
∗
Press. Ratto, R., & Hurley, J. R. (1995). Outcomes of
McNair, D. M., Lorr, M., & Droppleman, L. F. inpatient group psychotherapy associated with
(1981). Profile of mood states manual. San Diego, dispositional and situational affiliativeness.
CA: Educational and Industrial Testing Group, 19(3), 163–72.
∗
Services. Rice, A. H. (2001). Evaluating brief structured
McRoberts, C., Burlingame, G. M., & Hoag, M. J. group treatment for depression. Research on
(1998). Comparative efficacy of individual and Social Work Practice, 11(1), 53–78.
∗
group psychotherapy: A meta-analytic perspec- Roether, H. A., & Peters, J. J. (1972). Cohesiveness
tive. Group Dynamics, 2(2), 101–117. and hostility in group psychotherapy. American
Moos, R. H. (1986). Group Environment Scale Journal of Psychiatry, 128(8), 1014–1017.
manual (2nd ed.). Palo Alto, CA: Consulting Rosenberg, M. (1965). Society and the adolescent self-
Psychologists Press. image. Princeton, NJ: Princeton University Press.
∗
Moos, R., & Humphrey, B. (1974). Group Rugel, R. P., & Barry, B. (1990). Overcoming
environment scale, Palo Alto, CA: Consulting denial through the group: A test of acceptance
Psychologists Press. theory. Small Group Research, 21(1), 45–58.
∗
Mullen, B., & Copper, C. (1994). The relationship Ryum, T., Hagen, R., Nordahl, H. M., Vogel, P. A.,
between group cohesiveness and performance: & Stiles, T. C. (2009). Perceived group climate as
An integration. Psychological Bulletin, 115, a predictor of long-term outcome in randomized
210–27. controlled trial of cognitive-behavioral group
∗
Norton, P. J., Hayes, S. A., & Springer, J. R. therapy for patients with comorbid psychiatric
(2008). Transdiagnostic cognitive-behavioral disorders. Behavioural and Cognitive Psychotherapy,
group therapy for anxiety: Outcome and process. 37, 497–510.
International Journal of Cognitive Therapy, Santarsiero, L. J., Baker, R. C., & McGee, T. F.
1(3), 266–79. (1995). The effects of cognitive pretraining on
∗
Oei, T. P. S., & Browne, A. (2006). Components of cohesion and self-disclosure in small groups: An
group processes: Have they contributed to the analog study. Journal of Clinical Psychology,
outcome of mood and anxiety disorder patients 51(3), 403-409.
in a group cognitive-behaviour therapy program? Schauenburg, H., Sammet, I., Rabung, S., &
American Journal of Psychotherapy, 60(1), 53–70. Strack, M. (2001). Zur differentiellen Bedeutung
∗
Ogrodniczuk, J., & Piper, W. (2003). The effect of des Gruppenerlebens in der stationaren
group climate on outcome in two forms of short- Psychotherapie depressiver Patienten [On the
term group therapy. Group Dynamics: Theory, differential importance of group experience in
Research, and Practice, 7, 64–76. inpatient psychotherapy of depressive patients].
∗
Ogrodniczuk, J. S., Piper, W. E., & Joyce, A. S. Gruppenpsychotherapie und Gruppendynamik,
(2005). The negative effect of alexithymia on the 37, 349–64.
outcome of group therapy for complicated grief: Selvin, H. C., & Hagstrom, W. O. (1963). The
What role might the therapist play? Comprehensive empirical classification of formal groups.
Psychiatry, 46, 206–213. American Sociological Review, 28, 399–411.
∗
Ogrodniczuk, J. S., Piper, W. E., & Joyce, A. S. Silbergeld, S., Koenig, G. R., Manderscheid, R. W.,
(2006). Treatment compliance among patients Meeker, B. F., & Hornung, C. A. (1975).
with personality disorders receiving group Assessment of environment-therapy systems: The
psychotherapy: What are the roles of inter- Group Atmosphere Scale. Journal of Consulting
personal distress and cohesion? Psychiatry: and Clinical Psychology, 43, 460–69.
Interpersonal and Biological Processes, 69(3), Snijders, J. A., Trijsburg, R. W., De Groot, M. H.,
249–61. & Duivenvoorden, H. J. (2005, March). Group
Palmer, K. D., Baker, R. C., & McGee, T. F. (1997). cohesion, working alliance, and therapeutic
The effects of pretraining on group psychother- interventions as factors predicting outcome in
apy for incest-related issues. International Group group psychotherapy for personality disorders:
Psychotherapy, 47(1), 71–89. First results of a follow-along study (2003–2006).
Presented at AGPA Research Symposium, Questionnaire (GCQ). The Netherlands:
New York). University of Amsterdam/Rotterdam.
∗
Sternburg, S., & Trijburg, W. (2005). The relation- Tschuschke, V., & Dies, R. R. (1994) Intensive
ship between therapeutic interventions and thera- analysis of therapeutic factors and outcome in
peutic outcome. Unpublished manuscript. long-term inpatient groups. International Journal
Stokes, J. (1983). Components of Group Cohesion: of Group Psychotherapy, 44(2), 185–208.
∗
Intermember attraction, instrumental value van Andel, P., Erdman, R. A. M., Karsdorp, P. A.,
and risk taking. Small Group Research, 14 (2), Appels, A., & Trijsburg, R. W. (2003). Group
163-173. cohesion and working alliance: Prediction of
Strauss, B., Burlingame, G., & Bormann, B (2008). treatment outcome in cardiac patients receiving
Using the CORE-R batter in group psycho- cognitive behavioral group psychotherapy.
therapy. Journal of Clinical Psychology, 64(11), Psychotherapy and Psychosomatics, 72, 141–49.
∗
1225–37. Woody, S. R., & Adessky, R. S. (2002). Therapeutic
∗
Taft, C. T., Murphy, C. M., King, D. W., alliance, group cohesion, and homework compli-
Musser, P. H., & DeDeyn, J. M. (2003). Process ance during cognitive-behavioral group treatment
and treatment adherence factors in group of social phobia. Behavior Therapy, 33, 5–27.
∗
cognitive-behavioral therapy for partner violent Wright, T. L., & Duncan, D. (1986). Attraction
men. Journal of Consulting and Clinical Psychology, to group, group cohesiveness, and individual
71(4), 812–20. outcome: A study of training groups. Small
∗
Taube-Schiff, M., Suvak, M. K., Antony, M. M., Group Research, 17(4), 487–92.
∗
Bieling, P. J., & McCabe, R. E. (2007). Group Yalom, I., Houts, P. S., Zimerberg, S. M., &
cohesion in cognitive-behavioral group therapy Rand, K. H. (1967). Prediction of improvement
for social phobia. Behaviour Research and Therapy, in group therapy: An exploratory study. Archives
45, 687–98. of General Psychiatry, 17, 159–68.
Trijsburg, R. W., Bogaerds, H., Letiche, M., Yalom, I. D., & Leszcz, M. (2005). The theory and
Bidzjel, L., & Duivenvoorden, H. J. (2004). practice of group psychotherapy (5th ed.). New York:
De ontwikkeling van de Group Cohesion Basic Books.
C HA P TER
6 Empathy
Robert Elliott, Arthur C. Bohart, Jeanne C. Watson, and Leslie S. Greenberg
Psychotherapist empathy has had a long (Decety & Ickes, 2009), which we will
and sometimes stormy history in psycho- address briefly in the next section.
therapy. Proposed and codified by Rogers
and his followers in the 1940s and 1950s, Definitions and Measures
it was put forward as the foundation of Defining Empathy
helping skills training popularized in the The first problem with researching empathy
1960s and early 1970s. Claims concerning in psychotherapy is that there is no con-
its universal effectiveness were treated with sensual definition (Batson, 2009; Bohart
skepticism and came under intense scru- & Greenberg, 1997; Duan & Hill, 1996).
tiny by psychotherapy researchers in the Recent neuroscience research on empathy
late 1970s and early 1980s. After that, begins to clarify some of the conceptual
research on empathy went into relative confusion, as a result of the concerted
eclipse, resulting in a dearth of research efforts of researchers using a variety of
between 1975 and 1995 (Duan & Hill, methods ranging from performance tasks,
1996; Watson, 2001). self-report, and neuropsychological assess-
Since the mid-1990s, however, empathy ment to fMRI and transcranial stimulation.
has once again become a topic of scientific Research examining the brain correlates of
interest in developmental and social psy- different component subprocesses of empa-
chology (e.g., Bohart & Greenberg, 1997; thy (Decety & Ickes, 2009) has extended
Ickes, 1997), particularly because empathy the initial discovery of “mirror neurons” in
came to be seen as a major part of “emotional the motor cortex of macaque monkeys
intelligence” (Goleman, 1995). We believe (e.g., Gallese, Fadiga, Fogassi, & Rizzolatti,
the time is ripe for the reexamination and 1996) to a broader range of affective and
rehabilitation of therapist empathy as a key perspective-taking components of empa-
change process in psychotherapy (Bohart thy in humans (Decety & Lamm, 2009).
& Greenberg, 1997). Indeed, the meta- The result of this research has been to
analytic results we will present clearly sup- deepen and clarify our understanding of
port such a conclusion. The most important therapist empathic processes (Watson &
development in the past 10 years, however, Greenberg, 2009), resulting in a growing
is the emergence of active scientific research consensus (e.g., Eisenberg & Eggum, 2009)
on the biological basis of empathy, as part that it consists of three major subprocesses,
of the new field of social neuroscience each with specific sets of neuroanatomical
132
correlates. First, there is an emotional simu- automatic, bodily-based emotional simula-
lation process that mirrors the emotional tion processes.
elements of the other’s bodily experience Nevertheless, it is easy to see both
with brain activation centering in the processes in Rogers’ (1980) definition of
limbic system (amygdala, insula, anterior empathy:
cingulate cortex) and elsewhere (Decety &
“the therapist’s sensitive ability and
Lamm, 2009; Goubert, Craig, & Buysse,
willingness to understand the client’s
2009). Second, a conceptual, perspective-
thoughts, feelings and struggles from
taking process operates, particularly local-
the client’s point of view. [It is] this ability
ized in medial and ventromedial areas of
to see completely through the client’s
prefrontal cortex as well as the temporal
eyes, to adopt his frame of reference . . .”
cortex (Shamay-Tsoory, 2009). Third, there
(p. 85) . . .“It means entering the private
is an emotion-regulation process that people
perceptual world of the other . . . being
use to reappraise or soothe their personal
sensitive, moment by moment, to the
distress at the other person’s pain or dis-
changing felt meanings which flow in
comfort, allowing them to mobilize com-
this other person . . . It means sensing
passion and helping behavior for the other
meanings of which he or she is scarcely
(probably based in orbitofrontal cortex, as
aware. . . .” (p. 142).
well as in the prefrontal and right inferior
parietal cortex). Defined this way, empathy is a higher-
Interestingly, the two therapeutic order category, under which different sub-
approaches that have most focused on types, aspects, expressions, and modes can
empathy—client-centered therapy and be nested. There are different ways one
psychoanalytic—have emphasized its cog- can put oneself into the shoes of the other:
nitive or perspective-taking (Selman, 1980) emotionally, cognitively, on a moment-
aspects, as well as its feeling aspects. That is, to-moment basis, or by trying to grasp an
they have focused on empathy as connected overall sense of what it is like to be that
knowing (Belenky et al., 1986), under- person. Within these subtypes different
standing the client’s frame of reference or aspects of the client’s experience can become
way of experiencing the world. By some the focus of empathy (Bohart & Greenberg,
accounts, 70% or more of Carl Rogers’ 1997). Similarly, there are many different
responses were to meaning rather than to ways of expressing empathy, including
feeling, despite the fact that his mode of empathic reflections, empathic questions,
responding is typically called “reflection of experience-near interpretations, empathic
feeling” (Brodley & Brody, 1990; Hayes & conjectures, as well as the responsive use of
Goldfried, 1996; Tausch, 1988). However, other therapeutic procedures. Accordingly,
understanding clients’ frames of reference empathy is best understood as a complex
does include understanding their affective construct consisting of a variety of different
experiences. In addition, empathy and acts used in different ways.
sympathy have typically been sharply dif- We distinguish between three main
ferentiated, with therapists such as Rogers modes of therapeutic empathy: empathic
disdaining sympathy but prizing empathy rapport, communicative attunement, and
(Shlien, 1997). In affective neuroscience person empathy. First, for some thera-
terms, this means that therapists in this pists empathy is primarily the establish-
tradition have often emphasized conscious ment of empathic rapport and support. The
perspective-taking processes over the more therapist exhibits a compassionate attitude
e l l i ot t, b o h a rt, wats o n , g re e n b e rg 133

toward the client and tries to demonstrate measures have been developed. Within
that he or she understands the client’s expe- psychotherapy, the measures of therapist
rience, often in order to set the context for empathy fall into four categories: empathy
effective treatment. A second mode of rated by nonparticipant raters; client-rated
empathy consists of an active, ongoing effort empathy; therapists rating their own empa-
to stay attuned on a moment-to-moment thy; and empathic accuracy (congruence
basis with the client’s communications and between therapist and client perceptions of
unfolding experience. Client-centered and the client).
experiential therapists are most likely to Observer-Rated Empathy. Some of the
emphasize this form of empathy. The thera- earliest observer measures of empathy were
pist’s attunement may be expressed in many those of Truax and Carkhuff (1967) and
different ways, but most likely in empathic Carkhuff and Berenson (1967). These scales
responses. The third mode, person empathy asked raters to decide if the content of
(Elliott, Watson, Goldman, & Greenberg, the therapist’s response detracts from the
2003) or experience-near understanding of client’s response, is interchangeable with it,
the client’s world, consists of a sustained or adds to or carries it forward. Typically,
effort to understand the kinds of experi- trained raters listened to 2–5 minute sam-
ences the client has had, both historically ples from session tapes. Samples are usually
and presently, that form the background of drawn from the beginning, middle, and/or
the client’s current experiencing. The ques- the end of therapy. Scales such as these do
tion is: How have the client’s experiences not adequately reflect the client-centered
led him or her to see/feel/think and act as conception of empathy as an attitude
he or she does? This is the type of empathic because they focus narrowly on a particular
understanding emphasized by psychody- kind of response, often empathic reflec-
namic therapists. However, empathic rap- tions. Furthermore, the equation of a par-
port, communicative attunement, and person ticular response with empathy has also
empathy are not mutually exclusive, and the made these scales less appropriate for mea-
differences are a matter of emphasis. suring empathy in approaches other than
Many other definitions for empathy have client centered (Lambert, De Julio, & Stein,
been advanced: as a trait or response skill 1978).
(Egan, 1982; Truax & Carkhuff, 1967), as More recent observer empathy measures
an identification process of “becoming” the are based on broader understandings of
experience of the client (Mahrer, 1997), forms of empathic responding. Elliott and
and as a hermeneutic interpretive process colleagues’ (1982) measure breaks empathy
(Watson, 2001). Perhaps the most practical down into component elements and has
conception, and one that we will draw on shown good psychometric properties, but
in our meta-analysis, is Barrett-Lennard’s it has not been widely used. Watson and
(1981) operational definition of empathy Prosser (2002) developed a promising new
in terms of three different perspectives: that observer-rated measure of empathy that
of the therapist (empathic resonance), the assesses therapists’ verbal and nonverbal
observer (expressed empathy), and the behavior and shows convergent validity
client (received empathy). with client ratings on the Barrett-Lennard
Relationship Inventory.
Measuring Empathy In addition, the therapist’s general empa-
Reflecting the complex, multidimensional thy can also be rated by others who know or
nature of empathy, a confusing welter of have supervised the therapist. For instance,
therapists’ empathic capacities can be rated comparing these ratings to how clients
by their supervisors (Gelso, Latts, Gomez, actually rated themselves. For instance, one
& Fassinger, 2002). For purposes of our study compared how therapists rated clients
meta-analysis, we lumped together all on Kelly’s REP grid with how clients rated
observer perspective measures of empathy. themselves (Landfield, 1971). The measure
Client Ratings. The most widely used of empathy is the degree of congruence
client-rated measure of empathy is the between therapist and client ratings. This
empathy scale of the Barrett-Lennard can be referred to as predictive empathy,
Relationship Inventory (BLRI). Other because the therapist is trying to predict
client rating measures have been developed how clients will rate themselves. This is
(e.g., Hamilton, 2000; Lorr, 1965; Persons closer to a measure of the therapist’s ability
& Burns, 1985; Truax & Carkhuff, 1967). to form a global understanding of what it
Rogers (1957) hypothesized that clients’ is like to be the client (person empathy)
perceptions of therapists’ facilitative con- than it is to a process measure of ongoing
ditions (positive regard, empathy, and communicative attunement.
congruence) predict therapeutic outcome. Recent work on empathic accuracy,
Accordingly, the BLRI, which measures however, does provide a predictive mea-
clients’ perceptions, is an operational defi- sure of communicative attunement (Ickes,
nition of Rogers’ hypothesis. In several 1997, 2003). This line of research typically
earlier reviews, including our meta-analysis employs a tape-assisted recall procedure in
in the previous edition of this book, client- which therapists or observers’ moment-to-
perceived empathy predicted outcome better moment empathy is measured by compar-
than observer- or therapist-rated empathy ing their perceptions of client experiences
(Barrett-Lennard, 1981; Gurman, 1977; to clients’ reports of those experiences.
Bohart, Elliott, Greenberg, & Watson, Unfortunately, no process–outcome studies
2002; Orlinsky, Grawe, & Parks, 1994; using this promising but time-consuming
Orlinsky & Howard, 1978, 1987). method have yet been carried out.
Therapist Ratings. Therapist empathy Correlations among Different Empathy
self-rating scales are not so common, but Measures. Intercorrelations of different
the BLRI does have one. Earlier reviews empathy measures have generally been
(Barrett-Lennard, 1981; Gurman, 1977) weak. Low correlations have been reported
found that therapist-rated empathy nei- between cognitive and affective measures
ther predicted outcome nor correlated (Gladstein et al., 1987) and between pre-
with client-rated or observer-rated empa- dictive measures and the BLRI (Kurtz &
thy. However, we previously found that Grummon, 1972). Other research has
therapist-rated empathy did predict out- found that tape-rated measures correlate
come, but at a lower level than client or only moderately with client-perceived
observer ratings (Bohart et al., 2002). empathy (Gurman, 1977). These weak
Empathic Accuracy. Several studies use correlations are not surprising when one
measures of therapist–client perceptual con- considers what the different instruments
gruence, commonly referred to as “empathic are supposed to be measuring. Trying to
accuracy” (Ickes, 1997, 2003). These typi- predict how a client will fill out a symptom
cally consist of therapists rating clients check list is very different from responding
as they think the clients would rate them- sensitively and tentatively in a way that
selves on various measures, such as person- demonstrates subtle understanding of what
ality scales or lists of symptoms, and then the client is trying to communicate, while

checking and adjusting one’s emerging Clinical Example
understanding with that of the client. Mark presented to psychotherapy com-
Similarly, client ratings of therapist under- plaining of pervasive anxiety. He was a
standing may be based on many other 30-year-old unmarried man who had been
things than the therapists’ particular skill struggling since his early 20s to break into
in empathic reflection. Accordingly, we the movie business. When he entered ther-
should not expect different measures of this apy, he was working as a waiter. He came
complex construct to correlate (Gladstein from a traditional family, living in the
et al., 1987). southern United States. His brothers and
Confounding between Empathy and Other sisters all had successful careers and were
Relationship Variables. A related concern is married, with children. His parents were
the distinctiveness of empathy from other constantly pestering him about his not
relationship constructs. One early review being married and not having a stable
of more than 20 studies primarily using career. His anxiety attacks had begun a few
the BLRI found that, on average, empathy weeks after a visit home for the Christmas
correlated 0.62 with congruence and 0.53 holidays. When Mark came to his first
with positive regard, and 0.28 with uncon- appointment, he was clearly agitated. He
ditionality (Gurman, 1977). Factor analysis had previously called and had sounded
of scale scores found that one global factor desperate over the phone. The therapist
typically emerged, with empathy loading initially was concerned that Mark might be
on it along with congruence and posi- in a state of crisis.
tive regard (Gurman, 1977). Others have The therapist’s orientation was integra-
reported that the empathy scale loaded tive experiential/humanistic, based in the
0.93 on a global BLRI factor, with Positive principles of person-centered therapy. The
Regard loading 0.87 and Congruence load- therapist tried to understand the client’s
ing 0.92 (Blatt et al., 1996). Such results point of view actively and empathically
suggest that clients’ perceptions of empathy and to share that understanding, using her
are not clearly differentiated from their per- attunement to the client’s experience to
ceptions of other relationship factors. identify effective interventions, and to stay
On the other hand, reviews of several responsively attuned so that therapeutic
factor analytic studies where, instead of procedures could be adjusted to maximize
using scale scores, specific items were used learning. The following are two examples
have found empathy emerging as a separate of the therapist’s utilization of empathic
factor (Gurman, 1977). In addition, empa- responding during the first session:
thy tends to correlate more highly with
the bond component of the therapeutic C1: I’m really in a panic (anxious,
alliance than with the task and goal compo- looking plaintively at the therapist).
nents (Horvath & Greenberg, 1986). I feel anxious all the time. Sometimes
Thus, there is evidence both for and it seems so bad I really worry that
against the hypothesis that the Rogerian I’m on the verge of a psychotic break.
triad of empathy, unconditional positive I’m actually afraid of completely
regard, and congruence are separate and falling apart. Nothing like this has
distinct variables. We view empathy as a ever happened to me before. I always
relationship component that is both con- felt in charge of myself before, but
ceptually distinct and part of a higher-order now I can’t seem to get any control
relationship construct. over myself at all.
T1: So a real sense of vulnerability— During the first few sessions, the client
kind of like you don’t even know had repeatedly expressed the suspicion that
yourself anymore. something about his early relationships
C2: Yes! That’s it. I don’t know myself with his parents played an important role
anymore. I feel totally lost. The in his current problems. Initially, the thera-
anxiety feels like a big cloud that just pist had not taken this too seriously, since
takes me over, and I can’t even find progress was apparently being made through
myself in it anymore. I don’t even the collaborative use of other procedures.
know what I want, what I trust . . . Because the therapist was not psychoge-
I’m lost. netic and past oriented, she had not tuned
T2: Totally lost, like, “Where did Mark into this. The therapist’s lack of person
go? I can’t find myself anymore.” empathy (i.e., grasping of how figural this
C3: No, I can’t (sadly, and was for the client within the client’s frame
thoughtfully). of reference) for the larger meaning of the
client’s interest in this topic had effectively
The dialogue continued like this, and shut off this avenue of exploration.
soon the therapist’s empathic recognition Eventually, the therapist listened,
provided the client with a sense of being responded in an invitational way to the
understood. This fostered a sense of safety, client, and the client began to explore his
and gradually the client moved from agita- childhood. This illustrates how empathy
tion into reflective sadness. The client then not only gives permission, but also provides
began to reflect on his experience in a more active support for exploration. It also illus-
productive, exploratory manner. He talked trates how sensitive empathic understand-
about the basic conflict in his life: over ing of the client’s way of seeing the problem
whether to continue to pursue an acting is sometimes crucial for therapeutic prog-
career or to find a “real job” and life partner, ress (Hubble, Duncan, & Miller, 1999).
given that he was now 30 and had shown no This led to a breakthrough moment. In
signs of making a breakthrough in acting. reviewing his childhood, Mark became
Later, the client role-played a dialogue emotionally aware of how neglected he
between two sides of himself. One side, his had felt as a child by his high-achieving
critic or “should” side, said that he should parents, who were not mean and cruel, but
get a stable job and get married and criti- who were not themselves highly empathic.
cized him for not being married. The other As a child, the client had always been
side was the “want” side—or in this case, unusually interested in fantasy activities,
the “don’t want” side—which said “I don’t and was a rather “inner” person, in contrast
want to live an ordinary life; I want to live to his siblings, who were more conventional
a creative life.” This side came out in the and high-achievers at school. The parents
form of defensive rebellion. Empathic had not known what to make of their
sensitivity was used to help the therapist unique child and were unable to respond in
tune into the client’s point of view and an empathic and supportive way to his
to focus the client’s exploratory activities emerging uniqueness.
during the role-play. What emerged from The result was that he had had to adopt
this role-play was that there was a longing a defensive “I have a right to be different”
for a “normal” lifestyle underlying Mark’s attitude. He was rarely able to genuinely
defensive rebellion, in conflict with a desire consider whether he wanted to be conven-
to do something creative. tional or not. Underlying this was a longing

for conventionality. Accessing this in the “counseling,” or “counselling”. Additionally,
context of his family life helped him accept we consulted the tables of contents of rele-
that he was different and to mourn the fact vant journals such as: Psychotherapy, Person-
that he was not conventional (and, in effect, Centered Journal, Psychotherapy Research,
mourn that he might never be what his Journal of Counseling Psychology, and Person-
family wanted him to be). Over the course Centered and Experiential Psychotherapies.
of this work, Mark’s anxiety decreased.
Eventually he made a decision to continue Inclusion Criteria
to pursue an acting career, for a while at Our inclusion criteria were as follows:
least; and his crisis abated. (a) a specific measure of empathy was used,
(b) empathy was related to some measure
Meta-Analytic Review of therapy outcome, (c) the client sample
In this section we report the results of an involved genuine clinical problems, (d) the
original meta-analysis conducted on available average number of sessions was three or
research relating empathy to psychother- more, (e) the study was available in English,
apy outcome. We addressed the following (f ) the study included at least five clients,
questions: (a) What is the association (g) the study was available in published
between therapist empathy and client out- form, and (h) the study contained suffi-
come? (b) Do different forms of psycho- cient information to calculate a weighted
therapy yield different levels of association effect size.
between empathy and outcome? (c) Does
the type of empathy measure predict the Characteristics of the Studies
level of association between empathy To examine variables that might moderate
and outcome? (d) What other study and the empathy–outcome association, we eval-
sample characteristics predict an association uated the studies on a wide range of sample
between empathy and outcome (i.e., sample and methodological features. For measures
size, treatment setting, therapy format and of outcome, we included a study as long
length, level of client severity, therapist as there was some assessment of the effects
experience, type of outcome measure, unit of therapy, even if only at the session level
of process)? (immediate outcome). For example, we
included abstinence from drinking (Miller
Search Strategy et al., 1980), level of depression (Burns &
Articles were culled from previous reviews Nolen-Hoeksema, 1992), MMPI scores
(Beutler, Crago, & Arizmendi, 1986; (Kiesler et al., 1967), client satisfaction
Gurman, 1977; Lambert, DeJulio & Stein, (Lorr, 1965), supervisors’ ratings of client
1978; Mitchell, Bozarth, & Krauft, 1977; improvement (Gelso et al., 2002), client
Orlinsky & Howard, 1986; Orlinsky, and therapist posttherapy ratings of amount
Grawe, & Parks, 1994; Parloff et al., 1978; of change (Hamilton, 2000), and postses-
Truax and Mitchell, 1971; N. Watson, sion ratings of progress (Orlinsky & Howard,
1984). We also searched PsycINFO and 1967). There is some conceptual overlap
PsycLIT forward from 1992 (2 years between feeling understood and client
before the publication of the last major satisfaction, but this one outcome measure
review of empathy research in Orlinsky represented only 6% of effects; we subse-
et al., 1994), using the search terms, “empa- quently examined type of outcome mea-
thy” or “empathic” and “psychotherapy,” sure as a moderator variable. The resulting
sample consisted of 224 separate tests of Coding Procedure
the empathy–outcome association, aggre- The following variables were coded: therapy
gated into 59 different samples of clients format (individual or group); theoretical
(from 57 studies) and encompassing a total orientation; experience level of therapists;
of 3,599 clients. Table 6.1 summarizes rel- treatment setting (inpatient, outpatient);
evant study characteristics. number of sessions (typically the mean); type
of problems (mixed neurotic, depression,
Estimation of Effect Size anxiety, severe problems such as psychosis);
For effect sizes, we used Pearson correla- source of outcome measure (therapist rating,
tions if available. Our strategy was to extract client rating, objective, and other mea-
all possible effects. Therefore, we used the sures); when outcome was measured (e.g.,
following conventions (extensions of those postsession, posttherapy, followup); type
used in Smith, Glass, & Miller, 1980) to of outcome measured (symptom change,
estimate r : First, if we had a significance improvement, global); source of empathy
level, we converted it to r. If the result was measure (objective ratings, therapist, client,
nonsignificant, but we had enough infor- therapist/client congruence, trait measure);
mation to calculate a t and then convert, and unit of measure (2–5 minute samples,
we did so. If we had no other information session, therapy to date).
than that the effect was nonsignificant, we We conducted two sets of analyses: by
set r at 0. If the authors indicated a “nonsig- effects and by studies. First, we analyzed
nificant trend” but did not report a correla- the 224 separate effects in order to examine
tion (for instance, a key study, Kiesler et al., the impact of perspective of empathy mea-
1967, indicated several trends on MMPI surement and type of outcome. Second,
scales), we estimated the trend by assigning study-level analyses used averaged indi-
an ES of half the size of a significant r. vidual effects within client samples using
Table 6.1 Study Characteristics

Parametric characteristics: M SD Range
Sample size: 61 59.6 6–320
Length of therapy (sessions) 24 42.4 3–228
Effects per study 3.8 5.7 1–42
Categorical characteristics: Modal categories %
Time period (range: 1961–2008) Before 1980 49
Theoretical orientation Mixed, eclectic, or unknown 40
Modality Individual 74
Client presenting problem Mixed neurotic (mixed anxiety/depression) 40
Therapist experience level Recent PhD or M.D. 36
Outcome assessment time point Posttreatment 60
Empathy perspective Client (mostly Barrett-Lennard) 39
(Observer, mostly Truax-Carkhuff: 34%)
Empathy measurement unit Therapy to date 60

Fisher r-to-Z conversions to correct for dis- outcome. This effect size is on the same order
tributional biases before further analysis, of magnitude as, or slightly larger than, pre-
thus avoiding problems of nonindepen- vious analyses of the relationship between
dence and eliminating bias due to variable the alliance in individual therapy and treat-
numbers of effects reported in different ment outcome (i.e., Horvath, Flückiger, &
studies (Lipsey & Wilson, 2001; e.g., one Symonds, this volume, Chapter 2: 0.275;
study, Kurtz & Grummon, 1972, contri- Martin, Garske, & Davis, 2000: 0.22).
buted 42 effects). For analyses across stud- Overall, empathy typically accounts for
ies, we weighted studies by inverse error more outcome variance than do specific
and analyzed for heterogeneity of effects treatment methods (compare Wampold’s,
using Cochrane’s Q (following the Hunter- 2001, estimate of 1% to 8% for interven-
Schmidt method, using the program in tion effects).
Diener, Hilsenroth, & Weinberger, 2009), However, the 0.30 figure conceals sta-
and also I 2, an estimate of the proportion of tistically significant variability in effects,
variation due to true variability as opposed as indicated by a study-level Cochrane’s Q
to random error (Higgins, Thompson, Deeks, of 205.8 (p < 0.001); in addition, I 2 was
& Altman, 2003). Finally, where necessary 67%, considered to be a large value. This
in the correlational analyses of moderator means that a further examination of possi-
variables to correct for nonindependence, ble moderators of the empathy–outcome
we used effects weighted by the inverse of association is not only justified but is in fact
number of analyses per study. necessary (Lipsey & Wilson, 2001).
Results Moderators and Mediators

The single best summary value, as shown in We divide this section on moderators
Table 6.2, is the study-level, weighted r and mediators into two parts: meta-analytic
of 0.30, a medium effect size. Average analyses of moderator variables and thera-
effects were 0.22 for analyses of the 224 pist-mediating factors.
nonindependent separate effects, probably
an underestimate due to smaller effects Meta-Analytic Moderator Analyses
found in one study (Kurtz & Grummon, The significant Q and large I 2 statistics
1972). These values were very similar to our point to the existence of important moder-
previous review (Bohart et al., 2002) and ator variables or sources of heterogeneity
mean that in general empathy accounts but do not specify what those are. We began
for about 9% of the variance in therapy our search by testing the hypothesis that
Table 6.2 Empathy–Outcome Correlations: Overall Summary Statistics

Effect level (N = 224) Study level (N = 59)
N M SD M SD
Weighted mean r 0.22∗ 0.33 0.30∗ 0.13

Cochrane’s Q 646.22∗ 174.65∗
I2 65.49 66.79
∗
p < 0.001.
Note: Fisher’s r-to-z transformation used to calculate means and SDs. Weighted rs use inverse variance (i.e., n−3) as weights and are tested
against mean r = 0 following the Hunter-Schmidt method using Deiner’s (2010) program.
different empathy–outcome correlations than observer-rated measures (0.25; n = 27)
might be obtained for different theoretical and therapist measures (0.20; n = 11); each
orientations. For example, one might expect of these mean effects was significantly
the association to be larger in those thera- greater than zero (p < 0.001). In contrast,
pies for which empathy is held to be a key empathy accuracy measures were unrelated
change process, such as person-centered to outcome (0.08; n = 5, ns). Although the
therapies. However, our analyses, summa- overall Q value for between-group hetero-
rized in Table 6.3, turned up little evi- geneity was not significant, comparison of
dence of such a trend, but significant, large confidence intervals indicated that client-
amounts of nonchance heterogeneity within perceived empathy significantly predicted
the CBT and Other/Unspecified therapy outcome better than accuracy measures
samples. This finding contrasts with our (p < 0.05). A word of caution: All perspec-
previous meta-analysis (Bohart et al., 2002), tives except empathic accuracy are character-
where we found tantalizing evidence that ized by large (>50%), statistically significant
empathy might be more important to amounts of nonchance heterogeneity.
outcome in cognitive-behavioral therapies Clarification of the source of this heteroge-
than in others. However, our present analy- neity awaits further research; however, for
sis failed to confirm that conjecture but now it seems fair to say that clients’ feelings
points to important sources of variability of being understood and observer ratings
that need to be explored (and to a lesser extent, therapist impres-
In Table 6.4 we chart relations between sions) appear to carry significant weight as
specific types of empathy measures and far as outcome goes, but that empathic
outcome, using effect-level analyses aggre- accuracy measures do not, in spite of their
gated within studies (n = 82). As we intuitive appeal.
expected, and has been noted by previous Finally, in Table 6.5, we examine several
reviewers (e.g., Barrett-Lennard, 1981; other variables that might account for
Parloff, Waskow & Wolfe, 1978), the per- some of the heterogeneity of the effect sizes:
spective of the empathy rater made a differ- year of publication, sample size, outpa-
ence for empathy–outcome correlations. tient versus inpatient treatment, treatment
Specifically, client measures predicted out- format (individual vs. group), length of
come the best (mean corrected r = 0.32; therapy, client severity, therapist experi-
n = 38), slightly but not significantly better ence level, globalness of outcome measures
Table 6.3 Mean Effects across Theoretical Orientation

Theoretical orientation n Mean weighted r Within group Q I2
Experiential/ humanistic 8 0.26∗∗ 7.68 8.91
Cognitive-behavioral 10 0.31∗∗ 24.55∗ 63.34
Psychodynamic 4 0.19∗∗ 2.01 0
Other/unspecified 37 0.31∗∗ 138.01∗∗ 74.64
Between groups Q 2.39 (df = 3, 55, ns)
∗
p < 0.01; ∗∗p < 0.001.
Note: Mean correlations calculated using Fisher’s z-scores. Significance tests for mean correlations are against the null hypothesis of mean r = 0.
Q tests for heterogeneity are evaluated as a chi-square test, using Diener et al.’s (2009) program and the Hunter-Schmidt method. Within-
groups Q is analogous to a one-way ANOVA with study samples as levels; between-groups Q calculated as difference between total sample Q
and within-group Q, following Lipsey and Wilson (2001).

Table 6.4 Mean Effects across Empathy Measurement Perspectives
Measurement perspective n Mean weighted r Within group Q I2
Observer 27 0.25∗∗ 93.14∗∗ 72.09
Client 38 ∗∗ ∗∗ 69.00
0.32 119.35
Therapist 11 0.20∗∗ 21.05∗ 52.50
Empathic accuracy 5 0.08 5.91 32.35
Total 82 0.27 258.08∗∗ 68.61
Between-groups Q 2.39 (df = 3, 78, ns)
∗ ∗∗
p < 0.05; p < 0.001.
Note: See note for Table 6.3.
(individualized to satisfaction ratings), and possibility that empathy is slightly more

size of empathy unit (5 min segment to predictive of positive outcome in group
whole therapy). Using ordinary (that is, therapy, with more severely distressed cli-
unweighted correlations), none of these ents, in more recent studies, and with more
were statistically significant. On the other global outcome measures (i.e., satisfaction
hand, analyses using weighting for inverse ratings, which begin to overlap conceptu-
error (i.e., sample size minus 3) were sig- ally with empathy).
nificant for all variables except outcome On the other hand, it may be that the
globality and size of empathy unit; how- empathy relationship is slightly less predic-
ever, these suffer from nonindependence tive of positive outcome in inpatient set-
within studies and will require a substan- tings, and with more experienced therapists
tially larger set of studies or more sophisti- (study level mean r = −0.19; effect level =
cated, multilevel meta-analytic methods to −0.29); the latter is the largest of this set of
verify. Briefly, these analyses point to the correlations and is consistent with our 2002
Table 6.5 Correlations between Empathy–Outcome Effect Size and Selected Moderator Variables
Predictor Unweighted Weighted
r n r n
Year of publication 0.14 59 0.12∗ 3422
No. of clients in study 0.06 59 0.15∗ 3422
Setting (1 = outpatient; 2 = inpatient) −0.13 58 −0.08∗ 3305
Format (1 = individual; 2 = group) 0.12 54 0.15∗ 2807
Length of therapy (in sessions) 0.04 41 −0.08∗ 2074
Client severity (3-point scale) 0.10 41 ∗ 2320
0.14
Therapist experience level (6-point scale) −0.19 51 −0.29∗ 2820
Outcome globality (6-point scale: individualized to satisfaction ratings) 0.17 59 0.00 3360
Size of empathy unit (4-point scale) −0.06 59 -0.02 3443
∗
p < 0.001.
Note: Weighted analyses used inverse error (i.e., degrees of freedom) but are not corrected for nonindependence of participants within studies;
analyses of outcome globality and size of empathy unit analyses were also inverse weighted by number of effects per study to correct for
nonindependence of effects within studies.
meta-analysis. As we previously speculated, (Duan & Hill, 1996; Watson, 2001). Other
there are at least two possible reasons for research has shown that responses that are
this: To begin with, inexperienced thera- just ahead of the client seem to be more
pists may vary more in empathy, while effective than responses that are either at
smaller correlations for experienced thera- the same level as the client, or at a more
pists may reflect a restriction of range or global level (Sachse, 1990a, 1990b; Tallman
ceiling effect. Alternatively, experienced et al., 1994; Truax & Carkhuff, 1967). And
therapists may have developed additional a qualitative study of clients’ experience
skills such as effective problem solving, so of empathy, interrupting, failing to main-
that clients are more likely to forgive tain eye contact, and dismissing the client’s
empathic misattunements. position while imposing the therapist’s own
position were all perceived as unempathic
Therapist-Mediating Factors (Myers, 2000). Conversely, being nonjudg-
As noted earlier, affective neuroscience mental, attentive, open to discussing any
researchers have proposed that empathy topic, and paying attention to details were
involves three interlinked skills or processes: perceived as empathic.
affective simulation, perspective taking, and
regulation of one’s own emotions (Decety Client Contributions
& Jackson, 2004). Supporting this, research Clinical and research experience suggest that
has found a relationship between various the amount of therapist empathy varies as a
measures of cognitive complexity, such as function of the client. Early studies (Kiesler
those of perspective taking or abstract abil- et al.,1967), for example, found that levels
ity, and empathy in both developmental of empathy were higher with clients who
psychology and in psychotherapy (Eisenberg had less pathology, who were brighter, but
& Fabes, 1990; Henschel & Bohart, 1981; yet who were lower in self-esteem. Therefore,
Watson, 2001). With respect to affective the client him or herself almost certainly
simulation and emotion regulation, thera- influences therapist empathy. As Barrett-
pists who were open to conflictual, coun- Lennard (1981) pointed out, the client’s
tertransferential feelings were perceived revealing of their experiencing is an essential
as more empathic by clients (Peabody & link in the cycle of empathy. Clients who are
Gelso, 1982). more open to and able to communicate
The degree of similarity between therapist their inner experiencing will be easier to
and client (Duan & Hill, 1996; Gladstein empathize with. Empathy truly appears to
& Associates, 1987; Watson, 2001) also be a mutual process of shared communica-
influences the level of empathy. Similarity tive attunement (Orlinsky et al., 1994).
and familiarity between the target of empa- On the other hand, not all clients respond
thy and the empathizer have been found favorably to explicit empathic expressions. In
to be important modulators of empathy their review, Beutler, Crago, and Arizmendi
in neuroscientific studies of mirror neu- (1986, p. 279) cite evidence that suggests
rons (Watson & Greenberg, 2009). Another that “patients who are highly sensitive,
important factor is therapist nonlinguistic suspicious, poorly motivated, and reac-
and paralinguistic behavior. This encom- tive against authority perform relatively
passes therapists’ posture, vocal quality, abil- poorly with therapists who are particu-
ity to encourage exploration using emotion larly empathic, involved, and accepting.”
words, and the relative infrequency of talking Another study (Mohr & Woodhouse, 2000)
too much, giving advice, and interrupting found that some clients prefer businesslike

rather than warm, empathic therapists. It is with the research on empathy. In addition
worth noting, however, that when thera- to the well-known difficulty of inferring
pists are truly empathic they attune to their causality from correlational data, these
clients’ needs and accordingly adjust how entail: (a) the questionable validity of some
and how much they express empathy. outcome measures (e.g., client satisfaction);
More broadly, Duan and Hill (1996) (b) lack of appropriate, sensitive outcome
speculated that different types of empathy measures; (c) restricted range of predic-
may be hindering or helpful to clients at tor and criterion variables; (d) confounds
different times. Hill and her colleagues among variations in time of assessment,
(Hill et al., 1992; Thompson & Hill, 1991) experience of raters, and sampling meth-
found that when clients had negative ods; (e) reliance on obsolete diagnostic
in-session reactions to their therapists, the categories; and (f ) incomplete reporting
therapist’s awareness or understanding of of methods and results. In fact, these and
the reaction “led to interventions that were other problems are not restricted to empa-
perceived as less helpful than when the thy research but are common to all process–
awareness was absent” (p. 269). In such outcome research (Elliott, 2010).
relational ruptures, it is probably useful for The restricted range of predictor and
therapist empathy to be accompanied and criterion variables is particularly a problem.
deepened by genuine warmth, openness, In the Mitchell, Truax, Bozarth, and Krauft
and concern for the clients’ feelings, rather (1973) study, for instance, most of the
than defending oneself and blaming the therapists scored below the minimum con-
client (also see Safran, Muran, & Eubanks- sidered to be effective, and outcome was
Carter, this volume, Chapter 11). only modest to moderate in the study. It is
Keeping in mind the notion of empathy not surprising that no significant correla-
as not only getting inside the skin of the tions were found. Furthermore, in a few
client, but getting inside the skin of the cases, results were reported as either sig-
relationship (O’Hara, 1984), it may be that nificant in the positive direction or nonsig-
in some cases the therapist is more empathic nificant, possibly disguising weak negative
by not expressing empathy. Martin (2000, effects. This is particularly a problem for
pp. 184–185) notes: “Think of the insensi- calculating effect sizes based on limited
tive irony of a therapist who says, ‘I sense information, thus introducing error into
the sadness you want to hide. It seems like the process.
you don’t want to be alone right now The key question of whether empathy is
but you also don’t want somebody talking causally related to therapeutic outcome—as
to you about your sadness . . .’ ” This opposed to being merely a correlate of it—
response might technically seem empathic, cannot be answered definitively from our
but in fact at a higher level, it is unem- meta-analysis. This is the central limitation
pathic, controlling, and intrusive, because of the process–outcome research reviewed
it violates the client’s need for interpersonal here. However, data from several studies
distance. Variations among clients in desire shed light on the question. First, Burns and
for and receptivity to different expressions Nolen-Hoeksema (1992) and Cramer and
of empathy need further research. Takens (1992) have used causal modeling
(structural equation modeling, path analysis)
Limitations of the Research to explore the relationship between empa-
Many reviewers (e.g., Watson, 2001; thy and outcome. Second, in another study
Patterson, 1984) have discussed problems (Miller et al., 1980), ratings of therapist
empathy were made by supervisors before caring, transparent, empathic stance in all
and independent of knowing about out- professional contacts, it is therefore both
come data. Yet empathy showed a strong impractical and unethical to randomize
(r = 0.82) relationship to outcome in a clients to demonstrably empathic versus
cognitive-behavioral program for drinking. unempathic therapists. In such cases, meta-
Third, Anderson (1999) measured thera- analyses can provide a valid alternative
pists’ facilitative interpersonal skills, includ- to randomized clinical trials (Berman &
ing accurate empathy, before therapy, by Parker, 2002), providing that the identifi-
having them respond to videotapes of cation and analysis of observational studies
clients who presented in difficult interper- has been done carefully and systematically.
sonal ways. Anderson found statistically
significant relationships between this prior Therapeutic Practices
measure of therapist interpersonal skills The most consistent and robust evidence is
and client outcome in subsequent psycho- that clients’ perceptions of feeling under-
therapy, a finding recently replicated with a stood by their therapists relate favorably to
larger, practice-based sample of therapists outcome. As we have shown, empathy is
and clients (Anderson, Ogles, Patterson, a medium-sized predictor of outcome in
Lambert, & Vermeersch, 2009). psychotherapy. It also appears to be a gen-
On the other hand, Burns and Nolen- eral predictor across theoretical orientations,
Hoeksema (1992) note that structural treatment formats, and client severity levels.
equation modeling cannot definitely show This repeated finding, in both dozens of
causality but only explore and elaborate individual studies and now in multiple
particular causal models. Miller et al. (1980) meta-analyses, leads to a series of clinical
had supervisors rate supervisees’ levels of recommendations.
empathy, but it is possible that these ratings
were influenced by supervisees’ reports of • It is important for psychotherapists
how well therapy was going with the cli- to make efforts to understand their clients,
ents. In Anderson et al.’s (2009) study in and to demonstrate this understanding
which empathy was measured indepen- through responses that address the needs
dently of therapy, empathy is confounded of the client as the client perceives them
with other facilitative interpersonal skills. on an ongoing basis. The empathic
Even though empathy is the predominant therapist’s primary task is to understand
process in client-centered and related ther- experiences rather than words.
apy, it is not the only process. • Empathic therapists do not parrot
The evidence we have presented is clearly clients’ words back or reflect only the
compatible with a causal model implicat- content of those words; instead, they
ing therapist empathy as a mediating pro- understand overall goals as well as
cess leading to client change. It is true that moment-to-moment experiences,
correlational studies can only probe into or both explicit and implicit. Empathy
lend support for or against causal models entails capturing the nuances and
of therapeutic change. As is the case for implications of what people say, and
much of the behavioral sciences, establishing reflecting this back to them for their
conclusive evidence for particular hypoth- consideration.
esized causal processes is notoriously difficult • Empathic responses follow the
and may ultimately prove elusive. Insofar “moving point” of the focus of the
as codes of professional ethics stipulate a client’s concerns as therapy progresses.

• Research has identified a range of your boat from being sucked in
useful types of empathic responses, several or capsizing.
of which we illustrate here with a running
example. Empathic understanding responses Empathic explorations are attempts by
convey understanding of client experience. therapists to get at that which is implicit in
For example: clients’ narratives and focus on information
that has been in the background but not
Client: I have been trying to push yet articulated:
things away, but every time I sit
down to do something it is like C: I keep responding to him, like it’s
I forget what I am doing. against what I want to do.
Therapist: Somehow you are not in a T: Somehow you can’t let go. It is just
space to work, it’s hard for you to so hard to walk away.
concentrate. • Empathic therapists assist clients to
Empathic affirmations are attempts by the symbolize their experience in words and
therapist to validate the client’s perspective: track their emotional responses, so that
clients can deepen their experience and
C: And my cat is still lost, so we have reflexively examine their feelings, values,
been staying up at night in case he and goals. To this end they need to attend
returns, so last night was another to that which is not said, or that which is
night without sleep . . . and work has at the periphery of awareness as well as
been so busy and I have been so tired that which is said and is in focal awareness
and P needs my attention. I have (Watson, 2001).
been going around in circles and, oh, • Empathy entails individualizing
everything is just a big mess, you responses to particular patients. For
know? example, certain fragile clients may find
T: Yeah, really hard, being pulled in a the usual expressions of empathy too
million different directions and there intrusive, while hostile clients may find
hasn’t been time for you, no wonder empathy too directive; still other clients
it feels like things are a mess. may find an empathic focus on feelings
Empathic evocations try to bring the cli- too foreign (Kennedy-Moore &Watson,
ents’ experience alive using rich, evocative, 1999). Therapists therefore need to know
concrete, connotative language and often when—and when not—to respond
have a probing, tentative quality: empathically. When clients do not want
therapists to be explicitly empathic, truly
C: I don’t know what I’m going to do. empathic therapists will use their
I have two hundred dollars this perspective-taking skills to provide an
month, everything’s behind, there optimal therapeutic distance (Leitner,
isn’t enough work, and I have been 1995) in order to respect their clients’
doing other things, and then my Dad boundaries.
was here. Things are just swirling • There is no evidence that accurately
around me. I don’t know how to predicting clients’ own views of their
keep my stuff together enough for problems or self-perceptions is effective.
me even to survive. Therapists should neither assume that
T: It’s like being caught in a they are mind readers nor that their
whirlpool as if it is hard to keep experience of understanding the client
will be matched by the client feeling neuroscience of empathy (pp. 3–15). Cambridge,
understood. Empathy should always be MA: MIT Press.
Belenky, M. F., Clinchy, B. M., Goldberger, N. R.,
offered with humility and held lightly,
& Tarule, J. M. (1986). Women’s ways of know-
ready to be corrected. ing: The development of self, voice, and mind.
• Finally, because research has shown New York: Basic Books.
empathy to be inseparable from the other ∗
Bergin, A. E., & Jasper, L. G. (1969). Correlates of
relational conditions, therapists should empathy in psychotherapy: A replication. Journal
seek to offer empathy in the context of of Abnormal Psychology, 74, 477–81.
Berman, N.G., & Parker, R. (2002). Meta-analysis:
positive regard and genuineness. Empathy
Neither quick nor easy. BMC Medical Research
will not be effective unless it is grounded Methodology, 2(10). Accessed online on June
in authentic caring for the client. We 19, 2010 at http://www.biomedcentral.com/
encourage psychotherapists to value 1471-2288/2/10
empathy as both an “ingredient” of a Beutler, L. E., Crago, M., & Arizmendi, T. G.
healthy therapeutic relationship as well (1986). Research on therapist variables in psycho-
therapy. In S. L. Garfield & A. E. Bergin (Eds.),
as a specific, effective response that
Handbook of psychotherapy and behavior change
strengthens the self and deeper (3rd ed., pp. 257–310). New York: Wiley.
exploration. ∗
Beutler, L. E., Johnson, D. T., Neville, C. W., &
Workman, S. N. (1972). “Accurate empathy”
and the AB dichotomy. Journal of Consulting and
References Clinical Psychology, 38, 372–75.
An asterisk (∗) indicates studies included in Blatt, S. J., Zuroff, D. C., Quinlan, D. M., &
the meta-analysis. Pilkonis, P. A. (1996). Interpersonal factors in
Anderson, T. (1999). Specifying non-‘specifics’ in brief treatment of depression: Further analyses of
therapists: The effect of facilitative interpersonal the National Institute of Mental Health
skills in outcome and alliance formation. Paper Treatment of Depression Collaborative Research
presented at the 30th annual meeting of the Program. Journal of Consulting and Clinical
International Society for Psychotherapy Research, Psychology, 64, 162–71.
Braga, Portugal. Bohart, A. C., Elliott, R., Greenberg, L. S., &
Anderson, T., Ogles, B. M., Patterson, C. L., Watson, J. C. (2002). Empathy. In J. Norcross
Lambert, M. J., & Vermeersch, D. V. (2009). (Ed.), Psychotherapy relationships that work (pp.
Therapist effects: Facilitative interpersonal skills 89–108). New York: Oxford University Press.
as a predictor of therapist success. Journal of Bohart, A. C., & Greenberg, L. S. (1997). Empathy:
Clinical Psychology, 65, 755–68. Where are we and where do we go from here?
∗ In A. C. Bohart & L. S. Greenberg (Eds.),
Barrett-Lennard, G. (1962). Dimensions of thera-
pist response as causal factors in therapeutic Empathy reconsidered: New directions in psy-
change. Psychological Monographs, 76, 1–33, chotherapy (pp. 419–50). Washington, DC:
Whole Number 562. American Psychological Association.
Barrett-Lennard, G. T. (1981). The empathy cycle: Brodley, B. T., & Brody, A. F. (1990, August).
Refinement of a nuclear concept. Journal of Understanding client-centered therapy through
Counseling Psychology, 28, 91–100. interviews conducted by Carl Rogers. Paper pre-
∗ sented at the annual convention of the American
Barrington, B. L. (1967). The differential effective-
ness of therapy as measured by the Thematic Psychological Association, Boston, MA.
∗
Apperception Test. In C. R. Rogers, et al. (Eds.), Buckley, P., Karasu, T. B., & Charles, E. (1981).
The therapeutic relationship and its impact (pp. Psychotherapists view their personal therapy.
337–52). Madison: University of Wisconsin Psychotherapy: Theory, Research, and Practice, 18,
Press. [combined with Kiesler, Mathieu, Klein & 299–305.
∗
Schoeninger, 1967, and Van der Veen, 1967] Bugge, I., Hendel, D. D., & Moen, R. (1985).
Batson, C. D. (2009). These things called empa- Client evaluations of therapeutic processes and
thy: Eight related but distinct phenomena. In outcomes in a university mental health center.
J. Decety & W. Ickes (Eds.) (2009). The social Journal of American College Health, 33, 141–46.

∗ Egan, G. (1982). The skilled helper (2nd ed.).
Bullmann, F., Horlacher, K. D., & Kieser, B.
(2004). Clarifying as a mediator-variable in per- Monterey, CA: Brooks/Cole.
son-centered psychotherapy and therapists´ gender Eisenberg, N., & Eggum, N. D. (2009). Empathic
and therapy-school differences in empathy and posi- responding: Sympathy and personal distress. In
tive regard. Unpublished study, Psychological J. Decety & W. Ickes (Eds.) (2009). The Social
Institute, University of Heidelberg, Germany. Neuroscience of Empathy (pp. 71–83). Cambridge,
∗ MA: MIT Press.
Burns, D. D., & Nolen-Hoeksma, S. (1992).
Therapeutic empathy and recovery from depres- Eisenberg, N., & Fabes, R. A. (1990). Empathy:
sion in cognitive-behavioral therapy: A structural Conceptualization, assessment, and relations to
equation model. Journal of Consulting and prosocial behavior. Motivation and Emotion, 14,
Clinical Psychology, 60, 441–49. 131–49.
Carkhuff, R. R., & Berenson, B. (1967). Beyond Elliott, R. (2010). Psychotherapy change process
counseling and therapy. New York: Holt, Rinehart, research: Realizing the promise. Psychotherapy
& Winston. Research, 20, 123–35.
∗ Elliott, R., Filipovich, H., Harrigan, L., Gaynor, J.,
Cartwright, R. D., & Lerner, B. (1965). Empathy,
need to change, and improvement in psychother- Reimschuessel, C., & Zapadka, J. K. (1982).
apy. Journal of Consulting Psychology, 27, 138–44. Measuring response empathy: The development
∗ of a multi-component rating scale. Journal of
Clark, J. V., & Culbert, S. A. (1965). Mutually
therapeutic perception and self-awareness in a Counseling Psychology, 29, 379–87.
T-group. Journal of Applied Behavioral Science, 1, Elliott, R., Watson, J., Goldman, R., Greenberg, &
180–94. L. S. (2003). Learning emotion-focused therapy:
∗ The process-experiential approach to change.
Cooley, E. J., & Lajoy, R. (1980). Therapeutic
relationship and improvement as perceived by Washington, DC: APA.
∗
clients and therapists. Journal of Clinical Filak, J., & Abeles, N. (1984). Posttherapy congru-
Psychology, 36, 562–70. ence on client symptoms and therapy outcome.
∗ Professional Psychology: Research & Practice, 15,
Cramer, D., & Takens, R. (1992). Therapeutic
relationship and progress in the first six sessions 846–55.
∗
of individual psychotherapy: A panel analysis. Fretz, B. R. (1966). Postural movements in a coun-
Counselling Psychology Quarterly, 5, 25–36. seling dyad. Journal of Counseling Psychology, 13,
Decety, J., & Ickes, W. (Eds.) (2009). The social neu- 335–43.
∗
roscience of empathy. Cambridge, MA: MIT Press. Fuertes, J. N., Mislowack, A., Brown, S.,
Decety, J., & Jackson, P. L. (2004). The functional Gur-Arie, S., Wilkinson, S., & Gelso, C. J.
architecture of human empathy. Behavioral and (2007). Correlates of the real relationship in
Cognitive Neuroscience Reviews, 3, 71–100. psychotherapy: A study of dyads. Psychotherapy
Decety, J., & Lamm, C. (2009). Empathy versus Research, 17, 423–30.
∗
personal distress: Recent evidence from social Fuertes, J. N., Stracuzzi, T. I., Bennett, J.,
neuroscience. In J. Decety & W. Ickes (Eds.) Scheinholtz, J., Mislowack, A., Hersh, M., et al.
(2009). The Social Neuroscience of Empathy (pp. (2006). Therapist multicultural competency:
199–213). Cambridge, MA: MIT Press. A study of therapy dyads. Psychotherapy: Theory,
Diener, M. J., Hilsenroth, M. J. & Weinberger, J. Research, Practice, Training, 43, 480–490.
∗
(2009). A primer on meta-analysis of correlation Gabbard, C. E., Howard, G. S., & Dunfee, E. J.
coefficients: The relationship between patient- (1986). Reliability, sensitivity to measuring
reported therapeutic alliance and adult attach- change, and construct validity of a measure of
ment style as an illustration. Psychotherapy counselor adaptability. Journal of Counseling
Research, 19, 519–26. Psychology, 33, 377–86.
∗ Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G.,
Dormaar, J. M., Dijkman, C. I., & de Vries, M. W.
(1989). Consensus in patient-therapist inter- (1996). Action recognition in the premotor
actions: A measure of the therapeutic relation- cortex. Brain, 119, 593–609.
∗
ship related to outcome. Psychotherapy and Garfield, S. L., & Bergin, A. E. (1971). Therapeutic
Psychosomatics, 51, 69–76. conditions and outcome. Journal of Abnormal
Duan, C., & Hill, C. E. (1996). A critical review Psychology, 77, 108–114.
∗
of empathy research. Journal of Counseling Gelso, C. J., Latts, M. G., Gomez, M. J., &
Psychology, 43, 261–74. Fassinger, R. E. (2002). Countertransference
management and therapy outcome: An initial D. C. Brink, & P. M. Raskin (Eds.), The psycho-
evaluation. Journal of Clinical Psychology, 58, therapy of Carl Rogers (pp. 357–74). New York:
861–67. Guilford.
∗
Gillispie, R., Williams, E., & Gillispie, C. (2005). Henschel, D. N., & Bohart, A. C. (1981, August).
Hospitalized African American mental health The relationship between the effectiveness of a course
consumers: Some antecedents to service satis- in paraprofessional training and level of cognitive
faction and intent to comply with after- functioning. Paper presented at the American
care. American Journal of Orthopsychiatry, 75, Psychological Association, Los Angeles, CA.
254–61. Higgins, J. P. T., Thompson, S. G., Deeks, J. J., &
Gladstein, G. A., & Associates (1987). Empathy and Altman, D. G. (2003). Measuring inconsistency
counseling: Explorations in theory and research. in meta-analyses, British Journal of Medicine,
New York: Springer-Verlag. 327, 557–60.
∗
Goldman, R., Greenberg, L., & Angus, L. ( June, Hill, C. E., Thompson, B. J., & Corbett, M. (1992).
2000). The York II Psychotherapy Study on The impact of therapist ability to perceive
Experiential Therapy of Depression. Paper pre- displayed and hidden client reactions on imme-
sented at Society for Psychotherapy Research, diate outcome in first sessions of brief therapy.
Chicago, IL. Psychotherapy Research, 2, 143–55.
∗
Goleman, D. (1995). Emotional intelligence. Hoffart, A., Versland, S., & Sexton, H. (2002).
New York: Bantam. Self-understanding, empathy, guided discovery,
Goubert, L., Craig, K. D., & Buysee, A. (2009). and schema belief in schema-focused cogni-
Perceiving others in pain: Experimental and clin- tive therapy of personality problems: A process-
ical evidence of the role of empathy. In J. Decety outcome study. Cognitive Therapy and Research,
& W. Ickes (Eds.) (2009). The Social Neuroscience 26, 199–219.
of Empathy (pp. 153–65). Cambridge, MA: MIT Horvath, A. O., & Greenberg, L. S. (1986). The
Press. development of the Working Alliance Inventory.
∗
Greenberg, L. S., & Webster, M. (1982). Resolving In L. S. Greenberg & W. M. Pinsof (Eds.), The
decisional conflict by means of two-chair dia- psychotherapeutic process: A research handbook (pp.
logue: Relating process to outcome. Journal of 529–56). New York: Guilford.
∗
Counseling Psychology, 29, 468–77. Horvath, A. O., & Greenberg, L. S. (1989).
∗
Gross, W. F., & DeRidder, L. M. (1966). Significant Development and validation of the working alli-
movement in comparatively short-term counsel- ance inventory. Journal of Counseling Psychology,
ing. Journal of Counseling Psychology, 13, 98–99. 36, 223–33.
Gurman, A.S. (1977). The patient’s perception of Hubble, M. A., Duncan, B. L., & Miller, S. D.
the therapeutic relationship. In A. S. Gurman (1999). Directing attention to what works. In
& A. M. Razin (Eds.), Effective psychotherapy: M. A. Hubble, M. A. Duncan, & S. D. Miller
A handbook of research (pp. 503–43). New York: (Eds.), The heart and soul of change: What
Pergamon. works in therapy (pp. 407–48). Washington, DC:
∗
Hall, J. A., & Davis, M. H. (2000). Dispositional American Psychological Association.
empathy in scientist and practitioner psycholo- Ickes, W. (Ed.). (1997). Empathic accuracy. New York:
gists: Group differences and relationship to self- Guilford.
reported professional effectiveness. Psychotherapy, Ickes, W. (2003). Everyday mind reading: Under-
37, 45–56. standing what other people think and feel. Amherst,
∗
Hamilton, J. C. (2000). Construct validity of NY: Prometheus Books.
the core conditions and factor structure of the Kennedy-Moore, E., & Watson, J. C. (1999).
Client Evaluation of Counselor Scale. The Person- Expressing emotion: Myths, realities, and therapeu-
Centered Journal, 7, 40–51. tic strategies. New York: Guilford.
∗ ∗
Hansen, J. C., Moore, G. D., & Carkhuff, R. R. Kiesler, D. J., Klein, M. H., Mathieu, P. L., &
(1968). The differential relationships of objective Schoeninger, D. (1967). Constructive personal-
and client perceptions of counseling. Journal of ity change for therapy and control patients. In
Clinical Psychology, 24, 244–46. C. R. Rogers, et al. (Eds.), The therapeutic rela-
Hayes, A. M., & Goldfried, M. R. (1996). Rogers’ tionship and its impact (pp. 251–94), Madison:
work with Mark: An empirical analysis and cog- University of Wisconsin Press. (Combined with
nitive-behavioral perspective. In B. A. Farber, Barrington, 1967, and Van Der Veen, 1967)

∗ treatment: Their reliable identification and influ-
Kurtz, R. R., & Grummon, D. L. (1972). Different
approaches to the measurement of therapist ence on behaviour change. Clinical Psychology
empathy and their relationship to therapy and Psychotherapy, 9, 395–405.
outcomes. Journal of Consulting and Clinical Martin, D. G. (2000). Counseling and therapy
Psychology, 39, 106–115. skills (2nd ed.). Prospect Heights, IL: Waveland
∗ Press.
Lafferty, P., Beutler, L. E., & Crago, M. (1989).
Differences between more and less effective Martin, D. J., Garske, J. P., & Davis, M. D. (2000).
psychotherapists: A study of select therapist Relation of the therapeutic alliance with out-
variables. Journal of Consulting and Clinical come and other variables: A meta-analytic review.
Psychology, 57, 76–80. Journal of Consulting and Clinical Psychology, 68,
Lambert, M. J., DeJulio, S. J., & Stein, D. M. 438–50.
∗
(1978). Therapist interpersonal skills: Process, Martin, P. J., & Sterne, A. L. (1976). Post-hospital
outcome, methodological considerations, and adjustment as related to therapist’s in-therapy
recommendations for future research. Psycho- behavior. Psychotherapy: Theory, Research, and
logical Bulletin, 85, 467–89. Practice, 13, 267–273.
∗
Landfield, A. W. (1971). Personal construct systems Melnick, B., & Pierce, R. M. (1971). Client evalu-
in psychotherapy. Chicago: Rand McNally. ation of therapist strength and positive-negative
∗ evaluation as related to client dynamics, objec-
Langhoff, C., Baer, T., Zubraegel, D., &
Linden, M. (2008). Therapist-patient alliance, tive ratings of competence and outcome. Journal
patient-therapist alliance, mutual therapeutic of Clinical Psychology, 27, 408–410.
∗
alliance, therapist-patient concordance, and out- Miller, W., Taylor, C., & West, J. (1980). Focused
come of CBT in GAD. Journal of Cognitive versus broad spectrum behavior therapy for prob-
Psychotherapy, 22, 68–79. lem drinkers. Journal of Consulting and Clinical
Leitner, L. M. (1995). Optimal therapeutic dis- Psychology, 48, 590–601.
tance: A therapist’s experience of personal con- Mitchell, K. M., Bozarth, J. D., & Krauft, C. C.
struct psychotherapy. In R. A. Neimeyer & (1977). A reappraisal of the therapeutic effec-
M. J. Mahoney (Eds.), Constructivism in psy- tiveness of accurate empathy, nonpossessive
chotherapy (pp. 70). Washington, DC: American warmth, and genuineness. In A. S. Gurman &
Psychological Association. A. N. Razin (Eds.), Effective psychotherapy.
∗ New York: Pergamon.
Lerner, B. (1972). Therapy in the ghetto. Baltimore:
∗
Johns Hopkins University Press. Mitchell, K. M., Truax, C. B., Bozarth, J. D., &
∗ Krauft, C. C. (1973, March). Antecedents to
Lesser, W. M. (1961). The relationship between
counseling progress and empathic understand- psychotherapeutic outcome. NIMH Grant Report
ing. Journal of Counseling Psychology, 8, 330–36. (12306). Arkansas Rehabilitation Research and
Lipsey, M. W., & Wilson, D. B. (2001). Practical Training Center, Arkansas Rehabilitation Services,
meta-analysis. Thousand Oaks, CA: Sage. Hot Springs, Arkansas.
∗ Mohr, J. J., & Woodhouse, S. S. (2000, June).
Lorr, M. (1965). Client perceptions of therapists:
A study of therapeutic relation. Journal of Clients’ visions of helpful and harmful psychother-
Consulting Psychology, 29, 146–49. apy: An approach to measuring individual differ-
Mahrer, A. R. (1997). Empathy as therapist-client ences in therapy priorities. Paper presented at the
alignment. In A. C. Bohart & L. S. Greenberg 31st Annual Meeting of the Society for
(Eds.), Empathy reconsidered: New directions in Psychotherapy Research, Chicago, IL.
∗
psychotherapy (pp. 187–216). Washington, DC: Muller, J., & Abeles, N. (1971). Relationship of
American Psychological Association. liking, empathy and therapists’ experience to
∗ outcome in psychotherapy. Journal of Counseling
Marshall, W. L., Serran, G. A., Fernandez, Y. M.,
Mulloy, R., Mann, R. E., & Thornton, D. Psychology, 18, 39–43.
(2003). Therapist characteristics in the treatment Myers, S. (2000). Empathic listening: Reports on
of sexual offenders: Tentative data on their rela- the experience of being heard. Journal of
tionship with indices of behaviour change. Humanistic Psychology, 40, 148–73.
Journal of Sexual Aggression, 9, 25–30. O’Hara, M. M. (1984). Person-centered gestalt:
∗ Towards a holistic synthesis. In R. F. Levant &
Marshall, W. L., Serran, G. A., Moulden, H.,
Mulloy, R., Fernandez, Y. M., Mann, R. E., et al. J. M. Shlien (Eds.), Client-centered therapy and
(2002). Therapist features in sexual offender the person-centered approach: New directions in
theory, research and practice (pp. 203–21). ∗
Rabavilas, A. D., Boulougouris, J. C., & Perissaki, C.
New York: Praeger. (1979). Therapist qualities related to outcome
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). with exposure in vivo in neurotic patients.
Process and outcome in psychotherapy—noch Journal of Behaviour Therapy and Experimental
einmal. In A. E. Bergin & S. L. Garfield (Eds.), Psychiatry, 410, 293–94.
Handbook of psychotherapy and behavior change ∗
Roback, H. B., & Strassberg, D. S. (1975).
(4th ed., pp. 270–378). New York: Wiley. Relationship between perceived therapist-offered
∗
Orlinsky, D. E., & Howard, K. I. (1967). The good conditions and therapeutic movement in group
therapy hour: Experiential correlates of patients’ psychotherapy with hospitalized mental patients.
and therapists’ evaluations of therapy sessions. Small Group Behavior, 6, 345–52.
Archives of General Psychiatry, 12, 621–32. Rogers, C. R. (1957). The necessary and sufficient
Orlinsky, D. E., & Howard, K. I. (1978). The rela- conditions of therapeutic personality change.
tion of process to outcome in psychotherapy. In Journal of Consulting Psychology, 21, 95–103.
Garfield, S. L., & Bergin, A. E. (Eds.), Handbook Rogers, C. R. (1980). A way of being. Boston:
of psychotherapy and behavior change (2nd ed., pp. Houghton Mifflin.
283–330). New York: Wiley. Sachse, R. (1990a). Concrete interventions are
Orlinsky, D. E., & Howard, K. I. (1986). Process crucial: The influence of the therapist’s process-
and outcome in psychotherapy, In S. L. Garfield ing proposals on the client’s intrapersonal explo-
& A. E. Bergin (Eds.), Handbook of psychother- ration in client-centered therapy. In G. Lietaer,
apy and behavior change (3rd ed., pp. 311–84). J. Rombauts, & R. Van Balen (Eds.), Client-
New York: Wiley. centered and experiential psychotherapy in the nine-
∗
Pantalon, M. V., Chawarski, M. C., Falcioni, J., ties (pp. 295–308). Leuven, Belgium: Leuven
Pakes, J., & Schottenfeld, R. S. (2004). Linking University Press.
process and outcome in the community rein- Sachse, R. (1990b). The influence of therapist pro-
forcement approach for treating cocaine depen- cessing proposals on the explication process of
dence: A preliminary report. American Journal of the client. Person-Centered Review, 5, 321–44.
Drug and Alcohol Abuse, 30, 353–67. ∗
Saltzman, C., Leutgert, M. J., Roth, C. H.,
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Creaser, J., & Howard, L. (1976). Formation of
Research on therapist variables in relation to a therapeutic relationship: Experiences during
process and outcome. In Garfield, S. L., & the initial phase of psychotherapy as predictors
Bergin, A. E. (Eds.), Handbook of psychotherapy of treatment duration and outcome. Journal of
and behavior change (2nd ed., pp. 233–82). New Consulting and Clinical Psychology, 44, 546–55.
York: Wiley. ∗
Sapolsky, A. (1965). Relationship between patient-
Patterson, C. H. (1984). Empathy, warmth, and doctor compatibility, mutual perceptions, and
genuineness: A review of reviews. Psychotherapy, outcome of treatment. Journal of Abnormal
21, 431–38. Psychology, 70, 70–76.
∗ ∗
Payne, A., Liebling-Kalifani, H., & Joseph, S. Saunders, S. M. (2000). Examining the relation-
(2007). Client-centred group therapy for survi- ship between the therapeutic bond and the
vors of interpersonal trauma: A pilot investiga- phases of treatment outcome. Psychotherapy, 37,
tion. Counselling & Psychotherapy Research, 7, 206–218.
100–105. Selman, R. I. (1980). The growth of interpersonal
Peabody, S. A., & Gelso, C. J. (1982). understanding. Orlando, FL: Academic Press.
Countertransference and empathy: The complex Shamay-Tsoory, S. (2009). Empathic processing:
relationship between two divergent concepts Its cognitive and affective dimensions and neuro-
in counseling. Journal of Counseling Psychology, anatomical basis. In J. Decety & W. Ickes (Eds.),
29, 240–45. The social neuroscience of empathy (pp. 215–32).
∗
Peake, T. H. (1979). Therapist-patient agreement Cambridge, MA: MIT Press.
and outcome in group therapy. Journal of Clinical Shlien, J. (1997). Empathy in psychotherapy:
Psychology, 35(3), 637–46. A vital mechanism? Yes. Therapist’s conceit? All
Persons, J. B., & Burns, D. D. (1985). Mechanisms too often. In A. C. Bohart & L. S. Greenberg
of action of cognitive therapy: Relative contribu- (Eds.), Empathy reconsidered: New directions in
tion of technical and interpersonal intervention. psychotherapy (pp. 63–80). Washington, DC:
Cognitive Therapy and Research, 9, 539–51. American Psychological Association.

Smith, M. L., Glass, G. V., & Miller, T. I. (1980). ∗
Truax, C. B., & Wittmer, J. (1971). The effects
The benefits of psychotherapy. Baltimore: The of therapist focus on patient anxiety source and
Johns Hopkins University Press. the interaction with therapist level of accu-
∗
Staples, F. R., Sloane, R. D., Whipple, K., rate empathy. Journal of Clinical Psychology, 27,
Cristol, A. H., & Yorkston, N. (1976). Process 297–99.
and outcome in psychotherapy and behavior ∗
Truax, C. B., Wittmer, J., & Wargo, D. G. (1971).
therapy. Journal of Consulting and Clinical Effects of the therapeutic conditions of accurate
Psychology, 44, 340–50. empathy, nonpossessive warmth, and genuine-
∗
Strupp, H. H., Fox, R. E., & Lessler, K. (1969). ness on hospitalized mental patients during
Patients view their psychotherapy. Baltimore: Johns group therapy. Journal of Clinical Psychology,
Hopkins Press. 27, 137–42.
Tallman, K., Robinson, E., Kay, D., Harvey, S., & ∗
Van der Veen, F. (1967). Basic elements in the
Bohart, A. (1994, August). Experiential and non- process of psychotherapy: A research study.
experiential Rogerian therapy: An analogue study. Journal of Consulting Psychology, 31, 295–303.
Paper presented at the American Psychological (Combined with Barrington, 1967, and Kiesler,
Association Convention, Los Angeles. Mathieu, Klein & Schoeninger, 1967)
Tausch, R. (1988). The relationship between emo- Wampold, B. E. (2001). The great psychotherapy
tions and cognitions: Implications for therapist debate: Models, methods, and findings. Mahwah,
empathy. Person-Centered Review, 3, 277–91. NJ: Erlbaum.
Thompson, B., & Hill, C. (1991). Therapist percep- Watson, J. C. (2001). Re-visioning empathy. In
tion of client reactions. Journal of Counseling and D. Cain & J. Seeman (Eds.), Humanistic psy-
Development, 69, 261–65. chotherapies: Handbook of research and practice
∗
Truax, C. B. (1966). Therapist empathy, warmth, (pp. 445–71). Washington, DC: APA.
and genuineness and patient personality change ∗
Watson, J. C., & Geller, S. M. (2005). The relation
in group psychotherapy: A comparison between among the relationship conditions, working alli-
interaction unit measures, time sample measures, ance, and outcome in both process-experiential
and patient perception measures. Journal of and cognitive-behavioral psychotherapy. Psycho-
Clinical Psychology, 22, 225–29. therapy Research, 15, 25–33.
Truax, C. B., & Carkhuff, R. R. (1967). Toward Watson, J. C., & Greenberg, L. S. (2009). Empathic
effective counseling and psychotherapy: Training resonance: A neuroscience perspective. In
and practice. Chicago: Aldine. J. Decety & W. Ickes (Eds.) (2009). The social
∗
Truax, C. B., Carkhuff, R. R., & Kodman, F., Jr. neuroscience of empathy (pp. 125–38). Cambridge,
(1965). Relationships between therapist-offered MA: MIT Press.
conditions and patient change in group psycho- Watson, J. C., & Prosser, M. (2002). Development
therapy. Journal of Clinical Psychology, 21, 327–29. of an observer rated measure of therapist
Truax, C. B., & Mitchell, K. M. (1971). Research empathy. In J. C. Watson, R. Goldman, &
on certain therapist interpersonal skills in rela- M Warner (Eds.), Client-centered and experien-
tion to process and outcome. In A. E. Bergin & tial psychotherapy in the 21st century: Advances in
S. L. Garfield (Eds.), Handbook of psychother- theory, research and practice (pp. 303–314). Ross
apy and behavior change (1st ed., pp. 299–344). on Wye, UK: PCCS Books.
New York: Wiley. Watson, N. (1984). The empirical status of Rogers’
∗
Truax, C. B., Wargo, D. G., Frank, J. D. Imber, S. hypotheses of the necessary and sufficient condi-
D., Battle, C. C., Hoehn-Saric, R., et al. (1966). tions for effective psychotherapy. In R. F. Levant
Therapist empathy, genuineness and warmth and & J. M. Shlien (Eds.), Client-centered therapy
patient therapeutic outcome. Journal of Consulting and the person-centered approach (pp. 17–40).
Psychology, 30, 395–401. New York: Praeger.
C HA P TER
7 Goal Consensus and Collaboration
Georgiana Shick Tryon and Greta Winograd
This chapter focuses on one element of the Definitions and Measures

therapeutic contract between patient and Goal Consensus
psychotherapist—goal consensus—that sets At the beginning of treatment, psychother-
the parameters of treatment and one apists effect a contract with their patients
therapeutic operation—collaboration—that that outlines the conditions of their work
implements the contract that should contri- together. This “. . . therapeutic contract
bute to a satisfactory treatment outcome. In is their ‘understanding’ about their goals
support of this statement, our chapter in the and conditions for engaging each other as
first edition of this volume (Tryon & patient and therapist” (Orlinsky, Grawe,
Winograd, 2002) presented evidence that & Parks, 1994, p. 279). Agreement about
goal consensus and collaboration were posi- treatment goals and the processes by which
tively associated with measures of adult patient and therapist will achieve these
patient psychotherapy outcome. The current goals is the essence of goal consensus. Goal
chapter updates and improves upon this consensus is part of the pantheoretical
work by examining via meta-analyses results working alliance that includes patient–
of more recent studies, published from 2000 therapist agreement on the therapy goals
through 2009, that relate goal consensus and and the tasks to reach those goals as well as
collaboration to therapy outcome. formation of a bond between the members
Our chapter begins with definitions of of the therapeutic dyad (Bordin, 1979;
terms followed by a clinical example of goal Chapter II of this book).
consensus and collaboration. We then As in our chapter in the first edition of
describe and present results of two meta- this book, we define goal consensus as:
analyses (one on goal consensus and one on
collaboration). We also present results of a (a) patient therapist agreement on
small meta-analysis using data from studies goals; (b) the extent to which a therapist
that relate goal consensus and collaboration explains the nature and expectations of
to each other. Following this, we discuss therapy, and the patient’s understanding of
the patients’ contributions and the perspec- this information; (c) the extent to which
tives that they bring to these elements of goals are discussed, and the patient’s belief
the therapeutic relationship. We also dis- that goals are clearly specified; (d) patient
cuss the limitations of the research reviewed. commitment to goals; and (e) patient–
The chapter concludes with suggestions for therapist congruence on the origin of
therapeutic practice. the patient’s problem, and congruence
153
on who or what is responsible for problem earlier chapter (Tryon & Winograd, 2002)
solution (Tryon & Winograd, 2001, covered the goal consensus literature
pp. 385–386). prior to 2000. The instruments comprised
scales of measures that assess the working
The third column of Table 7.1 shows alliance, such as the Working Strategy
that the studies included in the goal Consensus scale of the California Psycho-
consensus meta-analysis for this chapter therapy Alliance Scale (CALPAS; Marmar,
measured goal consensus using several Gaston, Gallagher, & Thompson, 1989);
different instruments. These studies were the Goals and Tasks scale for patients, and
published between 2000 and 2009. Our the Shared Goals and Goal and Task
Table 7.1 Descriptions of Studies of Collaboration Outcome and Goal Consensus Outcome (2000–2009)
with Effect Sizes
Effect Size (95% CI)
Study N Goal consensus Collaboration Outcome measure(s) GC Cb
measure(s) measure(s)
Ablon et al. (2006) 17 Psychotherapy Symptom Checklist 90 .18
Process Q-Set - R, Anxiety Sensitivity (−.33
Index, Panic Disorder – 0.62)
Severity Scale
Abramowitz et al. 28 Therapy rationale Treatment Yale-Brown Obsessive .65 .68
(2002) acceptance rating compliance, Compulsive Scale (0.37 (0.42
homework – 0.83) – 0.84)
completion
Ackerman et al. 128 Combined Combined Session Evaluation .66 .55
(2000) Alliance Short Alliance Short Questionnaire (0.56 (0.42
Form–Evaluation Form–Evaluation – 0.76) – 0.67)
scales scales
Addis & Jacobson 150 Treatment Homework Beck Depression .35 .29
(2000) acceptance completion Inventory, Hamilton (0.20 (0.14
rating rating Rating Scale for – 0.49) – 0.43)
Depression
Bogalo & 30 Homework IBS-Symptom Severity .16
Moss-Morris (2006) Assessment Scale, Subject’s Global (−.22
Tool Assessment of Relief – 0.50)
Brocato & Wagner 124 Working Alliance Stages of Change .05
(2008) Inventory Readiness & Eagerness (−.13
for Treatment Scale, – 0.23)
treatment retention
Burns & Spangler 521 Homework Beck Depression .34
(2000) completion Inventory, Hopkins (0.27
Symptom Checklist – 0.42)
Busseri & Tyler 46 Patient & therapist Patient improvement .14
(2004) goal agreement ratings, Post Therapy (−.16–.42)
Questionnaire
(Continued)
Table 7.1 Continued
Effect Size (95% CI)

Caspar et al. (2005) 21 Plan Analysis Beck Depression .33
Inventory, Global (−.12
Assessment of – 0.66)
Functioning, Hamilton
Rating Scale for
Depression, Symptom
Checklist 90 - R
Clemence et al. (2005) 113 Combined Alliance Combined Help Received Scale, .37 .44
Short Form Alliance Short Patient’s Estimate of (0.20 (0.29
Form Improvement to Date – 0.53) – 0.58)
Cowan et al. (2008) 576 Homework Beck Depression .08
completion Inventory, Hamilton (0.00
Rating Scale for – 0.16)
Depression, Perceived
Social Support Scale,
Social Support
Instrument
Dunn et al. (2006) 29 Rating of Positive and Negative −.13
homework Signs of Schizophrenia (−.48
–0.25)
Fitzpatrick et al. 48 Working Alliance Session Impact Scale .28
(2005) Inventory (0.00
– 0.53)
Gabbay et al. (2003) 128 Problem agreement Dropout .02
(−.16
– 0.20)
Gonzalez et al. (2006) 123 Homework % positive urine, .12
completion treatment retention (−.06
rating – 0.29)
Graf et al. (2008) 44 Post writing Satisfaction rating, .45
questionnaire insight rating (0.18
– 0.67)
Hegel et al. (2002) 179 Rating of patient Homework Hamilton Rating Scale .25 .21
understanding of completion for Depression (0.11 (0.07
therapy – 0.38) – 0.35)
Lingiardi et al. (2005) 47 California California Dropout .42 .45
Psychotherapy Psychotherapy (0.15 (0.19
Alliance Scale Alliance Scale – 0.64) - 0.66)
Long (2001) 24 Working Alliance Causal Dimension Scale .21
Inventory, Goal II, Global Assessment (−.21
Statement Inventory of Functioning, Target – 0.57)
Complaint
questionnaire
(Continued)
155
Table 7.1 Continued
Effect Size (95% CI)
Principe et al. (2006) 91 Working Alliance Dropout .12
Inventory (−.09
– 0.32)
Schönberger et al. 45 c Working Alliance Client’s d2 Test of Concentration, .18 .52
(2006, 2007) Inventory Compliance European Brain Injury (−.05 (0.27
Scale Questionnaire, – 0.40) – 0.71)
Neurosensory Center
Comprehensive
Examination for Aphasia,
Ravens Advanced
Progressive Matrices,
success ratings, Trail
Making Test, WAIS-R,
Wisconsin Card Sorting
Test, word fluency
Stein et al. (2004) 53 Adherence to Hamilton Rating Scale .31
intervention for Depression (0.04
checklist – 0.54)
Wettersten et al. 64 Working Alliance Brief Symptom .18
(2005) Inventory Inventory, Counseling (−.07
Center Follow-up – 0.41)
Questionnaire
Whittal et al. (2004) 59 Homework Yale-Brown Obsessive .34
Compliance Compulsive Scale (0.09
– 0.56)
Woods et al. (2002) 82 Homework Behavioral Avoidance .05
hours, Test, target symptoms (−.17
homework – 0.27)
completion
Yovel & Safren (2007) 15 Homework AD/HD Rating Scale, .39
adherence, Clinical Global (−.15
symptom Impression Scale, Global – 0.76)
change Assessment of
Functioning, Hamilton
Anxiety Scale, Hamilton
Rating Scale for
Depression
Zane et al. (2005) 60 Goals measure, Global Assessment .24
Perceptual Rating of Functioning, (−.01
Scale Session Evaluation – 0.46)
Questionnaire
Note: Of the 15 goal consensus effect sizes reported in column 6, 11 were based on zero-order correlations, 3 were derived from partial beta
coefficients (Abramowitz et al., 2002; Brocato & Wagner, 2008; Zane et al., 2005), and 1 originated from a combination of zero order
correlations and partial beta coefficients (Wettersten et al., 2005). Of the 19 collaboration effect sizes reported in column 7, 15 were based
on zero-order correlations, and 4 (Ablon et al., 2006; Abramowitz et al., 2002; Cowan et al., 2008; Woods et al., 2002) were derived from
partial beta coefficients with two or more independent variables.
a
GC = Goal Consensus. b C = Collaboration. c Both Schönberger et al. (2006) and Schönberger et al. (2007) used the same study sample.
The 2006 paper assessed collaboration (N = 45). Both papers assessed goal consensus (N = 72).
156
Disagreement scales for therapists of the involvement of patient and therapist in a
Combined Alliance Short Form (CASF; helping relationship as well as (b) patient
Hatcher, 1999; Hatcher & Barends, 1996); cooperation and (c) role involvement.
and the Goal and Task scales of the Working Another indicator of, but not a measure of,
Alliance Inventory (WAI; Horvath & collaboration is (d) patient completion of
Greenberg, 1989). Other studies assessed assigned homework.
goal consensus using rating scales specifi- The fourth column of Table 7.1 presents
cally designed to assess actual patient– collaboration measures used by the studies
therapist goal agreement, such as the Goal in our meta-analysis. These studies were
Statement Inventory (GSI; McNair & Lorr, published between 2000 and 2009. Our
1964) and the Causal Dimensions Scale II earlier chapter (Tryon & Winograd, 2002)
(CDS-II; McAuley, Duncan, & Russell, covered the collaboration literature prior
1992). Still others (e.g., Gabbay et al., 2003; to 2000. Collaboration measures include
Hegel, Barrett, Cornell, & Oxman, 2002) scales from working alliance measures
used goal consensus measures unique to such as the Therapist Understanding and
their studies. Involvement scale of the CALPAS (Marmar
Column 5 in Table 7.1 shows that the et al., 1989); the Confident Collaboration
studies also used several measures of therapy scale of the patient CASF (Hatcher &
outcome associated with goal consensus. Barends, 1996); and the Patient Working
Most outcome measures assessed patient Engagement, Patient Confidence and
improvement (e.g., as in Busseri & Tyler, Commitment, and Therapist Confident
2004), while others examined treatment Collaboration scales of the therapist CASF
dropout (e.g., as in Gabbay et al., 2003), (Hatcher, 1999). Other studies used rat-
and others considered session impact (e.g., ings of patient treatment compliance (e.g.,
as in Fitzpatrick, Iwakabe, & Stalikas, as in Abramowitz, Franklin, Zoellner, &
2005). In our previous review of the goal DiBernardo, 2002), treatment acceptance
consensus literature (Tryon & Winograd, (e.g., as in Addis & Jacobson, 2000), and
2001), we found that 68% (n = 17) of the patient adherence to the treatment (e.g.,
studies reviewed (n = 25) “revealed a Stein et al., 2004) that were unique to the
positive relationship between goal consen- study. Studies that assessed homework
sus and outcome on at least one measure compliance also generally used measures
completed by patient, therapist, or observer” unique to their studies (e.g., as in Yovel &
(p. 386). Safren, 2007).
The studies in the collaboration outcome
Collaboration meta-analysis used several types of outcome
To implement the therapeutic contract, measures (see fifth column of Table 7.1).
patient and therapist must function as a These ranged from patient symptom
team. Collaboration represents the active improvement (e.g., as in Ablon, Levy, &
process of working together to fulfill Katzenstein, 2006), session evaluation (e.g.,
treatment goals. As with goal consensus, col- as in Ackerman, Hilsenroth, Baity, &
laboration is a pantheoretical concept that Blagys, 2000), treatment retention (e.g., as
applies to all types of therapies. Collaboration in Brocato & Wagner, 2008), and patient
“. . . is largely defined by the instruments satisfaction (e.g., as in Graf, Gaudiano, &
devised to assess the concept” (Bachelor, Geller, 2008). In our previous review of the
Laverdière, Gamache, & Bordeleau, 2007, collaboration outcome literature (Tryon &
p. 175). These instruments assess (a) mutual Winograd, 2001), “we combined results
t ryo n , w i n o g r a d 157
from 24 studies and found that 89% of basis while she continues to work toward
the time, collaborative involvement and recovery.
outcome were significantly positively related The excerpt below is from Hope and her
on at least one measure completed by therapist’s second session together. Elements
patient, therapist, or observer” (p. 387). of goal consensus and collaboration in their
interaction are indicated in brackets.
Relationship of Goal Consensus
and Collaboration Therapist: Last time we talked about the
By definition, consensus implies an agree- challenges you’ve been facing as you
ment based on the opinions of the par- return to school and reconnect with
ties involved.1 Achievement of consensus friends and family members who
requires that those involved work coopera- know about your recent episode and
tively, which is the definition of collabora- hospitalization. I asked you to jot
tion.2 Since therapist and patient work down some of the thoughts that have
together to establish agreement on the goals been running through your mind
and tasks of psychotherapy, one might expect when you are around people at
that measurements of goal consensus and school, and some of the ways people
collaboration would be related. Since 2000, act or the things they say that you
however, only seven of the published studies find upsetting.
that we found reported this relationship Hope: Yes, I did both of those
(Abramowitz et al., 2002; Ackerman et al., things [collaboration: homework
2000; Addis & Jacobson, 2000; Clemence, completion]. I realized that even
Hilsenroth, Ackerman, Strassle, & Handler, with classmates I hardly know,
2005; Hegel et al., 2002; Lingiardi, I find myself worrying constantly
Filippucci, & Baiocco, 2005; Schönberger, about whether they know what
Humle, & Teasdale, 2006, 2007). happened, or if they can just tell
by the way I act that I was in the
hospital, that I have a mental illness.
Clinical Example
And when I’m thinking this way,
Hope is a 21-year-old college senior and
I start to act nervous and insecure,
honors student who took a leave of absence
and just find it hard to concentrate.
from school last semester after being hospi-
But when I’m around friends and
talized for several weeks in a mental health
family members who know about my
crisis center. A psychotic episode—her
crisis, I feel even worse. They act so
first—led to Hope’s hospitalization and to
differently from the way they acted
a diagnosis of bipolar disorder. Hope has
before my diagnosis. This is what I
recently returned to college on a part-time
wrote down: “They either act like I’m
made of glass and about to break, or
they keep their distance.” And I don’t
1 consensus. (2010). In Merriam-Webster know how to get them to just treat
Online Dictionary. Retrieved June 6, 2010, from me like a regular person again.
http://www.merriam-webster.com/dictionary/ Therapist: It sounds to me like some
consensus
of the people you are close to have
2 collaboration. (2010). In Merriam-Webster
Online Dictionary. Retrieved June 6, 2010, from disappointed you. You’d like to
http://www.merriam-webster.com/dictionary/ reconnect with them but aren’t sure
collaboration how to do this. It also sounds to me
as though even when other students Presuming Hope continued to experi-
don’t necessarily act differently, you ence her psychotherapy as a collabora-
are worried that they are seeing you tive enterprise, she would likely strongly
differently, or would see you endorse collaboration items from the revised
differently if they knew about your Helping Alliance Questionnaire (Haq-II;
recent life events. This makes it hard Luborsky et al., 1996) such as, “the thera-
to relax around them and focus on pist and I have meaningful exchanges” and
your schoolwork. “I feel I am working together with the thera-
Hope: Yes, I am quite uncomfortable pist in a joint effort” (p. 271). Hope would
around other people now, almost also be likely to endorse consensus items
all of the time [goal consensus: from the Working Alliance Inventory such
congruence on patient problem]. as, “we agree about what is important for
Therapist: I also recall that you me to work on” and “my therapist and I are
mentioned being quite isolated. working towards mutually agreed upon
Hope: Yes, that’s true. Since I don’t goals” (Horvath & Greenberg, 1989a,
know what to say or how to act, p. 226).
I’ve started to avoid people.
I basically just go directly to my Meta-Analytic Review
classes and then straight home. Inclusion Criteria and Study Selection
But I don’t think spending so much We included goal consensus and collabo-
time alone is good for my mood. It ration studies in their respective meta-
doesn’t even feel like me. I used to be analysis if they used: (a) at least one measure
a really social person [goal consensus: of goal consensus and/or one measure of
further congruence on patient collaboration defined according to criteria
problem]. listed above; (b) at least one psychotherapy
Therapist: I was thinking that over the outcome measure; (c) a group design;
next few sessions, we could work (d) individually conducted psychotherapy;
together to come up with ideas about (e) adult clients (aged 18 and older). Finally,
how to talk about your each study (f ) reported a correlation, its
hospitalization and recovery with equivalent (standardized β weight), or other
your friends and family statistic (t, F, or d ) that could be converted
[collaboration: mutual involvement of to a correlation between goal consensus
patient and therapist in a helping and/or collaboration scores and outcome
relationship]. I was also thinking we scores; and (g) was published in English in a
might try out some relaxation and refereed journal from 2000 through 2009.
thought replacement strategies for We and our students (a psychology doc-
when you get anxious around toral student, a psychology masters student,
classmates in school. and an undergraduate psychology major) con-
Hope: I like the sound of that. ducted advanced Google Scholar searches for
And I’d also like you to help articles from 2000 through 2009 using the
me experiment with gradually following terms: patient–therapist collabora-
coming out of my shell as I work on tion (1,290 references), patient–therapist goal
getting healthy again [goal consensus: consensus (631 references), homework comp-
discussion and specification of goals; liance and psychotherapy outcome (3,760
covllaboration: patient role references), patient–therapist goal consensus
involvement]. and psychotherapy outcome (459 references),
patient–therapist agreement and psychother- in the course of therapy, and rater perspec-
apy outcome (796 references), and patient– tive. We also independently identified goal
therapist collaboration and psychotherapy consensus, collaboration, and outcome data
outcome (555 references). and calculated effect sizes for the relation-
We cross-tabulated the references, ships among these data. Our ratings and
inspected the article abstracts, and identi- calculations were in agreement for 498 out
fied 53 articles for in-depth examination by of 512 items (97%). When we disagreed,
the two authors of this chapter. After review- we discussed the item until we reached
ing each of the 53 studies independently, agreement. Table 7.1 presents the coded
the authors were in perfect agreement in information for the measures and effect
identifying 28 studies that met inclusion sizes.
criteria and 25 studies that did not. The We classified patient disturbance as mild
reasons for not meeting inclusion criteria (volunteers from college classes, no formal
were: other than individual therapy (n = 8), diagnosis, or seen at university-based train-
measure of collaboration or goal consensus ing clinics), moderate (formal diagnoses of
not used in outcome analysis (n = 8), no nonpsychotic mood disorder, seen in out-
measure of collaboration or goal consensus patient settings other than university train-
(n = 3), no outcome measure (n = 3), no ing clinics), and severe (seen in inpatient
treatment (n = 2), and results could not be settings, diagnosed with a psychosis, iden-
converted to effect size (n = 1). tified by study authors as severely disturbed).
Table 7.1 describes the 28 studies that met Eight studies had patients with severe
inclusion criteria. Thirteen of the 28 studies disturbances, 13 studies had patients with
(46%) provided effect sizes for collaboration moderate disturbances, and 7 studies had
and outcome only, 9 studies (32%) provided patients with mild disturbances.
effect sizes for goal consensus and outcome Psychotherapists were either experienced
only, and 7 studies (25%) provided effect (16 studies), trainees (2 studies), or a
sizes for both goal consensus and collabo- combination of experienced therapists and
ration with outcome. Thus, the goal consen- trainees (7 studies). Three studies did not
sus meta-analysis included results from specify therapist experience. Four studies
15 studies, and the collaboration meta- did not specify theoretical orientation. In
analysis included results from 19 studies. studies using a single theoretical orienta-
We were also in perfect agreement in catego- tion, therapy was behavioral (2 studies),
rizing the studies into these two groups. cognitive behavioral (9 studies), psychody-
namic (2 studies), or eclectic, interper-
Coding of Study Characteristics sonal, individualized, or solution focused
In keeping with editorial requirements for (1 study each). The remaining 7 studies
this chapter and similar to a prior meta- used therapies reflecting more than one ori-
analysis by the first author and colleagues entation (e.g., cognitive behavioral). We
(Tryon, Blackwell, & Hammel, 2007), we divided treatment length into 1–10 sessions
independently recorded and coded the (7 studies), 11–20 sessions (12 studies),
following information from each study: and 21 or more sessions (7 studies). Two
number of participants, severity of patient studies did not specify treatment length.
disturbance, therapist experience, treatment Table 7.1 shows the instruments used
theoretical orientation, treatment length, in studies for goal consensus (Column 3),
measures of goal consensus and/or collabo- collaboration (Column 4), and outcome
ration and outcome, time of measurement (Column 5). The majority (13 out of
15, 87%) of goal consensus studies used consensus and four for collaboration)
only one measure of goal consensus were derived from partial beta coefficients
(M = 1.20, SD = 0.56), and the majority of with two or more independent variables
collaboration studies (16 out of 19, 84%) (see table note). Partial beta coefficients
used only one collaboration measure underestimate the zero-order correlation
(M = 1.16, SD = 0.37). The studies tended coefficient.
to have more than one outcome measure When measures within a study were
(M = 2.14, SD = 1.53). Measures were completed or recorded by different numbers
often completed by more than one rater. In of participants or observers, we weighted
total, patients completed the most mea- each correlation by the number of patients
sures (n = 49), followed by therapists for whom the measures were completed and
(n = 37), and observers (n = 32). divided by the total number of patients
For each study, we coded the time in represented for each correlation.
therapy that measures were completed by
dividing the number of sessions in the study The Meta-Analyses
into thirds, and naming the thirds as early, We used a meta-analysis package (Schmidt
middle, and end of therapy (in all cases, & Le, 2005) that corrects for study artifacts
measures in the latter third were at the very (Hunter & Schmidt, 2004) such as the
end of treatment). Most measures were unreliability of measures used in each study.
completed at the end (54 measures) of ther- For the current meta-analyses, we corrected
apy, followed by early (17 measures) and for unreliability using coefficient alphas for
in the middle (3 measures) of therapy. the studies’ measures. Because each study
Twelve studies used at least one measure did not provide alphas for all measures used
(18 measures) that was completed conti- in the meta-analyses (i.e., collaboration,
nuously (in each third, often after each goal consensus, and outcome measures),
session) during therapy, and one study we used the artifact distribution option of
contained three measures completed at the program that allowed us to enter the
follow-up only. Finally, two studies did alphas that the studies provided as well as
not specify time of completion of measures alphas from other studies that used the
(7 measures). measures (contact corresponding author
for a reference list of additional studies
Estimation of Effect Sizes providing reliabilities). Effect sizes were
To obtain the effect sizes listed in Table 7.1, weighted by sample size and reliability so
for each study, we recorded correlations that effects that were more precise (i.e.,
or standardized β weights between goal derived from studies with larger sample
consensus measures and outcome measures sizes and greater reliability) were given
and/or collaboration measures and outcome more weight.
measures. We averaged correlations or stan-
dardized β weights in studies that had more Results: Goal Consensus
than one measure of goal consensus and/or Fifteen studies with a total sample size
collaboration or outcome to obtain one of 1,302 provided goal consensus–psycho-
effect size for the relationship of each thera- therapy outcome effect sizes for the meta-
peutic element (goal consensus, collabora- analysis (see Table 7.1). Weighting for
tion) to outcome for each study. Most effect sample size and unreliability of measures,
sizes were based on zero-order correlations; the meta-analysis yielded a mean correlation
however, a few effect sizes (three for goal of 0.34 (SD = 0.19) with a 95% confidence
interval of 0.23 to 0.45. The variability standard deviation (d = 0.68; Lyons, 2003)
between studies after removal of the effects improvement associated with a 1 standard
of unreliability of the measures and varia- deviation boost in collaboration. Thus,
tion in sample size was 0.02. Because this patient experience and well-being appear
procedure left virtually no variability in to be considerably enhanced with a better
effects due to differences between studies, quality collaborative relationship between
there was no variability to be explained by patient and therapist.
moderators. Therefore, as recommended
(Hunter & Schmidt, 2004), we did not Results: Relation of Goal Consensus
conduct a moderator analysis. The results and Collaboration
signify a medium (Cohen, 1992), unmod- Of the 7 studies that had measures of both
erated effect between goal consensus and goal consensus and collaboration, only 4
psychotherapy outcome. Because an r of reported effect sizes, or information from
0.34 is equivalent to a d of 0.72 (Lyons, which to calculate effect sizes, for the
2003), a 1 standard deviation improvement association between these two variables.
in goal consensus predicts nearly a 3/4 stan- Studies by Addis and Jacobson (2000;
dard deviation improvement in outcome. r = 0.17, n = 150), Abramowitz et al.
This is a substantial relationship, especially (2002; r = 0.17, n = 28), Clemence et al.
considering outcomes as meaningful as (2005; r = 0.09, n = 125), and Lingiardi
retention in treatment, symptom reduc- et al. (2005, r = 0.19, n = 37) provided
tion, and adaptive functioning. data for the goal consensus–collaboration
relationship. Collectively, these four studies
Results: Collaboration provided a sample of 340 patients. Weigh-
Nineteen studies involving a total sample ing for sample size and unreliability of
of 2,260 patients provided collaboration– measures, the meta-analysis on these four
psychotherapy outcome effect sizes for the studies yielded a mean correlation of 0.19
meta-analysis (Table 7.1). Weighing for (SD = 0), which represents an effect size
sample size and unreliability of measures, between small and medium (Cohen, 1992).
the meta-analysis yielded a mean correla- Variability between studies after removal of
tion of 0.33 (SD = 0.17) with a 95% the effects of unreliability of the measures
confidence interval of 0.25 to 0.42. The and variation in sample size was 0.
variability between studies after removal of
the effects of unreliability of the measures File Drawer Analyses
and variation in sample size was 0.02. It is possible that studies not included
Because this procedure left virtually no in the meta-analyses (e.g., unpublished
variability in effects due to differences papers, dissertations, book chapters)
between studies, there was no variability to could have null results that, if included,
be explained by moderators. Therefore, as would have reduced the effects we found.
recommended (Hunter & Schmidt, 2004), Thus, we conducted file drawer analyses
we did not conduct a moderator analysis. to determine how many studies with
The results denote a medium (Cohen, null results would reduce the effect sizes
1992), unmoderated effect between colla- substantially. For the goal consensus meta-
boration and psychotherapy outcome. Across analysis, we would need to have found 87
outcomes including service use, satisfaction studies with null results to reduce the effect
with services received, and patient improve- size to 0.05. For collaboration, the number
ment, this effect corresponds to a 2/3 of studies needed is 106. Thus, it is unlikely
that we would find such large numbers of input and feedback. The verbal interchanges
well-designed, unpublished studies. involved in goal consensus reflect a negoti-
For the goal consensus–collaboration ation in which patients and therapists
meta-analysis, however, we would need to together refine the goals and tasks of
find only 11 studies with null effects to therapy.
reduce the effect size to 0.05. Conse- In addition to completing homework,
quently, the results of the goal consensus– the patient’s contribution to the ongoing
collaboration analysis should be interpreted collaborative work of therapy takes the
cautiously. form of offering information, insights, self-
reflections, elaborations and explorations
Potential Moderators and Mediators of important themes, and “work(ing)
Although in meta-analyses there may be actively with the therapist’s comments”
moderators or mediators of statistical (Colli & Lingiardi, 2009, p. 723). Patients
relationships between variables, after may not recognize the importance of these
removal of the effects of unreliability of the behaviors and the role they play in goal
measures and variation in sample size, consensus and collaboration (Hatcher &
results of the current meta-analyses indi- Barends, 2006). Indeed, they tend to
cated unmoderated relationships between emphasize the importance of what the
the therapy elements of collaboration and therapist does, and even when prompted,
goal consensus and therapy outcome. We, downplay their contribution to the work of
therefore, did not analyze for moderators. therapy (Bedi, Davis, & Williams, 2005).
Perhaps some patients have so little confi-
Patient Contribution dence in their efficacy in relation to their
Patients, particularly those who are new to problems (after all, they have been unable
therapy, may have an inaccurate percep- to solve their troubles on their own) that
tion of the role they are expected to play they do not acknowledge the importance
in the treatment process. In their experi- of their part in collaborating with thera-
ences with other health professionals, such pists toward a successful outcome.
as physicians, patients tend to play a rela-
tively passive, submissive role, presenting Limitations of the Research
their symptoms and receiving treatment. This chapter only included studies with
The goals of such treatment typically do adults that were published in English and
not involve much discussion, and there in refereed journals, and it did not include
may be little collaboration regarding treat- studies with child or adolescent patients.
ment beyond patient compliance in follow- Because goal consensus and collaboration
ing professional directives. are often considered to be part of the
Psychotherapy, in contrast, requires working alliance, articles included in this
active involvement by patients from initial chapter may also have been included in
goal setting to termination. The therapist the analysis presented in Chapter 2 of this
cannot effect treatment alone. Patients volume concerning the working alliance in
bring their concerns to therapists, and adult psychotherapy. In contrast to analyses
together they conceptualize treatment goals reported in other chapters in this volume,
and ways to achieve them. Although thera- which covered a more extensive time period,
pists frequently reconceptualize patients’ the meta-analyses in this chapter used
problems, refine goals, and suggest ways to studies that were published in the past
achieve those goals, they do so with patients’ 10 years. For a review of studies published
prior to 2000, see our earlier chapter in the consensus and collaboration and measures
first edition of this volume (Tryon & that allow for their accurate assessment.
Winograd, 2002). Finally, while the results of the meta-
Although we included a goal consensus– analyses in this chapter indicate positive
collaboration meta-analysis using data from relationships between goal consensus and
articles in listed Table 7.1, we did not search outcome and between collaboration and
for additional articles that included these outcome, they do not provide proof that
two elements because their relationship was either goal consensus or collaboration
not the focus of the chapter. So, we advise causes positive outcomes.
readers to interpret the correlation between
goal consensus and collaboration reported
here cautiously.
The results of the primary meta-analyses
While the studies in the current meta-
indicate strong links between patient–
analyses represent improvements over those
therapist goal consensus and positive
included in our prior chapter, they were
therapy outcomes, as well as between their
not without limitations. Although many
collaboration and outcome. The results
studies reported results as effect sizes, several
point to a number of practices that psycho-
studies did not. Several studies also failed
therapists can profitably effect.
to report nonsignificant results. Statistical
correction of this problem through file • Begin work on client problems
drawer analyses or assignment of arbitrary only after you and the patient agree on
effect sizes is a poor substitute for having treatment goals and the ways you will go
the actual effect sizes. Also, some studies about reaching them.
reported results based on fewer participants • Rarely push your own agenda.
than indicated in their procedure sections. Listen to what your patients tell you and
Readers should bear these limitations, and formulate interventions with their input
those in the previous paragraph, in mind and understanding.
when interpreting results. The acceptance • Encourage patients’ contributions
of journal article reporting standards throughout psychotherapy by asking
(JARS; APA Publications and Communi- for their feedback, insights, reflections,
cations Working Board Group, 2008) by and elaborations. Regularly seek infor-
editors should address these difficulties mation from patients about their current
and allow for more precise meta-analytic functioning, motivation to change,
syntheses of research data. and social support and provide them
A glance at Table 7.1 shows the diver- with feedback about their progress
sity of goal consensus and collaboration (Harmon et al., 2007; Whipple et al.,
measures. These reflect the various defini- 2003; also see Chapter 10 by Lambert
tions of these elements in the literature, and & Shimokawa).
in the case of collaboration, the element • Educate patients about the
itself is defined by the measures used to importance of their collaborative
assess it (Bachelor et al., 2007). Instru- contribution to the success of therapy.
mentation frequently evolves from theoreti- Psychotherapists can do so by sharing with
cal advances, and we endorse continuing patients the results of research, such as
conceptualization of these elements. The those reviewed in this chapter, that link
goal should be to provide researchers and their collaborative contribution to
practitioners with clear definitions of goal successful outcomes.
• Encourage homework completion. research in psychology. American Psychologist,
To enhance homework completion, 63, 839–51.
Bachelor, A., Laverdière, O., Gamache, D., &
encourage patient collaboration in
Bordeleau, V. (2007). Clients’ collaboration in
formulating homework assignments; therapy: Self-perceptions and relationships with
assign homework that relates to client psychological functioning, interpersonal
treatment goals; begin with small, easily relations, and motivation. Psychotherapy: Theory,
accomplished assignments and build to Research, Practice, Training, 44, 175–192.
larger ones; define homework tasks clearly; Bedi, R. P., Davis, M. D., & Williams, M. (2005).
Critical incidents in the formation of the
give homework assignments in writing;
therapeutic alliance from the client’s perspective.
provide written reminders to complete Psychotherapy: Theory, Research, Practice, Training,
tasks; encourage and incorporate client 42, 311–323.
feedback on homework (Detweiler & ∗
Bogalo, L., & Moss-Morris, R. (2006). The
Whisman, 1999; Nelson, Castonguay, effectiveness of homework tasks in an irritable
& Barwick, 2007). bowel syndrome self-management programme.
New Zealand Journal of Psychology, 35, 120–125.
• Be “on the same page” with patients.
Bordin, E. S. (1979). The generalizability of the
Check frequently with patients to make psychoanalytic concept of the working alliance.
sure that you understand each other and Psychotherapy: Theory, Research, Practice, Training,
are working toward the same ends. 16, 252–260.
∗
• Modify your treatment methods and Brocato, J., & Wagner, E. F. (2008). Predictors of
relational stance, if ethically and clinically retention in an alternative-to-prison substance
abuse treatment program. Criminal Justice and
appropriate, in response to patient
Behavior, 35, 99–119.
feedback. ∗
Burns, D. D., & Spangler, D. L. (2000). Does
psychotherapy homework lead to improvements
in depression in cognitive-behavioral therapy or
References
does improvement lead to increased homework
References marked with an asterisk (∗) indi- compliance? Journal of Consulting and Clinical
cate studies included in the meta-analyses. Psychology, 68, 46–56.
∗ ∗
Ablon, J. S., Levy, R. A., & Katzenstein, T. (2006). Busseri, M. A., & Tyler, J. D. (2004). Client-
Beyond grand names of psychotherapy: Identi- therapist agreement on target problems, working
fying empirically supported change processes. alliance, and counseling outcome. Psychotherapy
Psychotherapy, Theory, Research, Practice, Training, Research, 14, 77–88.
43, 216–231. ∗
Caspar, F., Grossmann, C., Unmüssig, C., &
∗
Abramowitz, J. S., Franklin, M. E., Zoellner, L. A., Schramm, E. (2005). Complementary thera-
& DiBernardo, C. L. (2002). Treatment compli- peutic relationship: Therapist behavior, inter-
ance and outcome in obsessive-compulsive personal patterns, and therapeutic effects.
disorder. Behavior Modification, 26, 447–463. Psychotherapy Research, 15, 91–102.
∗ ∗
Ackerman, S. J., Hilsenroth, M. J., Baity, M. R., & Clemence, A. J., Hilsenroth, M. J., Ackerman, S. J.,
Blagys, M. D. (2000). Interaction of therapeutic Strassle, C. G., & Handler, L. (2005). Facets of
process and alliance during psychological assess- the therapeutic alliance and perceived progress in
ment. Journal of Personality Assessment, 75, psychotherapy: Relationship between patient
82–109. and therapist perspectives. Clinical Psychology
∗
Addis, M. E., & Jacobson, N. S. (2000). A closer and Psychotherapy, 12, 443–454.
look at the treatment rationale and homework Cohen, J. (1992). A power primer. Psychological
compliance in cognitive-behavioral therapy for Bulletin, 112, 155–159.
depression. Cognitive Therapy and Research, 24, Colli, A., & Lingiardi, V. (2009). The Collaborative
313–326. Interactions Scale: A new transcript-based
APA Publications and Communications Board method for assessment of therapeutic alliance
Working Group on Journal Article Reporting ruptures and resolutions in psychotherapy.
Standards. (2008). Reporting standards for Psychotherapy Research, 19, 718–734.
∗ ∗
Cowan, M. J., Freedland, K. E., Burg, M. M., Hegel, M. T., Barrett, J. E., Cornell, J. E., &
Saab, P. G., Youngblood, M. E., Cornell, C. E., Oxman, T. E. (2002). Predictors of response to
et al. (2008). Predictors of treatment response problem-solving treatment of depression in
for depression and inadequate social support– primary care. Behavior Therapy, 33, 511–527.
The ENRICHD randomized clinical trial. Horvath, A. O., & Greenberg, L. S. (1989a).
Psychotherapy and Psychosomatics, 77, 27–37. Development and validation of the Working
Detweiler, J. B., & Whisman, M. A. (1999). The Alliance Inventory. Journal of Counseling
role of homework assignments in cognitive ther- Psychology, 36, 223–233.
apy for depression: Potential methods for Horvath, A. O., & Greenberg, L. S. (1989b). The
enhancing adherence. Clinical Psychology: Science development of the Working Alliance Inventory.
and Practice, 6, 267–282. In L. S. Greenberg & W. M. Pinsof (Eds.), The
∗
Dunn, H., Morrison, A. P., & Bentall, R. P. (2006). psychotherapeutic process: A research handbook
The relationship between patient suitability, (pp. 529–556). New York: Guilford Press.
therapeutic alliance, homework compliance, Hunter, J. E., & Schmidt, F. L. (2004). Methods of
and outcome in cognitive therapy for psycho- meta-analysis: Correcting error and bias in research
sis. Clinical Psychology and Psychotherapy, 13, findings (2nd ed.). Thousand Oaks, CA: Sage.
∗
145–152. Lingiardi, V., Filippucci, L., & Baiocco, R. (2005).
∗
Fitzpatrick, M. R., Iwakabe, S., & Stalikas, A. Therapeutic alliance evaluation in personality
(2005). Perspective divergence in the working disorders psychotherapy. Psychotherapy Research,
alliance. Psychotherapy Research, 15, 69–79. 15, 45–53.
∗ ∗
Gabbay, M., Shields, C., Bower, B., Sibbald, M., Long, J. R. (2001). Goal agreement and early
King, M., & Ward, E. (2003). Patient- therapeutic change. Psychotherapy, 38, 219–232.
practitioner agreement: Does it matter? Luborsky, L., Barber, J. P., Siqueland, L., Johnson,
Psychological Medicine, 33, 241–251. S., Najavits, L. M., Frank, A., et al. (1996). The
∗
Gonzalez, V. M., Schmitz, J. M., & DeLaune, K. revised Helping Alliance Questionnaire (Haq-II):
A. (2006). The role of homework in cognitive- Psychometric properties. Journal of Psychotherapy
behavioral therapy for cocaine dependence. Practice and Research, 5, 260–271.
Journal of Consulting and Clinical Psychology, 74, Lyons, L. C. (2003, January 30). Meta-analysis:
633–637. Methods of accumulating results across research
∗
Graf, M. C., Gaudiano, B. A., & Geller, P. A. domains. Retrieved from http://www.lyonsmorris.
(2008). Written emotional disclosure: A con- com/MetaA/studyfx1.htm
trolled study of expressive writing homework in Marmar, C. R., Gaston, L., Gallagher, D., &
outpatient psychotherapy. Psychotherapy Research, Thompson, L. W. (1989). Alliance and outcome
18, 389–399. in late-life depression. Journal of Nervous and
Harmon, S. C., Lambert, M. J., Smart, D. M., Mental Disease, 177, 464–472.
Hawkins, E., Nielsen, S. L., Slade, K., et al. McAuley, E., Duncan, T. E., & Russell, D. W.
(2007). Enhancing outcome for potential (1992). Measuring causal attributions: The
treatment failures: Therapist-client feedback and revised Causal Dimension Scale (CDS-II).
clinical support tools. Psychotherapy Research, 17, Personality and Social Psychology Bulletin, 18,
379–392. 566–573.
Hatcher, R. L. (1999). Therapists’ views of treat- McNair, D., & Lorr, M. (1964). Three kinds of
ment alliance and collaboration in therapy. psychotherapy goals. Journal of Clinical
Psychotherapy Research, 9, 405–423. Psychology, 20, 390–593.
Hatcher, R. L., & Barends, A. W. (1996). Patients’ Nelson, D. L., Castonguay, L. G., & Barwick, F.
view of the alliance in psychotherapy: Exploratory (2007). Directions for the integration of home-
factor analysis of three alliance measures. Journal work in practice. In N. Kazantzis & L. L’ Abate
of Consulting and Clinical Psychology, 64, (Eds.), Handbook of homework assignments in
1326–1336. psychotherapy: Research, practice, and prevention
Hatcher, R. L., & Barends, A. W. (2006). How (pp. 425–444). New York: Springer.
a return to theory could help alliance research. Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994).
Psychotherapy: Theory, Research, Practice, Process and outcome in psychotherapy–Noch
Training, 43, 292–299. einmal. In A. E. Bergin & S. L. Garfield (Eds.),
Hanbook of psychotherapy and behavior change Psychotherapy relationships that work: Therapist
(4th ed., pp. 270–376). New York, NY: Wiley. contributions and responsiveness to patients
∗
Principe, J. M., Marci, C. D., Glick, D. M., & (pp. 109–125). New York: Oxford University
Ablon, J. S. (2006). The relationship among Press.
∗
patient contemplation, early alliance, and con- Wettersten, K. B., Lichtenberg, J. W., &
tinuation in psychotherapy. Psychotherapy: Theory, Mallinckrodt, B. (2005). Associations between
Research, Practice, Training, 43, 238–243. working alliance and outcome in solution-
∗
Schönberger, M., Humle, F., & Teasdale, T. (2006). focused brief therapy and brief interpersonal
Subjective outcome of brain injury rehabilitation therapy. Psychotherapy Research, 15, 35–43.
in relation to the therapeutic working alliance, Whipple, J. L., Lambert, M. J., Vermeersch, D. A.,
client compliance and awareness. Brain Injury, Smart, D. W., Nielsen, S. L., & Hawkins, E. J.
20, 1271–1282. (2003). Improving the effects of psychotherapy:
∗
Schönberger, M., Humle, F., & Teasdale, T. (2007). The use of early identification of treatment
The relationship between clients’ cognitive func- failure and problem-solving strategies in routine
tioning and the therapeutic working alliance in practice. Journal of Counseling Psychology,
post-acute brain injury rehabilitation. Brain 50, 59–68.
∗
Injury, 21, 825–836. Whittal, M. L., Thordarson, D. S., & McLean, P. D.
Schmidt, F. L., & Le, H. (2005). Software for the (2004). Treatment of obsessive-compulsive
Hunter-Schmidt meta-analysis methods (Version disorder: Cognitive behavior therapy vs. exposure
1.1) [Computer software]. Iowa City: University and response prevention. Behaviour Research and
of Iowa, Department of Management & Therapy, 43,1559–1576.
∗
Organization. Woods, C. M., Chambless, D. L., & Steketee, G.
∗
Stein, M. D., Solomon, D. A., Herman, D. S., (2002). Homework compliance and behavior
Anthony, J. L., Ramsey, S. E., Anderson, B. J., therapy outcome for panic with agoraphobia and
et al. (2004). Pharmacotherapy plus psychother- obsessive compulsive disorder. Cognitive
apy for treatment of depression in active injec- Behaviour Therapy, 31, 88–95.
∗
tion drug users. Archives of General Psychiatry, Yovel, I., & Safren, S. A. (2007). Measuring
61, 152–159. homework utility in psychotherapy: Cognitive-
Tryon, G. S., Blackwell, S. C., & Hammel, E. F. behavioral therapy for adult attention-deficit
(2007). A meta-analytic examination of client- hyperactivity disorder as an example. Cognitive
therapist perspectives of the working alliance. Therapy and Research, 31, 385–399.
∗
Psychotherapy Research, 17, 629–642. Zane, N., Sue, S., Chang, J., Huang, L.,
Tryon, G. S., & Winograd, G. (2001). Goal Huang, J., Lowe, S., et al. (2005). Beyond ethnic
consensus and collaboration. Psychotherapy: match: Effects of client-therapist cognitive match
Theory, Research, Practice, Training, 38, 385–389. in problem perception, coping orientation, and
Tryon, G. S., & Winograd, G. (2002). Goal consensus therapy goals on treatment outcomes. Journal of
and collaboration. In J. C. Norcross (Ed.), Community Psychology, 33, 569–585.
C HA P TER
8 Positive Regard and Affirmation
Barry A. Farber and Erin M. Doolin
Author Note. We gratefully acknowledge time—and still an enormously influen-

the invaluable research assistance provided tial one, though currently cast in some-
by Alex Behn, Sarah Bellovin-Weiss, and what different (e.g., more evidence-based)
Valery Hazanov. terms—was that technical expertise on the
part of the therapist, especially in terms of
The book [Client-Centered Therapy] . . . choice and timing of interventions, is the
expresses, I trust, our conviction that essential discriminating element between
though scientists can never make effective and noneffective therapy. Under
therapists, it can help therapy; that though the sway of Rogers’ burgeoning influence
the scientific finding is cold and abstract, in the late 1950s and throughout the 1960s,
it may assist us in releasing forces that are the notion that the relationship per se was
warm, personal, and complex; and that the critical factor in determining therapeu-
though science is slow and fumbling, it tic success, took hold.
represents the best road we know to the Over the years, a great many studies have
truth, even in so delicately intricate an attempted to investigate Rogers’ claims
area as that of human relationships. regarding the necessary and sufficient con-
—Rogers, 1951, p. xi ditions of therapy. There is, then, a sub-
stantial body of research to draw upon in
Over 50 years ago, in what is now looking at the association between the ther-
considered a classic paper, Carl Rogers apist’s positive regard for his or her patients
(1957) posited that psychotherapists’ pro- and therapeutic outcome. However, as
vision of positive regard (nonpossessive detailed below, drawing firm conclusions
warmth), congruence (genuineness), and from these efforts has been difficult. The
empathy were the necessary and sufficient problems that typically plague the investi-
conditions for therapeutic change. Rogers gation of complex psychological issues have
had been developing these views for many been played out in this area as well: incon-
years, some of which were expressed as early sistent findings, small sample sizes, lack of
as 1942 in his seminal work, Counseling standardized measures, and lack of opera-
and Psychotherapy. Still, the publication tional definitions of the concepts them-
of the 1957 article seems to have catalyzed selves. In addition, as the Rogerian influence
a shift in the way that many thought about on clinical practice has diminished in the last
the putative mechanisms of psychothe- three decades––or, more accurately, has been
rapeutic change. The prevailing view at the incorporated into the psychotherapeutic
168
mainstream with little awareness or This observation suggests that Freud was
explicit acknowledgment (Farber, 2007)–– unaware of, or at least underappreciated,
empirical studies based on Rogerian con- what may well have been the most potent
cepts have also waned. Similarly, the focus elements of his approach—that along with
of research and theory has shifted away whatever positive effects accrue as a result
from the individual contributions of each of accurate interpretations, psychoanalytic
of the participants in therapy toward a con- success has arguably always been based
sideration of the alliance or therapeutic substantially on the undervalued ability
relationship—what each member of the of the analyst to be empathic and, even
dyad contributes to the ongoing, interac- more to the point of this chapter, to be
tive process of the work. supportive and positively regarding of his
Whereas consideration of the therapeu- or her patients.
tic relationship as mutative began with In this chapter, we review Rogers’
Rogers, therapists of varying persuasions, ideas about the concept of positive regard
even those from theoretical camps that and discuss how the use of multiple terms
had traditionally emphasized more techni- (including positive regard, affirmation, respect,
cal factors, have begun to acknowledge the warmth, support, and prizing) has led to
importance of the relationship. Behaviorists conceptual confusion as well as empirical
and cognitive-behaviorists now suggest that difficulties in determining the link between
a good relationship may facilitate the pro- this phenomenon and therapeutic out-
vision of their technical interventions come. The emphasis of this chapter is on
(e.g., Beck, 1995), and many psychoana- meta-analytically reviewing the findings of
lysts have shifted their clinical perspective those empirical studies that have investi-
to emphasize “relational” factors (Mitchell gated the relation between therapist sup-
& Aron, 1999; Wachtel, 2009). But even port and treatment outcome in individual
before these relatively recent developments, psychotherapy. Most studies of positive
there is evidence to suggest that Freud’s regard are framed within a Rogerian (per-
psychoanalytic cases were only successful son-centered) paradigm; however, as noted
when he was supportive and positively above, nearly all schools of therapy now
regarding. As Breger (2009) has noted: either explicitly or implicitly promote
the value of this basic attitude toward
When Freud followed these [psychoanalytic] patients. Thus, the results of these studies
rules his patients did not make progress. have implications for the conduct not only
His well-known published cases are of person-centered therapists, but for virtu-
failures . . . in contrast are patients like ally all psychotherapists.
Kardiner and others—cases he never
wrote or publicly spoke about—all of Definitions and Measures
whom found their analyses very helpful.
With these patients, what was curative To the extent that the therapist finds
was not neutrality, abstinence, or himself experiencing a warm acceptance
interpretations of resistance, but a more of each aspect of the client’s experience
open and supportive relationship, as being a part of that client, he is
interpretations that fit their unique experiencing unconditional positive
experiences, empathy, praise, and the regard . . . it means there are no conditions
feelings that they were liked by their of acceptance . . . it means a ‘prizing’ of
analyst. (p. 105). the person . . . it means a caring for the
fa r b e r, d o o l i n 169
client as a separate person (Rogers, 1957, of therapist affirmation. We will use the
p. 101). phrase positive regard to refer to the general
constellation of attitudes encompassed by
From the beginning of his efforts to explicate this and similar phrases.
the essential elements of client-centered Further confusing the conceptual issues
therapy, Rogers focused on positive regard at play here, Rogers’ focus on accepting and
and warmth: “Do we tend to treat individ- affirming the client has, from the outset,
uals as persons of worth, or do we subtly been conflated with an emphasis on empa-
devaluate them by our attitudes and behav- thy and genuineness. The therapist’s attempt
ior? Is our philosophy one in which respect to “provide deep understanding and accep-
for the individual is uppermost?” (1951, tance of the attitudes consciously held at
p. 20). Implicit in this statement is his the moment by the client” could only be
disapproval of what he perceived as the accomplished by the therapist’s “struggle to
arrogance of, and strict hierarchical dis- achieve the client’s internal frame of refer-
tinctions between, psychotherapists and ence, to gain the center of his own percep-
patients held by the psychoanalytic com- tual field and see with him as perceiver”
munity at that time. Rogers did not believe (1951, pp. 30–31). Rogers seems to be sug-
that anyone, including a therapist, could gesting here that positive regard (including
be more expert or knowledgeable about the component of acceptance) can best be
a client than the client him or herself. He achieved through empathic identification
did not believe that a therapist’s neutrality, with one’s client. In a similar vein, Rogers
dispassionate stance, or even intellectual suggested that the therapist’s genuineness
understanding could facilitate a client’s or congruence was a prerequisite for his or
growth—no matter how astute the inter- her experience of positive regard and empa-
pretations emanating from such a therapy thy (Rogers & Truax, 1967).
might be. Instead, he believed that treat- Further problems with the concept of
ing clients in a consistently warm, highly positive regard have been identified (e.g.,
regarding manner would inevitably allow Lietaer, 1984). One is that there may be
them to grow psychologically, to fulfill their an inherent tension between this attitude
potential. and that of genuineness; that is, therapists’
To this day, agreeing on a single phrase own conflicts inevitably affect what they
to refer to this positive attitude remains can and cannot truly accept or praise in
problematic. It is most often termed posi- others. A second, related problem is that it
tive regard but early studies and theoretical is unlikely that any therapist can provide
writings preferred the phrase nonpossessive constant doses of unconditional positive
warmth. In his famous filmed work with regard in that we all reinforce selectively. As
Gloria (Shostrom, 1965), Rogers struggled Rogers himself anticipated:
to find a single phrase to illuminate this
concept: it is, he said, “real spontaneous The phrase ‘unconditional positive regard’
praising; you can call that quality accep- may be an unfortunate one, since it
tance, you can call it caring, you can call it a sounds like an absolute, an all-or-nothing
non-possessive love. Any of those terms tend dispositional concept . . . From a clinical
to describe it.” Some reviews of “accep- and experiential point of view I believe
tance, nonpossessive warmth, or positive the most accurate statement is that
regard” (Orlinsky, Grawe, & Parks, 1994, the effective therapist experiences
p. 326) grouped them under the category unconditional positive regard for the
client during many moments of his a person,” “I feel appreciated by her,” and
contact with him, yet from time to time “she is friendly and warm toward me.”
he experiences only a conditional positive Representative negative items include
regard—and perhaps at times a negative “I feel that she disapproves of me,” “She
regard, though this is not likely in is impatient with me,” and “At times she
effective therapy. It is in this sense that feels contempt for me.”
unconditional positive regard exists as Unconditionality of Regard is explained
a matter of degree in any relationship. by Barrett-Lennard (1986, p. 443) in terms
(p. 101). of its stability, “in the sense that it is
not experienced as varying with other or
How can one, then, assess a therapist’s otherwise dependently linked to particular
level of positive regard without implicitly attributes of the person being regarded.”
measuring empathy or genuineness as well? Examples of positively worded items: “How
In fact, reading transcripts of Rogers’ work much he likes or dislikes me is not altered
(e.g., Farber, Brink, & Raskin, 1996) makes by anything that I tell him about myself ”;
clear how difficult it is to tease out pure “I can (or could) be openly critical or appre-
examples of positive regard. Rogers is ciative of him without really making him
consistently “with” his clients, testing his feel any differently about me.” Examples
understanding, clarifying, and intent on of negatively worded items: “Depending
entering and grasping as much as possible on my behavior, he has a better opinion of
the client’s experiential world. For these me sometimes than he has at other times”;
reasons, most research focusing on the “Sometimes I am more worthwhile in his
effects of therapist positive regard have eyes than I am at other times.”
used measures, typically either the Barrett- Truax developed two separate instru-
Lennard Relationship Inventory (BLRI; ments for the measurement of Rogers’ facili-
1964, 1978) or the Truax Relationship tative conditions. One was a set of scales
Questionnaire (Truax & Carkhuff, 1967), to be used by raters in their assessment of
that include items reflecting multiple, over- these conditions as manifest in either live
lapping, relational elements. observations or through tape recordings of
The BLRI consists of 64 items across sessions. There are five stages on the scale
four domains (Level of Regard, Empathic that measures Nonpossessive Warmth. At
Understanding, Unconditionality of Regard, Stage 1, the therapist is “actively offering
Congruence). Eight items are worded posi- advice or giving clear negative regard”
tively, eight negatively in each domain; (Truax & Carkhuff, 1967, p. 60); at Stage 5,
each item is answered on a +3 (yes, strongly the therapist “communicates warmth with-
felt agreement) to −3 (no, strongly felt out restriction. There is a deep respect for
disagreement) response format. This ins- the patient’s worth as a person and his
trument can be used by the client, thera- rights as a free individual” (p. 66).
pist, or both. Both Level of Regard and The second instrument developed by
Unconditionality have been used in research Truax, The Relationship Questionnaire,
studies to investigate the influence of posi- was to be used by clients. This measure
tive regard. Level of Regard, according to consists of 141 items marked “true” or
Barrett-Lennard (1986, p. 440–441), “is “false” by the client. Of these items, 73 are
concerned in various ways with warmth, keyed to the concept of nonpossessive
liking/caring, and ‘being drawn toward’.” warmth; it is noteworthy, however, that
Positive items include “she respects me as many of these items are also keyed to the
other two facilitative conditions (genu- High scores on this dimension reflect the
ineness and empathy). That is, a “true” therapist’s ability to teach or encourage a
response on one item may count toward a patient in a kind or positive manner.
higher score on more than one subscale. Whereas the “bond” component of
Representative items on the Nonpossessive various alliance measures (e.g., Horvath &
Warmth scale: “He seems to like me no Greenberg, 1989; Tracey & Kokotovic,
matter what I say to him” (this item is also 1989) contains aspects of positive regard
on the “genuineness” scale); “He almost phenomena that have been elucidated
always seems very concerned about me”; above, its items primarily assume an inter-
“He appreciates me”; “I feel that he really action between patient and therapist, one
thinks I am worthwhile”; “even if I were to that reflects the contributions and charac-
criticize him, he would still like me”; and teristics of each. Thus, results from studies
“whatever I talk about is OK with him.” using alliance measures were not included
In addition to these scales, therapist pos- in our meta-analysis.
itive regard has been assessed via instru-
ments designed primarily to measure the Clinical Examples
strength of the alliance. In particular, the The case examples below have been pur-
Vanderbilt Psychotherapy Process Scale posely drawn from disparate theoretical
(VPPS) has been used in this manner. perspectives. Although the concept of posi-
The VPPS is “a general-purpose instru- tive regard originated with Rogers, the pro-
ment designed to assess both positive and vision of this facilitative condition can and
negative aspects of the patient’s and the does occur in the work of practitioners of
therapist’s behavior and attitudes that are quite distinct therapeutic traditions.
expected to facilitate or impede progress
in therapy” (Suh, Strupp, & O’Malley, Case Example 1
1986, p. 287). Each of 80 items is rated by
clinical observers on a 5-point, Likert-type Client: I can outsmart people. I won’t
scale, either from the actual therapy sessions be taken advantage of. I call the
or from video- or audiotapes of therapy. shots.
Factor analyses of these items have yielded Therapist: It seems important for you to
eight subscales, one of which, Therapist be dominant in every relationship.
Warmth and Friendliness, closely approxi- Client: Yes. I don’t show emotion and
mates the concept of positive regard. The I don’t put up with it in anyone else.
specific therapist attributes rated in this I don’t want someone to get all
subscale include “involvement” (the thera- hysterical and crying with me. I don’t
pist’s engagement in the patient’s experi- like it.
ence), “acceptance” (the therapist’s ability Therapist: How did you learn that
to help the patient feel accepted), “warmth being emotional is a sign of
and friendliness,” and “supportiveness” (the weakness?
therapist’s ability to bolster the patient’s Client: I don’t know.
self-esteem, confidence, and hope). Therapist Therapist: What if you meet your
positive regard has also sometimes been intellectual match, if you can’t
measured through the use of Structural “outsmart” them?
Analysis of Social Behavior (SASB; Client: (silence)
Benjamin, 1984), specifically through the Therapist: Okay, what if someone got
dimension of Helping and Protecting. to you through your feelings?
Client: Last week you did. It bothered therapist might do) but in doing so, explic-
me all day. itly conveys the fact that he cares for this
Therapist: That you were weak? patient far more than she imagines to be
Client: Yeah. the case.
Therapist: I didn’t see you as submissive
or weak. In fact, since showing
emotion is so difficult for you I saw it Case Example 3
as quite the opposite. Client: It really hurts when I think
In this example, the therapist, primarily about the fact that it is over.
psychodynamic in orientation, initially Therapist: Yes, of course it hurts. It
tries to get the patient to open up about hurts because you loved him and it
his past and discuss his “faulty strategy” did not work out. It shows, I think,
of dominating relationships. It appears as your capacity to love and to care. But
if they are about to discuss transference it also hurts to have that ability.
issues. However, the therapist shifts at the Client: I don’t think I’ll ever feel
end, perhaps intuitively sensing that what that way.
would be most effective for this patient (at Therapist: Right now it may be
least at this moment) is a statement of true important for you to protect yourself
positive regard. Thus, the therapist is affirm- with that feeling. Perhaps we can
ing, suggesting that she views the client not look at what you have learned about
as weak or submissive but rather the oppo- yourself and your needs and the kind
site, as perhaps brave for doing something of man who would be right for you.
that was difficult for him. Client: What do you mean?
Therapist: I mean that you have a great
ability to love. But what can you
Case Example 2 learn about what you need in a man
that [Tom] lacked?
“You’re reading me entirely wrong. I don’t Client: I guess I learned not to get
have any of those feelings. I’ve been pleased involved with a married man.
with our work. You’ve shown a lot of Therapist: What do you think led you
courage, you work hard, you’ve never to think you’d be able to handle
missed a session, you’ve never been late, being involved with a married man?
you’ve taken chances by sharing so many Client: Well, after my marriage ended,
intimate things with me. In every way I guess I didn’t want to get too
here, you do your job. But I do notice that attached. So I thought that being
whenever you venture a guess about how involved with someone who is
I feel about you, it often does not jibe with married would keep me from being
my inner experience, and the error is hurt.
always in the same direction: You read Therapist: Perhaps you’ve learned that
me as caring for you much less than I do” you have such a strong ability to love
(Yalom, 2002, p. 24). that you can’t compartmentalize your
feelings that way (Leahy, 2001, p. 82).
In this example, Yalom, an existential
therapist, not only offers assumedly accu- In this example, Leahy, a cognitive thera-
rate feedback to his patient on her interper- pist, is not only consistently empathic (“of
sonal tendencies (much like a psychoanalytic course it hurts”) and not only attempting
to teach his patient something about her- Client: I actually surprised myself.
self and her needs and choices, but he It didn’t even feel so risky. I just
also makes sure that he contextualizes his “went with it.” He smiled,
interventions in a supportive, caring way, I smiled back . . .
emphasizing his patient’s “strong ability to Therapist: Good. And that was so
love.” courageous of you . . ..
Client: Yeah, that’s me, I guess. My
Case Example 4 mother used to say that my refusal
to not give in irritated the hell out
Client: (smiles) I think I’m having male of my father. I’m sure I disappointed
menopause. him as a son . . . But sometimes,
Therapist: (smiles) OK, but I think I’m glad he’s gone. I feel guilty
you’ll need to explain that condition thinking and saying it but if I had
to me. to choose which parent would go
Client: (laughs) I met this great guy first, I’m not sorry it was him and
coming out of the supermarket. glad it wasn’t my mother. I really
And we chatted right there on the still need her.
street and we exchanged phone Therapist: I know. And you know,
numbers. He probably won’t call. I want to tell you how much
He’s a lot younger than I am so I appreciate your honesty in
I don’t think he’s really interested, allowing yourself to think about
but, hey, I actually had a daylight these things that are sometimes
conversation with an attractive man hard to think about.
and he knows my name. Now that’s Client: Thanks. (Pause.) I hope he
something, huh? calls me.
Therapist: Yes, it is. And something Therapist: I hope so too.
different for you. Here, a relationally oriented psychody-
Client: Yeah, I’m feeling less creature- namic clinician banters somewhat (“you’ll
like. More human these days. have to explain that condition to me”) as
Like coming out a means to be connected and supportive.
from under a rock. Oh, I finished my Moreover, she values her patient’s efforts
painting . . . the one with the lost to change (“that was so courageous of
boy. I thought about what you said you”) and gives him credit for the work
about the boy feeling lost . . . When he’s doing in the here-and-now of the clini-
I was finishing the painting, I felt cal setting (“how much I appreciate your
like . . . it’s almost like you and honesty”).
I came up with that together.
Therapist: I feel that, too. I think that Case Example 5
we each contribute to our work here
together. The accomplishment of Client: Yeah, I don’t feel like it’s [filling
finishing the painting is all yours, out diary cards every day] for me.
though. And talking to a man you I don’t want to wake up every day,
find attractive, giving him your name and go “Oh I felt like suicide last
and phone number . . . Sounds well, Tuesday! Oh my god, I was sad last
something commonplace, but not week!” I don’t want to keep
for you. Not in a long time. remembering!
Therapist: Oh, okay, so you want one Case Example 6
of those therapies where you don’t
remember things? Client: I feel like there are people who
Client: No, I just don’t want to keep do care and accept me. I do, but.
bringing it up all the time. “Oh, Therapist: But the person that
I was raped on so-and-so date, can’t accept and value you, is
let me remember what I felt at actually you.
the time.” Client: Yes, mostly.
Therapist: Yeah, it’s so painful to bring Therapist: It seems the person who
up this stuff. Why would anyone is hardest on you is you.
want that? Client: Yes. No one else would be as
Client: Yeah, exactly! cruel to me as I am.
Therapist: Now here’s the dilemma. Therapist: And make such harsh
We could not talk about your judgments, you’re pretty tough
problems, and if this would take on yourself.
away your pain and misery, I’d be Client: Yes, I wouldn’t judge my friends
all for it. On the other hand, if we the way I judge myself.
help you figure out how to tolerate Therapist: No, you’re not a very good
your bad feelings, then you won’t friend to yourself.
have to rely on your pain medicine Client: No, I wouldn’t treat anyone
or resort to thinking of killing the way I treat me.
yourself when these feelings Therapist: Maybe because you can
come up. see what is lovable in them, but
Client: But the feelings are horrible! not in yourself. To you, you’re
What am I supposed to do, just unlovable.
wave a magic wand to make Client: Maybe there are small pieces
them go away? You make it sound of me that are lovable.
so easy. Therapist: (pause) So there are parts
Therapist: It’s not easy at all. This is of you that you see as OK, as worthy
incredibly tough and painful for you, of being loved.
and I also believe you have what it Client: Yes, I guess. The child in me,
takes to do it. (Adapted from the child that struggled and survived.
McMain, Korman, & Dimeff, 2001, She, I, can still be playful and fun
p. 196). and warm.
In this dialogue, a dialectical behavior Therapist: Those are very wonderful
therapist offers a supportive qualities.
statement (“I believe you have Client: She’s strong, a survivor.
what it takes to do it”) that has Therapist: She’s a part of you that you
much in common with the can hold on to.
comments of the relationally Client: Yes.
oriented therapist in the previous Therapist: Do you think she’d judge
example. Here, the therapist’s you so harshly?
empathic response (“it’s not easy Client: No, she loves me.
at all”) is followed up by a more Therapist: To this special child part of
explicit statement of positive you, none of you is unforgivable.
regard. Client: No, she loves all of me.
In this final example, the client-centered Wolfe, 1978) later pointed out—namely,
therapist is clearly conveying to the patient that there are multiple ways of understand-
that she is worthy of respect and love. The ing such complex data. For example, of 108
therapist’s positive regard for the patient correlations noted in Truax and Mitchell’s
may allow her to begin to view herself report, only 34 were reported as significantly
as the therapist does. These last few exam- positive. While none of these correlations
ples are prime illustrations of the multi- were significantly negative, relatively few
ple aspects of positive regard, including were significantly positive.
affirmation, trust, understanding, warmth, In a follow-up review, Mitchell, Bozarth,
interest, and respect. and Krauft (1977) evaluated 11 studies
that investigated the relationship between
Previous Reviews positive regard (here again termed nonpos-
Before describing the results of our meta- sessive warmth) and treatment outcome.
analysis, we summarize several previous According to these authors, at most four
reviews of the association between positive of these studies offered support for the
regard and outcome. The first such effort proposition that higher levels of therapist-
was by Truax and Carkhuff (1967) in their provided warmth lead to better outcome.
book, Toward Effective Counseling and The following year, Orlinsky and Howard
Psychotherapy. Many of the studies they (1978) reviewed 23 studies, concluding
cited failed to report the separate associa- that approximately two thirds of these
tions of each of Rogers’ facilitative condi- indicated a significant positive association
tions to outcome, focusing instead on the between therapist warmth and therapeutic
aggregate results of all three conditions outcome, with the remaining one third
taken together. They did, however, review showing mostly null results. However, they
10 studies from which conclusions could also added several caveats, notably that
be drawn on the effects of positive regard the uneven quality and methodological
alone on therapeutic outcome, finding that flaws in the research made any firm con-
8 of these supported the hypothesis that clusions suspect. Their conclusion: “If they
nonpossessive warmth (the preferred term [warmth and empathy] do not by them-
at that time) is significantly associated with selves guarantee a good outcome, their
therapeutic improvement. presence probably adds significantly to the
Next, Bergin and Garfield’s first (1971) mix of beneficial therapeutic ingredients,
edition of Handbook of Psychotherapy and and almost surely does no harm” (p. 293).
Behavior Change included a chapter by As part of a comprehensive review of
Truax and Mitchell that summarized the process and outcome in psychotherapy,
results of 12 studies (involving 925 clients) Orlinsky and Howard (1986) conducted
that included nonpossessive warmth. The separate reviews of studies evaluating the
authors contended that the evidence was effects of therapist support and therapist
quite positive in regard to the relation- affirmation. They identified 11 studies that
ship between warmth and therapeutic out- included a support/encouragement variable;
come, noting that there was a statistically within this group of studies they focused
significant relationship between this vari- on 25 separate findings. Their conclusion:
able and a total of 34 specific outcome “Although 6 of the 25 are significantly
measures. Nevertheless, it is important to positive findings and none are negative,
reiterate what others (Mitchell, Bozarth, more than three-quarters show a null asso-
Truax, & Krauft, 1973; Parloff, Waskow, & ciation between specific therapist efforts to
give support and patient outcome” (p. 326). and outcome. Second, the results of the
In addition, the authors identified 94 find- 16 studies analyzed in that chapter were
ings on the association between thera- essentially evenly split between positive and
pist affirmation (essentially warmth, caring, nonsignificant effects. That is, 49% (27/55)
and acceptance) and outcome, with more of all reported associations were signifi-
than half (53%) demonstrating a significantly positive and 51% (28/55) were non-
cant relationship between these sets of significant. However, the authors noted
variables. Underscoring their emphasis on that the majority of nonsignificant find-
considering the perspective of raters, they ings occurred when an objective rater
noted that “the proportion of positive find- (rather than the therapist or patient) evalu-
ings is highest across all outcome catego- ated therapeutic outcome. Third, confirm-
ries when therapist warmth and acceptance ing the pattern noted by previous reviewers,
are observed from the patient’s process Farber and Lane found that when the
perspective” (p. 348). That is, in 30 cases patient rated both the therapist’s positive
where the patient’s ratings of therapist posi- regard and treatment outcome, a positive
tive regard were used, 20 outcome scores association between these variables was
were positively correlated with these ratings especially likely. Lastly, the effect sizes for
(aggregated over the outcome perspectives the significant results tended to be modest,
of patient, therapist, rater, and objective with the larger effect sizes occurring when
score), and no outcome scores (regardless positive regard was assessed in terms of its
of the source) were significantly negatively association to length of stay in therapy
correlated with patient ratings of therapist rather than outcome per se (for example,
positive regard. Najavits & Strupp, 1994).
In 1994 Orlinsky and colleagues studied
this general phenomenon under the rubric Meta-Analytic Review
of therapist affirmation, explained by the Literature Search and Study Selection
authors as a variable that includes aspects To find studies that documented a relation-
of acceptance, nonpossessive warmth, or ship between positive regard and outcome
positive regard. They found that 56% of in psychotherapy, we used the PsycINFO
the 154 results reviewed were positive, and database. Main root terms searched in the
that, again, the findings based on patients’ title or the abstract were positive regard,
process perspective (the patient’s rating warmth, nonpossessive warmth, therapist
of the therapist’s positive regard) yielded affirmation, unconditional positive regard,
even a higher rate of positive therapeutic affirmation, acceptance, and unconditional
outcomes, 65%. “Overall,” Orlinsky et al. regard. All these terms were crossed with
(p. 326) concluded, “nearly 90 findings psychotherapy, searching for the following
indicate that therapist affirmation is a sig- root terms in the title or the abstract: psy-
nificant factor, but considerable variation chotherapy, therapy, counseling, and client-
in ES [effect size] suggests that the contri- centered. Additional studies were located by
bution of this factor to outcome differs running a search with the root term “Barrett-
according to specific conditions.” Lennard” since this is the most widely used
Lastly, in the previous review of positive instrument to assess positive regard.
regard for this volume, the authors (Farber Our specific inclusion criteria were as
& Lane, 2002) highlighted several patterns. follows: (a) the study identified positive
First, no post-1990 study reported a nega- regard as either unconditional regard, posi-
tive relationship between positive regard tive regard, warmth, nonpossessive warmth,
affirmation, or acceptance; (b) positive 1991) was excluded from the revised analy-
regard (in any of these forms) was consis because it could no longer be located.
sidered as a predictor of outcome in the After scanning the literature with these
study; (c) the study reported quantitative criteria in mind, 44 studies were selected
outcome data and relevant statistics (e.g., for review of which 18 were found to be
correlations between positive regard ratings entirely consistent with these criteria and
and treatment outcome or mean outcome thus were included in the meta-analysis.
comparisons between groups with differen-
tial positive regard ratings) that could be Coding Potential Moderators
used to calculate effect sizes; (d) patients were The moderator variables were broken into
adults or adolescents; and (e) treatment was three categories: study characteristics, char-
individual psychotherapy. In addition, stud- acteristics of sample/treatment, and thera-
ies that reported the contribution of posi- pist factors. All studies were coded by the
tive regard to other process or relational junior author; a coding manual is available
variables and thus indirectly to treatment upon request.
outcome were excluded from our analysis. Study characteristics included: (a) publi-
For example, studies that looked at positive cation status (e.g., published article,
regard as a component of other predictor book chapter, or unpublished dissertation);
variables (e.g., empathy, therapeutic alli- (b) sampling (whether the sample was
ance) or were part of an aggregated factor random or a convenience sample); (c) rater
associated with outcome were not included perspective for both the independent and
in the analysis. In fact, many early studies dependent variables; and (d) total number of
looked at positive regard in the context participants. Characteristics of the sample/
of the entire constellation of facilitative treatment that were coded were: mean age,
conditions posed by Rogers—congruence, percentage of women, percentage of racial/
empathy, positive regard—without explic- ethnic minorities, frequency of treatment,
itly reporting the impact of positive regard average number of sessions involved in
as an individual variable. treatment, measure of relationship element,
In addition to searching for relevant point in time that the relationship element
studies in this manner, we consulted was assessed, and the theoretical orienta-
the 2002 chapter to determine which of tion that informed said treatment.
those 16 studies met our current criteria. The specific therapist factors coded for
Six potential articles (Gaston et al., 1990; this analysis involved mean age, percentage
Hynan, 1990; Klein, 2002; Meyer, 1990; of women, number of therapists used, and
Rothman, 2007; Schauble & Pierce, 1974; composition of therapists (e.g., trainees;
Schut, Castonguay, Flanagan, & Yamasaki, four years postgraduation).
2005) were excluded because there was
not enough information presented in the Effect Size Coding
results to compute the appropriate effect Because the purpose of this meta-analysis
size, and any contact with the original was to examine the relation between thera-
authors yielded no assistance. One article pist positive regard and treatment outcome,
(Russell, Bryant, & Estrada, 1996) included a simple correlation, r, was obtained to
in our 2002 review was excluded because measure the effect for each study. The effect
it did not explicitly examine the relation- sizes for several studies had to be recom-
ship between positive regard and therapeu- puted using the data the authors provided
tic outcomes. Another article (Schindler, and then converted to r (per Cooper,
Hedges, & Valentine, 2009). After each of publication, total sample size, and 95%
study was coded for the moderator vari- confidence limits are provided in Table 8.1.
ables, effect sizes were again computed for The aggregate effect size was 0.27, indicat-
each of the 18 studies. Additionally, if there ing that positive regard has a moderate
was more than one effect size per study, association with psychotherapy outcomes;
within-study aggregation was performed. only two of the 18 studies had negative
A new statistical package available online effect sizes. Additionally, the 95% confi-
aided in the statistical analysis for this dence interval (CI) did not include zero
project (see Del Re, 2010; Del Re & Hoyt, (CI = 0.16, 0.38), which demonstrates that
2010). the effect of positive regard on outcome is
significantly different from zero.
Results To assess whether there were differences
A total of 18 effect sizes were yielded (after between these 18 studies above and beyond
aggregation) and were included in the sampling error, a homogeneity test was
analysis. A complete list of the studies conducted. Using the homogeneity statis-
included in the analysis, their authors, date tic, Q (Hedge, 1982), the assumption that
Table 8.1 Effect Sizes between Positive Regard and Outcome

Study n Study effect 95% Confidence limits for r
size (r) [lower, upper]
Bachelor (1991) 47 .49 [.24, 0.68]
Chisholm (1998) 173 −.04 [−.19, 0.11]
Coady (1991) 9 .71 [.09, 0.93]
Conte, Ratto, Clutz, & Karasu (1995) 138 .29 [.13, 0.44]
Cramer & Takens (1992) 37 .37 [.05, 0.62]
Eckert, Abeles, & Graham (1988) 77 .35 [.14, 0.53]
Garfield & Bergin (1971) 38 −.15 [−.45, 0.18]
Green & Herget (1991) 11 .83 [.46, 0.96]
Hayes & Strauss (1998) 32 .31 [−.04, 0.59]
Henry, Schacht, & Strupp (1990) 14 0 [−.53, 0.53]
Keijsers, Hoogduin, & Schaap (1994) 40 .12 [−.20, 0.42]
Litter (2004) 8 .53 [.29, 0.71]
Najavits & Strupp (1994) 12 .75 [.32, 0.93]
Quintana & Meara (1990) 48 .02 [−.26, 0.31]
Rabavilas, Boulougouris, & Perissaki (1979) 36 .09 [−.25, 0.41]

Sells, Davidson, Jewell, Falzer, & Rowe (2006) 83 .33 [.12, 0.51]
Williams & Chambless (1990) 33 .20 [−.15, 0.51]
Zuroff & Blatt (2006) 191 .20 [.06, 0.33]
Overall n or r 1067 .27 [.16, 0.38]
n of studies/samples 18
the studies selected were sampled from the willingness to deviate from the conventions
same population (i.e., were homogenous) of psychodynamic treatment. However,
was rejected, Q (17) = 50.52, p = 0.000. given that all of these moderators were
This indicates that there is a large amount significant, they all contribute in some way
of heterogeneity in the studies due to dif- to the large amount of heterogeneity in the
ferences among the studies. This degree overall effect of positive regard on thera-
of heterogeneity implies that the over- peutic outcomes.
all effect varies as a function of study
characteristics. Continuous Moderator Variables
We also conducted several univariate con-
Moderators tinuous moderator analyses; none were sig-
Categorical Moderator Analyses nificant at the .05 level However, because
In order to account for the systematic vari- of its increasing importance in contempo-
ance present in this sample of studies, we rary psychological research, we note that
conducted several univariate categorical the percentage of racial-ethnic minorities
moderator analyses. As Table 8.2 indicates, as a patient characteristic (R2 = .42, F (1, 7)
the following moderators were significant = 4.37, p = .08) approached significance.
(i.e., demonstrated significant heterogeneity The numbers indicated that as the percent-
in their aggregate effect sizes): publication age of racial/ethnic minorities increases in
status, rater perspective, origin of sample, the patient sample, the overall effect size
measure used to assess positive regard, time also increases. If this finding proves to be
in treatment when positive regard was mea- robust, it has implications for the field of
sured, and type of treatment. These vari- therapist-client matching, as well as multi-
ables moderate the overall effect of positive cultural competence (see Smith, Rodriguez,
regard on therapeutic outcome. In other & Bernal, Chapter 16, this volume).
words, each of these moderators accounts
for some portion of the unaccounted-for Patient Contribution
variance discussed in the previous section Although no patient characteristics emerged
on the overall effect. Most notably, as indi- as significant moderators in our analyses
cated in Table 8.2, the overall effect of posi- of the data, we hypothesize that some
tive regard on outcome tends to be higher patient factors, not assessed in the studies
when studies are published in journal arti- we examined, are likely to affect the thera-
cles, or when the type of treatment pro- pist’s provision of positive regard and the
vided is psychoanalytic/psychodynamic. extent to which this increases the likelihood
In regard to this last finding, our hypothesis of therapeutic success. First, most thera-
is that patients engaged in traditional pists’ behavior is a function, among other
(rather than more contemporary, relational) things, of the characteristics of the patients
psychodynamic treatment were particularly they work with. Simply put, some patients
affected by their therapists’ occasional and are more easily liked and therefore elicit
perhaps unexpected displays of support more affirmation than others. Patients who
and positive regard. In a manner analo- themselves are warm, empathic, and dis-
gous to the power of relatively infrequent closing are more easily liked and affirmed.
therapist disclosures wherein “less is more” Just as disclosure begets disclosure (Jourard,
(Knox & Hill, 2003), these patients’ treat- 1971), it is quite likely that warmth begets
ment outcomes may have been influenced warmth. Conversely, demanding, resistant,
significantly by their therapists’ ability and or angry patients can be difficult to like or
Table 8.2 Significant Moderators
Moderator No. of studies (k) Effect size (r) 95% Confidence limits for r
[lower, upper]
Publication status
Journal article 16 .26 [.20, 0.32]
Unpublished dissertation 2 .09 [−.05, 0.22]
Rater perspective
Patient 7 .29 [.20, 0.37]
Non-participant rater 6 .05 [−.06, 0.17]
More than one perspective 4 .46 [.29, 0.60]
Not specified 1 .20 [.17, 0.28]
Origin of Sample
University setting 4 .07 [−.04, 0.18]
Part of larger study 7 .27 [.17, 0.37]
Hospital setting 1 .29 [.13, 0.44]
General outpatient setting 4 .26 [.11, 0.40]
K–12 setting 1 .53 [.28, 0.70]
Not specified 1 .09 [−.25, 0.41]
Measure Used
VPPS: TWFS 1 .49 [.24, 0.68]
SASB: H&P 3 .23 [.02, 0.42]
TSS 1 .29 [.13, 0.44]
VPPS 1 .53 [.28, 0.70]
Measure created for study 3 .30 [.08, 0.50]
RSTCP 1 .00 [−.53, 0.53]
TRS 1 .20 [.15, 0.51]
BLRI 2 .24 [.12, 0.35]
More than one measure 2 .01 [−.14, 0.15]
Relationship 1 .12 [−.20, 0.42]
Inventory
Other 2 .20 [.01, 0.37]
Time when positive regard was measured
Pre- to post- 3 .16 [−.02, 0.32]
Post-treatment 7 .37 [.27, 0.46]
After second and sixth sessions 1 .83 [.46, 0.96]
One month and three years 1 .31 [−.05, 0.59]
More than one of the above-listed time 3 .01 [−.11, 0.14]
periods
Intake, four, eight, twelve, and sixteen 1 .20 [.06, 0.33]
weeks
Six and twelve months 1 .33 [.12, 0.51]
Follow-up 1.4 years later 1 .09 [−.25, 0.41]
Theoretical orientation/Type of treatment
Psychoanalytic/psychodynamic 4 .52 [.35, 0.65]
Combination/eclectic 7 .12 [.03, 0.20]
Not specified 3 .31 [.20, 0.42]
Peer-based 1 .33 [.12, 0.51]
In-vivo/exposure 3 .14 [−.06, 0.32]
VPPS: TWFS = Vanderbilt Psychotherapy Process Scale: Therapist Warmth and Friendliness; SASB: H&P = Structured Analysis of Social
Behavior: Helping & Protecting; TSS = Therapist Satisfaction Scale; VPPS = Vanderbilt Psychotherapy Process Scale; RSTCP = The Rating
Scale of Therapy Change Processes; TRS = Therapist Rating Scale; BLRI = Barrett-Lennard Relationship Inventory.
181 fa r b e r, d o o l i n 181
affirm (see Winnicott, 1949). Thus, we data indicate a greater overall treatment
suspect that those with Axis II pathologies, effect of positive regard when more than
especially individuals with borderline or one perspective (e.g., patient, therapist, and
narcissistic disorders, are less likely to con- nonparticipant rater) was assessed. These
sistently evoke positive regard from their multiple perspectives may serve as reliabil-
therapists. Many difficult patients are test- ity checks on the accuracy, including the
ing their sense of the world (e.g., their potential underevaluation, of the patient’s
lovableness), simultaneously desperate to perspective on the provision of therapist
have their worst fears unconfirmed but positive regard.
overdetermining through their behavior Our data indicated that the effects of
that they will be reconfirmed (Weiss & positive regard increased as a function of
Sampson, 1986). the racial-ethnic composition in a study,
A related client characteristic that may although this association was statistically
influence a therapist’s tendency to be posi- weak. Thus, we tentatively hypothesize that
tively regarding is the nature of the client’s therapists’ provision of positive regard may
needs at a particular point in therapy. For be a salient factor in treatment outcome
example, patients suffering acutely from when non-minority therapists work with
any of the many variants of depression, minority clients. In such cases, the possibil-
or dealing with the aftermath of a recently ity of mistrust and of related difficulties—
experienced trauma, may explicitly ask stemming in large part from our nation’s
for or more subtly indicate their need for troubled racial history as well as traditional
intensive doses of positive regard and affir- neglect of minority clients by the mental
mation. These requests may range from health community--may be attenuated by
“please tell me I’m OK,” to “I really need clear indications of the therapist’s positive
your support now,” to “no one cares about regard, in turn facilitating the likelihood of
me at all.” A third possible factor here a positive outcome (Sue & Sue, 1999).
is motivational status—that is, a patient’s
current stage of change (see Norcross, Krebs, Limitations of the Research
& Prochaska, Chapter 14, this volume). Our database was restricted to 18 studies, a
Patients who are more highly motivated to relatively small basis for conclusions about
do the work, who appear to be courageous a variable that has been part of psychother-
or risk taking, are more likely to evoke their apeutic lore for more than 50 years. In part,
therapist’s positive regard. this reflects the stringent criteria we used in
A consideration of how various patient deciding which studies were to be entered
characteristics may contribute to the expres- in the meta-analysis; in part, it represents
sion of positive regard in the therapeutic the fact that positive regard has been stud-
relationship also illuminates the relevance ied primarily within the realm of client-
of the perspective from which positive centered therapy, an orientation that no
regard is rated. Although Rogers (1957) longer attracts the attention of many prom-
believed that it is only the client’s perspec- inent researchers. In this respect, there have
tive that matters––that it is the client’s been very few studies of positive regard
experience of positive regard (or genuine- within the past 20 years. We believe that
ness or empathy) that “counts” and that the concept of positive regard hasn’t so
the therapist’s belief as to whether he or she much gone away in recent years as it has
has been positively regarding is essentially been folded into newer concepts in the field,
moot in regard to positive outcome––our particularly measures of the therapeutic
alliance (see Chapters 2, 3, and 4 in this types of problems at which point in
volume). therapy, is the provision of therapist regard
The restricted range of theoretical orien- most important?
tations in which positive regard has been
studied leads to another limitation: the Therapeutic Practice
possibility that the action of this variable The psychotherapist’s ability to provide
is restricted to a specific form of therapy positive regard is significantly associated
(person-centered) or interacts with a spe- with therapeutic success. However, our meta-
cific aspect of this therapy. Thus, Orlinsky analysis indicates a moderate relationship,
and Howard (1978) raised the possibility suggestive of the fact that, like many other
that empathy and warmth “interact differ- relational factors, it is a significant but not
entially with other aspects of therapist style” exhaustive part of the process–outcome
(p. 293). That is, they suggested that these equation. Extrapolating from the data, we
qualities might be significantly associated offer the following recommendations for
with outcome only when therapist direc- clinical practice:
tiveness is low––as is usually true of person-
centered or psychodynamic therapists––and • Therapists’ provision of positive
may not be the case when therapists prac- regard is strongly indicated in practice.
tice from a more heterogeneous or directive At a minimum, it “sets the stage” for
perspective that might reflect a CBT orien- other mutative interventions and that,
tation. In fact, our data allow us to know at least in some cases, it may be sufficient
only partially the answer to this question. by itself to effect positive change.
Whereas we found a significant moderat- • There is virtually no research-driven
ing effect for psychodynamic treatment–– reason to withhold positive regard. We are
patients in this form of therapy tended reminded of the oft-heard sentiment in
to improve more than others as a func- contemporary psychoanalytic circles that
tion of receiving positive regard from their one of Kohut’s major contributions was to
therapists––our database included no stud- provide a theoretical justification for being
ies of patients in any CBT-related treatment. kind to one’s patients.
It is nevertheless noteworthy that meta- • Positive regard serves many valuable
analytic results on the association between functions across the major forms of
therapist empathy and treatment outcome psychotherapy. From a psychodynamic
reveal no differences in effect sizes for perspective, positive regard serves to
different forms of therapy (Elliott, Bohart, strengthen the client’s ego (sense of self
Watson, & Greenberg, Chapter 6, this or agency) and belief in his or her capacity
volume). Nor were there any differential to be engaged in an effective relationship;
effect sizes as a function of the type of treat- from a behavioral perspective, the
ment in the relation between treatment therapist’s positive regard functions as a
outcome and the therapeutic alliance positive reinforcer for clients’ engagement
in individual therapy for adults (Horvath, in the therapeutic process, including
Del Re, & Flückiger, Symonds, Chapter 2, difficult self-disclosures; and from a more
this volume) or youth (Shirk & Karver, purely humanistic perspective, the
Chapter 3, this volume). therapist’s stance of caring and positive
More generally, the extant research has regard facilitates the client’s natural
not addressed the question of specificity: tendency to grow and fulfill his or her
For which patients, presenting with which potential.
• Positive regard may be especially alliance as seen by client and therapist.
indicated in situations wherein a non- Psychotherapy, 28, 534–49.
Barrett-Lennard, G. T. (1964). The Relationship
minority therapist is working with a
Inventory. Form OS-M-64 and OS-F-64 Form
minority client. MO-M-64 and MO-F-64. Armidale, New South
• Therapists cannot be content with Wales, Australia: University of New England.
feeling good about their patients but Barrett-Lennard, G. T. (1978). The Relationship
instead should ensure that their positive Inventory: Later development and applications.
feelings are communicated to them. This JSAS: Catalog of selected documents in psychology,
8, 68 (Ms. No. 1732, p. 55).
does not have to translate to a stream of
Barrett-Lennard, G. (1986). The relationship
compliments or to a gushing of positive inventory now: Issues and advances in theory,
sentiment that, in fact, may overwhelm or method and use. In L. S. Greenberg &
even terrify some clients; rather, it speaks W. M. Pinsof (Eds.), The Psychotherapeutic process:
to the need for therapists to communicate A research handbook (pp. 439–76). New York:
a caring, respectful, positive attitude that Guilford.
Beck, J. S. (1995). Cognitive therapy: Basics and
serves to affirm a client’s basic sense of
beyond. New York: Guilford.
worth. To many, if not most, clients, the Benjamin, L. (1984). Principles of prediction using
conviction that “my therapist really cares Social Analysis of Structural Behavior (SASB).
about me” likely serves a critical function, In R. A. Zucker, J. Aronoff, and A. J. Rabin
especially in times of stress. (Eds.), Personality and the prediction of behavior
• Therapists need to monitor their (pp. 121–73). New York: Academic Press.
Bergin, A. E., & Garfield, S. L. (Eds.) (1971).
positive regard and adjust it as a
Handbook of psychotherapy and behavior change:
function of the needs of particular An empirical analysis. New York: Wiley.
patients and specific clinical situations. Beutler, L. E., Machado, P. P. P., & Neufeldt, S. A.
The research demonstrates that therapists (1994). Therapist variables. In S. L. Garfield &
vary in the extent to which they are A. E. Bergin (Eds.), Handbook of psychother-
able to convey positive regard to their apy and behavior change (4th ed., pp. 229–69).
New York: Wiley.
patients, and clients vary in the extent
Chisholm, S. M. (1998). A comparison of the
to which they need, elicit, and/or benefit therapeutic alliances of premature terminators
from a therapist’s positive regard. In versus therapy completers. Unpublished doctoral
this regard, too, we suspect that the dissertation, Kent State University, Kent, OH.
inevitable ruptures in the therapeutic Coady, N. F. (1991). The association between client
alliance (Safran, Muran, & Eubanks- and therapist interpersonal processes and out-
comes in psychodynamic psychotherapy. Research
Carter, Chapter 11, this volume) that
on Social Work Practice, 1, 122–38.
occur over the course of therapy are Conte, H. R., Ratto, R., Clutz, K., & Karasu, T. B.
the result not only of a therapist’s (1995). Determinants of outpatients’ satisfaction
technical errors but also of the therapist’s with therapists. Journal of Psychotherapy Practice
occasional inability to demonstrate and Research, 4, 43–51.
minimally facilitative levels of positive Cooper, H., Hedges, L. V., & Valentine, J. C. (Eds.).
(2009). The handbook of research synthesis and
regard and support.
meta-analysis (2nd ed.). New York: Russell Sage
Foundation.
Cramer, D., & Takens, R. J. (1992). Therapeutic
References
relationship and progress in the first six sessions
An asterisk (∗) indicates studies included of individual psychotherapy: A panel analysis.
in the meta-analysis. Counseling Psychology Quarterly, 5, 25–36.
Bachelor, A. (1991). Comparison and relationship Del Re, A. C. (2010). RcmdrPlugin.MAc: Meta-
to outcome of diverse dimensions of the helping Analysis with correlations (MAc) Rcmdr Plug-in.
R package version 1.0.7. http://CRAN.R-project. Jourard, S. M. (1971). The transparent self. New York:
org/package=RcmdrPlugin.MAc Van Nostrand.
Del Re, A. C., & Hoyt, W. T. (2010). MAc: Meta- Keijsers, G. P., Hoogduin, C. A., & Schaap, C. P.
Analysis with correlations. R package version (1994). Predictors of treatment outcome in the
1.0.6. http://Cran.R-project.org/package=Mac behavioral treatment of obsessive-compulsive
Eckert, P. A., Abeles, N., & Graham, R. N. (1988). disorder. The British Journal of Psychiatry, 165,
Symptom severity, psychotherapeutic process, 781–86.
and outcome. Professional Psychology: Research Knox, S., & Hill, C. E. (2003). Therapist self-
and Practice, 19, 560–64. disclosure: Research-based suggestions for prac-
Farber, B. A. (2007). On the enduring and substan- titioners. Journal of Clinical Psychology/In Session,
tial influence of Carl Rogers’ not-quite essential 59, 529–40.
nor necessary conditions. Psychotherapy: Theory, Leahy, R. L. (2001). Overcoming resistance in cogni-
Research, Practice, Training, 44, 289–294. tive therapy. New York: Guilford.
Farber, B. A., Brink, D. C., & Raskin, P. M. (1996). Lietaer, G. (1984). Unconditional positive regard:
The psychotherapy of Carl Rogers: Cases and com- A controversial basic attitude in client-centered
mentary. New York: Guilford. therapy. In R. F. Levant & J. M. Shlien (Eds.),
Garfield, S. L., & Bergin, A. E. (1971). Therapeutic Client-centered therapy and the person-centered
conditions and outcome. Journal of Abnormal approach: New directions in theory, research, and
Psychology, 77, 108–114. practice (pp. 41–58). New York: Praeger.
Gaston, L., Marmar, C. R., Gallagher, D., & Litter, M. (2004). Relationship-based psycho-
Thompson, L. W. (1991). Alliance prediction of therapy with court-involved youth: The therapy
outcome beyond in-treatment symptomatic relationship’s effect on outcome. Dissertation
change as psychotherapy processes. Psychotherapy Abstracts International, 65 (12), 4474. (AAT no.
Research, 1, 104–113. 3156920).
Green, R. J., & Herget, M. (1991). Outcomes of McMain, S., Korman, L. M., & Dimeff, L. (2001).
systemic/strategic team consultation: The impor- Dialectical behavior therapy and the treatment
tance of therapist warmth and active structuring. of emotion dysregulation. Journal of Clinical
Family Process, 30, 321–36. Psychology, 57(2), 183–96.
Gurman, A. (1977). The patient’s perception of Meyer, A. E. (1990). Nonspecific and common
the therapeutic relationship. In A. S. Gurman factors in treatment outcome: Another myth? Paper
& A. M. Razin (Eds.), Effective psychotherapy presented at the Annual Meeting, Society for
(pp. 503–43). New York: Pergamon. Psychotherapy Research, Wintergreen, VA.
Hayes, A. M., & Strauss, J. L. (1998). Dynamic sys- Mitchell, S. A., & Aron, L. (1999). Relational psy-
tems theory as a paradigm for the study of change choanalysis. New York: Analytic Press.
in psychotherapy: An application to cognitive Mitchell, K., Bozarth, J., & Krauft, C. (1977).
therapy for depression. Journal of Consulting and A reappraisal of the therapeutic effectiveness of
Clinical Psychology, 66, 939–47. accurate empathy, non-possessive warmth and
Hedges, L. V. (1982). Estimating effect sizes from a genuineness. In A. S. Gurman & A. M. Razin
series of independent experiments. Psychological (Eds.), Effective psychotherapy (pp. 482–502).
Bulletin, 92, 490–99. New York: Pergamon.
Henry, W. P., Schacht, T. E., & Strupp, H. H. Mitchell, K., Bozarth, J., Truax, C., & Krauft, C.
(1990). Patient and therapist introject, interper- (1973). Antecedents to psychotherapeutic outcome.
sonal process, and differential psychotherapy Arkansas Rehabilitation Research and Training
outcome. Journal of Consulting and Clinical Center, University of Arkansas (NIMH Final
Psychology, 58, 768–74. Report, MH 12306).
Horvath, A. O., & Greenberg, A. (1989). Devel- Najavits, L. M., & Strupp, H. H. (1994). Differences
opment and validation of the Working Alliance in the effectiveness of psychodynamic therapists: A
Inventory. Journal of Counseling Psychology, 36, process-outcome study. Psychotherapy, 31, 114–23.
223–33. Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994).
Hynan, D. J. (1990). Client reasons and experi- Process and outcome in psychotherapy—noch
ences in treatment that influence termination of einmal. In A. E. Bergin & S. L. Garfield (Eds.),
psychotherapy. Journal of Clinical Psychology, 46, Handbook of psychotherapy and behavior change
891–95. (4th ed., pp. 27–376). New York: Wiley.
Orlinsky, D. E., & Howard, K. (1978). The rela- prozessforschung in der verhaltenstherapie [The
tion of process to outcome in psychotherapy. empirical analysis of the therapeutic relation:
In S. L. Garfield & A. E. Bergin (Eds.), Handbook Contribution to process research in behavior
of psychotherapy and behavior change (2nd ed., therapy]. Berlin, Heidelberg, Germany: Springer-
pp. 283–329). New York: Wiley. Verlag.
Orlinsky, D. E., & Howard, K. (1986). Process and Sells, D., Davidson, L., Jewell, C., Falzer, P., &
outcome in psychotherapy. In S. L. Garfield, & Rowe, M. (2006). The treatment relationship in
A. E. Bergin (Eds.), Handbook of Psychological peer-based and regular case management for
Behavior and Change (pp. 311–81) New York: clients with severe mental illness. Psychiatric
Wiley. services, 57, 1179–84.
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Sue, D. W., & Sue, D. (1999). Counseling the cultur-
Research on therapist variables in relation to ally different: Theory and practice (3rd edition).
process and outcome. In S. L. Garfield & New York: Wiley.
A. E. Bergin (Eds.), Handbook of psychotherapy Suh, C. S., Strupp, H. H., & O’Malley, S. S.
and behavior change (2nd ed., pp. 233–82). (1986). The Vanderbilt Process Measures:
New York: Wiley. The Psychotherapy Process Scale (VPPS) and
Quintana, S. M., & Meara, N. M. (1990). the Negative Indicators Scale (VNIS). In
Internalization of therapeutic relationships in L. S. Greenberg & W. M. Pinsof (Eds.), The
short-term psychotherapy. Journal of Counseling psychotherapeutic process: A research handbook
Psychology, 2, 123–30. (pp. 285–323). New York: Guilford.
Rabavilas, A. D., Boulougouris, J. C., & Tracey, T. J., & Kokotovic, A. M. (1989). Factor
Perissaki, C. (1979). Therapist qualities related structure of the working alliance inventory.
to outcome with exposure in vivo in neu- Psychological Assessment, 1, 207–210.
rotic patients. Journal of Behavior Therapy and Truax, C. B., & Carkhuff, R. R. (1967). Toward
Experimental Psychiatry, 10, 293–94. effective counseling and psychotherapy: Training
Rogers, C. R. (1942). Counseling and psychotherapy. and practice. Chicago: Aldine.
Boston: Houghton Mifflin. Wachtel, P. L. (2008). Relational theory and the prac-
Rogers, C. R. (1951). Client-centered therapy. tice of psychotherapy. New York: Guilford.
Boston: Houghton Mifflin. Wampold, B. (2001). The great psychotherapy debate.
Rogers, C. R. (1957). The necessary and sufficient Mahwah, NJ: Erlbaum.
conditions of therapeutic personality change. Weiss, J., Sampson, H., and the Mt. Zion
Journal of Consulting Psychology, 21, 95–103. Psychotherapy Research Group. (1986). The
Rogers, C. R. (1986). A client-centered/person- psychoanalytic process: Theory, clinical observa-
centered approach to therapy. In I. Kutash & tions and empirical research. New York: Guilford
A. Wolf (Eds.), Psychotherapist’s casebook (pp. Press.
197–208). San Francisco: Jossey-Bass. Williams, K. E., & Chambless, D. L. (1990). The
Rogers, C. R., & Truax, C. B. (1967). The thera- relationship between therapist characteristics
peutic conditions antecedent to change: A theo- and outcome of in vivo exposure treatment for
retical view. In C. R. Rogers, E. T. Gendlin, agoraphobia. Behavior Therapy, 21, 111–116.
D. J. Kiesler, & C. B. Truax (Eds.), The therapeu- Winnicott, D. W. (1949). Hate in the counter-
tic relationship and its impact: A study of psycho- transference. International Journal of Psycho-
therapy with schizophrenics. Madison: University Analysis, 30, 69–74.
of Wisconsin Press. Yalom, I. D. (2002). The Gift of therapy: An Open
Russell, R. L., Bryant, F. B., & Estrada, A. U. letter to a new generation of therapists and their
(1996). Confirmatory P-technique analysis of patients. New York: HarperCollins.
therapist discourse: High- versus low-quality Zuroff, D., & Blatt, S. J. (2006). The therapeutic
child therapy sessions. Journal of Consulting and relationship in the brief treatment of depres-
Clinical Psychology, 64, 1366–76. sion: Contributions to clinical improvement
Schindler, L. (1991). Die empirische analyse and enhanced adaptive capacities. Journal of
der therapeutischen Beziehung. Beitrage zur Consulting and Clinical psychology, 74, 130–40.
C HA P TER
9 Congruence/Genuineness
Gregory G. Kolden, Marjorie H. Klein, Chia-Chiang Wang, and Sara B. Austin
Congruence or genuineness is a relational accurately represented by his awareness of

quality that has been highly prized through- himself ” (Rogers, 1957, p. 97). These days
out the history of psychotherapy but of we might say that the therapist is mindfully
diminished research interest in recent years. genuine in the therapy relationship, under-
In this chapter, we offer definitions and scoring present personal awareness as well
examples of this attribute of the therapy as genuineness or authenticity.
relationship as well as an original meta- The second facet of congruence refers
analytic review of the empirical litera- to the therapist’s capacity to communicate
ture showing its relation to improvement. his or her experience with the client to
Moderators of the association between con- the client. This requires careful reflection
gruence and treatment outcome are exam- and considered judgment on the part of
ined, and limitations of the extant research the therapist. While the aim is not for the
are discussed as well. In closing, we advance therapist to indulge in indiscriminate self-
several therapist practices that are likely to disclosure or ventilation of feelings, the
foster congruence and thus improve psy- therapist must not deceive the client about
chotherapy outcomes. his or her feelings, especially if they stand
in the way of progress. Neither empathy nor
Definitions and Measures regard can be conveyed unless the therapist
Definitions is perceived as genuine. As such, congru-
In 1957 Carl Rogers characterized the nec- ence occupies a central position in Rogers’
essary and sufficient conditions of thera- conceptualization.
peutic change as the client being in a “state The concept of congruence can at times
of incongruence,” the client and therapist seem abstract and elusive. Consider how
in “psychological contact,” and the thera- this relational quality might appear in
pist as “congruent or integrated in the rela- everyday interactions with people in your
tionship” and experiencing “positive regard world. Insurance agent Jones is quite formal
for the client” and “an empathic under- and proper while appearing to be playing
standing of the client’s internal frame of a prescribed role. Mr./Ms Jones interacts in
reference” (p. 96). This characterization a relationally incongruent manner. Coffee
underscores two facets of congruence. The barista Brian warmly greets you by your
first refers to the therapist’s personal inte- first name, attentively asks after your family,
gration in the relationship, that “he is freely and openly shares his opinion about a
and deeply himself, with his experience movie he recently took in. Brian engages
187
you, makes contact, and sincerely expresses “the ability to and willingness to be what
himself in the brief time it takes to pour one truly is in the relationship . . .” (Gelso &
and pay for a cup of coffee. Brian interacts Carter, 1994, p. 297). Genuineness is
in a relationally congruent fashion. also related to other terms, such as authen-
In psychotherapy, this means that the ticity, openness, honesty, and nonphoni-
therapist is openly “being the feelings and ness (Gelso & Hayes, 1998). Realistic
attitudes which at the moment are flowing experiences of the therapeutic relation-
within him” (Rogers et al., 1967, p. 100) ship pertain to perceptions that are not
and not hiding behind a professional role or distorted by transference alterations and
holding back feelings that are obvious in the defense mechanisms.
encounter. Congruence thus involves mind- In the current literature, genuineness
ful self-awareness and self-acceptance on the is frequently considered the most impor-
part of the therapist, as well as a willingness tant of the three Rogerian facilitative
to engage and tactfully share perceptions. conditions. Moreover, Lietaer (1993) has
To quote Rogers, the congruent therapist offered a conceptualization of genuineness
“comes into a direct personal encounter with with both an internal and external facet.
his client by meeting him on a person-to- The internal facet “. . . refers to thera-
person basis. It means that he is being him- pists’ own internal experiencing with their
self, not denying himself ” (p. 101). clients . . . To the extent that therapists
One reason why congruence plays so are able to be in touch with their own
central a role in Rogers’ thinking is that he experience they may be termed congruent”
defines the problems that clients bring to (Watson, Greenberg, & Lietaer, 1998,
therapy in terms of their incongruence p. 9). The external facet “. . . refers to the
and sees the therapy process as one of help- therapists’ ability to reveal their experience
ing the client to become more congruent; to their clients. This is termed transpar-
that is, to develop the capacity to own and ency . . . it is not necessary to share every
express thoughts and feelings without fear. aspect of their experience but only those
Thus, therapist congruence can model for that they feel would be facilitative of
the client a “realness” and can facilitate their clients’ work. Transparency is always
the client becoming more open to his/her used in an empathic climate” (Watson,
own experiencing; this makes the therapist– Greenberg, & Lietaer, 1998, p. 9).
client relationship deeper and the psycho- There has also been some broadening of
logical contact more immediate. the definitions of congruence to include
Although most fully developed with the therapeutic presence (Geller & Greenberg,
client-centered tradition, therapist congru- 2002). In an interview, Carl Rogers said,
ence is highly prized in many theoretical “Over time, I think I have become more
orientations. The notion of the therapist real aware of the fact that in therapy I do use
relationship (Gelso & Carter, 1985; Gelso & myself. I recognized that when I am
Hayes, 1998), for example, is conceptually intensely focused on a client, just my
similar to congruence/genuineness and is presence seems to be healing” (Baldwin,
consistent with ideas initially offered by psy- 1987, p. 45). In addition to the three basic
choanalysts (e.g., Greenson, 1967). The real conditions “perhaps it is something around
relationship is seen as primarilyundistorted the edges of those conditions that is really the
by transferential material and comprised of most important element of therapy–when
two defining features: genuineness and real- myself is very clearly, obviously present”
istic perceptions. Genuineness is viewed as (p. 30). In sum, presence implies a dual
level of mindful awareness whereby the while in session (e.g., “ I am willing to tell
therapist balances contact with his or her him my own thoughts and feelings”) or the
own experience and contact with the cli- patient to describe his or her experience of
ent’s experience to maintaining a “place of the therapist (e.g., “He is willing to tell me
internal and external connection” (Geller his own thoughts and feelings”).
& Greenberg, 2002, p. 83). The original 92-item version of the BLRI
Authenticity and transparency are also included five scales: level of regard, empathic
components of congruence that involve the understanding, unconditionality, genuine-
therapist’s awareness of his or her internal ness, and willingness to be known. This
experience and willingness to communicate last scale was merged into the genuineness/
this awareness to the client (Greenberg & congruence scale in the 64-item 1964 revi-
Watson, 2005). “The communicative aspects sion (Barrett-Lennard, 1978). Likert scal-
of congruence involve the ability to trans- ing ranging from −3 (I strongly feel that it
late intrapersonal experience into certain is not true) to +3 (I strongly feel that it
types of interpersonal responses” (p. 127). is true) is used to rate each question (see
A congruent interactional response involves Table 9.1). A shortened, 30-item version of
the conveyance of “attitudes or intentions the BLRI was developed later (Gurman,
of being helpful, understanding, valuing, 1973a, 1973b).
respecting and being nonintrusive or non- Truax also developed a version of the
dominant” (p. 129). Thus, congruence is BLRI entitled the Truax Relationship
more than avoiding formality on the one Questionnaire (TRQ; see Truax & Carkuff,
hand or phoniness on the other; it entails 1967) as a self-report assessment of the core
the therapist’s attentive recognition and conditions. While Truax (1968) stated that
nonjudgmental acceptance of feelings, per- client self-report was less valid for assess-
ceptions, and thoughts, both positive and ing the core conditions and impractical
negative. with certain populations (such as psychotic
patients), he argued that the economical
Measures and wide-ranging uses for the instrument
The first measures of Rogers’ facilitative justified its development.
conditions were developed at the University As Barrett-Lennard was developing and
of Chicago Counseling Center. Halkides revising the BLRI, Rogers’ group at the
(1958) designed separate scales for each University of Wisconsin was engaged in
condition for use by independent raters. extensive development of scales for raters
This groundbreaking work was followed to assess the conditions from audiotape
by numerous studies examining the rela- recordings of sessions. Early versions (Hart,
tion between judge’s ratings of the core 1960; Gendlin & Geist, 1962) were fol-
conditions and patient outcomes (Barrett- lowed by Truax’s development of the 1962
Lennard, 1998). Self-Congruence Scale (Rogers et al., 1967)
Barrett-Lennard (1959) developed what for use in the Wisconsin Schizophrenia
has become the most recognized and Project. Independent observers rate how
validated therapist- or patient self-report the therapist “appears” in tape-recorded ses-
assessment of the core conditions: the sion samples; Table 9.1 lists the descrip-
Barrett-Lennard Relationship Inventory tors for the five stages of the Truax (1966a)
(BLRI; see Barrett-Lennard, 1962). Parallel Self-Congruence Scale. Further revisions
forms of the BLRI ask the therapist to of the measure were made for the final rat-
describe his or her feelings toward the client ings of the Wisconsin Schizophrenia Project
ko l d e n , k l e i n , wa n g , au s t i n 189
(Rogers et al., 1967) because of difficul- BLRI and the Truax scale—has gener-
ties obtaining acceptable reliability with the ally been adequate. Most internal and test–
Truax version. This modification consisted retest reliability coefficients for the BLRI
of a five-point scale ranging from “a point range between 0.75 and 0.95 with the
where there is obvious discrepancy between majority exceeding 0.80 (Barrett-Lennard,
the therapist’s feelings about the patient 1998). An extensive review reported inter-
and his concurrent communication to the nal reliability coefficients for congruence
patient (stage 1) to a high point where the ranging from 0.76 to 0.92 with a mean
therapist communicates both his positive coefficient of 0.89 (Gurman, 1977). On
and negative feelings about the patient the Truax scale, reliability coefficients for
openly and freely, without traces of defen- congruence/genuineness ranged from 0.34
siveness or retreat into professionalism to 0.85 with most over 0.65 (Mitchell,
(stage 5)” (Kiesler, 1973, p. 229; Rogers Bozrath, & Krauft, 1977). Internal reli-
et al., 1967, pp 581–583). In contrast to abilities for the Real Relationship Inventory
the Truax version, ratings of the Kiesler in the professional and graduate student
scale only applied to session segments samples were 0.79 for Realism and 0.83
in which the client either “explicitly or for Genuineness (Gelso et al., 2005).
implicitly questioned the therapist’s feel- Confirmatory factor analysis compared
ings or opinions about him” (Rogers et al., 1-factor and 2-factor solutions; results were
p. 140). slightly more supportive of a single factor,
Carkhuff (1969) also developed a scale but it was decided to retain the two scales
of genuineness derived from the Truax scale for further study because “the theory from
for broad application to interpersonal inter- which the measures emanated is embedded
actions beyond those occurring between a in this dual notion” (p. 647).
therapist and client. Aside from its broader
application, the Carkhuff scale differs Clinical Examples
from Truax’s version in that it includes Therapist Perspectives
more of an emphasis on negative reactions The following excerpts are examples of
resulting from moderate to low levels of Rogers’ description of how his work with
genuineness. individuals who are schizophrenic led him to
More recently, patient and therapist ver- refine the experiential component of client-
sions of a Real Relationship Inventory have centered therapy (Rogers et al., 1967). In
been published (Gelso, et al., 2005). Items the first example, Rogers explains how he
for the Real Relationship Inventory were may use his feelings about the difficulty
solicited from professional therapists and ending a session to provide the “vehicle for
graduate students. A pool of 130 items was therapeutic responding” (p. 389):
reduced to 44 items, which was adminis-
tered to randomly selected members of Some of my feelings about him (the
APA Divisions 29 (Psychotherapy) and 42 patient) in the situation are a good source
(Independent Practice) with instructions of responses, if I tell them in a personal,
to rate if they had used that item in their detailed way. . . . . One whole set of
last session with a client. This resulted in a feelings I have for others in situations
final scale of 24 items, 12 for “Realism” and comes at first as discomfort. As I look to
12 for “Genuineness.” see why I am uncomfortable I find content
The reliability of the two most frequently relevant to the person I am with, to what
used measures of the core conditions—the we just did or said. Often it is quite
personal. I was stupid, rude, hurrying, wrong (as I could tell from the patient’s
embarrassed, avoidant, on the spot: reaction), I would not know what to do.
I wished I didn’t have to go since he Now I spend moments letting my feelings
wants me to stay. I wish I hadn’t hurried clarify themselves, but once they feel clear,
him out of the store in front of all those I no longer wonder so much whether it is
people. Or, “I guess you’re mad at me right or wrong to express them. Rather,
because I’m leaving. I don’t feel very good I have open curiosity, sensitivity, and a
about it either. It just never feels right readiness to meet whatever reaction I will
to me to go away and leave you in here get. This may tell me what I said was
[hospital ward]. I have to go, or else I’ll “wrong,” but all will be well if now
be late for everything I have to do all I respond sensitively to what I have stirred.
day today, and I’ll feel lousy about that.” I now say whatever I now sense which
Silence. “In a way, I’m glad you don’t makes what I said before “wrong.” (It is not
want me to go. I wouldn’t like it at all my admission that I was wrong which
if you didn’t care one way or the other.” matters here. I rarely make a point of
(p. 390). having been wrong. That matters only to
me. I am the only one who cares how
In reflecting on these moments, Rogers often I am right or wrong. But whatever
explains that: it is in him which I now sense and which
makes what I said wrong - I now see it
These . . . have in common that I express
in his further reaction - that is what
feelings of mine which are at first
I have to respond to in the next moment.)
troublesome or difficult, the sort I would
(p. 391).
at first tend to ignore in myself. It requires
a kind of doubling back. When I first notice
A final example illustrates the key role of
it, I have already ignored, avoided, or
the therapist’s self-experiencing in building
belied my feelings - only now do I notice
mutual congruence:
what it was or is. I must double back to
express it. At first, this seems a sheer
We tend to express the outer edges of our
impossibility! How can I express this
feelings. That leaves us protected and
all-tied-up, troublesome, puzzling feeling?
makes the other person unsafe. We say,
Never! But a moment later I see that it
“This and this (which you did) hurt me.”
is only another perfectly human way to
We do not say, “This and this weakness
feel, and in fact includes much concern
of mine made me be hurt when you did
for the patient, and empathic sensitivity
this and this.”
to him. It is him I feel unhappy about - or
To find this inward edge of my feelings,
what I just did to him. (p. 390).
I need only ask myself, “Why?” When
Another example shows how Rogers I find myself bored, angry, tense, hurt,
uses an “openness to what comes next” at a loss, or worried, I ask myself, “Why?”
to increase his sensitivity, even to repair a Then, instead of “You bore me,” or “This
breach in the interaction. He notes that by makes me mad,” I find the “why” in me
being open to what comes next, a positive which makes it so. That is always more
feeling will usually emerge: personal and positive, and much safer to
express. Instead of “You bore me,” I find,
I used to ponder whether I was about to “I want to hear more personally from you,”
say a right or wrong thing. Then, if it was or, “You tell me what happened, but I want
to hear also what it all meant to you.” and then I’d do what they wanted to
(pp. 390–391). do. And I’d think, “Aw, hell!” It
just—never works out, you know.
Patient Perspectives C: Always somebody there, isn’t there?
How is congruence offered by the therapist (Laughs)
perceived by the patient? One way to cap- T: Yeah, just somebody goofing up
ture this is to review the items that a patient the works all the time. (Pause)
might endorse in the BLRI, as shown in Yeah, if you’re dependent on
Table 9.1. The patient’s experience of the somebody else, you’re under their
highly self-congruent therapist is that the control, sort of.
therapist is fully at ease within the relation- C: To a certain extent . . .
ship and is openly him or herself. Being T: Yeah, that’s what I was going to
attuned to his or her experience in the say—yeah, you’re right. (Pause).
moment, the therapist is open to honestly So you just sit around the ward
sharing this experience with the patient and and you read a little bit, and then
does not avoid sharing uncomfortable feel- you go out and play horseshoes
ings and impressions that are important to and—boy, that sounds like a drag!
treatment. Because of this personal attune- (p. 72).
ment and genuineness, the therapist’s words
The next example comes from the
accurately capture his or her momentary
transcripts of Carl Rogers’s filmed demon-
experience.
stration session with the client “Gloria”
(Shostrum, 1966) where he clearly expresses
Observer Perspectives his feeling of closeness to Gloria:
A third perspective on genuineness is pro-
vided by samples from session transcripts. Gloria: That is why I like substitutes.
The following example of Stage 5 high- Like I like talking to you and I like
congruence comes from the training mate- men that I can respect. Doctors,
rial for the Truax (1966a) scale: and I keep sort of underneath feeling
like we are real close, you know, sort
C: I guess you realize that, too, don’t of like a substitute father.
you? Or do you? (Laughs) Rogers: I don’t feel that is pretending.
T: Do I realize that? You bet I do! Sure Gloria: Well, you are not really my
yeah—I always wanted somebody father.
to take care of me, you know, but Rogers: No. I meant about the real close
I also wanted them to let me do business.
what I wanted to do! Well, if you Gloria: Well, see, I sort of feel that’s
have somebody taking care of you, pretending too because I can’t expect
then you’ve got to do what they you to feel very close to me. You
want you to do. don’t know me that well.
C: That’s right. (Pause) Rogers: All I can know is what I am
T: So, I never could kind of get it so feeling and that is I feel close to you
that I’d have both, you know, both in this moment.
things at once: either I’m doing
what I want to do and taking care Three aspects of congruence are its com-
of myself or, you know, I used to munication, transparency (in a disciplined
have somebody taking care of me manner), and affirmation of the client’s
Table 9.1 Rating Scales for Congruence
Congruence items on the BLRI a
Positively valenced items
He is comfortable and at ease in our relationship.
I feel that he is real and genuine with me.
I nearly always feel that what he says expresses exactly what he is feeling and thinking as he says it.
He does not avoid anything that is important for our relationship.
He expresses his true impressions and feelings with me.
He is willing to express whatever is actually in his mind with me, including any feelings about himself or about me.
He is openly himself in our relationship.
I have not felt he tries to hide anything from himself that he feels with me.
Negatively valenced items

I feel that he puts on a role or front with me.
It makes him uneasy when I ask or talk about certain things.
He wants me to think that he likes me or understands me more than he really does.
Sometimes he is not at all comfortable but we go on, outwardly ignoring it.
At times I sense that he is not aware of what he is really feeling with me.
There are times when I feel that his outward response to me is quite different from the way he feels underneath.
What he says to me often gives a wrong impression of his whole thought or feeling at the time.
I believe that he has feelings he does not tell me about that are causing difficulty in our relationship.
Stages of the Truax Self-Congruence Scale b

Stage 1
The therapist is clearly defensive in the interaction, and there is explicit evidence of a very considerable discrepancy
between what he says and what he experiences. There may be striking contradictions in the therapist’s statements, the
content of his verbalization may contradict the voice qualities or nonverbal cues (i.e., the upset therapist stating in a
strained voice that he is “not bothered at all” by the patient’s anger).
Stage 2
The therapist responds appropriately but in a professional rather than in a personal manner, giving the impression that
his responses are said because they sound good from a distance but do not express what he really feels or means. There
is a somewhat contrived or rehearsed quality or an air of professionalism present.
Stage 3
The therapist is implicitly either defensive or professional, although there is no explicit evidence.
Stage 4
There is neither implicit nor explicit evidence of defensiveness or the presence of a façade. The therapist shows no
self-incongruence.
(Continued )
193
Table 9.1 Continued
Stage 5
The therapist is freely and deeply himself in the relationship. He is open to experiences and feelings of all types—both
pleasant and hurtful—without traces of defensiveness or retreat into professionalism. Although there may be
contradictory feelings, these are accepted or recognized. The therapist is clearly being himself in all of his responses,
whether they are personally meaningful or trite. At Stage 5 the therapist need not express personal feelings, but whether
he is giving advice, reflecting, interpreting, or sharing experiences, it is clear that he is being very much himself, so that
his verbalizations match his inner experiences.
a
Barrett-Lennard (1962).
b
Truax (1966a, pp. 68–72).
perspective (Greenberg & Watson, 2005). As such, we conducted PsycINFO and

These are illustrated in a vignette where MEDLINE searches using the keywords
a therapist described being told by a client “congruence,” “genuineness,” and “psycho-
that she saw the therapist as phony and therapy.” In our previous review (Klein
presumptuous. The therapist responded by et al., 2002), we identified 20 articles
telling the client that he felt afraid of her meeting the above criteria. For the present
anger and how hurt he felt. This disclosure review, we identified five additional poten-
led to a change in their interactions. tial articles.
Meta-Analytic Review Inclusion Criteria

The empirical evidence for the relation In order to be included in the meta-analysis,
between therapist congruence or genuineness a study had to include quantitative informa-
and patient outcome has been previously tion adequate to calculate an effect size (e.g.,
reviewed by at least 10 sets of research- a correlation coefficient). This procedure
ers (in chronological order): Meltzoff & resulted in 14 articles reporting 16 studies
Kornreich, 1970; Truax & Mitchell, 1971; that were included in our meta-analysis
Luborsky, Chandler, Auerbach, Cohen, & (see Table 9.2). Eleven of the 25 identified
Bachrach, 1971; Kiesler, 1973; Lambert, articles were excluded due to insufficient
DeJulio, & Stein, 1978; Mitchell, Bozarth, information. Table 9.2 lists studies included
& Krauft, 1977; Parloff, Waskow, & Wolfe, in our meta-analytic review and provides
1978; Orlinsky & Howard, 1978, 1986; summary information with respect to
Watson, 1984; Orlinsky, Grawe, & Parks, (a) aggregate effect size (for those studies that
1994). The consensus of these reviews is included multiple reports of congruence–
that empirical support for the contribution outcome relations), (b) type and perspective
of congruence to patient outcome is mixed of congruence measure, and (c) type and
but leaning toward the positive. perspective of outcome measure.
Search Strategy Methodological Decisions

In order to identify studies to include in The effect size (ES) we used in this chapter
the present review, we narrowed our focus is r, the correlation coefficient for the rela-
to published studies (in English) and dis- tion between congruence and outcome.
sertation research on individual or group Each study was reviewed and coded by
therapy with adults or adolescents (thereby two raters (coauthors Wang and Austin).
excluding studies of psychotherapy with Discrepancies in original coding were nego-
children and unpublished research reports). tiated in a consensus discussion involving
Table 9.2 Studies Included in Meta-Analytic Review
Reference Effect size (r) CM CP OM OP
Athay (1973) 0.24 2 1 4 1,2
Buckley et al. (1981) 0.06 3 1 1,2,5 1
Fretz (1966) 0.25 1 1,2 4 1,2
Fuertes et al. (2007) 0.34 3 1, 2 4 1,2
Garfield & Bergin (1971) −0.26 3 3 1,4,5 1,2,3
Hansen et al. (1968) 0.69 1,3 1,3 5 1
Jones & Zoppel (1982) −0.02 3 1 1,3,4 1
Marmarosh et al. (2009) 0.41 3 1, 2 1 1
Melnick & Pierce (1971) 0.42 3 3 5 1
Ritter et al. (2002) 0.21 1 1 1,2,5 1
Rothman (2007) 0.50 1 1 2,4 2
Staples et al. (1976) 0.16 3 3 4 3
Truax (1966a) 0.38 1,3 1,3 4 1
Truax (1971) −0.02 2 1 1,5 1
Truax (1971) 0.28 2 1 1,5 1
Truax (1971) 0.11 2 1 5 1
Note: CM = congruence measure (1 = Barrett-Lennard Relationship Inventory [BLRI], 2 = Truax Relationship Questionnaire [TRQ], 3 = other
scales/checklists); CP = congruence perspectives (1 = patient, 2 = therapist, 3 = observer); OM = outcome measure (1 = symptoms [e.g.,
anxiety, SCL-90-R], 2 = functioning [e.g., GAFS, adaptive skills/coping], 3 = well-being [e.g., overall success], 4 = global [a measure focusing
on general change without any particular dimension], 5 = other [e.g., MMPI, satisfaction, self-concept, goal attainment, personality, Q-sort,
self-efficacy, self-esteem]); OP = outcome perspectives (1 = patient, 2 = therapist, 3 = observer).
the first author. If r was not available or the inverse of variance (Hedges & Olkin,
was nonsignificant (and not reported), we 1985).
adopted the strategy of entering zero as the A test of homogeneity, using Hedges
effect size (Lipsey & Wilson, 2001). and Olkin’s Q statistic, was conducted to
For studies reporting multiple correlations determine if the effect sizes among studies
and using multiple measures, we aggregated were homogeneous. We adopted a random-
within each study by accounting for the effects model for determining overall effect
dependencies of measures. This aggregation size (ES) since the studies we identified were
used the correlation matrix among mea- quite heterogeneous (Q = 35.32, p < 0.01),
sures if reported. Otherwise, we assumed thus violating the assumptions required for
that the correlation was 0.50 when the same fixed-effects ES modeling (e.g., homogene-
method was used (e.g., self-report congruity of sample, variation in study ES due
ence and self-report outcome) and a corre- only to sampling error; Hedges & Vevea,
lation of 0.25 when different methods 1998). In addition, random-effects modeling
were used (e.g., self-report vs. observation; allows for greater generalizability. Moreover,
Gleser & Olkin, 1994). Overall, the cor- if the analysis showed between-study het-
relation from each study was used to cal- erogeneity, weighted univariate regression
culate an aggregated correlation using a or weighted between-group tests were used
weighted average where the weights were to examine moderator variables.
Results outcome (r = 0.29) produced a significantly
Estimates of effect sizes (ESs) in the 16 higher ES than therapist-rated outcome (r =
studies representing 863 participants ranged 0.07) (QB = 8.05, p < 0.05). This may be
from −.26 to 0.69. The weighted aggregate due to the fact that both congruence and
ES for congruence with psychotherapy out- outcome were more often assessed from the
come was 0.24 (95% CI = 0.12 to 0.36). patient perspective and is consistent with the
The overall ES of 0.24 for congruence is observation that relations within perspective
considered a medium effect (Cohen, 1988) (patient-rated process and patient-rated
and accounts for approximately 6% of outcome) are often more robust (Hill &
the variance in treatment outcome. This Lambert, 2004). Of course, it is also impor-
provides evidence for congruence as a note- tant to consider that this is simply an artifact
worthy facet of the psychotherapy relation- of method variance. At the same time, these
ship. Yet, this finding must be cautiously constructs are highly phenomenological
interpreted, as publication bias favors sig- in nature, and relations like this are likely
nificant results; thus, this ES may be an to be best captured by within-perspective
overestimation of the true congruence– self-report.
outcome relation in psychotherapy. At the
same time, this ES could also be an under- Measurement
estimation as we used the conservative The specific congruence instrument used
assumption of treating unreported, nonsig- did not influence the magnitude of the
nificant results as zero. congruence–outcome relation, as ESs ranged
from 0.19 to 0.24. In other words, congru-
Moderators ence measures yielded similar results across
The finding of heterogeneity of ESs among studies.
studies led us to examine the extent Similarly, aggregate ESs did not differ
to which potential moderators accounted across four types of outcome measures
for the variability in magnitude of the (symptoms, functioning, global, other) rang-
congruence–outcome association across the ing from 0.14 to 0.25. Outcome measures
studies. As noted above, weighted univari- produced similar results across studies.
ate regression or weighted between-group
tests were used to examine moderator Therapist Variables
variables. Specifically, we examined poten- This chapter highlights the importance of
tial moderator influences in the form of therapist-offered congruence as an impor-
measurement-related variables (rater per- tant influence in effective psychotherapy.
spective, congruence measurement, outcome As such, therapist variables may emerge
measurement), therapist variables, patient as moderators of the congruence–outcome
variables, and treatment variables (dura- link. Unfortunately, limited therapist-
tion, orientation, setting, and format). relevant information was available in the
studies included in our analyses, although
Rater Perspective there was sufficient data to examine five
We found no consistent pattern regarding the therapist variables: age, experience, gender,
influence of rater perspective on the mea- race/ethnicity, and training status.
surement of congruence on the congruence– The mean number of years of therapist
outcome association. ESs ranged from 0.13 clinical experience across five studies was 7.2
to 0.31 across perspectives and were rela- years (clinical experience ranged from
tively homogeneous. However, patient-rated 0 years for trainees to 13.6 years with a
median of 5.6 years). Results from a weighted Studies examining the congruence–outcome
univariate regression analysis indicated a pos- relation in adolescents (r = 0.42) attained
itive relation between therapist clinical expe- a significantly higher ES that those using
rience and the congruence–outcome ES (B = adult patients (r = 0.19) (QB = 7.15, p <
0.05, p < 0.01). Age as well as gender of ther- 0.01). Thus, it appears that therapist con-
apist (coded as percent female), therapist gruence may be more important for out-
minority status, and therapist’s training status come in adolescent patients.
did not significantly moderate the congru- Gender of patient (coded as percent
ence–outcome relationship. female) and patient minority status (coded as
percent minority) did not significantly mod-
Patient Variables erate the congruence–outcome relationship.
Perhaps because most of the studies
included in this meta-analytic review were Treatment Variables
published prior to 1990, patient descrip- We coded four potential moderators of the
tive information was seldom reported. congruence–outcome association in terms
We could code only four patient variables: of treatment parameters: duration, orienta-
education, age, gender, and minority status. tion, setting, and format. The number of
Educational attainment for patients was sessions of therapy attended, an index of
11.6 years on average across the studies duration, did not moderate the congru-
included for this review. This is somewhat ence–outcome relationship.
low (when compared with typical adult In terms of theoretical orientation, stud-
outpatient psychotherapy samples) due to ies in a mixed category (described as eclec-
the inclusion of three studies involving tic, client-centered, or interpersonal) (r =
adolescent patients. The mean years of 0.36) attained significantly higher ESs than
education was 9.2 for the adolescent stud- those characterized as psychodynamic (r =
ies and 14.5 for the adult studies, indicat- 0.04) (QB = 8.76, p < 0.01). One could
ing that most of the adults had completed speculate that congruence is more important
at least some college, which is consistent for outcome in a more present-oriented,
with the adult outpatient therapy research problem-focused therapy in contrast to
literature (cf. Vessey & Howard, 1993). psychodynamic approaches.
According to the weighted univariate regres- The ESs among therapeutic settings
sion analysis, patient education moderated showed significant differences. School coun-
the magnitude of the congruence–outcome seling centers (r = .43), inpatient settings
relation. As education decreased, the congru- (r = .27), and mixed settings (2 or more
ence–outcome relation increased (B = −0.09, settings) (r = .23) had a significantly higher
p < 0.001). Patients with less education were ES than outpatient mental health settings
more likely to demonstrate a greater con- (r = −.02) (QB = 16.47, p < .01). School
gruence–outcome relation; in other words, counseling centers also had a significantly
therapist congruence is more important for higher ES than mixed settings. This finding
outcome with less educated patients. is difficult to interpret without resorting to
Patient age as a continuous variable was conjecture.
not a significant moderator, but we dichoto- Finally, we examined the effect of psy-
mized age as adolescent versus adult in order chotherapy format (group vs. individual)
to clarify the previous finding regarding on the congruence–outcome association.
education. Adolescent versus adult mod- Group therapy studies (r = 0.36) obtained a
erated the congruence–outcome relation. higher ES than those examining individual
therapy (r = 0.18) (QB = 5.55, p < 0.05). more structured relationship in which the
Congruence may be more important for therapist takes on a more formal, directive,
outcome in group therapy. However, this and authoritative (i.e., less congruent) role
finding may have more to do with the char- (Sue & Sue, 2003). Congruence match
acteristics of the patients involved than the between patient and therapist (Zane et al.,
format per se given that adolescents (see 2004) may be of great consequence for the
findings for age and education) and inpa- therapy relationship.
tients were highly represented in the group A patient who has greater needs and
therapy condition. expectations for congruence is likely to
find comfort and satisfaction (an emo-
Patient Contribution tional bond) with a highly congruent ther-
Congruence/genuineness is both intraper- apist. These patients require a therapist to:
sonal and interpersonal. It can be seen as a be comfortable and at ease; be “real and
personal characteristic (intrapersonal) of the genuine;” say tactfully what s/he is feeling
psychotherapist as well as a mutual, experi- and thinking; naturally express honest/
ential quality of the relationship (interper- authentic impressions; and not avoid, hide,
sonal). Thus, the patient contribution to hold back, or fail to be direct when the
our understanding of congruence refers “elephant in the room” requires confronta-
to the experiential, interpersonal quality of tion. Patients in a congruent therapy rela-
genuineness and authenticity (i.e., congru- tionship learn that they are capable and
ence) in the therapy relationship, modeled worthy of time and attention, that they
by the therapist and experienced by the matter as a person with strengths and weak-
patient. nesses, and regrets as well as hopes and
All of us have needs, preferences, and dreams for the future. Therapist commit-
expectations for relationships; patients ment to truthfulness promotes patient
bring these to the therapy relationship (see acceptance of the problems they face as
Chapters 15 and 16). One can assume well as efforts to change.
that the need for congruence varies across
patients as well. Some would like a more Limitations of the Research
congruent therapist, some less. Cultural Any inferences arising from our meta-
background may importantly influence analysis of congruence–outcome relations
patient predilection for congruence. must be mindful of the methodological
Members of other cultures often approach limitations of the studies included as well
psychotherapy in fundamentally different as the meta-analytic methods used. Previous
ways than their Westernized counterparts researchers have noted limitations of stud-
(Patterson, 1996). Culturally informed ther- ies included in our meta-analysis: studies
apy considers patient qualities such as per- not limited to clients in need of change;
sonality, values, political heritage, social low levels or restricted ranges of congru-
structure, and communication style. For ence; different rating perspectives; use of
example, the value in Western psycho- ratings from audiotapes that do not allow
therapy for autonomy and independence nonverbal behaviors to be considered; vary-
may not hold true for interdependently ing qualifications and/or training of raters;
oriented patients from Eastern cultures inadequate and variable sampling methods;
(Tseng, 1999). Personal autonomy and and small sample sizes; (see Lambert et al.,
self-differentiation is often discouraged in 1978; Parloff et al., 1978; Patterson, 1984;
such patients; instead, they may desire a Watson, 1984). It is also important to note
the paucity of recent studies examining Therapeutic Practices
the congruence–outcome association and the In closing, we offer several recommendations
lack of any randomized controlled trials for clinical practice to foster congruence:
investigating the causal impact of congru-
ence. Caution is therefore warranted. • Therapists can strive for genuineness
Moreover, it is important not to over- with their patients. This involves
generalize. Positive findings for congru- acceptance of and receptivity to
ence/genuineness have appeared primarily experiencing with the patient as well
in studies investigating client-centered, as a willingness to use this information
eclectic, and interpersonal therapies. As in discourse. The congruent therapist
such, researcher bias (an allegiance effect) is is responsible for his or her feelings
one possible explanation for our results. and reactions, and this “ownership of
Additionally, congruence/genuineness may feelings is specified” (Rogers et al.,
not be as potent a change process in all 1967, p. 377). This might include the
types of therapy nor with all kinds of therapist’s thinking out loud about why
patients. Finally, congruence may only be he or she said or did something. This
important for patient change in the con- experiential stance serves an attachment
text of the other facilitative conditions, for function (i.e., bonding) as well as a role
example, as a precondition for the impact function (i.e., guides behavior) for the
of either empathy or positive regard. therapy relationship.
While meta-analytic techniques hold • Therapists can mindfully develop
great utility in quantitatively integrating and the intrapersonal quality of congruence.
summarizing results across studies, careful As with all complex skills, this will
consideration is also warranted. Concerns require discipline, practice, and effort.
for the present review include: quality Solicitation of feedback from colleagues,
of studies, comparability of studies, and supervisors, peers, and perhaps patients
limited number of studies including the (when appropriate) might also enhance
exclusion of 11 due to lack of information the development of the capacity for
sufficient to calculate an ES. Given these relational authenticity.
limitations, the finding of a medium ES • What can therapists “do” to foster as
in the present quantitative review, and well as augment the interpersonal
affirmative impressions from our previous experience of congruence? Therapists
qualitative review (Klein et al., 2002), leads must model congruence. Congruent
us to reaffirm our previous conclusion that responding may well involve considered
the evidence is likely to be more strongly self-disclosure of personal information
supportive than appears at first glance of and life experiences. It could also entail
a positive relation between congruence articulation of thoughts and feelings,
and psychotherapy outcome. Orlinsky and opinions, pointed questions, and feedback
Howard (1978, pp. 288–289) noted, “If regarding patient behavior. Congruent
study after flawed study seemed to point in responses are honest. Congruent
the same general direction, we could not responses are not disrespectful, overly
help believing that somewhere in all that intellectualized, or insincere, although
variance there must be a reliable effect.” they may involve irreverence. They are
A consistent pattern of positive findings authentic and consistent with the therapist
is quite unlikely to be explained by study as a real person with likes, dislikes, beliefs,
flaws. and opinions. Genuine therapist responses
are cast in the language of personal References
pronouns (e.g., “I feel . . .,” “My view An asterisk (∗) indicates studies included in
is . . .,” “This is how I experience . . .”). the meta-analysis.
∗
• The maintenance of congruence Athay, A. L. (1973/1974). The relationship between
requires that therapists be aware of counselor self-concept, empathy, warmth, and
instances when congruence falters. genuineness, and client rated improvement.
(Doctoral dissertation, University of Utah, 1973).
Rogers and colleagues (1967) speak Dissertation Abstracts International, 34, 3976A.
of feeling “twisted . . . perhaps I am Baldwin, M. (1987). Interview with Carl Rogers on
responding socially, smiling, while the use of self in therapy. Journal of Psychotherapy
actually I know we are avoiding and the Family , 3, 45–52.
something” (p. 396), and then using Barrett-Lennard, G. T. (1959). Dimensions of
the twisted feeling as a cue for the perceived therapist response related to therapeu-
tic change. Doctoral dissertation (psychology),
need for self-examination and a University of Chicago.
return to a more genuine and direct Barrett-Lennard, G. T. (1962). Dimensions of
way of relating. therapist response as causal factors in therapeu-
• It is important for therapists to tic change. Psychological Monographs: General
identify and become aware of their and Applied, 76 (43, Whole No. 562).
congruence style and to discern the Barrett-Lennard, G. T. (1978). The Relationship
Inventory: Later development and adaptations.
differing needs, preferences, and JSAS Catalog of Selected Documents in Psychology,
expectations that patients have for 8, 68.
congruence. Effective therapists will Barrett-Lennard, G. T. (1998). Carl Rogers’ helping
modify and tailor their congruence system: Journey and substance. Beverly Hills, CA:
style according to patient presentation Sage.
∗
(e.g., culture, age, education). Buckley, P., Karasu, T. B., & Charles, E. (1981).
Psychotherapists’ view their personal therapy.
• Congruence may be especially Psychotherapy: Theory, Research and Practice, 18,
important in younger, less educated, 299–305.
and perhaps less sophisticated patients Carkhuff, R. R. (1969). Helping and human rela-
(e.g., adolescents, college students, tions: A primer for lay and professional helpers
young adults). The congruent therapist (2 vols.). New York: Holt, Rinehart & Winston.
communicates acceptance and the Cohen, J. (1988). Statistical power analysis for the
behavioral sciences (2nd ed.). Hillsdale, NJ:
possibility of engaging in an authentic Lawrence Erlbaum.
relationship, something not easily ∗
Fretz, B. R. (1966). Postural movements in a
expected from the often formal and counseling dyad. Journal of Counseling Psychology,
authoritarian adults in the lives of 13, 335–43.
∗
these patients. Fuertes, J. N., Mislowack, A., Brown, S., Gur-Arie,
• Congruence appears to be especially S., Wilkinson, S., & Gelso, C. J. (2007). Correlates
of the real relationship in psychotherapy: A study
apparent in psychotherapy with more of dyads. Psychotherapy Research, 17, 423–30.
experienced (often older) practitioners. ∗
Garfield, S., & Bergin, A. E. (1971). Therapeutic
Perhaps therapists come to relax the conditions and outcome. Journal of Abnormal
pretense of role-bound formality and Psychology, 77, 108–114(a).
give themselves permission to genuinely Geller, S., & Greenberg, L. (2002). Therapeutic pres-
engage their patients as they gain ence: Therapists experience of presence in the psy-
chotherapy encounter in psychotherapy. Person
experience and confidence. Moreover, Centered & Experiential Psychotherapies, 1, 71–86.
experienced therapists may recognize Gelso, C. J., & Carter, J. A. (1985). The relationship
and more carefully discern a patient’s in counseling and psychotherapy. The Counseling
need for relational congruence. Psychologist, 13, 155–244.
Gelso, C. J., & Carter, J. A. (1994). Components Hedges, L. V., & Vevea, J. L. (1998). Fixed-
of the psychotherapy relationship: Their interac- and random-effects models in meta-analysis.
tion and unfolding during treatment. Journal of Psychological Methods, 3, 486–504.
Counseling Psychology, 41, 296–306. Hill, C. E., & Lambert, M. J. (2004). Methodological
Gelso, C. J., & Hayes, J. A. (1998). The psycho- issues in studying psychotherapy processes
therapy relationship: Theory, research, and practice. and outcomes. In J. M. Lambert (Ed.), Bergin and
New York: Wiley. Garfield’s handbook of psychotherapy and behavior
Gelso, C. J., Kelley, F. A., Fuertes, J. N., change (pp. 84–135). New York: Wiley and Sons.
∗
Marmarosh, C., Holmes, S. E., & Costas, C. Jones, E. E., & Zoppel, C. L. (1982). Impact of
(2005). Measuring the real relationship in psycho- client and therapist gender on psychotherapy
therapy: Initial validation of the Therapist Form. process and outcome. Journal of Consulting and
Journal of Counseling Psychology, 52, 640–49. Clinical Psychology, 50, 259–72.
Gendlin, E. T., & Geist, M. (1962). The relation- Kiesler, D. J. (1973). The process of psychotherapy:
ship of therapist congruence to psychological Empirical foundations and systems of analysis.
test evaluations of personality change. Wisconsin Chicago: Aldine.
Psychiatric Institute, Brief Research Reports, 24. Klein, M. H., Kolden, G. G., Michels, J., &
Gleser, L. J., & Olkin, I. (1994). Stochastically Chisholm-Stockard, S. (2002). Congruence. In
dependent effect sizes. In H. M. Cooper & J. C. Norcross (Ed.), Psychotherapy relationships
L. V. Hedges (Eds), The Handbook of Research that work: Therapist contributions and responsive-
Synthesis (pp. 339–55). New York: Russell Sage ness to patients (pp. 195–215). New York: Oxford
Foundation. University Press.
Greenberg, L., & Watson, J. (2005). Emotion- Lambert, M. J., DeJulio, S. S., & Stein, D. M.
focused therapy of depression. Washington, DC: (1978). Therapist interpersonal skills: Process,
APA Press. outcome, methodological considerations, and
Greenson R. R. (1967). The technique and prac- recommendations for future research. Psycho-
tice of psychoanalysis (Vol. 1). Madison, CT: logical Bulletin, 85, 467–89.
International Universities Press. Lietaer, G. (1993). Authenticity, congruence, and
Gurman, A. S. (1973a). Effects of the therapist transparency. In D. Brazier (Ed.), Beyond Carl
and patient mood on the therapeutic function- Rogers: Toward a psychotherapy for the twenty-first
ing of high and low-facilitative therapists. Journal century (pp. 17–46). London: Constable.
of Consulting and Clinical Psychology, 40, 48–58. Lipsey, M. W., & Wilson, D. B. (2001). Practical
Gurman, A. S. (1973b). Instability of the thera- meta-analysis. Thousand Oaks, CA: Sage.
peutic conditions in psychotherapy. Journal of Luborsky, L., Chandler, M., Auerbach, A. H.,
Counseling Psychology, 20, 16–24. Cohen, J., & Bachrach, H. M. (1971). Factors
Gurman, A. S. (1977). The patient’s perception of influencing the outcome of psychotherapy:
the therapeutic relationship. In A. S. Gurman A review of quantitative research. Psychological
& A. M. Razin (Eds.), Effective psychotherapy: Bulletin, 75, 145–85.
∗
A handbook of research (pp. 503–43). Oxford, Marmarosh, C., Gelso, C., Majors, R., Markin, R.,
UK: Pergamon. Mallery, C., & Choi, J. (2009). The real relation-
Halkides, G. (1958). An experimental study of four ship in psychotherapy: Relationships to adult
conditions necessary for therapeutic change. attachments, working alliance, transference,
Unpublished doctoral dissertation, University of and therapy outcome. Journal of Counseling
Chicago. Psychology, 56, 337–50.
∗ ∗
Hansen, J. C., Moore, G. D., & Carkhuff, R. R. Melnick, B., & Pierce, R. M. (1971). Client evalu-
(1968). The differential relationships objective ation of therapist strength and positive-negative
and client perceptions of counseling. Journal of evaluation as related to client dynamics, objec-
Counseling Psychology, 24, 244–46. tive ratings of competency and outcome. Journal
Hart, J. T. (1960). A replication of the Halkides of Clinical Psychology, 27, 408–11.
study. Unpublished manuscript, University of Meltzoff, J., & Kornreich, M. (1970). Research in
Wisconsin. Madison, WI. psychotherapy. New York: Atherton.
Hedges, L. V., & Olkin, I. (1985). Statistical Mitchell, K. M., Bozrath, J. D., & Krauft, C. C.
methods for meta-analysis. Orlando, FL: Academic (1977). A reappraisal of the therapeutic effec-
Press. tiveness of accurate empathy, nonpossessive
warmth, and genuineness. In A. S. Gurman & Sue, D. W., & Sue, D. (2003). Counseling the
A. M. Razin (Eds), Effective psychotherapy: culturally diverse: Theory and practice, 4th edition.
A handbook of research. (pp. 503–43). Oxford, New York: John Wiley & Sons.
UK: Pergamon. Truax, C. B. (1966a). Toward a tentative measure-
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). ment of the central therapeutic ingredients.:
Process and outcome in psychotherapy–Noch Arkansas Rehabilitation Research and Training
einmal. In A. E. Bergin & S. L. Garfield (Eds), Center and University of Arkansas, Fayetteville.
∗
Handbook of psychotherapy and behavior change Truax, C. B. (1966b). Therapist empathy, warmth
(4th ed.) (pp. 270–376). New York: Wiley. and genuineness and patient personality change
Orlinsky, D. E., & Howard, K. I. (1978). The rela- in group psychotherapy: A comparison between
tion of process to outcome in psychotherapy. In interaction unit measures, time sample measures,
S. L. Garfield & A. E. Bergin (Eds.), Handbook patient perception measures. Journal of Clinical
of psychotherapy and behavior change (2nd ed.) Psychology, 22, 225–29.
(pp. 283–329). New York: Wiley. Truax, C. B. (1968). Therapist interpersonal rein-
Orlinsky, D. E., & Howard, K. I. (1986). Process forcement of client self-exploration and thera-
and outcome in psychotherapy. In S. L. Garfield peutic outcome in group psychotherapy. Journal
& A. E. Bergin (Eds.), Handbook of psychother- of Counseling Psychology, 15, 225–31.
∗
apy and behavior change (3rd ed.) (pp. 311–84). Truax, C. B. (1971). Perceived therapeutic condi-
New York: Wiley. tions and client outcome. Comparative Group
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Studies, 2, 301–10.
Research on client variables in psychotherapy. In Truax, C. B., & Carkhuff, R. R. (1967). Toward
S. L. Garfield & A. E. Bergin (Eds.), Handbook effective counseling and psychotherapy: Training
of psychotherapy and behavior change (2nd ed., and practice. Chicago: Aldine.
pp. 233–82). New York: Wiley. Truax, C. B., & Mitchell, K. M. (1971). Research
Patterson, C. H. (1984). Empathy, warmth, and on certain therapist interpersonal skills in rela-
genuineness in psychotherapy: A review of tion to process and outcome. In A. E. Bergin &
reviews. Psychotherapy, 21, 431–38. S. L Garfield (Eds.), Handbook of psychotherapy
Patterson, C. H. (1996). Multicultural counsel- and behavior change (pp. 299–344). New York:
ing: From diversity to universality. Journal of Wiley.
Counseling & Development, 74, 227–31. Tseng, W. S. (1999). Culture and psychother-
∗
Ritter, A., Bowden, S., Murray, T., Ross, P., Greeley, J., apy: Review and suggested practical guidelines.
& Pead, J. (2002). The influence of the therapeu- Transcultural Psychiatry, 36, 131–79.
tic relationship in treatment for alcohol depen- Vessey, J. T., & Howard, K. I. (1993). Who seeks
dency. Drug and Alcohol Review, 21, 261–68. psychotherapy? Psychotherapy, 30, 546–53.
Rogers, C. R. (1957). The necessary and sufficient Watson, J. C., Greenberg, L. S., & Lietaer, G.
conditions of therapeutic personality change. (1998). The experiential paradigm unfolding:
Journal of Consulting Psychology, 21, 95–103. Relationship and experiencing in therapy. In
Rogers, C. R., Gendlin, E. T., Kiesler, D. J., & L. S. Greenberg, J. C. Watson, & G. Lietaer
Truax, C. B. (Eds.). (1967). The therapeutic rela- (Eds.), Handbook of experiential psychotherapy
tionship and its impact: A study of psychother- (pp. 3–27). New York: Guilford.
apy with schizophrenics. Madison: University of Watson, N. (1984). The empirical status of Rogers’
Wisconsin Press. hypotheses of the necessary and sufficient condi-
∗
Rothman, D. B. (2007). The role of the therapeutic tions for effective psychotherapy. In R. F. Levant
alliance in psychotherapy with sexual offenders. & J. M. Schlein (Eds.), Client-centered therapy
Unpublished doctoral dissertation, University of and the person centered approach: New direc-
Manitoba, Canada. tions in theory, research, and practice (pp. 17–40).
Shostrum, E. L. (Producer). (1966). Three approaches New York: Praeger.
to psychotherapy. [Film]. (Available from Psycho- Zane, N., Hall, G. N., Sue, S., Young, K., & Nunez, J.
logical and Educational Files, Santa Ana, CA). (2004). Research on psychotherapy with cultur-
∗
Staples, F. R., & Sloane, R. B. (1976). Truax fac- ally diverse populations. In M. J. Lambert (Ed.),
tors, speech characteristics, and therapeutic out- Bergin and Garfield’s handbook of psychotherapy
come. Journal of Nervous and Mental Disease, and behavior change, 5th edition (pp. 767–804).
163, 135–40. New York: Wiley & Sons.
C HA P TER
10 Collecting Client Feedback
Michael J. Lambert and Kenichi Shimokawa
Hundreds of studies have now been of disorders and have determined that
conducted on the effects of psychother- some disorders (e.g., phobias, panic) yield
apy, including research on psychodynamic, to treatment more easily than others (e.g.,
humanistic, behavioral, cognitive, and varia- obsessive-compulsive disorder).
tions and combinations of these approaches. An often ignored but critical consider-
Reviews of this research, both qualitative ation in psychotherapy and related inter-
and quantitative, have shown that about ventions is the degree to which they
75% of those who enter treatment show have negative rather than positive conse-
some benefit (Lambert & Ogles, 2004). quences for clients. An estimated 5%–10%
This finding generalizes across a wide range of adult clients participating in clinical
of disorders with the exception of severe bio- trials leave treatment worse off than they
logically based disturbances, such as bipolar began (Lambert & Ogles, 2004). In rou-
disorder and the schizophrenias, where the tine care the situation is more problematic.
impact of psychological treatments is sec- Outcomes for more than 6,000 patients
ondary to psychoactive medications. treated in routine practice settings suggest
Quantitative reviews (meta-analyses) of that the clients did not fare nearly as well as
psychotherapy efficacy support these con- those in clinical trials, with only about one
clusions and provide a numerical index for third showing improvement or recovery
treatment effects. Early applications of meta- (Hansen, Lambert, & Forman, 2002). The
analysis to psychotherapy efficacy (e.g., situation for child psychotherapy in rou-
Smith, Glass, & Miller, 1980) addressed tine care is even more sobering. The small
the broad question of the extent of benefit body of outcome studies in community-
associated with psychotherapy. For exam- based, usual-care settings has yielded a mean
ple, an average effect size of 0.85 standard effect size near zero (Weisz, 2004; Weiss,
deviation units was found over 475 studies Catron, Harris, & Phung, 1999; Weisz,
comparing treated and untreated groups. Donenberg, Han, & Weiss, 1995), yet
This indicates that, at the end of treat- millions of youths are served each year in
ment, the average treated person is better these systems of care (National Advisory
off than 80% of the untreated control Mental Health Council, 2001; Ringel &
sample. Subsequent meta-analytic findings Sturm, 2001). In a comparison of chil-
(e.g., Shadish et al., 1997; Wampold, 2008) dren being treated in community mental
have supported the consistent benefit of health (N = 936) or through managed care
treatment over control for a broad variety (N = 3075), estimates of deterioration were
203
24% and 14%, respectively (Warren et al., obvious; in others a certain degree of blind-
2010). ing occurs, such that the association is not
There is no doubt that all of the deterio- so temporally connected and the effects
ration that occurs during the time a patient of performance are harder to discern (such
is in treatment cannot be causally linked as in psychotherapy), making it much more
to therapist activities. Certainly, a portion difficult to learn and improve. In obvious
of patients are on a negative trajectory at as well as more subtle situations, provid-
the time they enter treatment and the dete- ing feedback to improve performance has
riorating course cannot be stopped. Some been studied quite extensively in a variety
patients are prevented from taking their of areas and confirms our common sense
own lives as a result of effective practices, expectations that it is helpful.
even if they do not show overall progress. In this chapter, we examine the effects
Just as positive psychotherapy outcomes of feedback broadly and then present evi-
depend largely on patient characteristics, so dence of its effects in relation to two widely
do the negative changes that occur in used feedback systems applied specifi-
patients who are undergoing psychological cally in psychotherapy. Following presenta-
treatments. Even so, positive as well as tion of the effects of feedback, we turn
negative patient change can be affected by our attention to implications for clinical
therapist actions and inactions. Research practice.
reviews find that the major contribution of
the therapist to negative change is usually Previous Reviews of Feedback
found in the nature of the therapeutic rela- In a meta-analysis of the effects of feedback
tionship, with rejections of either a subtle on human performance published since the
or manifest nature being the root cause 1930s, Kluger and DeNisi (1996) found a
(e.g., Safran, Muran, Samstang, & Winston, small to medium effect size (d = .41) for
2005). In fact, the research on negative interventions utilizing feedback, suggest-
outcomes find very little negative change ing about two-thirds of individuals receiv-
as a result of the misapplication of thera- ing feedback performed better than those
peutic techniques, while relationship fac- who received no feedback. Unfortunately,
tors loomed large across treatment formats most of the studies examined in this
(e.g., couple, family, group, individual) and meta-analysis were analog situations involv-
theoretical orientation (Lambert, Bergin, ing motor performance, puzzle solutions,
& Collins, 1977). memory tasks, and the like, rather than
A recent trend in clinical practice involves based in clinical practice. Further, of the
regularly monitoring and tracking client few that tested feedback to professionals, not
treatment responses with standardized one study provided feedback on patient’s
scales throughout the course of treatment status to health professionals. Nevertheless,
and then providing clinicians with this infor- this review does suggest that a broad array of
mation (Lambert & Burlingame, 2007). feedback interventions consistently improve
The basic rationale behind collecting client performance and encourages the idea that
feedback is based on common sense. If feedback will enhance performance.
we get information about what seems to In a more comprehensive meta-analysis,
be working, and more importantly, what Sapyta (quoted in Sapyta, Riemer, and
is not working, our responsiveness to cli- Bickman, 2005) examined 30 randomized
ents will improve. In many situations, per- clinical trials conducted in community set-
formance and feedback are intertwined and tings that assessed the effectiveness of client
health status feedback to health profession- flagged samples in the Sapyta and Bickman
als. The nature of feedback interventions (as quoted in Sapyta, et al., 2005) meta-
and methods of their delivery varied from analysis achieved an effect size (d ) of 0.31,
giving general practice physicians depres- which indicates that the average client in
sion or anxiety screening information about the treatment group had better outcomes
their patients to repeatedly and routinely than 62% of the flagged control group. At
providing clinicians with their patients’ the same time, it appears that the feedback
mental health status feedback. The average of client’s health status is mainly beneficial
client in the feedback group was better off to clients who may require changes to their
than 58% of the control group (d = .21, a current treatment.
small effect). This finding is consistent with feedback
In general, this research supports the theories that suggest feedback will only
conclusion that feedback in clinical prac- change behavior when the information
tice improves patient outcome. provided indicates the individual is not
A recent meta-analysis that focused meeting up to an established standard of
directly on mental health status feedback practice (e.g., Riemer & Bickman, 2004).
in psychotherapy (Knaup, Koesters, Becker, Riemer and Bickman (Riemer & Bickman,
& Puschner, 2009) based on 12 studies also 2004; Riemer, Rosof-Williams, & Bickman,
found a statistically significant, albeit small 2005) have developed a contextual feed-
effect (d = 0.10) for progress feedback. This back intervention theory to explain how
particular analysis suggested that the effects feedback is interpreted and made useful.
may be short lived, although few studies Basic tenets of this theory are that clinicians
had a follow-up. (and professionals, generally) will benefit
The Sapyta, Riemer, and Bickman (2005) from feedback if they are committed to the
review indicated that the effectiveness of goal of improving their performance; they
feedback is likely to vary as a function of are aware of a discrepancy between the goal
the degree of discrepancy between therapists’ and reality (particularly if the goal is attrac-
views of progress and measured progress, tive and the clinician believes it can be accom-
and that the greater the discrepancy, the plished ); the feedback source is credible; and
more likely feedback will be helpful. This if feedback is immediate, frequent, systematic,
finding is consistent with feedback theories cognitively simple (such as graphic in nature),
and the role of negative feedback in regula- unambiguous, and provides clinicians with
tory systems (e.g., Bandura, 1997; Lord & concrete suggestions of how to improve.
Hanges, 1987); feedback about poor prog- If clinicians do not consider feedback
ress is expected to have a greater impact as credible, valid, informative, or useful,
than feedback indicating positive progress. they are more likely to dismiss it whenever
A key element of effective feedback is bring- it does not fit their own preferences. As
ing into the recipient’s awareness the discrep- we know from research on cognitive disso-
ancy between what is thought and what is nance, people can change attitudes rather
reality, thereby prompting corrective action. than persevere toward goals, thus regard-
When comparing the effect of feedback ing the goal as less important, or see a
on “flagged” clients who were not pro- client as too resistant or injured to bene-
gressing well versus clients who were pro- fit from treatment (e.g., disown personal
gressing through treatment as expected, responsibility for meeting the goal of posi-
flagged clients responded more favorably tive functioning; Riemer, Rosof-Williams,
to the feedback intervention. Feedback to & Bickman, 2005). As feedback research
l a m b e rt, s h i m o k awa 205

suggests, the value of monitoring and Helping the therapist become aware of
systematic feedback through psychologi- negative change and discussing such prog-
cal assessments hinges on the degree to ress in the therapeutic encounter are much
which the information provided goes beyond more likely when formal feedback is pro-
what a clinician can observe and under- vided to therapists. Such feedback helps the
stand about patient progress without such client communicate and helps the therapist
information. It is important for the infor- to become aware of the possible need to
mation to add something to the psycho- adjust treatment or alter or addresses prob-
therapist’s view of patients’ well-being and lematic aspects of the treatment as appro-
future actions. priate (e.g., problems in the therapeutic
Unfortunately, clinicians may have an relationship or in the implementation of
overly optimistic view of their patients’ the goals of the treatment). By contrast, for
progress (Walfish et al., 2010). Clinicians clients who are progressing well in treat-
overlook negative changes and have a lim- ment, progress feedback delivered to thera-
ited capacity to make accurate predictions pists is not expected to help therapists be
of the final benefit clients will receive during more responsive, a finding that is consis-
treatment, particularly with clients who are tent with the meta-analysis by Knaup et al.
failing to improve. One study, for example, (2009) previously mentioned.
found that even when therapists were pro-
vided with the base rate of deterioration in Definitions and Feedback Systems
the clinic where they worked (8%), and Clients can complete a brief measure of
were asked to rate each client that they saw their psychological function by using stan-
at the end of each session (with regards to dardized rating scales, and then this infor-
the likelihood of treatment failure and if mation can be delivered to psychotherapists
the client was worse off at the current ses- in real time. In addition to alerting thera-
sion in relation to their intake level of func- pists to deviations from expected treat-
tioning), they rated only 3 of 550 clients as ment response, the information gathered
predicted failures and seriously underesti- from patients provides novel information
mated worse functioning for a significant to therapists. Collecting this information
portion of clients (Hannan et al., 2005). from the client on a session-by-session basis
A retrospective review of case notes of cli- provides the clinician with a systematic
ents who had deteriorated during treatment way of monitoring life functioning from
found infrequent mention of worsening the client’s point of view. A brief formal
even when its degree was dramatic (Hatfield assessment can provide a summary of life
et al., 2010). functioning that is not otherwise available
Such results are not surprising, given to the therapist, unless the therapist spends
psychotherapist optimism, the complexity time within the treatment hour to system-
of persons, and a treatment context that atically inquire about all the areas of func-
calls for considerable commitment and tioning covered by the self-report scale.
determination on the part of the therapist, Several psychotherapy outcome manage-
who actually has very little control over the ment systems that provide progress feed-
patient’s life circumstances and personal back have been developed and implemented
characteristics. Patients’ response to treat- in clinical settings worldwide. Although
ment is, especially in the case of a worsening the specific procedures employed in each of
state, a likely place where outside feedback these systems vary, their common features
might have the greatest chance of impact. involve the monitoring of client outcome
throughout the course of treatment, shar- this system is clinician friendly and insures
ing the client data with clinicians, and discussion of assessment results by the
using these data to improve outcomes. client and therapist in session because
The first system to arrive was that devel- rating of mental health status and thera-
oped by Howard and colleagues using the peutic alliance are normally collected in the
COMPASS Treatment Outcome Systems presence of the therapist. The therapeutic
(Lueger et al., 2001). Kordy, Hannover, relationship as measured by the SRS is
and Richard (2001) developed a computer- based on the concept of the therapeutic
assisted, feedback-driven psychotherapy alliance by Bordin (1979), a similar con-
quality management system in Germany. cept of therapeutic alliance by Gaston
Barkham and colleagues (2001) created the (1990), and the construct Duncan, Miller,
Clinical Outcomes in Routine Evaluation and Sparks (2004) termed “client’s theory
(CORE) system widely used in the United of change.” These interrelated alliance
Kingdom, while Kraus and Horan (1997) theories emphasize three aspects of the
developed the Treatment Outcome Package helping relationship: the affective bond,
(TOPS) system in use in the USA. In gen- agreement on tasks during sessions, and
eral, these later two systems have empha- agreement about the ultimate goals of the
sized the administrative use of data rather encounter.
than feedback to therapists during the Miller and colleagues (2005) developed
course of psychotherapy. Administrative three items to rate these constructs and a
use allows managers of mental health ser- fourth item that provides an overall rating
vices to examine the periodic and final out- of the relationship. Figure 10.1 provides a
come of treatments and compare outcomes hypothetical example of feedback to both
to appropriate benchmarks. clinician and client for a client falling under
Two systems have gone beyond measuring benchmark predictions. ORS scores are
progress and outcome, investing consider- graphically portrayed compared with the
able energy in collecting and feeding back 50th percentile trajectory based on the cli-
client ratings of their therapist in the hopes ent’s intake score. Verbal feedback messages
of maximizing final treatment outcome. interpret the scores, taking the alliance
measure (SRS) into account, and encourage
Partners for Change Outcome client and provider discussion about next
Management System possible steps to avert a negative outcome.
The Partners for Change Outcome Clinical use of the ORS and the SRS
Management System (PCOMS; Miller, is gaining in popularity, and a growing
Duncan, Sorrell, & Brown, 2005) is a psy- number of published studies have exam-
chotherapy assurance system that employs ined the psychometric properties of these
two brief scales (four items each). The measures. Given the emphasis of this report
Outcome Rating Scale (ORS; Miller, on treatment outcome, we summarize here
Duncan, Brown, Sparks, & Claud, 2003) the psychometric properties of the ORS.
focuses on mental health functioning, The ORS is a visual analog scale that
modeled after the domains of outcome requires clients to rate their functioning
measured by subscales of the Outcome on four items (subjective well-being, inter-
Questionnaire-45 (Lambert et al., 2004). personal relations, social functioning, and
The Session Rating Scale (SRS; Duncan overall sense of well-being). Miller and
and Miller, 2008) is aimed at assessing the Duncan (2003) reported test–retest corre-
therapeutic alliance. Because of its brevity, lations among nonclinical samples ranging

User Signed in: Provider1
0056
Results:
You report that things are getting worse. There is strong reason for
concern. You are also reporting concerns about the provider and/or the
service.
Activity:
Strongly consider changing the frequency, type, or provider of services.
Talk about what your provider can do to improve the items marked with
a red hand.
Individually: 2 out of 10
Interpersonally: 2 out of 10
Socially: 2 out of 10
Overall: 1.1 out of 10
Total Score: 7.1
Outcome rating scale

40
Outcome score
32
24 18
16
14
16
13.2 15 7.1
8
0
0 1 2 3 4 5 6 7 8 9
Session number
Intake score Predicted score 25th Percentile 75th percentile
Clinical cutoff Actual score
Fig. 10.1 Hypothetical example of feedback to both clinician and client for a client falling under
benchmark predictions.
from 0.49 to 0.66 with Cronbach’s alpha of question as to the degree of potential influ-
0.93. Miller and Duncan (2003) also tested ence exerted by therapists’ presence on client
correlations between the ORS and OQ-45 response (e.g., demand characteristic).
over four waves of repeated administra- The ORS incorporates expected trajecto-
tions among 86 nonclinical individuals ries of change based on the initial score and
and found correlations ranging from 0.53 the change at a given session in relation to
to 0.69. When considering the intent of the initial score (Miller, Duncan, Sorrell, &
the ORS developers to create “a brief alter- Brown, 2005). Consistent with the recom-
native” to the OQ-45 (Miller et al., 2003, mendations given in the administration
p. 92), the modest degree of common vari- and scoring manual of the ORS, studies
ance among these two measures (i.e., r 2 = by Reese et al. (2009) classified clients as
0.28 to r 2 = 0.48) raises questions as to the being at risk, or not progressing, if they
extent to which the developers’ intent was failed to improve five ORS points or more
achieved. by the third session. In the study, the iden-
The outcome studies reviewed in this tification of at-risk clients was generated
chapter had the clients complete the ORS by a different system, employing a web-
in the presence of the therapist in session. based software that “calculates trajectories of
As Anker, Duncan, and Sparks (2009) noted, change at the 25th, 50th, and 75th percen-
such an administrative protocol raises a tile levels” based on a large ORS database
(Anker et al., 2009, p. 697). Clients whose problems, and social role functioning.
ORS score at the third session fell below Consistent with this conceptualization of
the 50th percentile mark of expected tra- outcome, the OQ-45 provides a Total
jectory of progress based on individual Score, based on all 45 items, as well as
response were identified as at risk. Symptom Distress, Interpersonal Relations,
Concurrent validity of SRS and Helping and Social Role subscale scores. Each of
Alliance Questionnaire-II (HAQ-II) corre- these subscales contains some items related
late at 0.48, with each item ranging from to the positive quality of life of the indi-
0.39 to 0.44. The SRS authors argue that vidual. Higher scores on the OQ-45 are
it is essentially measuring the same thing indicative of greater levels of psychological
as HAQ-II and other therapeutic alliance disturbance.
scales, although these moderate correlations Research has indicated that the OQ-45
suggest less overlap than would be hoped. is a psychometrically sound instrument,
with strong internal consistency (Cronbach’s
OQ Psychotherapy Quality alpha = 0.93), adequate 3-week test–retest
Management System reliability (r = 0.84), and strong concurrent
Lambert and colleagues developed the OQ validity estimates ranging from 0.55 to
system, which emphasizes the measure- 0.88 when the total score and the subscale
ment of mental health functioning and, scores were correlated with scores from the
like the PCOMS, includes a measure of MMPI-2, SCL-90R, BDI, Zung Depres-
the therapist-client relationship. In distinc- sion Scale, Taylor Manifest Anxiety Scale,
tion to the PCOMS, the OQ system goes State-Trait Anxiety Inventory, Inventory
beyond feedback on the therapeutic alli- of Interpersonal Problems, and Social
ance and includes additional assessments Adjustment Scale, among others (Lambert
to aid problem solving. In addition, the et al., 2004). Furthermore, the items that
relationship and problem-solving approach make up the OQ-45 have been shown
is only employed with specific patients to be sensitive to changes in multiple
who are experiencing a negative response to client populations over short periods of
psychotherapy rather than with all persons time while remaining relatively stable in
who enter treatment. In the remainder untreated individuals (Vermeersch, Lambert,
of this chapter and our meta-analysis, the & Burlingame, 2002; Vermeersch et al.,
OQ psychotherapy system and to a lesser 2004). In addition, evidence from factor
extent the PCOMS will be emphasized analytic studies suggests it measures an
because of the evidence base surrounding overall psychological distress factor as well
these systems. as factors consistent with the three subscales
The Outcome Questionnaire-45 (OQ-45; (Bludworth, Tracey, & Glidden-Tracey,
Lambert et al., 2004) is a 45-item, self- 2010; de Jong et al., 2007; Lo Coco et al.,
report measure designed for repeated admin- 2008). In short, the OQ-45 is a brief mea-
istration throughout the course of treatment sure of psychological disturbance that is
and at termination with adult patients. reliable, valid, and sensitive to changes cli-
In accordance with several reviews of the ents make during psychotherapy. It provides
literature (e.g., Lambert, 1983), the OQ clinicians with a mental health vital sign.
was conceptualized and designed to assess Similar measures have been developed for
three domains of client functioning: symp- use with children (www.oqmeasures.com).
toms of psychological disturbance (particu- A core element of outcome management
larly anxiety and depression), interpersonal systems is the prediction of treatment failure.

In order to improve outcomes of clients treatment). This 40-item measure does not
who are responding poorly to treatment, produce a total score but, rather, provides
such clients must be identified before subscale score feedback and item feedback
termination, and ideally, as early as for therapists to consider in problem solving.
possible in the course of treatment. The The first 11 items of the ASC require the
OQ system plots a statistically generated client to reflect on the therapeutic relation-
expected recovery curve for differing levels ship and report his or her perceptions.
of pretreatment distress and uses this as a In Figure 10.2 a sample feedback report
basis for identifying clients who are not is provided. Items that fall below an empir-
making expected treatment gains and are at ically based cutoff score (about one stan-
risk of having a poor outcome. The accu- dard deviation from the mean rating on
racy of this signal-alarm system has been the item) are brought to the therapist’s
evaluated in a number of empirical investi- attention.
gations (Ellsworth, Lambert, & Johnson, The ASC is central to the Clinical
2006; Lambert, Whipple, Bishop, et al., Support Tool (CST). The CST is composed
2002; Lutz et al., 2006; Percevic, Lambert, of a problem-solving decision tree designed
& Kordy, 2006; Spielmans, Masters, & to systematically direct therapists’ attention
Lambert, 2006) and has been found to to certain factors that have been shown to
be highly sensitive. It accurately predicts be consistently related to client outcome in
deterioration in 85%–100% of cases that the empirical literature, such as the thera-
actually end with a negative outcome and peutic alliance, social support, readiness to
far exceeds clinical judgment in its ability change, diagnostic formulation, and need
to identify clients who are at risk of having for medication referral. The ASC provides
a negative treatment outcome (Hannan the information necessary to go through
et al., 2005). the decision tree and focus the therapist’s
A sample feedback report for the OQ-45 attention on the quality of the therapeutic
is displayed in Figure 10.2. This report alliance, client motivation, and client per-
displays the client’s progress at the ninth ceptions of social support as well as the
session of psychotherapy in relation to a possible need for medication. Furthermore,
horizontal line at a score of 64/63 marking the CST provides specific intervention
normal functioning, and a solid dark line strategies that could be used by therapists if
displaying the expected treatment response. problems were detected in the aforemen-
Most important is the “Red” alert signal in tioned domains. Figure 10.3 depicts the
the upper left hand corner which indicates CST problem-solving decision tree pro-
the client is responding so poorly to ther- vided to therapists in cases in which their
apy that he or she is predicted to leave ther- clients were predicted to have a poor
apy with a negative treatment response. outcome.
In conjunction with identifying Alarm The signal-alarm system alerts clinicians
status, an instrument Assessment for signal to potential treatment failures and allows
cases (ASC; Lambert, Whipple, et al., 2004) them to modify their treatment approach.
was developed to assist clinicians to problem- Once a client takes the OQ-45, commences
solve with the clients who backslide during treatment, and completes a session of treat-
treatment (i.e., when a therapist receives a ment, the signal-alarm system can be used
warning message, indicating that the client to generate feedback regarding the client’s
is not responding or is deteriorating in progress.
Name: 12, case ID: 12 Alert Status: Red
Session Date: 12/21/2005 Session: 9 Most Recent Score: 79
Clinician: Maristany, Mariana Clinic: Aigle Initial Score: 58
Diagnosis: Unknown Diagnosis Change From Initial: Reliably Worse
Algorithm: Empirical Current Distress Level: Moderate
Most Recent Critical Item Status: Output. Comm.

Subscales Current
8. Sucide - I have thoughts of ending my Never Norm Norm
life. Symptom Distress: 45 49 25
11. Substance Abuse - After heavy Never
drinking, I need a drink the next morning to Interpersonal
18 20 10
get going. Relations:
26. Substance Abuse - I feel annoyed by Never Social Role: 16 14 10
people who criticize my drinking.
Total: 79 83 45
32. Substance Abuse - I have trouble at Never
work/school because of drinking or drug
use
44. Work Violence - I feel angry enough at Rarely
work/school to do something I might regret.
Total Score by Session Number
12/21/05 12/21/05 12/21/05 12/21/05 12/21/05

120
110
100
90
Score
81 (R) 78 (R)
80
75 (R)
70
62.000
60 58.0
50
1 3 5 7 9
Session Number
Graph Label Legend:
(R) = Red: High change of negative outcome (Y) = Yellow: Some chance of negative outcome
(G) = Green: Making expected progress (W) = White: Functioning in normal range
Feedback Message:
The patient is deviating from the expected response to treatment. They are not on track to realize substantial benefit
from treatment. Chances are they may drop out of treatment prematurely or have a negative treatment outcome.
Steps should be taken to carefully review this case and identify reasons for poor progress. It is recommended that you be
alert to the possible need to improve the therapeutic alliance, reconsider the client’s readiness for change and the need to
renegotiate the therapeutic contact, intervene to strengthen social supports, or possibly alter your treatment plan by
intensifying treatment, shifting intervention strategies, or decide upon a new course of action, such as referral for
medication. Continuous monitoring of future progress is highly recommended.
Fig. 10.2 OQ-Analyst screen shot illustrating feedback report of client progress provided to therapist.
211
Therapeutic Alliance Interventions
• Discuss therapeutic alliance with patient
• Give and ask for feedback on relationship
• Spend more time exploring patient’s experience
Therapeutic alliance: Does the YES • Discuss shared experiences
client report concerns with the • Reassess/agree on therapeutic tasks and goals
therapeutic alliance? • Clarify possible misunderstandings
• Give more positive feedback
NO • Use more empathic engagements
• Discuss therapist and therapeutic style match
• Process transference
Readiness to Change Interventions

• Discuss readiness to change with patient
• Give and ask for feedback about readiness for change
Motivation: Does the client report YES
being in a precontemplation or • Adjust goals and tasks to be challenging but not too
contemplation stage of readiness difficult
to change? • Discuss consequences of changing or not changing
• Discuss the processes involved with change and
specific skills that help
NO
Social Support Interventions

• Refer to group therapy
• Refer to biofeedback treatment
• Refer to assertiveness training
• Role play social situations
• Assign related homework
Social Support: Does the client YES
report low social support? • Assess patient’s self beliefs
• Bring others to sessions
• Encourage activities with others (e.g., family, friends,
NO significant others
• Work on concerns related to trusting others
• Encourage greater involvement in organizations
characterized by social interaction (e.g., clubs)
Reassess the diagnostic

formulation. Is there an effective YES Consult relevant resources and
treatment option that has not been alter the treatment plan.
attempted?
NO
Is medication an effective YES Refer for psychiatric

treatment option? consultation.
Fig. 10.3 Clinical Support Tool (CST) problem-solving decision tree.
Meta-Analytic Review feedback to clinicians, several critical dif-

Inclusion Criteria and Search Strategy ferences also existed. The OQ system was
In the following sections, we present designed to enhance the outcome of clients
a meta-analysis of the outcomes of two predicted to experience treatment failure at
real-time feedback-based psychotherapy termination. Accordingly, the studies exam-
systems. Although studies utilizing the ining the effects of the OQ systems con-
PCOMS provided some client-reported ducted separate analyses for at-risk clients
and on-track clients. In contrast, while the studies based on the PCOMS have been
PCOMS incorporates a method of identi- reported in two articles to date. To obtain
fying nonprogressing cases as at-risk clients, overall estimates of the effects of the
the studies employing the PCOMS, except PCOMS, we meta-analytically aggregated
for the study by Anker et al. (2009), did the published results of those studies.
not investigate or report differential effects
of the PCOMS feedback system on client Dependent Measures and
outcome based on on-track versus not-on- Computation of Effect Sizes
track classification of client progress. Due Because different effect size units were
to this difference in methodology, we pres- employed in original studies, we applied
ent separate summaries of these two quality uniform units of effect size. For each
assurance systems. comparison of mean posttreatment outcome
Included in the meta-analysis were stud- measurement scores between an experi-
ies from just two well-developed systems mental condition and a treatment as usual
that have published the effects of feedback. (TAU), we employed Hedges’ standard dif-
A few scattered efforts to examine feedback ference g (Hedges & Olkin, 1985) and cor-
were not reviewed in this meta-analysis relation r, to be consistent with meta-analytic
because they have yet to rise to the level findings presented in other chapters of
of well-developed systems. For example, this book. Correlation r was obtained by
Brodey et al. (2005) examined the effect of converting the standardized mean differ-
providing feedback to clinicians on half ence d, utilizing a commonly used formula
of a total sample of 1,374 clients using 11 of r = d/sqr(d 2 + 4) (e.g., Wolf, 1986).
anxiety and depression items from the A key element in psychotherapy research
Symptom Checklist-90. Rather than pro- is operationalizing the concepts of positive
vide mental health status (depressive & and negative outcome for the individual
anxiety symptoms) at every session of care, client. Jacobson and Truax (1991) offered
thus allowing clinicians to track progress, a methodology by which client changes
the second administration (taken by 954 of on an outcome measure can be classified in
the original sample; 69%) took place at a the following categories: recovered, reliably
prespecified time and after many patients improved, no change, deteriorated. There
had completed treatment. Even the initial are two necessary pieces of information to
report was not provided in the most timely make these client outcome classifications: a
manner with only 14% of clinicians report- Reliable Change Index (RCI) and a normal
ing they received it prior to the inception functioning cutoff score.
of treatment. Even so, the results suggested Clinical and normative data were ana-
the intervention improved patient outcomes lyzed by Lambert and colleagues (2004) to
at a statistically significant level (r = 0.05; establish an RCI and a cutoff score for the
d = 0.11). Another example is provided OQ-45. The RCI obtained on the OQ-45
by Berking, Orth, and Lutz (2006) who was 14 points, indicating that client changes
provided evidence that feedback enhanced of 14 or more points on the OQ-45 can be
outcome in a 30-day inpatient program. considered reliable (i.e., not due to mea-
Because a meta-analytic review of surement error). The cutoff score for normal
the OQ 45 was recently conducted (see functioning on the OQ-45 was calculated
Shimokawa, Lambert, & Smart, 2010), to be 63, indicating that scores of 64
we provide here a summary of the meta- or higher are more likely to come from a
analytic findings. Only three well-designed dysfunctional population than a functional

population, and scores of 63 or lower are two studies reported in a single article com-
more likely to come from a functional pop- paring the treatment outcome of clients
ulation than a dysfunctional population. receiving the PCOMS feedback interven-
Support for the validity of the OQ-45’s tion and those receiving no feedback. The
reliable change and cutoff score has been first study, conducted at a university coun-
reported by Lunnen and Ogles (1998) and seling center, was initially comprised of
Beckstead et al. (2003). 131 clients of which 74 (56%) were
Similar to the OQ system, based on included in the final analyses. Clients were
the same methods developed by Jacobson randomly assigned to either a PCOMS-
and Truax (1991), Miller and Duncan based feedback condition or no-feedback
(2004) reported the RCI and clinical cutoff condition (TAU) to investigate the effects
scores for the ORS. A change of five points of the feedback intervention on client out-
or greater in either direction in comparison come. The authors reported the effect size
to the pretreatment ORS score is consid- of d = 0.54 when the feedback group and
ered a reliable change. The clinical cutoff TAU were compared on the basis of pre-
score used in studies examining the effects treatment to posttreatment changes on
of the ORS was 25 points. Clinical signifi- the ORS total scale scores (using typical
cance classification of the ORS has not methods as applied to differences in means
been cross-validated with that of OQ-45 at posttest, the effect size was reduced to
or socially validated with other measures, Hedges’ g of 0.25). They further reported
thus empirical evidence on the meaning of that 80% of clients in the feedback group
clinical significance classification of the experienced reliable change, while 54% of
ORS is limited at this time. clients in TAU achieved the same criteria.
To contrast the rates and odds of client Only 4% of those in the feedback group
deterioration and significant improvement met the criteria for deterioration, while 13%
between feedback groups and TAU, we cal- of their TAU counterparts met the same.
culated combined odds ratios (OR) as a The second study in the Reese et al.
measure of effect size. Specifically, when (2009) article, conducted at a graduate
examining the odds of deterioration, we training clinic, was initially comprised of
dichotomized clients into either the deteri- 96 clients of which 74 (77%) were included
oration group or nondeterioration group in the final analyses. In this study, 17 trainee
and calculated the odds ratio of deteriora- therapists in graduate practicum, rather
tion for a given comparison. Similarly, when than clients, were randomly assigned to
comparing the odds of improvement in either a PCOMS-based feedback condition
two groups, the odds ratio was calculated or TAU. The authors (Reese et al., 2009)
based on the odds of improvement versus reported an effect size of d = 0.49 when
those of nonimprovement. comparing the feedback group and TAU
on the basis of the pre-post change in the
Meta-Analytic Results ORS scores (the standard mean difference
Results Based on the PCOMS. Three method- posttest comparison produced a Hedges’ g
ologically sound psychotherapy outcome of 0.58). The authors of the study reported
studies investigating the effects of the that 16 clients (36%) in the feedback group
PCOMS have been published in English and 11 clients (38%) in TAU were identi-
to date (Anker, Duncan, & Sparks, 2009; fied as “not progressing,” therefore at risk
Reese, Norsworthy, & Rowlands, 2009). of poor outcome. In terms of clinical sig-
Reese and colleagues (2009) conducted nificance, 67% of those in the feedback
condition achieved reliable change status, response. With regard to marital adjust-
while 4% deteriorated. This was contrasted ment, feedback was found to be somewhat
to 41% of clients in TAU achieving reliable helpful. The posttreatment d between groups
change, with 3% (n = 1) deteriorating. was 0.29 (an r = 0.14).
Anker, Duncan, and Sparks (2009) Findings from the three studies were
conducted a randomized controlled trial aggregated to provide estimated weighted
investigating the effects of PCOMS-based mean effect sizes. Because of the small
feedback intervention on clients in couple number of studies (k = 3) included in the
therapy at a community family counseling analyses, tests of moderators and mediators
clinic. Of the 906 Norwegian individuals were not performed. Given the equivalence
who initially sought couple therapy, 410 of pretreatment scores reported for each
individuals met the inclusion criteria study in the original articles, we did not
(n = 103 in experimental and n = 102 in test for pretreatment score differences in
TAU). The authors of the study reported this analysis. As shown in Table 10.1, when
an effect size of d = 0.50 when comparing mean posttreatment ORS scores of the
the posttreatment ORS scores. The authors feedback group were compared to those
reported posttreatment outcome classifi- of TAU, the combined effect size was
cation (based on the notion of clinical sig- g = 0.48, p < 0.001, 95% CI [0.31, 0.65].
nificance) of couples at posttreatment. The The r equivalent of 0.23 favored the feed-
reported n and percentage of outcome clas- back group, suggesting that the average
sification were based on couples where both client in the feedback group was better off
individuals in the couple met the same out- than approximately 68% of those in TAU.
come classification. Based on these inclu- It should be noted that the type of effect
sion criteria, 66% of couples in the feedback size we report here is different from the
group and 50% of couples in the TAU were effect sizes reported in the original article
included in the analyses. The outcome by Reese and colleagues (2003) in which
classifications at the individual level were the effect sizes were based on the group dif-
not reported. Of those included in the ferences in pretreatment to posttreatment
analysis, the authors reported 51% of cou- change.
ples in the feedback condition achiev- As discussed earlier, the methods of
ing either clinically significant change or classifying treatment outcome at posttreat-
reliable change, while 2% deteriorated. In ment were different between the Reese
contrast, 23% of couples in the TAU group et al. study and the Anker et al. study (i.e.,
reached either clinically significant change outcome classification based on individu-
or reliable change and 4% experienced als’ response or couples as a unit). However,
deterioration. Anker and colleagues (2009) given the consistency in the classification
reported the percentages of couples identi- systems within studies, we calculated com-
fied as being at risk at the third session. bined effect sizes for differential outcome
Among those in the feedback condition, classification between the feedback and
54% of couples were identified as being at TAU groups. When the odds of reliable
risk, while 75% of couples in TAU were improvement over the odds of not achiev-
classified as the same. These results indicate ing reliable improvement were compared
that 54% of couples in the feedback group across groups, the results indicated that
and 75% of couples in TAU were below those in the feedback group had 3.5 times
the 50th percentile mark of expected prog- higher odds of experiencing reliable change,
ress for client progress based on individual OR = 3.52, p < 0.001, 95% CI [2.08, 5.96],

r = 0.32, while having less than half the As Anker et al. (2009) pointed out, the dif-
odds of experiencing deterioration, OR = ferential rates of at risk cases suggest that
0.44, p =0.149, 95% CI[0.14, 1.35], r = the PCOMS may be effective in preventing
−0.22. The summary of these effect sizes poor outcome. Partial support for the social
in comparison to those found in the OQ validity of the feedback effect was suggested
system are presented in Table 10.1. in the authors’ report that, among those
One aspect of the above studies based who responded to 6-month follow-up (149
on the PCOMS system is worth noting couples out of 205 couples), the feedback
in terms of comparative conclusions and group had a lower rate of separation or
implications for clinical practice. First, the divorce (18.4%) than TAU (34.2%), which
rates of “at-risk” cases reported in the three indicates the couples in TAU had approxi-
PCOMS studies are considerably higher mately 1.9 times higher probability of
(36% to 75%) than studies based on separation or divorce (relative risk = 1.86)
the OQ system (11%–33%; Shimokawa than those in the feedback condition. These
et al., 2010). Although Anker et al. (2009) same outcomes, however, indicate that,
reported a higher percentage of at-risk cou- despite the low deterioration rate at post-
ples responding favorably to treatment than treatment (2% in feedback and 4% in
those in TAU (29% vs. 9%), the meaning TAU), a substantial number of couples
and clinical implication of this classifica- experienced separation or divorce rela-
tion seem unclear because the majority tively shortly after terminating treatment.
of cases were identified as being at risk. Although the occurrence of separation or
Table 10.1 Effect Sizes a of Client Feedback in Comparison with TAU (Efficacy Analysis)
Feedback k Posttreatment score Reliable improvement Deterioration
system
Hedges’ g r OR r OR rb
[95% CI] [95% CI] [95% CI]
OQ System
NOT Fb 4 0.53∗∗∗ 0.25 2.55∗∗∗ 0.23 0.44∗ −0.21
[0.28, 0.78] [1.64, 3.98] [0.23, 0.85]
NOT P/ 3c 0.55∗∗∗ 0.25 2.87∗∗∗ 0.27 0.68 −0.10
T Fb [0.36, 0.73] [1.93, 4.27] [0.42, 1.13]
CST Fb 3d 0.70∗∗∗ 0.33 3.85∗∗∗ 0.34 0.23∗∗∗ −0.37
[0.52, 0.88] [2.65, 5.60] [0.12, 0.44]
PCOMS
Feedback 3 0.48∗∗∗ 0.23 3.52∗∗∗ 0.32 0.44∗ −0.22
[0.31, 0.65] [2.08, 5.96] [0.14, 1.35]
∗
p < 0.05. ∗∗p < 0.01. ∗∗∗p < 0.001.
Note: k = number of studies; r = correlation r; CI = confidence interval; NOT Fb = not-on-track clients whose therapists received client
progress feedback; NOT P/T Fb = not-on-track clients where both clients and therapists received client progress feedback. CST Fb =
not-on-track clients whose therapists received client progress feedback and Clinical Support Tools feedback.
a
Effect sizes (Hedges’ g and OR) of OQ system-based feedback interventions were meta- and mega-analytically calculated and reported in
Shimokawa. et al. (2010).
b
Negative correlations indicate greater effect in reducing treatment failure at termination.
c
Original data from three studies employing the P/T Fb groups were aggregated and compared with the aggregated TAU data from four
studies, using a mega-analytic approach.
d
Original data from three studies employing the CST Fb groups were aggregated and compared with the aggregated TAU data from four
studies, using a mega-analytic approach.
divorce alone should not be used to assume diagnosis or comorbid conditions (rather
the quality of treatment outcome (some than being disorder specific); (b) random
couples sought treatment to “[end] their assignment of clients to experimental con-
relationship in the best possible way” or to ditions (various feedback interventions)
“seek clarification regarding whether the and treatment-as-usual conditions (no
relationship should continue”; Anker et al., feedback) was made in 4 of the 6 studies,
2009, p. 695), the discrepancy between the while reasonable measures were taken in
proportion of clients classified as deterio- two studies to ensure equivalence in experi-
ration cases and the actual occurrences of mental and control conditions at pre-
separation/divorce raises a general question treatment; (c) psychotherapists provided a
about the suitability of the clinically sig- variety of theoretically guided treatments,
nificant classification in addressing this with most adhering to cognitive-behavioral
dimension of treatment outcome. and eclectic orientations and fewer repre-
Results Based on the OQ System. In the senting psychodynamic and experiential
most recent meta-analytic review of the orientations; (d) a variety of therapist expe-
OQ system, Shimokawa et al. (2010) rean- rience—postgraduate therapists and gradu-
alyzed the combined data set (N = 6,151) ate students each accounted for about 50%
from all six OQ feedback studies published of clients seen; (e) therapists saw both
to date (Harmon et al., 2007; Hawkins, experimental (feedback) and no-feedback
Lambert, Vermeersch, Slade, & Tuttle, cases, thus limiting the likelihood that out-
2004; Lambert, Whipple, et al., 2001; come differences between conditions could
Lambert, Whipple, Vermeersch, et al., be due to therapist effects; (f ) the outcome
2002; Slade, Lambert, Harmon, Smart, & measure as well as the methodology rules/
Bailey, 2008; Whipple et al., 2003). Each standards for identifying signal-alarm
of the studies required about 1 year of daily clients (failing cases) remained constant;
data collection and evaluated the effects (g) the length of therapy (dosage) was
of providing feedback about each client’s determined by client and therapist rather
improvement through the use of progress than by research design or arbitrary insur-
graphs and warnings about clients who ance limits; and (h) client characteristics
were not demonstrating expected treatment such as gender, age, and ethnicity were
responses (signal-alarm cases). A primary generally similar across studies and came
question in each of these studies was: Does from the same university counseling center,
feedback to psychotherapists about their with the exception of Hawkins et al. study
clients’ progress improve psychotherapy (2004).
outcomes compared with progress feedback The meta-analysis involved both intent-
only, and treatment without feedback (e.g., to-treat (ITT) and efficacy analyses on the
treatment as usual)? The hypothesis in each effects of various feedback interventions in
of these studies was: Clients identified relation to TAU. These two distinct sets
as signal-alarm cases (those predicted to of analyses were performed to investigate
have a poor final treatment response) whose the amount of effects expected for each
therapists received feedback will show feedback intervention based on treatment
better outcomes than similar clients whose assignment alone, including clients who
therapists did not receive feedback. left treatments before the effects of feed-
The six studies shared many design and back interventions could be experienced or
methodological features: (a) consecutive measured (ITT analysis) and the amount
cases seen in routine care regardless of client of effects expected among clients who were

beneficiaries of treatments (efficacy analy- feedback, 20% deteriorated while 22%
sis). Furthermore, we investigated the clinically significantly improved. When the
incremental benefits of two newer feedback odds of deterioration and clinically signifi-
interventions: provision of formal prog- cant improvement were compared, results
ress feedback directly to both clients and indicated that those in the feedback group
therapists (patient/therapist feedback; P/F had less than half the odds of experienc-
Fb), and provision of formal progress feeding deterioration (OR = 0.44, p < 0.05,
back to therapists combined with Clinical 95% CI[0.23, 0.85], r = −.21), while
Support Tools feedback (CST Fb). having approximately 2.6 times higher
Because the PCOMS studies resembled odds of experiencing reliable improvement
the efficacy analyses, we present here the (OR = 2.55, p < 0.001, 95% CI[1.64,
results of the efficacy analyses. In these 3.98], r = 0.23).
analyses, only those clients who received Effects of Patient/Therapist Feedback (P/T
and completed the treatments were com- Fb) on at-risk Clients. The effect size
pared to treatment as usual. Based on these of posttreatment OQ score averaged g =
inclusion criteria, the following percent- 0.55, p < 0.001, 95% CI[0.36, 0.73],
age of clients out of those initially assigned equivalent of r = 0.25—effects very similar
to treatment conditions were included to that of the therapist-only feedback
in the meta-analyses by Shimokawa and group. However, direct feedback to both
colleagues (2010): not-on-track feedback psychotherapists and clients appeared to
(61.6%); not-on-track feedback to thera- have had polarizing effects, resulting in
pists and patients, (79.7%); and not-on- deterioration rates and odds comparable
track feedback plus the use of Clinical to treatment-as-usual clients. The rates
Support Tools, (52.2%). It should be noted, of deterioration and clinically significant
as expected in any comparison between improvement when both participants
efficacy analysis and ITT analysis, the ITT received feedback were 15% and 45%,
effect sizes were generally smaller than those respectively. The results suggest that clients
of efficacy analyses. The summary of effect who received feedback along with their
sizes are presented in Table 10.1. therapist had approximately 0.7 times the
Effects of OQ Progress Feedback (Fb) on odds of deterioration, OR = 0.68, p = 0.134,
at-risk Clients. When the at-risk clients 95% CI[0.42, 1.13], r = −0.10, while
whose therapists received feedback were having approximately three times higher
compared with not-on-track clients whose odds of achieving clinically significant
therapist did not receive feedback, the effect improvement, OR = 2.87, p < 0.001, 95%
size for posttreatment OQ score difference CI[1.93, 4.27], r = 0.27. These results
averaged g = 0.53, p < 0.001, 95% CI[0.28, suggest that, although the average client
0.78], equivalent of r = 0.25. These results who received feedback along with his
suggest that the average client whose thera- or her therapist was better off than 71% of
pist received feedback was better off than clients in treatment as usual, there may
approximately 70% of clients in the have been moderators that facilitated out-
no-feedback condition (routine care). In come enhancement in some clients while
terms of the clinical significance at termi- failing to prevent, or possibly contributing
nation, 9% of those receiving feedback to, outcome worsening.
deteriorated while 38% achieved clinically Effects of Clinical Support Tools Feedback
significant improvement. In contrast, among (CST Fb) on at-risk Clients. When the
clients whose therapists did not receive outcome of clients whose therapist received
the Clinical Support Tool feedback was As can be seen, reduction of deterioration
compared with that of the treatment-as- and increases in positive outcomes are
usual clients, the effect size for the differ- rather dramatic in relation to treatment
ence in mean posttreatment OQ scores was as usual even though the same therapists
g = 0.70, p < 0.001, 95% CI[0.52, 0.88], offered both interventions.
r = 0.33. These results indicate that the
average client in the Clinical Support Tool Limitations of the Research
feedback group, who stay in treatment Major limitations of feedback research are
to experience the benefit of this interven- the small number of studies evaluating effec-
tion, are better off than 76% of clients in tiveness, the limited number of researchers
treatment as usual. The rates of deteriora- responsible for the findings reviewed here,
tion and clinically significant improvement and the sole reliance on self-report measures.
among those receiving Clinical Support It is likely that future research will be done
Tools were 6% and 53%, respectively. The across a wider range of treatment settings
results suggest that clients whose thera- and patient populations, thus illuminating
pists used Clinical Support Tools have less the limits of these procedures and clarifying
than a fourth the odds of deterioration, the factors that maximize patient gains.
OR = 0.23, p < 0.001, 95% CI[0.12, 0.44], The research reviewed here utilized two
r = −0.37, while having approximately self-report measures of improvement and
3.9 times higher odds of achieving clini- therefore provides only one view of the
cally significant improvement, OR = 3.85, impact of therapy. Decisions regarding the
p < 0.001, 95% CI [2.65, 5.60]. continued provision of treatment, modifi-
The above findings from the OQ-based cation of ongoing treatment, and the like,
feedback studies indicate that three forms cannot be made on the basis of a single
of feedback interventions in the OQ system questionnaire or independently from clini-
are effective in enhancing the treatment cal judgment.
effects of clients who are at risk of leaving
therapy worse off than when they came in. Therapeutic Practices
Table 10.2 displays clinically significant These two meta-analyses demonstrate the
change and reliable change in order to effectiveness of two well-developed feed-
make the impact of feedback more clear. back systems. In this closing section, we
Table 10.2 Percent of Not-on-Track (Signal-Alarm) Cases Meeting Criteria for Clinically Significant
Change at Termination Summed across Six Studies (Efficacy Sample)
Outcome classification CST Fba P/T Fbb Fbc TAU d
n (%) n (%) n (%) n (%)

Deteriorated e 12 (5.5%) 26 (14.7%) 24 (9.1%) 64 (20.1%)
No change 91 (41.9%) 71 (40.1%) 140 (53.2%) 183 (57.5%)
Reliable/Clinically significant change f
114 (52.5%) 80 (45.2%) 99 (37.6%) 71 (22.3%)
a
CST Fb = patients who were not on track and whose therapist received feedback and used clinical support tools.
b
P/T Fb = patients who were not on track and both patients and their therapist received feedback.
c
Fb = patients who were not on track and whose therapist received feedback.
d
TAU = patients who were not on track and whose therapist was not given feedback.
e
Worsened by at least 14 points on the OQ from pretreatment to posttreatment.
f
Improved by at least 14 points on the OQ or improved and passed the cutoff between dysfunctional and functional populations.

offer practice suggestions based on our inconclusive results about additive
findings. effectiveness of direct client feedback.
• Routine use of a feedback system In addition, even when progress feedback
to augment clinical decision making— was provided only to clinicians, they
a “lab test” to be used by clinicians, were left free to discuss progress reports
rather than a replacement for the with clients or to problem-solve on their
clinician’s judgment. Feedback systems own. We know that clinicians received
can be used with adults, adolescents, feedback, but we do not know if or how
and children and in individual as well it was provided to clients.
as couples therapy. • Employ real-time client feedback to
• Beware of those situations in which compensate for therapist’s limited ability
clients feel it may be in their interest to to accurately detect client worsening in
understate (or overstate) their problems psychotherapy. Despite considerable
and produce inaccurate ratings on evidence that psychotherapists are not
feedback systems. The systems are alert to treatment failure (e.g., Hannan
predicated on accurate self-reporting of et al., 2005; Hatfield, et al., 2010), and
levels of disturbance and corresponding strong evidence that clinical judgments
changes. are usually inferior to actuarial methods
• Supplement with clinical support (Meehl, 1954), therapists’ confidence in
tools. As suggested by the general their clinical judgment stands as a barrier
literature on feedback and the evidence to implementation of monitoring and
presented here, problem-solving and feedback systems.
decision-enhancement tools prove helpful • Consider using electronic versions
to clinicians and, most importantly, of feedback systems that expedite and
clients. Brief assessment of the therapeutic ease practical difficulties. Adding
alliance and suggestions for its monitoring measures to busy practices
modification, along with assessment of can also be a barrier to implementation.
outside forces such as life events and the Fortunately, the brevity of the PCOMs
strength of social supports, can make a and the recent software for the OQ can
substantial impact on routine care for provide instantaneous feedback to
individuals whose treatment response clinicians. The electronic PCOMs takes
is in doubt. only a few minutes in sessions, while if
• For some patients, it is preferable to the client takes the OQ immediately
provide feedback to both patient and prior to the scheduled psychotherapy
therapist. As yet we are uncertain of the session, electronic feedback is available
necessity of sharing progress feedback to the therapist prior to beginning that
directly with clients. In the PCOMS session.
system, progress and relationship
information is gathered within session References
and discussed routinely. Since both are An asterisk (∗) indicates studies included in
a part of the session, either may account the meta-analysis.
∗
Anker, M. G., Duncan, B. L., & Sparks, J. A.
for the therapeutic effects that have (2009). Using client feedback to improve couple
occurred. In contrast, the OQ system therapy outcomes: A randomized clinical trial in
has examined therapist feedback and a naturalistic treatment setting. Journal of
direct client feedback separately with Consulting & Clinical Psychology, 77, 693–704.
Bandura, A. (1997). Social cognitive theory of self Hannan, C., Lambert, M. J., Harmon, C.,
regulation. Organizational Behavior and Human Nielsen, S. L., Smart, D. W., Shimokawa, K.,
Decision Processes, 50, 248–87. et al. (2005). A lab test and algorithms for iden-
Barkham, M., Margison, F., Leach, C., Lucock, M., tifying clients at risk for treatment failure. Journal
Mellor-Clark, J., Evans, C., et al. (2001). Service of Clinical Psychology: In Session, 61, 155–63.
profiling and outcomes benchmarking using the Hansen, N. B., Lambert, M. J., & Forman, E. V.
CORE_OM: Toward practice-based evidence in (2002). The psychotherapy dose-response effect
the psychological therapies. Journal of Consulting and its implications for treatment delivery
and Clinical Psychology, 69, 184–96. services. Clinical Psychology: Science and Practice,
Beckstead, D. J., Hatch, A. L., Lambert, M. J., 9, 329–43.
∗
Eggett, D. L., Goates, M. K., & Vermeersch, D. A. Harmon, S. C., Lambert, M. J., Smart, D. W.,
(2003). Clinical significance of the Outcome Hawkins, E. J., Nielsen, S. L., Slade, K., et al.
Questionnaire (OQ-45.2). The Behavior Analyst (2007). Enhancing outcome for potential treat-
Today, 4, 79–90. ment failures: Therapist/client feedback and clin-
Berking, M., Orth, U., & Lutz, W. (2006). Wie ical support tools. Psychotherapy Research, 17,
effekiv sind systematische Ruckmeldungen des 379–92.
therapieverlaufs an den therapeuten? Eine Hatfield, D., McCullough, L., Plucinski, A., &
empirishe studie in einem stationar-verhalten- Krieger, K. (2010). Do we know when our cli-
stherapeutischen setting, Zeitchrift fur Klinishe ents get worse? An investigation of therapists’
Psychologie und Psychotherapie, 35, 21–29. ability to detect negative client change. Clinical
Bludworth, J., Tracey, T. J. G., & Glidden-Tracey, C. Psychology & Psychotherapy, 17, 25–32.
∗
(2010). The bi-level structure of the Outcome Hawkins, E. J., Lambert, M. J., Vermeersch, D. A.,
Questionnaire-45. Psychological Assessment, 22, Slade, K., & Tuttle, K. (2004). The effects of
350–355. providing patient progress information to thera-
Bordin, E. S. (1979). The generalizability of the pists and patients. Psychotherapy Research, 14,
psychoanalytic concept of the working alliance. 308–27.
Psychotherapy, 16, 252–60. Hedges, L. V., & Olkin, I. (1985). Statistical meth-
Brodey, B. B., Cuffel, B., McCulloch, J., Tani, S., ods for meta-analysis. New York: Academic Press.
Maruish, M., Brodey, I., et al. (2005). The Jacobson, N. S., & Truax, P. (1991). Clinical signifi-
acceptability and effectiveness of patient-reported cance: A statistical approach to defining mean-
assessments and feedback in managed behavioral ingful change in psychotherapy research. Journal
healthcare setting. The American Journal of of Consulting and Clinical Psychology, 59, 12–19.
Managed Care, 11, 774–80. Kluger, A. N., & DeNisi, A. (1996). The effects
de Jong, K., Nugter, M. A., Polak, M. G., of feedback interventions on performance:
Wagenborg, J. E. A., Spinhoven, P., & Heiser, W. J. A historical review, a meta-analysis, and a prelimi-
(2007). The Outcome Questionnaire (OQ-45) nary feedback intervention theory. Psychological
in a Dutch population: A cross-cultural valida- Bulletin, 119, 254–84.
tion. Clinical Psychology & Psychotherapy, 14 , Knaup, C., Koesters, M., Schoefer, D., Becker, T.,
288–401. & Puschner, B. (2009). Effect of feedback of
Duncan, B. L., & Miller, S. D. (2008). The Outcome treatment outcome in specialist mental health-
and Session Rating Scales: The revised administra- care: Meta-analysis. The British Journal of
tion and scoring manual, including the Child Psychiatry, 195, 15–22.
Outcome Rating Scale. Chicago: Institute for the Kordy, H., Hannöver, W., & Richard, M. (2001).
Study of Therapeutic Change. Computer-assisted feedback-driven quality
Ellsworth, J. R., Lambert, M. J., & Johnson, J. (2006). management for psychotherapy: The Stuttgart-
A comparison of the Outcome Questionnaire-45 Heidelberg model. Journal of Consulting and
and Outcome Questionnaire-30 in classification Clinical Psychology, 69, 173–83.
and prediction of treatment outcome. Clinical Kraus D. R., & Horan, F. P. (1997). Outcomes
Psychology and Psychotherapy, 13, 380–91. roadblocks: Problems and solutions. Behavioral
Gaston, L. (1990). The concept of the alliance Health Management, 17, 22–26.
and its role in psychotherapy: Theoretical and Lambert, M. J. (1983). Introduction to assessment
empirical considerations. Psychotherapy, 27, of psychotherapy outcome: Historical perspective
143–53. and current issues. In M. J. Lambert,

E. R. Christensen, & S. S. DeJulio (Eds), The development and applied implications. Behavioral
assessment of psychotherapy outcome. New York: Science, 32, 161–78.
Wiley. Lueger, R. J., Howard, K. I., Martinovich Z.,
Lambert, M. J., Bergin, A. E., & Collins, J. L. Lutz, W., Anderson, E. E., & Grissom, G.
(1977). Therapist induced deterioration in (2001). Assessing treatment progress of individ-
psychotherapy patients. In A. S. Gurman, & ual clients using expected treatment response
A. M. Razin (Eds.), Effective psychotherapy: models. Journal of Consulting and Clinical
A handbook of research (pp. 452–81). New York: Psychology, 69, 150–58.
Pergamon Press. Lunnen, K. M., & Ogles, B. M. (1998). A multi-
Lambert, M. J., & Burlingame, G. M. (2007). perspective, multivariable evaluation of reliable
Measuring outcomes in the State of Utah: change. Journal of Consulting and Clinical
Practice-based evidence. Behavioral Healthcare, Psychology, 66, 400–410.
27, 16–20. Lutz, W., Lambert, M. J., Harmon, S. C.,
Lambert, M. J., Morton, J. J., Hatfield, D., Harmon, Tschitsaz, A., Schurch, E., & Stulz, N. (2006).
C., Hamilton, S., Reid, R. C., et al. (2004). The probability of treatment success, failure
Administration and scoring manual for the and duration- what can be learned from empiri-
Outcome Questionnaire -45. Salt Lake City, cal data to support decision making in clinical
UT: OQ Measures. practice? Clinical Psychology & Psychotherapy, 13,
Lambert, M. J., & Ogles, B. M. (2004). The 223–32.
efficacy and effectiveness of psychotherapy. Meehl, P. E. (1954). Clinical versus statistical pre-
In M. J. Lambert (Ed.), Bergin and Garfield’s diction. Minneapolis: University of Minnesota
handbook of psychotherapy and behavior change Press.
(5th ed., pp. 139–93). New York: Wiley. Miller, S. D., & Duncan, B. L. (2004). The Outcome
Lambert, M. J., Whipple, J. L., Bishop, M. J., and Session Rating Scales: Administration and
Vermeersch, D. A., Gray, G. V., & Finch, A. E. scoring manual. Chicago: Institute for the Study
(2002). Comparison of empirically derived and of Therapeutic Change.
rationally derived methods for identifying clients Miller, S. D., Duncan, B. L., Brown, J., Sparks, J. A.,
at risk for treatment failure. Clinical Psychology & Claud, D. A. (2003). The outcome rating
and Psychotherapy, 9, 149–64. scale: A preliminary study of the reliability, valid-
Lambert, M. J., Whipple, J. L., Harmon, C., ity, and feasibility of a brief visual analog mea-
Shimokawa, K., Slade, K., & Christofferson, C. sure. Journal of Brief Therapy, 2, 91–100.
(2004). Clinical support tools manual. Provo, UT: Miller, S. D., Duncan, B. L., Sorrell, R., &
Department of Psychology, Brigham Young Brown, G. S. (2005). The Partners for Change
University,. Outcome System. Journal of Clinical Psychology:
∗
Lambert, M. J., Whipple, J. L., Smart, D. W., In Session, 61, 199–208.
Vermeersch, D. A., Nielsen, S. L., & National Advisory Mental Health Council. (2001).
Hawkins, E. J. (2001). The effects of provid- Blueprint for change: Research on child and adoles-
ing therapists with feedback on client progress cent mental health. A report by the National
during psychotherapy: Are outcomes enhanced? Advisory Mental Health Council’s Workgroup on
Psychotherapy Research, 11, 49–68. Child and Adolescent Mental Health Intervention
∗
Lambert, M. J., Whipple, J. L., Vermeersch, D. A., Development and Deployment. Bethesda, MD:
Smart, D. W., Hawkins, E. J., Nielsen, S. L., National Institutes of Health/National Institute
et al. (2002). Enhancing psychotherapy out- of Mental Health.
comes via providing feedback on client prog- Percevic, R., Lambert, M. J., & Kordy, H. (2006).
ress: A replication. Clinical Psychology and What is the predictive value of responses to
Psychotherapy, 9, 91–103. psychotherapy for its future course? Empirical
Lo Coco, G., Chiappelli, M., Bensi, L., Gullo, S., explorations and consequences for outcome
Prestano, C., & Lambert, M. J. (2008). The facto- monitoring. Psychotherapy Research, 16, 364–73.
∗
rial structure of the Outcome Questionnaire-45: Reese, R. J., Norsworthy, L., & Rowlands, S.
A study with an Italian sample. Clinical Psychology (2009). Does a continuous feedback system
& Psychotherapy, 15, 418–23. improve psychotherapy outcome? Psychotherapy:
Lord, R. G., & Hanges, P. J. (1987). A control system Theory, Research, Practice, & Training, 46,
model of organizational motivation: Theoretical 418–31.
Riemer, M., & Bickman, L. (2004) The Vermeersch, D. A., Lambert, M. J., &
Contextualized Feedback Intervention Theory: Burlingame, G. M. (2000). Outcome Ques-
A theory of guided behavior change. Unpublished tionnaire: Item sensitivity to change. Journal of
manuscript. Personality Assessment, 74, 242–61.
Riemer, M., Rosof-Williams, J., & Bickman, L. Vermeersch, D. A., Whipple, J. L., Lambert, M. J.,
(2005). Theories related to changing clinician Hawkins, E. J., Burchfield, C. M., & Okiishi, J. C.
practice. Child Adolescent Psychiatric Clinics of (2004). Outcome Questionnaire: Is it sensitive
North America, 14, 241–54. to changes in counseling center clients? Journal
Ringel, J. S., & Sturm, R. (2001). National esti- of Counseling Psychology, 51, 38–49.
mates of mental health utilization and expendi- Walfish, S., McAlister, B., O’Donnell, P., &
tures for children in 1998. Journal of Behavioral Lambert, M. J. (2010). Are all therapists from Lake
Health Services & Research, 28, 319–33. Wobegon?: An investigation of self-assessment bias in
Safran, J. D., Muran, J. C., Samstang, L. W., & mental health providers. Unpublished manuscript.
Winston, A. (2005). Evaluating alliance-focussed Wampold, B. (2008). The great psychotherapy debate
intervention for potential treatment failures: (2nd.ed.). Mahwah, NJ: Erlbaum.
A feasibility and descriptive analysis. Psycho- Warren, J. S., Nelson, P. L., Mondragon, S. A.,
therapy: Theory, Research, Practice, & Training, Baldwin, S. A., & Burlingame, G. M. (2010).
42, 512–531. Youth psychotherapy change trajectories & out-
Sapyta, J., Riemer, M., & Bickman, L. (2005). come in usual care: Community mental health
Feedback to clinicians: Theory, research, and versus managed care. Journal of Consulting &
practice. Journal of Clinical Psychology, 62, Clinical Psychology, 78, 144–55.
145–53. Weiss, B., Catron, T., Harris, V., & Phung, T. M.
Shadish, W. R., Matt, G. E., Navarro, A. M., (1999). The effectiveness of traditional child
Siegle, G., Crits-Christoph, P., Hazelrigg, M. D., psychotherapy. Journal of Consulting and Clinical
et al. (1997). Evidence that therapy works Psychology, 67, 82–94.
in clinically representative conditions. Journal Weiss, B., Catron, T., Harris, V., & Phung, T. M.
of Consulting and Clinical Psychology, 65, (1999). The effectiveness of traditional child
355–65. psychotherapy. Journal of Consulting and Clinical
Shimokawa, K., Lambert, M. J., & Smart, D. W. Psychology, 67, 82–94.
Enhancing treatment outcome of patients at risk Weisz, J. R. (2004). Psychotherapy for children and
of treatment failure: Meta-analytic and mega- adolescents: Evidence-based treatments and case
analytic review of a psychotherapy quality assur- examples. New York: Cambridge University
ance system. (2010). Journal of Consulting & Press.
Clinical Psychology, 78, 298–311. Weisz, J. R., Donenberg, G. R., Han, S. S., &
∗
Slade, K., Lambert, M. J., Harmon, S. C., Weiss, B. (1995). Bridging the gap between
Smart, D. W., & Bailey, R. (2008). Improving laboratory and clinic in child and adolescent psy-
psychotherapy outcome: The use of immediate chotherapy. Journal of Consulting and Clinical
electronic feedback and revised clinical support Psychology, 63, 688–701.
tools. Clinical Psychology & Psychotherapy, 15, Whipple, J. L., Lambert, M. J., Vermeersch, D. A.,
287–303. Smart, D. W., Nielsen, S. L., & Hawkins, E. J.
Smith, M. L., Glass, G. V., & Miller, T. L. (1980). (2003). Improving the effects of psychotherapy:
The benefits of psychotherapy. Baltimore: The The use of early identification of treatment failure
Johns Hopkins University press. and problem solving strategies in routine prac-
Spielmans, G. I., Masters, K. S., & Lambert, M. J. tice. Journal of Counseling Psychology, 58, 59–68.
(2006). A comparison of rational versus empirical Wolf, F. M. (1986). Meta-analysis: Quantitative meth-
methods in prediction of negative psychotherapy ods for research synthesis. Sage University Paper
outcome. Clinical Psychology & Psychotherapy, series on quantitative applications in the social
13, 202–214. sciences. Beverly Hills, CA: Sage Publications.

C HA P TER
11 Repairing Alliance Ruptures
Jeremy D. Safran, J. Christopher Muran, and Catherine Eubanks-Carter
One of the most consistent findings emerg- Definitions and Measures

ing from psychotherapy research is that the A rupture in the therapeutic alliance can be
quality of the therapeutic alliance is a robust defined as a tension or breakdown in the
predictor of outcome across a range of dif- collaborative relationship between patient
ferent treatment modalities (e.g., Chapters and therapist (Safran & Muran, 2000,
2, 3, and 4) and that, conversely, weak- 2006). Although the term rupture may
ened alliances are correlated with unilateral imply to some a dramatic breakdown in
termination by the patient (e.g., Horvath collaboration, ruptures vary in intensity
& Bedi, 2002; Martin, Garske, & Davis, from relatively minor tensions, which one
2000; Samstag, Batchelder, Muran, Safran, or both of the participants may be only
& Winston, 1998; Tryon & Kane, 1990, vaguely aware of, to major breakdowns
1993, 1995). In the last two decades, there in collaboration, understanding, or com-
has emerged what we have characterized munication. Concepts that are similar or
as a “second generation” of alliance research overlapping with the construct of the
that attempts to clarify the factors leading alliance rupture include empathic failure
to the development of the alliance, as well (Kohut, 1984), therapeutic impasse, and
as those processes involved in repairing misunderstanding event (Rhodes et al.,
ruptures in the alliance when they occur 1994). Alliance ruptures and repairs can
(Safran & Muran, 2006; Safran et al., be measured from patient, therapist, and
2002). In this chapter, we provide a review observer perspectives. They can focus
of this research and meta-analyses of two on rupture repair events that take place
different types of relevant studies. The first either within session or over the course of
meta- analysis examines the association treatment.
between the presence of rupture repair epi-
sodes and treatment outcome. The second Patient Self-Report of
examines the impact of rupture resolution Within-Session Ruptures
training or supervision on patient outcome One method of identifying alliance rup-
by assessing patient change from therapy tures and repairs involves obtaining patient
intake to termination. We will also briefly and therapist reports of shift in quality of
review the task analytic studies investigat- the alliance, or perception of alliance rup-
ing processes associated with alliance rupture and degree of resolution within ses-
ture repair. sion, using session impact questionnaires.
224
For example, in a study comparing the (CALPAS; Marmar, Weiss, & Gaston, 1989)
efficacy of brief relational therapy, cognitive that were at least as large as the mean stan-
behavioral therapy, and short-term dynamic dard deviation of alliance scores across the
therapy with personality-disordered patients sample. Of the patients with at least three
(Muran et al., 2005), patients completed alliance assessments, rupture repair sequences
postsession questionnaires (PSQ; Muran, occurred in 56% of the cases. For another
Safran, Samstag, & Winston, 1992), which example, Stevens and associates (2007)
included self-report measures of the alli- developed criteria for identifying rupture
ance (12-item Working Alliance Inventory, repair sequences from fluctuations in WAI
WAI; Horvath & Greenberg, 1989; Tracey scores in a sample of 44 patients drawn from
& Kokotovic, 1989), as well as self-report the personality disorder cases. Ruptures were
indices measuring the occurrence of rup- defined as decreases of at least one point on
tures, rupture intensity, and the extent the WAI; ruptures were deemed to be
to which ruptures were resolved. Ruptures resolved if the alliance score rose to within
occurred frequently across the three treat- 0.25 points of the prerupture score in three
ments: in the first six sessions of treatment, to five sessions. Fully 50% of the cases
ruptures were reported by 37% of patients included episodes that met these rupture
and 56% of therapists (Muran et al., 2009). repair criteria.
Ruptures were also found to be significantly
related to outcome. Higher rupture inten- Observer-Based Methods
sity, as reported jointly by patients and Differences between patient and therapist
therapists, was associated with poor out- perspectives of the alliance ruptures raise
come on measures of interpersonal func- the concern that patients may underreport
tioning. Failure to resolve these ruptures ruptures due to a lack of awareness of them
was predictive of dropout. Another study or discomfort with acknowledging them.
(Eames & Roth, 2000) also administered One way to address this problem is to use
the WAI items and the rupture indices from observer-based measures to detect ruptures
the PSQ to 30 patients receiving treatment and resolution processes. An early example
as usual. Therapists reported ruptures more of this is the study of “weakenings” in
often, reporting them in 43% of sessions, the therapeutic alliance (Lansford, 1986).
while patients reported them in 19%. Using audiotapes of six sessions of time-
limited dynamic therapy, raters identified
Patient Fluctuations in Alliance ruptures and repairs and then examined
Measures across Sessions the relationship between occurrence of
Another method of identifying alliance rup- ruptures and treatment outcome. The study
tures and their repairs has been to track fluc- found that the most successful outcomes
tuations in patients’ alliance scores across the were associated with patients and therapists
course of therapy. For example, Strauss and who actively dealt with alliance ruptures
colleagues (2006) sought to identify rupture (Lansford, 1986).
repair episodes in a sample of 30 patients In a study of 151 sessions from five patients
with avoidant and obsessive-compulsive in psychodynamic therapy (Sommerfeld
personality disorders who received up to a et al., 2008), the difference between patient
year of cognitive therapy. They developed self-report of ruptures and observer-based
criteria for rupture and resolution sessions report was directly examined. Patients
by looking for fluctuations in scores on completed a brief version of PSQ after each
the California Psychotherapy Alliance Scale session that included the alliance measure,
s a f r a n , mu r a n , e u b a n k s - c a rt e r 225
self-reports of ruptures and resolution, and reliability with graduate student raters
items tapping into the depth and smooth- (Colli & Lingiardi, 2009). The patient
ness of the session from the Session rupture markers and therapist intervention
Evaluation Questionnaire (SEQ; Stiles, items were largely derived from the Rupture
1980). Patients reported ruptures in 42% Resolution Scale (Samstag, Safran, &
of the sessions. Using transcripts of these Muran, 2004).
same sessions, judges identified confronta- Given that most observer-based meth-
tion and withdrawal ruptures using Harper’s ods for coding ruptures and resolutions
unpublished coding system (1989a, 1989b); rely on the use of transcripts or the use
rupture markers were identified by observ- of highly experienced clinicians as judges
ers in 77% of sessions. There was no signifi- (e.g., Aspland et al., 2008; Bennett et al.,
cant association between the observer and 2006), our research team has sought to
client perspectives. But sessions where develop a coding system that is accessi-
both patient and observer saw a rupture ble to graduate student raters and does
were rated as having greater depth by not require transcription of sessions. The
the patient. As ruptures that are identified Rupture Resolution Rating System (3RS;
by both self- and observer-report are likely Eubanks-Carter, Muran, & Safran, 2009)
ones that are explicitly discussed in the draws on Harper’s (1989a, 1989b) manual
session, this finding suggests that patients for coding confrontation and withdrawal
find therapy more helpful when thera- ruptures, as well as the Rupture Resolution
pists are sensitive to subtle indications of Scale (Samstag, Safran, & Muran, 2004).
ruptures and encourage patients to explore Preliminary findings from the 3RS are
them. consistent with the evidence that alli-
The researchers also found a significant ance ruptures identified through observer-
association between the occurrence of based coding systems are more frequent
ruptures and the appearance of dysfunc- than those identified by patient self-report
tional interpersonal schemas involving the (Mitchell et al., 2010).
therapist, identified using the core con-
flictual relationship theme method (CCRT; Prevalence of Ruptures
Luborsky & Crits-Christoph, 1998). This However measured, ruptures in individ-
finding suggests that when ruptures occur, ual psychotherapy are quite frequent.
dysfunctional interpersonal schemas are Table 11.1 summarizes the frequency of alli-
likely to be active. Thus, ruptures provide ance ruptures and rupture repair sequences
critical opportunities to identify, explore, across eight studies employing either
and change patients’ self-defeating patterns patient report, therapist report, or observer
of thought and behavior. ratings. These studies demonstrate that
Colli and Lingiardi (2009) have devel- (1) patients report ruptures in 19% to 42%
oped an observer-based method that codes of sessions, (2) therapists report them
transcribed sessions for both alliance rup- in 43% to 56% of sessions, and (3) third-
tures and resolutions: the Collaborative party raters observe ruptures anywhere
Interaction Scale (CIS). A strength of from 41% to 100% of sessions. In studies
the CIS is that it assesses both patients’ that examined postsession alliance ratings
and therapists’ positive and negative con- to identify the prevalence of rupture repair
tributions to the therapeutic process. The sequences, patients reported such sequences
CIS has also demonstrated good interrater in 22% to 56% of cases.
Table 11.1 Prevalence of Alliance Ruptures and Rupture Repair Sequences
Study Method Frequency
Colli & Lingiardi (2009) 16 patients, 32 sessions Observer-based Indirect ruptures: 100%
of sessions
Direct ruptures: 43%
of sessions
Eames & Roth (2000) 30 patients, sessions 2-5 In session self-report, 19% of sessions
patient
In session self-report, 43% of sessions
therapist
Mitchell et al. (2010) 9 patients, 24 sessions In-session self-report, 41% of sessions

patient
In-session self-report, 59% of sessions
therapist
Observer-based Withdrawal ruptures: 74%
of sessions
Confrontation ruptures:
44% of sessions
Muran et al. (2009) 128 patients, Sessions 1–6 In-session self-report, 37% of cases
patient
In-session self-report, 56% of cases
therapist
Sommerfeld et al. (2008) 5 patients, 151 sessions In-session self-report, 42% of sessions
patient
Observer-based 77% of sessions
Stevens et al. (2007) 44 patients In-session self-report Rupture repair sequences:
50% of cases
Stiles et al. (2004) 79 patients In-session self-report Rupture repair sequences:
21.5% of cases
Strauss et al. (2006) 25 patients In-session self-report Rupture repair sequences:
56% of cases
Clinical Examples treatment goals, or (3) strains in the

Following Bordin’s (1979) understanding patient–therapist bond. An example of
of the alliance, we find it useful to concep- a disagreement about the goal dimension
tualize ruptures in the alliance as consist- would be a situation in which the patient
ing of (1) disagreements about the tasks begins treatment seeking immediate relief
of therapy, (2) disagreement about the from his or her panic symptoms, but the
therapist believes the goal should be one of a rupture ensues when a therapist attempts
obtaining insight rather than immediate to challenge a patient’s dysfunctional
symptom relief. An example of a disagree- thinking style. In response the therapist
ment about the task dimension would be a shifts to validating his experience rather
situation in which the patient believes that than challenging his perception. A patient
it is important to spend time reviewing and is frustrated by the therapist’s attempt
making sense of his or her history, but the to explore his feelings and asks for more
therapist has a present-focused, pragmatic direct guidance. In response the therapist
orientation. An example of a strain in the shifts to providing direct advice or
bond dimension would be a situation in engaging in collaborative problem solving
which the patient feels patronized or mis- with the patient.
understood by the therapist. 3. Clarifying Misunderstandings at a
These three types of ruptures are, of Surface Level. For example, a therapist
course, not mutually exclusive. For example, notices that her patient seems withdrawn
the patient whose therapist is unwilling to and initiates an exploration of what is
negotiate the tasks or goals of treatment going on in the here and now of their
may feel misunderstood or disrespected. relationship. The patient admits to feeling
Conversely, a patient who feels mistrusting criticized by the therapist. The therapist
of his or her therapist will be more likely to responds in a nondefensive fashion and
disagree with the therapist about a thera- acknowledges that she can see how the
peutic task or goal. patient might have felt criticized by what
Understanding the typical clinical mani- she said.
festations of alliance ruptures leads natu- 4. Exploring Relational Themes Associated
rally to common rupture repair interventions with the Rupture. In some situations, the
on the part of the psychotherapist: process of clarifying factors leading to a
rupture can lead to an exploration of
1. Repeating the Therapeutic Rationale. underlying relational themes. For example,
Outlining the therapeutic rationale at the a patient may experience the therapist’s
beginning of treatment can play an questions about her inner experience as
important role in developing the alliance intrusive. Exploring the meaning and
at the outset. Reiterating the rationale nature of this experience for the patient
throughout treatment can help to repair may reveal that it is related to a more
a strained alliance. For example, the general experience on her part of feeling
therapist can help to repair an alliance intruded upon by others. A patient who
rupture resulting from his or her attempt fails to do his homework assignments in
to make a transference interpretation by cognitive therapy may have a particular
reiterating that exploring parallels between sensitivity to feeling dominated and
the therapeutic relationship and other controlled by others. A patient’s feeling
relationships can help the patient to of being misattuned to by the therapist
become aware of self-defeating patterns. may reflect a narcissistic sensitivity that
2. Changing Task or Goals. In this becomes a major focus of the treatment.
intervention, the therapist responds to 5. Linking the Alliance Rupture to
ruptures resulting from disagreements Common Patterns in a Patient’s Life.
about tasks or goals by modifying In some situations resolving a rupture can
his behaviors in a fashion that feels involve explicitly exploring the link
meaningful to the patient. For example, between the rupture that occurs
in the session. For example, a therapist was published in English in a peer-reviewed
explores similarities between the control journal, and (b) it included a quantifiable
struggles occurring the therapeutic measure of outcome at the beginning and
relationship and the patient’s parallel termination of treatment. To be included
tendency to become involved in controls in the meta-analysis of rupture repair epi-
struggles with others in his or her life. sodes, a study also had to use quantitative
6. New Relational Experience. The criteria to identify patients who experienced
therapist acts in a way that he or she discrete ruptures and rupture repairs or res-
hypothesizes will provide the patient olutions over the course of treatment.
with an important new relational In order to be included in the meta-
experience without explicitly exploring analysis of rupture resolution training and
the underlying meaning of the interaction. supervision, a study also had to constitute
This intervention is particularly important an investigation of therapist training or
when the patient has difficulty exploring supervision focused on improving thera-
the therapeutic relationship in the here pists’ abilities to build and/or maintain
and now. For example, a therapist decides good alliances with their adult patients in
to answer a patient’s request for advice individual, in-person psychotherapy. Many
because she formulates the situation as psychotherapy treatments include atten-
one in which the decision to do so will tion to the alliance; in order to be included
provide a corrective contrast to the in this analysis, the alliance-focused train-
patient’s abandoning mother. ing or supervision had to include a specific
focus on helping therapists to manage alli-
Meta-Analytic Review ance ruptures or problems in the therapeu-
Two sets of meta-analyses were conducted. tic relationship.
The first set of analyses examined the asso- The literature search identified four stud-
ciation between the presence of rupture ies that met the inclusion criteria for the
repair episodes and treatment outcome. The rupture repair analysis. Three of these stud-
second set of analyses examined the impact ies (Stiles et al., 2004; Stevens et al., 2007;
of rupture resolution training or supervi- Strauss et al., 2006) defined rupture repair
sion on patient outcome. episodes based on session-to-session fluc-
tuations in alliance scores, and they exam-
Search Strategy and Inclusion Criteria ined the relation between the presence of
To identify potential studies, we searched these episodes and outcome. A fourth study
the reference sections of several recent (Muran et al., 2009) examined ruptures
reviews of the alliance rupture literature and repairs that occurred within the first
(i.e., Eubanks-Carter, Muran, & Safran, six sessions of treatment for patients with
2010; Eubanks-Carter, Muran, Safran, & Cluster C and Personality Disorder NOS
Hayes, 2011; Safran, Muran, Samstag, & diagnoses, based on patient and therapist
Stevens, 2002). In addition, we conducted self-reports. This study reported findings
a computerized search of the PsycINFO regarding the relation between rupture
database. Using the search terms alliance repair episodes and outcome, namely that
and outcome, and the terms alliance and higher rupture intensity was associated
rupture, a list of 578 journal articles was with poor outcome on measures of inter-
generated on April 15, 2010. These arti- personal functioning (r = −0.35, p < 0.01),
cles were inspected for studies meeting the and rupture repair was predictive of reten-
following inclusion criteria: (a) the study tion in treatment (r = 0.29, p < 0.05).
However, due to the significant method- different patients before and after thera-
ological difference of examining rupture pists received rupture resolution training
repairs within sessions, rather than between or supervision. The first of these was
sessions, this study was excluded from the the Vanderbilt II study conducted by
meta-analysis. Hans Strupp and colleagues (Bein et al.,
The literature search identified nine 2000). The second was conducted by
studies that met the inclusion criteria for Crits-Christoph et al. (2006). One study
the rupture resolution training analysis. (Hilsenroth et al., 2007) did not include
However, one study (Safran et al., 2005) a control group. In order to include all of
was excluded from the meta-analysis due the eligible studies, we chose to first con-
to its markedly different design, which duct a meta-analysis of all eight studies
included unique selection criteria and a using standardized mean gain effect sizes
change in treatment conditions during the comparing pretreatment to posttreatment
course of the study. Specifically, a subset of scores. However, given that pre-post com-
patients in the CBT and dynamic supervi- parisons typically yield very large effect
sion conditions of the Muran et al. (2005) sizes due to their failure to control for con-
study were identified as potential treat- founds such as the passage of time, we also
ment failures based on patient and thera- conducted a meta-analysis of the standard-
pist postsession questionnaire ratings, and ized mean difference scores of the seven
these patients were given the opportunity studies that included control conditions.
to switch midtreatment to one of the other
treatment conditions. Those who agreed to Methodological Considerations
switch were randomly assigned to either Not all studies reported effect sizes, and
rupture resolution supervision condition those that did varied as to the effect size
or the other standard treatment condition statistic used as well as the data on which it
(CBT or dynamic therapy). This study was based (e.g., termination vs. follow-up
found that patients who switched into the data, all outcome measures or a subset of
rupture resolution condition were signifi- outcome measures). In order to achieve
cantly more likely to remain in treatment greater methodological consistency, effect
than those who switched into another treat- sizes were recalculated for all studies. First,
ment condition. standardized mean differences (or, in the
The remaining eight studies in the rupture case of pre-post effect sizes, standardized
resolution training analysis all presented mean gains) were calculated based on
intake and termination data for therapists means and standard deviations or F ratios
who received some form of rupture resolu- provided in the articles or directly from
tion training and/or supervision. Seven of the authors. The standardized mean scores
the eight studies also included a control were then converted into r effect sizes.
condition; however, these control groups When studies reported more than one out-
varied considerably. They included a wait come measure or finding for more than one
list control (Castonguay et al., 2004), an subgroup, effect sizes were calculated for
unsupervised active treatment (Bambling each outcome measure or each subgroup,
et al., 2006), supervised active treatments and then averaged to form one effect size per
(Constantino et al., 2008; Muran et al., study. The meta-analyses were conducted
2005; Newman et al., 2008), and thera- using random-effects models, with each
pists serving as their own controls in stud- effect size weighted by the inverse of its
ies that compared outcomes obtained with variance. Comprehensive Meta-Analysis,
Version 2.0, was used for all analyses from therapists who were not trained in
(Borenstein, Hedges, Higgins, & Rothstein, rupture resolution.
2005). A meta-analysis of the between-group
effect sizes for the seven studies with con-
Results trol conditions yielded a mean weighted
Table 11.2 presents the correlations between effect size of 0.15, z = 2.66, 95% CI (0.04–
the presence of rupture repair episodes and 0.26), p = 0.01. When one outlier study
treatment outcome in three studies. The was removed (Castonguay et al., 2004),
aggregated correlation was 0.24, z = 3.06, the mean weighted effect size was reduced
95% CI (0.09–0.39), p = 0.002, a medium to 0.11, z = 2.24, 95% CI (0.01–0.21),
size effect that indicates that the presence p = 0.03. These results indicate that rup-
of rupture repair episodes was positively ture resolution training/supervision leads
related to good outcome. to small but statistically significant patient
Our second set of meta-analyses exam- improvements relative to treatment by
ined the impact of rupture resolution train- therapists who did not undergo such
ing or supervision on patient outcome in training.
eight published studies. Both pre-post and
group contrast effect sizes were calculated; Moderators and Mediators
the results are presented in Table 11.3. The The meta-analysis examining the relation
mean weighted pre-post r for the rupture between rupture repair and outcome
resolution training studies was 0.65, z = 5.56, included only three studies, which precludes
95% CI (0.46–0.78.), p < 0.001. Given most moderator analyses. Furthermore,
the particularly large effect sizes produced across these three studies, mean weighted
by two studies (Bambling et al., 2006; effect sizes were not significantly heteroge-
Castonguay et al., 2004), the results were neous, Q (2) = 0.99, p = 0.61.
recalculated excluding these studies, yield- For the analysis of pre-post effect sizes of
ing an effect size of 0.52, z = 6.94, 95% CI rupture resolution training, mean weighted
(0.40–0.63), p < 0.001. These results effect sizes across the studies were sig-
provide evidence that rupture resolution nificantly heterogeneous, Q (7) = 203.85,
training/supervision led to significant patient p < 0.001. Potential moderators that might
improvement; however, with a pre-post explain this variability were examined. To
design, we cannot determine whether examine whether effect sizes varied as a
this improvement was greater than what function of patient diagnosis, studies were
patients would experience with treatment divided into two groups: a group of studies
Table 11.2 Correlation between Rupture Repair and Outcome

Study Treatment Patient N Outcome r Lower Upper Z-value p-value
diagnostic measure limit (r) limit (r) (r) (r)
criteria
Stiles CBT and PI Depression 79 BDI, GSI, IIP, 0.19 –0.04 0.39 1.64 0.10
et al. (2004) SAS, Self-esteem
Stevens BRT, CBT, Cluster C or 44 GAS, GSI, IIP, 0.26 −0.03 0.50 1.77 0.08
et al. (2007) and STDP PDNOS TC, WISPI
Strauss CT for PDs AVPD and 25 BDI, SCID II, 0.39 0.03 0.66 2.12 0.03
et al. (2006) OCPD WISPI
2 32
Table 11.3 Rupture Resolution Training/Supervision Effect Sizes

Study Treatment Patient diagnostic Outcome Pre-post r 95% CI Z-value Control group Between- 95% CI Z-value
criteria measure groups r
Bambling et al. PST with process- Major depression BDI 0.89 0.84–0.92 15.62∗∗ PST with no 0.18 0.02–0.33 2.13∗
(2006) focused supervision supervision
(N = 34) (N = 38)
Bambling et al. PST with skills-focused Major depression BDI 0.85 0.79 – 0.89 13.56∗∗
(2006) supervision
(N = 31)
Bein et al. (2000) TLDP (N = 32) Interpersonal GAS, GSI, IDI, 0.30 0.20 –0.39 5.82∗∗ Pretraining 0.09 -0.16 – 0.32 0.70
problems MCMI (n = 32)
Castonguay et al. ICT (N =11) Major depression BDI, GAF, 0.86 0.77 – 0.91 9.94∗∗ Wait list (n = 10) 0.54 0.21 – 0.76 3.02∗∗
(2004) HAM-D
Constantino et al. ICT (N = 11) Major depression BDI, BSI 0.66 0.51– 0.78 6.64∗∗ CT (n = 11) 0.19 −0.22 – 0.54 0.92
(2008)
Crits-Christoph Alliance-fostering Major depression BDI, HAM-D, 0.57 0.43 –0.69 6.72∗∗ Pretraining 0.15 −0.23 – 0.49 0.76
et al. (2006) therapy (N = 14) QOLI (n = 14)
Hilsenroth et al. STPP (N = 33) Major depression MDE criteria, 0.51 0.44 – 0.58 11.51∗∗ N/A
(2007) and BPD BSI, GARF
Muran et al. BRT (N = 33) Cluster C PD or PD GAS, GSI, IIP, 0.38 0.27 – 0.48 6.41∗∗ CBT (n = 29) and 0.02 −0.16 – 0.20 0.19
(2005) NOS TC, WISPI STDP
(n = 22)
Newman et al. CBT with I/EP GAD Assessor severity, 0.69 0.59–0.77 9.86∗∗ CBT with SL 0.14 −0.27 – 0.51 0.65
(2008) (N = 18) daily diary, (n = 3)
HAM-A, PSWQ,
RRAQ, STAI-T
Note: Although the two conditions of the Bambling et al. (2006) study are listed separately in the table, the conditions were combined to create one effect size for use in the meta-analyses.
focused on patients with Axis I disorders studies into the two groups compared in
(depression and anxiety), and a group of the moderator analysis above. The briefer,
studies that targeted patients with Axis II predominantly CBT treatments targeting
disorders or interpersonal problems. We Axis I disorders again yielded a higher mean
observed that these groupings also reflected weighted effect size (r = 0.22, z = 3.18,
treatment length: studies targeting Axis I p = 0.001) than the longer, dynamic
disorders provided between 8 and 20 ses- and relational treatments targeting Axis II
sions of treatment, while studies targeting disorders (r = 0.04, z = 0.57, p = 0.57);
Axis II disorders or interpersonal prob- however, the difference between the two
lems provided 25 or more sessions. With groups failed to reach statistical signifi-
the exception of one 16-session dynamic cance, Q(1) = 3.14, p = 0.08.
treatment targeting major depression
(Crits-Christoph et al., 2006), these group- Task-Analytic Studies
ings also reflected the theoretical orienta- Task analysis is a programmatic research
tion of the treatments administered by the paradigm that employs a combination of
therapists receiving rupture resolution train- qualitative and quantitative research meth-
ing: the briefer, Axis I treatments were cog- odologies. Since most of the task analysis
nitive behavioral, while the longer treatments research on the rupture resolution process
targeting personality and interpersonal is still at the qualitative stage, it cannot be
problems were dynamic and/or relational. reviewed through meta-analytic procedures.
Mean weighted effect sizes were computed Nevertheless, the review of this emerging
for each of these groups, and they differed literature is valuable for the purpose of pro-
significantly, Q(1) = 10.96, p = 0.001, with viding a preliminary view of ways in which
briefer, predominantly cognitive-behavioral the process of rupture resolution may vary
treatments that targeted Axis I disorders in different treatments.
showing more patient improvement from From a series of task-analytic studies in
intake to termination (r = 0.76, z = 6.62, the 1990s (Safran & Muran, 1996; Safran
p < 0.001) than longer dynamic and rela- et al., 1990, 1994), we developed a four-
tional treatments that targeted Axis II disor- part stage model: (1) attending to the
ders or interpersonal problems (r = 0.40, rupture marker, (2) exploring the rup-
z = 5.34, p < 0.001). The smallest pre-post ture experience, (3) exploring the avoid-
effect size found was for the Vanderbilt II ance, and (4) emergence of wish/need. We
study. In some respects, this is not surpris- distinguished between two types of patient
ing given the fact that the study found that communications or behaviors marking
therapists’ skills in some respects actually a rupture—withdrawal and confrontation
deteriorated after training, and that a major- markers. In withdrawal markers, the patient
ity of the therapists had not achieved basic withdraws or partially disengages from the
competence in TLDP (Bein, 2000). therapist, his or her own emotions, or some
In contrast to the pre-post effect sizes, aspect of the therapeutic process. In confron-
the between-groups effect sizes comparing tation ruptures, the patient directly expresses
rupture resolution supervision/training anger, resentment, or dissatisfaction with the
to a control condition were not signifi- therapist or some aspect of the therapy in an
cantly heterogeneous, Q(6) = 7.65, p = 0.27. attempt to control the therapist.
However, to facilitate comparison with We observed that the type of rupture
the pre-post meta-analysis, we examined marker dictated differences in the resolu-
potential moderators by dividing the tion process. The common progression in
the resolution of withdrawal ruptures con- alliance rupture to situations outside of
sists of moving through increasingly clearer therapy and discussing new ways to handle
articulations of discontent to self-assertion, those situations.
in which the need for agency is realized and Bennett et al. (2006) used task analysis
validated by the therapist. The progression to examine rupture resolution in cognitive-
in the resolution of confrontation rup- analytic therapy (CAT: Ryle, 1997) for
tures consists of moving through feelings borderline personality disorder. The task
of anger, to feelings of disappointment analysis was performed using 6 cases - 4
and hurt over having been failed by the with good outcome and 2 with poor.
therapist, to contacting vulnerability and Rupture sessions were selected based on
the wish to be nurtured and taken care of. deviations in scores on the Therapy Expe-
Typical avoidant operations that emerge, rience Questionnaire (TEQ: Ryle, 1995),
regardless of rupture type, concern anxiet- which was completed by the patient after
ies and self-doubts resulting from the fear every session. Bennett et al. found that in
of being too aggressive or too vulnerable good-outcome cases, therapists recognized
associated with the expectation of retalia- and focused attention on the majority of
tion or rejection by the therapist. ruptures, while in poor-outcome cases they
Building on this research, three studies, usually failed to notice or draw attention to
using the task-analytic paradigm have the alliance threat. They also found that
found preliminary evidence of somewhat therapists in good-outcome cases tended to
similar rupture resolution procedures in have a nondefensive stance. However, in
other treatment modalities. Agnew, Harper, contrast to Safran and Muran’s findngs that
Shapiro, and Barkham (1994) tested a therapists in resolved ruptures tend to focus
psychodynamic-interpersonal model of res- on the immediate process and progressive
olution of confrontation ruptures using clarification of the patient’s underlying
one good-outcome case of eight-session needs, Bennett et al.’s therapists in good-
psychodynamic-interpersonal therapy from outcome cases placed greater emphasis on
the Sheffield study of treatment for depres- linking the rupture to a preestablished case
sion. One rupture and one resolution ses- formulation and to the patient’s other
sion were selected based on changes in relationships.
alliance scores measured using the Agnew Aspland, Llewelyn, Hardy, Barkham,
Relationship Measure (ARM; Agnew- and Stiles (2008) used task analysis proce-
Davies, Stiles, Hardy, Barkham, & Shapiro, dures to begin the process of clarifying the
1998), which was completed by the patient way ruptures are successfully resolved in
after each session. Consistent with Safran CBT. They examined ruptures and resolu-
and Muran’s resolution model, Agnew and tion in two good-outcome cases of CBT
colleagues observed that the resolution pro- for depression from the Second Sheffield
cess involves the therapist acknowledging Psychotherapy Project (Shapiro et al., 1994).
the rupture and then exploring the rupture Cases were identified based on changes in
collaboratively with the patient in order alliance scores, measured using the ARM,
to reach a shared understanding. However, following a naturalistic observation (Stiles
whereas Safran and Muran’s model depicts et al., 2004). Two experienced clinicians
resolution as a progression toward clarifi- examined transcripts of the sessions and
cation of the patient’s underlying wish identified confrontation and withdrawal
or need, therapists in Agnew et al.’s study markers. After close examination of rup-
tended to place greater focus on linking the ture and resolution markers, Aspland and
colleagues (2008) observed that most rup- in number and at an early stage of develop-
tures appeared to arise from unvoiced dis- ment. The most well-established research
agreements about the tasks and goals of program in this area (Safran and colleagues)
therapy, which led to negative complemen- has some verification or hypothesis-testing
tary interactions in which the therapist data in support of its model, but even these
focused on the task and the patient with- findings are based on small samples and
drew. Resolution occurred when therapists have not been replicated in multiple sam-
shifted their focus from the therapy task ples or by independent investigators.
to issues that were salient for the patient.
Consistent with the other rupture resolu-
tion models described above, Aspland
In the first edition of Psychotherapy
et al.’s (2008) preliminary model empha-
Relationships That Work (Norcross, 2002),
sized the therapist’s collaborative stance,
we reviewed the existing evidence suggest-
but in contrast to the other resolution
ing that the process of repairing alliance
models, Aspland et al.’s (2008) model did
ruptures may be related to good positive
not include any overt recognition or dis-
outcome. In this chapter, we have reviewed
cussion of the rupture itself. The authors
the growing body of evidence indicating
noted that the lack of explicit discussion of
that repairing ruptures in the alliance
the rupture may have been due to the pre-
is related to positive outcome. We have
dominance of withdrawal ruptures over
also reviewed some of the similarities and
confrontation ruptures in their sample.
differences in principles relevant to alliance
Limitations of the Research rupture resolution in different treatment
There are a number of limitations to the approaches. On the basis of these reviews,
studies included in our meta-analyses. At we describe below research-supported ther-
this point in time, there are a limited apeutic practices:
number of relevant studies. A number • Practitioners should be aware that
of the studies are correlational in nature. patients often have negative feelings about
The studies included were heterogeneous the psychotherapy or the therapeutic
with respect to design, treatment modality relationship that they are reluctant to
tested, treatment length, and client popula- broach for fear of the therapist’s reactions.
tion. Some of the outcome studies included It is thus important for therapists to be
were not randomized clinical trials. The attuned to subtle indications of ruptures
majority of the outcome studies included in the relationship and to take the
evaluated the efficacy of alliance-focused initiative in exploring what is transpiring
treatments (or treatments enhanced with in the relationship when they suspect that
alliance-focused interventions) but did not a rupture has occurred.
directly test the hypothesis that training in • It is probably helpful for patients to
the implementation of an alliance-focused express negative feelings about the therapy
treatment improved therapists’ ability to to the therapist should they emerge or to
work with challenging patients. Finally, assert their perspective on what is going
none of the studies included in the meta- on when it differs from the therapist’s.
analyses investigated the processes through • When this takes place, it is important
which alliance ruptures are resolved. The for therapists to attempt to respond in
task-analytic research programs investigat- an open or nondefensive fashion, and to
ing these processes are at this point limited accept responsibility for their contribution
to the interaction, as opposed to blaming British Journal of Medical Psychology, 67,
the patient for misunderstanding or 155–170.
Agnew-Davies, R., Stiles, W. B., Hardy, G. E.,
distorting.
Barkham, M. & Shapiro, D. A. (1998). Alliance
• It is also important for therapists to structure assessed by the Agnew Relationship
empathize with their patients’ experience Measure (ARM). British Journal of Clinical
and to validate them for broaching a Psychology, 37, 155–172.
potentially divisive topic in session. Aspland, H., Llewelyn, S., Hardy, G. E.,
• In some forms of treatment, the Barkham, M., & Stiles, W. (2008). Alliance
ruptures and rupture resolution in cognitive-
primary intervention may consist of the
behavior therapy: A preliminary task analysis.
therapist changing the tasks or goals of Psychotherapy Research, 18, 699–710.
treatment without necessarily explicitly ∗
Bambling, M., King, R., Raue, P., Schweittzer, R.,
addressing the rupture with the patient. & Lambert, W. (2006). Clinical supervision: Its
• In other forms of treatment, resolving influence on client-rated working alliance and
alliance ruptures may involve more client symptom reduction in the brief treatment
of major depression. Psychotherapy Research, 16,
in-depth exploration of what is transpiring
317–331.
between the therapist and patient as well ∗
Bein, E., Anderson, T., Strupp, H., Henry, W. P.,
as in-depth exploration of the patient’s Schacht, T. E., Binder, J., et al. (2000). The
experience. effects of training in time-limited dynamic psy-
• There is also preliminary evidence to chotherapy: Changes in therapeutic outcome.
suggest that in some approaches it may Psychotherapy Research, 10, 119–132.
Bennett, D., Parry, G., & Ryle, A. (2006). Resolving
be useful for the therapist to explicitly
threats to the therapeutic alliance in cognitive
establish a link between the rupture event analytic therapy of borderline personality disor-
and characteristic interpersonal patterns der: A task analysis. Psychology and Psychotherapy:
in the patient’s life. This evidence should, Theory, Research, and Practice, 79, 395–418.
however, be interpreted cautiously in light Bordin, E. (1979). The generalizability of the psy-
of the growing body of evidence indicating choanalytic concept of the working alliance.
Psychotherapy: Theory, Research, and Practice, 16,
that frequent transference interpretations
252–260.
linking what is taking place in the Borenstein, M., Hedges, L., Higgins J., &
therapeutic relationship to other Rothstein, H. (2005). Comprehensive meta-analysis:
relationships in the patient’s life can exert Version 2, Biostat, Englewood NJ (2005).
∗
negative effects (e.g., Crits-Christoph & Castonguay, L. G., Schut, A. J., Aikins, D.,
Gibbons, 2002; Henry, Strupp, Schacht, Constantino, M. J., Lawrenceau, J.P., Bologh, L.,
& Burns, D. (2004). Integrative cognitive ther-
& Gaston, 1994). The quality of the
apy for depression: A preliminary investigation.
interpretation (as opposed to the quantity) Journal of Psychotherapy Integration, 14, 4–20.
and the collaborative style in which it is Colli, A., Lingiardi, V. (2009). The Collaborative
tendered appear to make the difference Interactions Scale: A new transcript-based
between a positive and negative effect on method for the assessment of the therapeutic
the patient. alliance ruptures and resolutions in psychother-
apy. Psychotherapy Research, 19, 718–734.
∗
Constantino, M. J., Marnell, M. E., Haile, A. J.,
Kanther-Sista, S. N., Wolman, K., Zappert, L.,
References & Arnow, B. A. (2008). Integrative cognitive
An asterisk (∗) indicates studies included in therapy for depression: A randomized pilot com-
the meta-analysis. parison. Psychotherapy: Theory, Research, Practice,
Agnew, R. M., Harper, H., Shapiro, D. A., & Training, 45, 122–134.
∗
Barkham, M. (1994). Resolving a challenge to Crits-Christoph, P., Gibbons, M. B., Crits-
the therapeutic relationship: A single case study. Christoph, K., Narducci, J., Schramberger, M.,
& Gallop, R. (2006). Can therapists be trained Marmar, C. R., Weiss, D. S., & Gaston, L. (1989).
to improve their alliances? A preliminary study Toward the validation of the California
of alliance-fostering psychotherapy. Psychotherapy Therapeutic Alliance Rating System. Psychological
Research, 16, 268–281. Assessment, 1, 46–52.
Eames, V., Roth, A. (2000). Patient attachment ori- Martin, D. J., Garske, J. P., & Davis, M. K. (2000).
entation and the early working alliance: A study Relation of the therapeutic alliance with out-
of patient and therapist reports of alliance qual- come and other variables: A meta-analytic review.
ity and ruptures. Psychotherapy Research, 10, Journal of Counseling and Clinical Psychology, 68,
421–434. 438–450.
Eubanks-Carter, C., Muran, J. C., & Safran, J. D. Mitchell, A., Eubanks-Carter, C., Safran, J. D., &
(2009). Rupture Resolution Rating System Muran, J.C. (2010). Rupture resolution rating
(3RS): Manual. Beth Israel Medical Center. system, version 2: An observer-based measure.
New York: New York. Society for Psychotherapy Research Annual
Eubanks-Carter, C. Muran, J.C., Safran, J.D., & Conference. June, Asilomar, California.
∗
Hayes, J. A. (2011). Interpersonal interventions Muran, J. C., Safran, J. D., Gorman, B. S.,
for maintaining an alliance. In L.M. Horowitz Samstag, L. W., Eubanks-Carter, C., & Winston, A.
& S. Strack (Eds.), Handbook of interpersonal (2009). The relationship of early alliance rup-
psychology: Theory, research, assessment, and thera- tures and their resolution to process and out-
peutic interventions (pp. 519–531). Hoboken, come in three time-limited psychotherapies
NJ: John Wiley & Sons. for personality disorders. Psychotherapy: Theory,
Harper, H. (1989a). Coding guide I: Identification Research, Practice, Training, 46, 233–248.
of confrontation challenges in exploratory therapy. Muran, J. C., Safran, J. D., Samstag, L. W., &
Sheffield, England: University of Sheffield. Winston, A. (2005). Evaluating an alliance-
Harper, H. (1989b). Coding guide II: Identification focused treatment for personality disorders.
of withdrawal challenges in exploratory therapy. Psychotherapy: Theory, Research, Practice, Training,
Sheffield, England: University of Sheffield. 42, 532–545.
∗
Henry, W. P., Schacht, T. E., Strupp, H. H., Newman, M. G., Castonguay, L. G., Borkovec, T. D.,
Butler, S. F., & Binder, J. L. (1993). Effects of Fisher, A. J., & Nordberg, S. S. (2008). Anopen
training in time-limited dynamic psychother- trial of integrative therapy for generalized
apy: Mediators of therapists’ responses to train- anxiety disorder. Psychotherapy: Theory,Research,
ing. Journal of Consulting and Clinical Psychology, Practice, Training, 45,135–147.
61, 441–447. Norcross, J. C. (2002). Empirically supported
∗
Hilsenroth, M., Defife, J., Blake, M., & Cromer, T. therapy relationships. In J. C. Norcross (Ed.),
(2007). The Effects of Borderline Pathology on Psychotherapy relationships that work (pp. 3–16).
Short-Term Psychodynamic Psychotherapy for New York: Oxford University Press.
Depression. Psychotherapy Research, 17, 175–188. Rhodes, R., Hill, C., Thompson, B., & Elliot, R.
Horvath, A. O., & Bedi, R. P. (2002). The alliance. (1994). Client retrospective recall of resolved and
In J. C. Norcross (Ed.), Psychotherapy relationships unresolved misunderstanding events. Counseling
that work. New York: Oxford University Press. Psychology, 41, 473–483.
Horvath, A. O., & Greenberg, L. S. (1989). Ryle, A. (1995). Transference and counter-
Development and validation of the Working transference variations in the course of cognitive-
Alliance Inventory. Journal of Consulting and analytic therapy of two borderline patients: The
Clinical Psychology, 36, 223–233. relation to the diagrammatic reformulation of
Kohut, H. (1984). How does analysis cure? Chicago: self-states. British Journal of Medical Psychology,
University of Chicago Press. 68, 109–124.
Lansford, E. (1986). Weakenings and repairs of the Ryle, A. (1997). Cognitive analytic therapy and
working alliance in short-term psychotherapy. borderline personality disorder: The model and the
Professional Psychology: Research and Practice, 17, method. Chichester, UK: Wiley.
364–366. Safran, J. D., Crocker, P., McMain, S., & Murray, P.
Luborsky, L., Crits-Christoph, P. (1998). Under- (1990). The therapeutic alliance rupture as
standing transference: The core conflictual relation- a therapy event for empirical investigation.
ship theme method (2nd ed.). Washington DC: Psychotherapy: Theory, Research, and Practice, 27,
American Psychological Association. 154–165.
Safran, J. D., & Muran, J. C. (1996). The resolu- postsession evaluation. Psychotherapy Research,
tion of ruptures in the therapeutic alliance. 18, 377–388.
∗
Journal of Counseling and Clinical Psychology, 64, Stevens, C. L., Muran, J. C., Safran, J. D.,
447–458. Gorman, B. S., & Winston, A. (2007). Levels
Safran, J. D., & Muran, J. C. (2000). Negotiating and patterns of the therapeutic alliance in brief
the therapeutic alliance: A relational treatment psychotherapy. American Journal of Psychotherapy,
guide. New York: Guilford Press. 61, 109–129.
Safran, J. D., & Muran, J. C. (2006). Has the Stiles, W. B. (1980). Measurement of the impact of
concept of the alliance outlived its usefulness? psychotherapy sessions. Journal of Consulting and
Psychotherapy, 43, 286–291. Clinical Psychology, 48, 176–185.
∗
Safran, J. D., Muran, J. C., & Samstag, L. W. Stiles, W. B., Glick, M. J., Osatuke, K., Hardy, G. E.,
(1994). Resolving therapeutic alliance ruptures: Shapiro, D. A., & Agnew-Davies, R. (2004).
A task analytic investigation. In A. O. Horvath Patterns of alliance development and the rup-
& L. S. Greenberg (Eds.), The working alliance: ture-repair hypothesis: Are productive relation-
Theory, research, and practice. (pp. 225–255). ships U-shaped or V-shaped? Journal of Counseling
New York: Wiley. Psychology, 51, 81–92.
∗
Safran, J. D., Muran, J. C., Samstag, L. W., & Strauss, J. L., Hayes, A. M., Johnson, S.L.,
Stevens, C. (2002). Repairing alliance ruptures. Newman, C. F., Brown, G. K., Barber, J. P., et al.
In J. C. Norcross (Ed.), Psychotherapy relation- (2006). Early alliance, alliance ruptures, and
ships that work. New York: Oxford University. symptom change in a nonrandomized trial of
Samstag, L. W., Batchelder, S., Muran, J. C., cognitive therapy for avoidant and obsessive-
Safran, J. D., & Winston, A. (1998). Predicting compulsive personality disorders. Journal of
treatment failure from in-session interpersonal Consulting and Clinical Psychology, 74, 337–345.
variables. Journal of Psychotherapy Practice and Tracey, T. J., & Kokotovic, A. M. (1989). Factor
Research, 5, 126–143. structure of the Working Alliance Inventory.
Shapiro, D., Barkham, M., Rees, A., Hardy, G. E., Psychological Assessment, 1, 207–210.
Reynolds, S., & Startup, M. (1994). Effects Tryon, G. S., & Kane, A. S. (1990). The helping
of treatment duration and severity of depression alliance and premature termination. Counseling
on the effectiveness of cognitive-behavioral and Psychology Quarterly, 3, 233–238.
psychodynamic interpersonal psychotherapy. Tryon, G. S., & Kane, A. S. (1993). Relationship
Journal of Consulting and Clinical Psychology, 62, of working alliance to mutual and unilateral
522–534. termination. Journal of Counseling Psychology,
Sommerfeld, E., Orbach, I., Zim, S., & 40, 33–36.
Mikulincer, M. (2008). An in-session exploration Tryon, G. S., & Kane, A. S. (1995). Client
of ruptures in working alliance and their associa- involvement, working alliance, and type of
tions with clients’ core conflictual relationship therapy termination. Psychotherapy Research, 5,
themes, alliance-related discourse, and clients’ 189–198.
C HA P TER
12 Managing Countertransference
Jeffrey A. Hayes, Charles J. Gelso, and Ann M. Hummel
Authors’ Note. We are grateful to Mark relationship. During those early days, CT
Hilsenroth and Marc Diener for providing attained the status of a taboo topic.
consultation on the meta-analyses. Newer conceptions of CT, to be dis-
The concept of countertransference is cussed subsequently, began emerging in the
about as old as psychotherapy itself. Like 1950s. During this decade, the first empiri-
so many fundamental constructs in psy- cal studies on the topic also emerged (e.g.,
chotherapy, the term was created by Freud Cutler, 1958; Fiedler, 1951). From that
(1910), shortly after the turn of the twenti- point on, there has been a steady increase
eth century. Although Freud did not write in both clinical and theoretical writing on
extensively about countertransference (CT), CT. As is so often the case, however, empir-
it was clear that he viewed it as problematic ical efforts lagged markedly behind theo-
and needing to be managed. For example, retical work. Research on CT has gained
Freud (1910, pp. 144–145) commented momentum during the past quarter cen-
that: “We have begun to consider the tury, however, to the point where substan-
‘counter-transference’ which arises in the tive reviews of the empirical literature have
physician as a result of the patient’s influ- become possible (e.g., Fauth, 2006; Hayes,
ence on his unconscious feelings, and have 2004; Hayes & Gelso, 2001; Rosenberger
nearly come to the point of requiring the & Hayes, 2002a).
physician to recognize and overcome this In the present chapter, we present the
countertransference in himself.” Freud noted results of an original meta-analysis on the
that anyone who fails to overcome CT may relation of CT to psychotherapy outcome.
at once “give up any idea of being able to Particular emphasis is placed on the man-
treat patients by analysis.” agement of CT and its role in enhancing
Freud’s negative view of CT likely led to the success of psychotherapy. The chapter
the field’s neglect of the topic for many incorporates studies conducted within all
decades. It became simply something to be research traditions: field, laboratory, survey,
done away with, not something to be exam- experimental, correlational, and qualitative.
ined or even used beneficially. The good We begin by describing varying defini-
analyst was, in fact, thought to be capable tions of CT, identifying its common mea-
of maintaining objectivity and keeping sures, and then examining research as it
personal conflicts out of the therapeutic bears upon the question of the effects of
239
CT and CT management on treatment by the patient in others. Ideally, thera-
outcome. pists restrain their “eye-for-an-eye” impulse
and seek to understand what the patient is
Definitions and Measures doing to stir up these reactions. This allows
Four conceptions of CT have emerged for an understanding of the patient’s inter-
as the most prominent over the years: the personal style of relating and for the effec-
classical, the totalistic, the complementary, tive framing of therapeutic interventions.
and the relational (Gelso & Hayes, 2007). Finally, the relational perspective views
In the classical definition, originated by CT as mutually constructed by the patient
Freud (1910), CT is seen as the therapist’s and the therapist (Mitchell, 1993). The
unconscious, conflict-based reaction to the needs, unresolved conflicts, and behaviors
patient’s transference. Unresolved conflicts of both contribute to the manifestation of
originating in the therapist’s childhood CT in session.
are triggered by the patient’s transference In the literature and in everyday dia-
and are acted out by the therapist in one logue, these four conceptions of CT are
way or another. Advocates of this view of all used, confusingly and sometime in con-
CT see little or no benefit to CT. They tradiction. The problem is that it is often
do not generally believe CT can be used unclear which of the four, or which combi-
to enhance understanding, nor to promote nation, is intended at any given time. Many
therapeutic gain. literature reviews and clinical exchanges
The totalistic conception of CT origi- are marred by the disparate definitions of
nated in the 1950s (Heimann, 1950; Little, the phenomenon.
1951). According to this conception, CT Beyond ambiguity of usage, each of the
represents all of the therapist’s reactions four views of CT possesses fundamental
to the patient. All reactions are important, limitations. The classical view is overly
all should be studied and understood, and restrictive in that its focus is solely on
all are usefully placed under the broad the therapist’s reactions to transference. In
umbrella of CT. This definition served addition, it construes CT in almost exclu-
to legitimize CT and make it an object of sively negative terms, and it ignores the
the therapist’s self-investigation and use. natural, inevitable reactions of the therapist
Accordingly, as the totalistic view gained that are tied to powerful “pulls” by the
ascendancy, CT became seen more and patient. Similarly, the totalistic position, in
more as potentially beneficial to the work, its attempt to encompass all of the thera-
if therapists studied their reactions and pist’s reactions, may render the concept of
used them to further their understanding CT meaningless. If all reactions are con-
of patients. strued as CT, then no reactions are not CT,
In the third, or complementary, concep- and there is no need for the term; CT is
tion of CT, the therapist’s reactions are simply redundant with the phrase “thera-
viewed as a complement to the patient’s pist reactions.” However, there are varying
style of relating (Levenson, 1995; Racker, kinds and causes of therapist reactions
1957; Teyber, 1997). That is, the patient and it is helpful to distinguish them. For
exhibits certain “pulls” on the therapist. example, adverse reactions that are due
The well-functioning therapist, however, to therapists’ unresolved personal conflicts
does not act out lex talionis (an eye for an may be usefully addressed in the therapist’s
eye, a tooth for a tooth), even though it is supervision or personal treatment, whereas
the typical, and expected, reaction elicited reactions due to inexperience or fatigue at
the end of a long day require different inter- Rather, it incorporates the therapist’s reac-
ventions. The complementary conception tion to all clinically relevant material,
is limited in the sense that, in its focus on including the patient’s personality style, the
CT as “pulls” emanating from the patient, actual content that the patient is present-
it does not take into enough account the ing, and even the patient’s appearance.
therapist’s personality, including his or her
own interpersonal style and unresolved Measuring CT
conflicts. This same limitation applies to Despite the history of definitional incon-
the relational view. sistency, most empirical studies on CT
Although each of the common concep- have employed a definition that implicates
tions is seriously limited, all four point to the therapist’s unresolved conflicts as the
important elements of CT. We favor an source, often with one or more characteris-
integrative definition that includes lessons tics of the patient as the trigger. Furthermore,
from all four (Gelso & Hayes, 2007). We most research on CT has examined either
believe it is important to limit the defini- internal or external manifestations of CT.
tion of CT to reactions in which unresolved As an internal emotional state, CT may be
conflicts of the therapist are implicated. reflected in therapist anxiety, anger, bore-
All of the therapist’s reactions are signifi- dom, despair, arousal, disgust, and so on.
cant and worthy of investigation, clinically These feelings not only range in valence
and empirically, but the definition of CT from highly pleasant to highly unpleas-
must be narrower than the totalistic one ant, but they also vary in intensity, from
if it is to be scientifically useful and clini- too intense to not strong enough. To mea-
cally meaningful. Whereas our conception sure CT emotions, researchers have used
of CT is similar to the classical view in its instruments such as the State Anxiety
focus on the therapist’s unresolved conflicts Inventory (cf. Hayes & Gelso, 1991, 1993)
as the source of CT, it is different in that and the Therapist Appraisal Questionnaire
CT is seen as a potentially useful phenom- (cf. Fauth & Hayes, 2006), as well as quali-
enon if therapists successfully understand tative methods (Hayes et al., 1998).
their reactions and use them to help under- As an internal cognitive phenomenon,
stand the patient. Thus, in seeing CT as CT may manifest itself in therapists’ fail-
both a hindrance and a potential benefit ure to accurately recall therapy-specific
to treatment, an integrative definition events and to misperceive clients as overly
picks up on the two thematic constructs similar to or dissimilar from themselves
that have been intertwined, like a double (Cutler, 1958; Gelso, Fassinger, Gomez, &
helix, throughout the history of thought Latts, 1995). To measure these cognitive
about CT (Epstein & Feiner, 1988). In aspects of CT, researchers have compared
addition, like the totalistic position, we the therapist’s perception of the session or
consider CT to be inevitable. This is so client with objective or factual data (e.g.,
because all therapists, by virtue of their from transcripts of sessions or personality
humanity, have unresolved conflicts, per- assessments).
sonal vulnerabilities, and unconscious “soft In terms of overt behavior, CT typically
spots” that are touched upon in one’s has been studied as the therapist’s with-
work. Finally, like the complementary and drawal, underinvolvement, or avoidance of
relational views, our integrative definition the patient’s material or, at times, as thera-
of CT does not simply focus on the therapists’ overinvolvement with patients. These
pist’s reaction to the patient’s transference. assessments typically are made by trained
h aye s , g e l s o , h u mm e l 241
raters who assess transcripts and/or tapes management: self-insight, self-integration,
of therapy sessions (e.g., Rosenberger & anxiety management, empathy, and con-
Hayes, 2002b). CT behavior also has been ceptualizing ability. Self-insight refers to the
measured via raters’ or clinical supervisors’ extent to which the therapists are aware
assessments of therapists’ behavior on a of and understand their own feelings, atti-
one-item measure (the CT Index; Hayes, tudes, personalities, motives, and histories.
Riker, & Ingram, 1997), the 21-item Index Self-integration taps the therapist’s posses-
of Countertransference Behavior (ICB; sion of an intact, basically healthy character
Friedman & Gelso, 2000), or the 10-item structure. In the therapy interaction, self-
Countertransference Behavior Measure integration manifests itself as an ability to
(CBM; Mohr, Gelso, & Hill, 2005). The differentiate self from other. Anxiety man-
ICB contains two subscales reflecting agement refers to therapists allowing them-
CT behaviors that have a positive (e.g., selves to experience anxiety and possessing
befriending the client) or negative (e.g., pun- the internal skill to control and understand
ishing the client) valence. The CBM con- anxiety so that it does not bleed over into
tains items that reflect therapist behaviors their responses to patients. Empathy, or the
that are dominant, hostile, or distant. More ability to partially identify with and put
recently, a Countertransference Measure was one’s self in the other’s shoes, permits the
developed via experts’ ratings of prototypical therapist to focus on the patient’s needs
CT reactions (Hofsess & Tracey, 2010). despite difficulties he or she may be experi-
Generally, CT behavior is seen as hinder- encing in the moment. Finally, conceptual-
ing because it is based more on the therapist’s izing ability reflects the therapist’s ability to
needs than the patient’s. Internal CT, on the draw on theory and understand the patient’s
other hand, is seen as potentially helpful. role in the therapeutic relationship.
If therapists can understand these reactions Although the CFI, CFI-R, and CFI-D
and how they may relate to the patient’s can be completed via self-report (e.g.,
inner life and behavior, then this under- Rosenberger & Hayes, 2002b), in most
standing can facilitate work with the patient. research supervisors have assessed trainee CT
Such management of CT has been mea- management. In fact, trainees seem to have
sured almost exclusively with the 50-item difficulty accurately assessing their own CT
Countertransference Factors Inventory (CFI; management abilities (Hayes et al., 1997;
Van Wagoner, Gelso, Hayes, & Diemer, Hofsess & Tracey, 2010). In some studies,
1991) or one of two abbreviated, revised ver- typically those conducted prior to the devel-
sions (CFI-D; Gelso, Latts, Gomez, & opment of the CFI, several of the con-
Fassinger, 2002; CFI-R, Hayes et al., 1997). structs measured by the CFI were assessed
The CFI contains items tapping therapists’ via other instruments, such as the Empathy
CT-relevant tendencies in general, whereas Scale of the Barrett-Lennard Relationship
the CFI-R contains 27 items from the CFI Inventory (Peabody & Gelso, 1982), a mea-
that were rated highly by experts in a content sure of the ability to conceptually under-
validity study (Hayes, Gelso, Van Wagoner, stand CT reactions (Robbins & Jolkovski,
& Diemer, 1991). The CFI-D contains 21 1987), and a self-awareness index (Williams,
items only about the therapist’s functioning Hurley, O’Brien, & Degregorio, 2003).
during psychotherapy.
All three versions of the CFI contain Clinical Example
five subscales reflecting therapist attributes Examples of countertransference are
thought to be important to successful CT ubiquitous. They range from the obvious,
aggressive acting out to more subtle, seduc- The therapist’s eventual success managing
tive comments. We suspect that in many or her CT may have been captured in items
even most cases in which the therapist’s on the CFI that reflected her ability “to dis-
intense reaction is a “natural” response to tinguish between the client’s needs and her
the patient, therapist unresolved conflicts own needs,” her capacity to “restrain herself
are implicated. An example helps illustrate from excessively identifying with clients’
this point. conflicts,” and her ability to “conceptualize
The trainee was in her fourth practicum her role in what transpires in the counsel-
of a doctoral training program and by every ing relationship.”
indication appeared to have extraordinary
potential as a psychotherapist. In the early Meta-Analytic Review
part of her treatment with a 20-year-old In the following sections, we present meta-
male patient, she experienced continued analytic findings pertaining to CT and its
strong irritation, and she reacted to the management, particularly in relation to psy-
patient in a controlled, muted, and metal- chotherapy outcome. First, we establish that
lic manner. For his part, the patient was CT reactions themselves plausibly influence
an angry, obsessional young man suffering outcome by reviewing research evidence on
from many borderline features. He negated the association of CT to outcome t. Second,
the therapist’s attempts to help him under- we examine studies that address whether
stand how his conflicts might be contribut- CT management reduces actual CT reac-
ing to his ongoing problems with women, tions. Third, we review the research that
and he denied that the treatment could bears directly on the relationship between
have any impact. Also, he usually negated CT management and outcome. For each of
the therapist’s observations about what he these questions, both quantitative and qual-
might be feeling. Clearly the therapist’s itative research is reviewed, and the findings
emotional reactions were “natural,” given from quantitative research are summarized
the patient’s negativity and hostility. Yet, in meta-analytic form.
the therapist’s unresolved anxieties about
not being good enough, about fearing that Search Strategy
she could not take care of others sufficiently, The literature search for identifying possi-
and about some transference-based fears ble articles for the meta-analyses involved
of her supervisor’s evaluation of her were three strategies. First, the reference lists of
clearly implicated in her irritation and her two comprehensive reviews of the literature
muted reaction to the patient. As she came were consulted (Gelso & Hayes, 2007;
to understand these dynamics, her irrita- Rosenberger & Hayes, 2002a). Second, a
tion with the patient lessened, and she search of PsycINFO and the Psychology
empathically grasped the terrifying emo- and Behavioral Sciences Collection was
tions that were underlying much of his conducted by title, keyword, and sub-
negativity. ject for the term “countertransference.”
Had the supervisor rated this trainee’s Finally, EBSCO databases offers a fea-
in-session behavior on the CFI early in her ture called “related articles,” which suggests
work with the client, CFI items that might articles similar to an article being viewed;
have captured her difficulties managing CT suggested articles that had not already
include “does not become overly anxious in been found through the first two search
the presence of most client problems” and methods were considered for the meta-
“is able to manage her need for approval.” analyses.
A total of 126 articles and dissertations published articles were included in the
were identified in these ways. We subse- meta-analyses in place of the dissertations.
quently analyzed whether they were quan-
titative studies. Fifty-five published articles Study Coding
and 21 dissertations were categorized as Characteristics that were coded for each
quantitative studies. study included research design (correla-
These studies were then analyzed by tional or experimental), research setting
three coders coming to consensus to deter- (field or lab), outcome type (proximal or
mine suitability for the meta-analyses. For distal), therapist type, client type, participant
inclusion, each study needed to include demographics, and the inferential statistics
at least two of three variables: CT, CT reported. Using an established method
management, and psychotherapy outcome. (Chatzisarantis & Stoica, 2009), modera-
Thus, studies that focused simply on pre- tors were tested in each meta-analysis when
dictors of CT that did not include outcome there were a sufficient number of studies
or CT management were not considered in each category. Mediators were consid-
suitable for the meta-analyses. Some stud- ered for coding, but there were no reported
ies included variables that could be con- instances of mediation in the studies.
sidered proxies for the three variables of
interest (see the distinction between big O Statistical Methods
and little o outcome in Greenberg & Pinsof, Because the goal of the meta-analysis was
1986). For example, working alliance and to obtain overall weighted rs for the rela-
session evaluation could be considered tionships between each set of variables,
proximal outcomes; self-awareness could results that that did not report correlation
be considered a type of CT management. coefficients were converted to rs (see Diener
Studies that used proxies of criterion vari- et al., 2008; Rosenthal & DiMatteo, 2001).
ables were included and coded. The studies In some cases, authors were contacted
were coded by one male doctoral-level to provide additional data, such as second-
psychologist (second author), one female ary tests or additional statistics when non-
counseling psychology doctoral student significant findings were reported. These
(third author), and one female postbacca- instances are noted in Tables 12.1, 12.2,
laureate research assistant. All published and 12.3, which present the coded studies.
articles were coded simultaneously by the Average rs were calculated for studies that
three coders, although each coder also pre- reported multiple results, so that each study
viewed articles independently and then had one r to contribute to the overall meta-
coded the articles together with the other analyses (Diener et al., 2008). The Hunter-
two coders. Dissertations were coded inde- Schmidt random-effects model was used to
pendently by the postbaccalaureate coder calculate an average weighted r and 95%
and then reviewed by all three coders, who confidence intervals. A χ2 was calculated to
reached consensus on the coding for each test for homogeneity of effect sizes in each
dissertation. Disagreements on coding were meta-analysis. We also report the standard
resolved through discussion until consen- deviation of the weighted rs. Note that
sus was achieved. Of the 55 published when confidence intervals do not overlap,
articles, 22 were found to be suitable for p < 0.05.
the meta-analysis; of the 21 dissertations, Although unpublished dissertations were
5 were suitable. Two dissertations had been included in the meta-analyses, it is possi-
subsequently published as articles, so the ble that other studies exist that were not
Table 12.1 Summary of Studies Relating Countertransference to Outcome
Authors Year Design Setting Predictor Criterion Outcome Moderator N Therapists Clients % Male Ethnicity Age mean (SD) CT rater r p value
(% white) type 1-tailed
Bandura, 1960 Corr Field Approach- Hostility Proximal 12 TH Prof Parents 83 TH Not Not reported Observer −0.53 0.04
Lipsher, & avoidance 17 CL reported
Miller
Yeh & Hayes 2010 Exp Lab TH CL rated Proximal 116 Video Video 22 88 21 (18–44) Student- −0.38 0.00
disclosure TH quality & participant
session quality
Williams & 2005 Corr Lab Negative Session Proximal 18 TH Mix Students 28 TH 94 TH TH ‘36(10.3) TH −0.37 0.07
Fauth ∗ stress eval. 18 CL 11 CL 75 CL CL ‘ 22 (8.2)
Hayes, Yeh, 2007 Corr Field Missing TH empathy, Proximal 69 TH Bereavement Clients 19 TH 89 TH 53.5 (8.6) TH TH −0.03 0.40
& Eisenberg∗ (subscale of CERS, 69 CL TH 10 CL 93 CL 46.5 (12.8) CL
TRIG) WAI-S,
SEQ-D
Hayes, Riker, 1997 Corr Field CT behavior Composite Distal Case 20 TH; Trainees Students 25 TH 80 TH 30.8 (7.4) TH TH −0.33 0.08
& Ingram CSAB success 20 CL 50 CL 85 CL 25.4 (7.0) CL Supervisor
observer
Ligiéro & 2002 Corr Field Negative CT Working Proximal 50 Trainees Students 26 70 Not reported TH −0.32 0.01
Gelso alliance
Rosenberger 2002b Corr Field CT avoidance WAI, EAT, Proximal None 1 TH Prof Student 0 TH 100 TH TH = 34 TH −0.06 0.42
& Hayes behavior SEQ 1 CL 0 CL 100 CL CL = 21
Cutler 1958 Corr Field CT (conflicted/ Task vs. Proximal 2 TH Trainees Students, 100 TH Not Not reported Observer −0.24 0.30
nonconflicted ego−oriented 5 CL Veterans reported
area) responses
Myers & 2006 Exp Lab CT EAT; session Proximal Working 224 Video Students 33 89 20.4 Exp. −0.04 0.28
Hayes quality alliance Manipulation
Mohr, Gelso, 2005 Corr Lab CT behavior Session Proximal 88 CL Trainees Students 13 TH Not Not reported Supervisor −0.04 0.37
& Hill∗ eval− depth, 27 TH 7 Super reported
smoothness 12 Sup 93 CL
245
Note: diss = dissertation; corr = correlational, exp = experimental; TH = therapist, CL = client, prof = professional; sup = supervisor.
∗
Additional data were provided by the authors.
246
Table 12.2 Summary of Studies Relating Countertransference Management to Countertransference

Authors Year Design Setting Predictor Criterion Outcome Moderator N Therapists Clients % Ethnicity Age mean CT rater r p value
Male (% white) (SD) type 1-tailed
Friedman 2000 Corr Field CFI-R Inventory of Distal 149 Prof Clients 58 91 49 (8.7) Supervisor −0.59 0.00
& Gelso CT Behavior
Hofsess & 2010 Corr Field CFI Experiences Distal 35 TH Trainees In 20 TH 54 TH 28 TH TH −0.57 0.00
Tracey with CT 12 Sup general 42 Sup 67 Sup 38 Sup
(SD not
reported)
Latts & 1995 Corr Lab Awareness of Avoidance Distal Awareness 47.00 Trainees Video 65.00 25 29 Observer −0.45 0.00
Gelso feelings; use of feelings (SD not
of theory and use of reported)
theory
Williams 2005 Corr Lab Self- Negative stress Proximal 18 TH Mix Students 28 TH 94 TH 36 (10.3) TH −0.43 0.00
& Fauth ∗ awareness 18 CL 11 CL 75 CL TH
22 (8.22)
CL
Williams, 2003 Corr Field Management Hindering Proximal 301 Prof In 44.00 92 51.2 (9.3) TH 0.29 0.04
Hurley, strategies self-awareness general
O’Brien, &
Degregorio
Peabody & 1982 Corr Field Therapist CT behavior Proximal 20 Trainees Students 100 Not Not Observer −0.24 0.15
Gelso empathy TH-CL and TH reported reported
pairs video 0 CL
Hayes, 1997 Corr Field CFI-R CT Index; Distal CT rater 20 TH Trainees Students 25 TH 80 TH 31 (7.4) TH −0.18 0.22
Riker, & avoidance 20 CL 50 CL 85 CL TH sup
Ingram 25 (7) CL observer
(SD not
reported)
Kholooci 2007 Corr Field Mindfulness CT Distal 203 Mix Client 30 TH 90 TH 42 TH TH −0.15 0.19
Questionnaire 36 CL 80 CL 40 CL
(SD not
reported)
Forester 2001 Corr Field Body Vicarious Proximal 96 Prof In 33 60 39.4 (SD TH −0.10 0.17
awareness traumatization general not
reported)
Robbins & 1987 Corr Lab Awareness of Withdrawal Distal Awareness 58 Trainees Video 53 91 29 (5.3) Observer −0.04 0.38
Jolkovski feelings; use of of feelings
of theory involvement and use of
theory
Gelso, 1995 Exp Lab CFI Cognitive, Distal Sexual 68 Trainees Video 29 56 Not TH, −0.04 0.40
Fassinger, affective, orientation reported observer
Gomez, & behavioral CT of client
Latts
Note: diss = dissertation; corr = correlational; exp = experimental; TH = therapist, CL = client; prof = professional; sup = supervisor; CFI = the Countertransference Factors Inventory.
∗
247
248
Table 12.3 Summary of Studies Relating Countertransference Management to Outcome
Authors Year Design Setting Predictor Criterion Outcome Moderator N Therapists Clients % Male Ethnicity Age mean CTMGMT r p value
(% white) (SD) rater type 1-tailed
Latts 1996 Corr Field CFI TH Proximal None 77 TH Trainees Not 29 TH 69 TH 29 TH Sup 0.89 0.00
effectiveness 77 Sup reported 42 sup 74 Sup 41 Sup
(SD not
reported)
Van Wagoner, 1991 Exp Lab CFI TH excellence Proximal None 122 Prof None 36 TH Not 47.90 Study 0.55 0.00
Gelso, Hayes, reported (SD not participants
& Diemer reported)
Peabody & 1982 Corr Field Openness TH empathy Proximal None 20 Trainees Students 100 Not Not Observer 0.42 0.03
Gelso to CT TH–CL TH 0 CL reported reported
feelings pairs
Gelso, Latts, 2002 Corr Field CFI TH-rated CL Distal None 32 TH Trainees Students 34 TH Not 29 (5.5) TH Sup 0.39 0.01
Gomez, & outcome; 15 Sup 40 Sup reported Sup not
Fassinger Sup rated CL 63 CL CL not reported
outcome reported CL not
reported
Williams & 2005 Corr Lab Self- Session eval Proximal None 18 TH Mix Students 28 TH 94 TH 36 (10.3) TH 0.18 0.25
Fauth∗ awareness 18 CL 11 CL 75 CL TH
22 (8.22)
CL
Fauth & 2005 Corr Lab Self- CL rated TH Proximal None 17 TH Trainees; Students 35 TH 65 TH 24.3 (6.26) TH 0.17 0.00
Williams awareness helpfulness 17 CL undergrads 24 CL 82 CL TH
with 21.9 (3.76)
counseling CL
experience
Rosenberger & 2002b Corr Field CFI-R WAI, EAT, Proximal None 1 TH Prof Student 0 TH 100 TH TH = 34 TH 0.38 0.00
Hayes SEQ 1 CL 0 CL 100 CL CL = 21
Note: diss = dissertation, corr = correlational, exp = experimental; TH = therapist, CL = client, prof = professional; sup = supervisor; CFI = the Countertransference Factors Inventory; MGMT = management;
WAI = Working Alliance Inventory, SEQ = Session Evaluation Questionnaire, and EAT = Expertness, Attractiveness, Trustworthiness.
∗
included, such as masters theses or studies the average weighted r to be increased to
that were not published. Thus, a file drawer −.10. The test for heterogeneity was not
analysis (Diener et al., 2008) was used to significant (χ2 = 16.31, p = 0.06), with 46%
calculate how many studies with null results of the observed variance due to true differ-
would be necessary to bring the weighted ences between studies (I2 = 46.15%). The
average r to equal 0.10 and 0.20. The statis- standard deviation of the rs from the 10
tical program used for the meta-analysis can studies was 0.16, and the range of rs
be accessed at www.informaworld.com/ was −.53 ≤ r ≤ −.03. The hypothesis that
mpp/uploads/metaanalysisprogramv.3.4.xls CT is inversely related to outcome was sup-
(Diener et al., 2008). ported by the meta-analysis, though the
relationship appears modest.
Do CT Reactions Affect Type of outcome (proximal vs. distal) was
Psychotherapy Outcomes? found to moderate the overall weighted r.
Treatment outcomes exist on a continuum Studies with distal outcomes had a higher
from immediate to distal. Immediate out- (p < 0.05) average r (k = 3; r = −.36; 95%
comes pertain to the effects of or on a given confidence interval: −.40 to −.32; I2 = 0%)
phenomenon within the session, whereas than studies with proximal outcomes (k = 7;
distal outcomes address the effects of treat- r = −.09; 95% confidence interval: −.18 to
ment on client functioning at the end of 0.01; I2 = 4.11%).
treatment or after termination. In between In what follows, we illustrate these meta-
immediate and distal outcomes reside a analytic results with select studies. One of
wide range of what might be called proxi- the first studies of CT demonstrated that
mate outcomes: those pertaining to a given CT can have detrimental effects (Cutler,
session or series of sessions, as well those 1958). The work of two therapist-trainees
presumed to be the way station for more was examined in depth, and it was found
distal outcomes. For example, increased that when patient material was related to
patient experiencing may be seen as proximal areas of unresolved conflict in the therapist,
to improvement in patient interpersonal the therapist’s interventions were judged
functioning, itself a more distal outcome. by supervisors to be less effective. These
A striking feature of the empirical CT findings are comparable with those from
literature is the limited amount of research a more recent study that detected that
seeking to connect CT to more distal out- therapist experience of stress in session was
comes. We shall have more to say about associated with poorer evaluations of the
this later, but for now we simply note that sessions (Williams & Fauth, 2005). Similar
nearly all research on CT effects focuses on results were found in a case study of 13
immediate or proximate outcomes. therapy sessions conducted by an experi-
The 10 quantitative studies permitted a enced female therapist (Rosenberger &
numerical estimate of the relationship Hayes, 2002b). In particular, the more the
between CT and outcome. A summary of patient talked about topics that were related
the studies is presented in Table 12.1. The to unresolved conflict in the therapist, the
average weighted r of CT with outcome less the therapist perceived herself as expert,
in these studies was significant and small attractive, and trustworthy, and the more
(r = −.16; 95% confidence interval = −.26 she viewed sessions as shallow.
to −.06). File drawer analysis indicates that The aforementioned case study also
6 studies with null findings with an average reported an inverse relation between
sample size of 60 would have to exist for CT behavior and the working alliance
(Rosenberger & Hayes, 2002b). Specifically, supervisors, and clients. For the more suc-
the more avoidance behavior the therapist cessful cases, however, no relationship was
exhibited in the work, the lower were her found. The authors speculated that in the
ratings of the overall alliance. Also, trained more successful cases, a strong working alli-
judges’ ratings of avoidant behavior in ses- ance mitigated the negative effects of CT.
sions were related to the bond aspect of Several qualitative studies, not included
the working alliance, as assessed by both in our meta-analyses, also have examined the
the therapist and the client. The greater the effects of CT. In one of these, (Hill, Nutt-
amount of avoidance behavior, the weaker Williams, Heaton, Thompson, & Rhodes,
was both participants’ experience of their 1996), factors that cause impasses, or dis-
working alliance bond. One possible expla- agreements between clients and therapists,
nation for these findings is that avoid- that end in termination were examined.
ance behavior on the therapist’s behalf can The researchers studied 12 experienced
suppress expression of affect by patients therapists of varying theoretical persua-
(Bandura, Lipsher, & Miller, 1960), thus sions, and each reported on one case with
limiting the emotional bond between ther- such an impasse. As expected, many factors
apists and patients. were found to be implicated in the impasses,
CT also was found to be negatively and CT was among the most prominent.
related to the alliance in a field study of 51 Most therapists indicated that their own
doctoral students and their clinical supervi- personal conflicts were involved in the
sors (Ligiéro & Gelso, 2001). Supervisors impasses. For example, two therapists had a
rated therapists’ CT behavior during the parent who had committed suicide, which
middle phase of treatment. Both supervi- led the therapists to feel especially vulnera-
sors and therapists rated the working alli- ble when their clients threatened suicide.
ance between therapists and their clients. In another qualitative study, 11 experi-
Results indicated that negative CT behav- enced therapists were interviewed to exam-
ior was inversely related to both supervisor ine their reactions to transference in cases
(r = −0.58) and therapist (r = −0.34) rat- of successful, long-term dynamic therapy
ings of the working alliance. Also, positive (Gelso, Hill, Mohr, Rochlen, & Zack, 1999).
CT (e.g., too much support, colluding with Even within these successful cases, thera-
the client) related negatively to supervisors’ pists reported many instances of CT.
ratings of the bond aspect of the alliance An example of CT was a case in which
(r = −0.36). Thus, negative CT behavior, the “therapist reported that he admired the
and perhaps positive CT as well, seems client for doing things in his life that the
to be associated with the development of therapist could not do, and at the same
weaker alliances. time enjoyed being idealized by the client.
In a rare study that examined the relation Tied to these reactions, the therapist felt he
of CT to distal outcome, 20 cases of brief did not give the client enough permission
therapy conducted by therapist-trainees were to express negative transference feelings”
examined (Hayes et al., 1997). Supervisors (p. 264).
observed each session and rated CT. For In another qualitative study, the focus
the less successful cases, a strong negative was entirely on CT (Hayes et al., 1998).
relationship was found between CT behav- Eight experienced therapists identified as
ior as rated by both supervisor (r = −0.87) experts by peers each treated one patient
and counselor (r = −0.69) and a composite for between 12 and 20 sessions. Therapists
measure of outcome rated by therapists, identified CT as operative in fully 80% of
their 127 sessions, and it appeared that CT We conducted a meta-analysis to sum-
was prominent in each case. Such find- marize findings from 11 quantitative stud-
ings support the proposition that CT is a ies pertaining to the relationship between
universal phenomenon in therapy, even CT management and CT. Here, studies
when defined from an integrative rather were selected that explicitly employed the
than totalistic perspective (Gelso & Hayes, term CT management, or incorporated
1998, 2007). The findings, of course, run processes that were clearly indicative of
counter to what may be seen as the Freudian therapists’ control of and effort to manage
myth that good therapists do not experi- their countertransference reactions (e.g.,
ence CT. As regards the effects of CT therapist self-awareness, therapist use of
on treatment outcome, the data contained mindfulness). The average weighted r for
evidence of both hindering and facilita- the relation between CT management and
tive effects. For example, one therapist was CT was significant and small, r = −.14,
too immersed in her CT issues of strength 95% confidence interval: −0.30 to −0.03.
and independence to connect with her Four studies with null findings with an
dependent client and help her work through average sample size of 89 would have
her problems. On the other hand, another to exist for the average weighted r to be
therapist was able to make use of her increased to −0.10. The test for heterogene-
CT-based needs to nurture and be a good ity was significant (χ2 = 76.89, p < 0.01),
parent by appropriately supporting and with about 89% of the observed vari-
being patient with her client. The research- ance due to true between-study differences
ers offered the working hypothesis that “the (I2 = 88.79%). The standard deviation of
more resolved an intrapsychic conflict is the rs was 0.28, and the range was −0.59 ≤
for a therapist, the greater the likelihood r ≤ 0.18, indicating variability in effect sizes
that the therapist will be able to use his among the studies. Overall, the expectation
or her countertransference therapeutically that CT management is negatively related
(e.g., to deepen one’s understanding of the to CT was supported by the results of the
client)” (p. 478). Conversely, the less resolved meta-analysis.
the conflict, the greater the likelihood that Operationalization of CT manage-
CT will have antitherapeutic effects. ment was found to moderate the overall
In sum, then, the results of both quan- weighted r. Studies using the CFI had a
titative and qualitative studies support larger (p < 0.05) average r (k = 4; r = −0.46;
the idea that CT does adversely affect 95% confidence interval: −0.68 to −0.24;
outcomes. I 2 = 80.15%) than studies using other CT
management measures (k = 7; r = −0.03;
Does CT Management Reduce 95% confidence interval of −0.17 to 0.12;
CT Manifestations? I 2 = 79.32%). CT rater also was found
Since the 1950s, the clinical literature on to moderate the overall weighted r. When
CT has been replete with writing about therapists rated their own CT, the average r
its potential to aid therapy. A fundamen- (k = 5; r = −0.02; 95% confidence interval
tal concept in this literature is that if CT of −0.21 to 0.17; I2 = 31.58%) was smaller
is to be a help rather than a hindrance, (p < 0.05) than when supervisors or other
the therapist must do something to, with, observers rated CT (k = 3; r = −0.53; 95%
or about CT, other than acting it out in confidence interval: −0.65 to −0.41;
therapy. A significant aspect of all of this I 2 = 44.79%). A summary of the studies is
may be termed CT management. presented in Table 12.2. Taken together,
these studies suggest that CT management resting, reading for pleasure, exercising, and
ability aids therapists in containing their not overscheduling patients, were helpful in
CT responses in sessions. reducing the incidence and intensity of CT
Here we present the findings of several behaviors during sessions.
studies to amplify and illustrate these meta- Robbins and Jolkovski (1987) similarly
analytic results. Reich (1960) observed reasoned that CT management would be
many years ago that therapy involves a par- aided by therapists’ openness to CT feel-
tial identification with the patient. In this ings (referred to as CT awareness). However,
identification, the therapist seeks to under- they also theorized that another of Reich’s
stand and feel, perhaps even viscerally, the factors, ability to analyze and make sense of
patient’s experience. In this way, then, the these feelings, was implicated. They opera-
therapist’s internal experience is an impor- tionalized the latter as what they called the
tant clue to what is going on with the “theoretical framework” or the extent to
patient beyond the surface. If the process which the therapist uses formal or informal
is to work effectively, however, the thera- theory to explain the events of the hour, or
pist has to swing to an “outside” position make them intelligible. Fifty-eight trainees
and inspect what is being experienced. In listened and responded verbally at certain
Reich’s words, therapists “should be alert to stopping points to an audiotaped client
our own feelings, stop to investigate them, as if they were working with the client. It
and analyze what is going on” (p. 392). In was found that awareness of CT feelings,
this statement, three factors are highlighted as expected, was associated with fewer CT
that may be fundamental to CT manage- behaviors. Also, CT awareness and theo-
ment: therapist empathy (partial identifi- retical framework interacted, such that high
cation), awareness of CT feelings, and the awareness and high theoretical frame-
ability to make sense of these feelings. work resulted in the least CT behavior,
An initial study on CT management whereas use of theory in conjunction with
addressed these three factors (Peabody & low awareness resulted in the most CT
Gelso, 1982). In a sample of 20 trainees, behavior. These findings were replicated
therapists’ general empathic ability related in a subsequent study despite the fact that
negatively to CT behavior under condi- different stimulus materials were used (Latts
tions the researchers speculated to be the & Gelso, 1995). Similarly, a recent study
most threatening to therapists. Further, the (Hofsess & Tracey, 2010) found that trainees’
greater their empathic ability, the more conceptual abilities, in and of themselves,
openness these therapists had to their CT were not predictive of actual CT behavior.
feelings. Empathic ability and openness to Thus, a merely intellectual approach to
CT feelings (two of Reich’s factors) appear managing CT appears ineffective, at best.
to be interrelated as expected, and empathy However, when one is aware of CT feelings
may prevent CT behavior when the thera- and utilizes theory to understand them,
pist deals with threatening material. CT management seems most effective.
In support of these laboratory findings
with trainees, the central role that empathy Does CT Management Improve
plays in managing CT also was identified in Psychotherapy Outcomes?
a qualitative study of 12 highly experienced The previous two sections of the chap-
therapists (Baehr, 2004). This same study ter have established that CT is associated,
suggested that therapists’ self-care behav- though modestly, with less desirable psy-
iors between sessions, such as meditating, chotherapy outcomes and that effective CT
management is associated with decreased interval: 0.43 to 0.85; I2 = 94.53%) was
CT reactions. By extension, then, it could higher (p < 0.05) than that of studies with
be inferred that effective CT management smaller samples (k = 4; r = 0.27, 95% con-
enhances psychotherapy outcomes. fidence interval: 0.15 to 0.38; I2 = 0%).
We conducted a meta-analysis on the Finally, CT management rater was found
seven identified studies that addressed to moderate the overall weighted r as well.
the relation between CT management and When therapists rated their own CT man-
treatment outcome. A summary of the agement, the average r (k = 3; r = 0.18; 95%
studies is presented in Table 12.3. The aver- confidence interval: 0.13 to 0.24, I2 = inde-
age weighted r of CT management to out- terminate) was lower (p < 0.05) than when
come was significant and large, r = 0.56 others rated the therapists’ CT management
(95% confidence interval: 0.40 to 0.73). (k = 4; r = 0.62; 95% confidence interval:
Thirty-two studies with null findings with 0.44 to 0.81; I2 = 39%).
an average sample size of 41 would have Below we review a couple of studies in
to exist for the average weighted r to be the meta-analysis to flush out the numeri-
reduced to 0.10. The test for heterogeneity cal results. In a mail survey, 122 therapists
was significant (χ2 = 31.8, p < 0.01), with rated either a particular therapist whom
88% of the variance reflecting real differ- they considered to be excellent or therapists
ences in effect size (I2 = 88.03%). The stan- in general on the CFI (Van Wagoner et al.,
dard deviation of the rs was 0.23, and the 1991). Excellent therapists were rated more
range was 0.17 ≤ r ≤ 0.89., indicating that favorably on all five CFI factors than thera-
there is variability in effect sizes among pists in general. These results are consistent
the studies. An analysis of the distribution with a study that found that overall CT
suggested that the r = 0.89 could be an out- management ability was directly associated
lier. When that finding was excluded from with 77 supervisors’ ratings of their train-
the meta-analysis, the average weighted ees’ therapeutic excellence (Latts, 1996).
r still was significant and large (r = 0.45; Although such studies present results
95% confidence interval: 0.34 to 0.56; relevant to outcome, each captures only
I 2 = 0%). Even with this reduction in data, proximal aspects of outcome. Only one
the hypothesis that CT management is study to date has directly assessed the rela-
positively related to outcome is supported. tion of CT management to distal therapy
Operationalization of CT management outcome (Gelso et al., 2001). In this study,
was found to moderate the relationship 32 therapist-trainees and their supervisors
between CT management and outcome. rated treatment outcomes of one case, while
The average r for studies using the CFI the supervisors also evaluated the supervi-
(k = 4, r = 0.64; 95% confidence interval: sees’ CT management ability on the CFI.
0.45 to 0.82; I2 = 36.69%) was higher Outcome ratings by both therapists and
(p < 0.05) than for studies using other mea- supervisors (controlling for initial level of
sures of CT management (k = 3, r = 0.26; client disturbance) were positively related
95% confidence interval of 0.13 to 0.40; to supervisor-rated CFI scores overall and
I 2 = 0%). Sample size also was a moderator. on three of the five subscales.
For testing the effects of sample size, the dis- Taken together, the quantitative find-
tribution of sample size appeared bimodal, ings on CT management are promising. It
so the studies were split at the median appears that such management is a charac-
(n = 32). The average r of studies with larger teristic of therapists seen as excellent by their
samples (k = 3; r = 0.64, 95% confidence peers, aids in controlling the manifestation
of CT, and is probably related to treatment between CT and outcome may be under-
outcome. In addition to these quantita- stood as due, in part, to patient variables.
tive studies, five qualitative investigations Although the three meta-analyses pre-
(Baehr, 2004; Gelso et al., 1999; Hayes sented in this chapter did not identify
et al., 1998; Hill et al., 1996; Williams consistent patient characteristics that con-
et al., 1997) all point to the importance of tributed to CT reactions or interfered with
CT management in the judgment of thera- their management, this is somewhat under-
pists whose views about particular clients standable given the relatively small number
were studied. These qualitative studies are of studies reviewed. We will expand on this
especially significant in that four of them point in the next section.
sampled highly experienced therapists.
Limitations of the Research
Patient Contribution CT does seem to adversely affect impor-
Perhaps as much as any relationship ele- tant outcomes of psychotherapy, but the
ment addressed in this book, CT manage- link between CT behavior and outcome is
ment is largely up to psychotherapists. a tenuous one. Effects of CT on outcome
If therapists evidence specific attributes may be inferred from the data. However,
(e.g., self-awareness) or engage in particular there is precious little direct empirical sup-
activities (e.g., meditation), they may be port for such conclusions. In other words,
better positioned to manage CT. At the it seems obvious that if CT behavior is
same time, there are certain patients who negatively related to sound working alli-
are simply more difficult for most thera- ances and to supervisors’ evaluations of
pists to work with, and these patients are trainees, then it also seems safe to suggest
likely to evoke CT reactions that are chal- that uncontrolled CT is harmful to treat-
lenging to manage. The patient influences ment. At the same time, we could locate
the therapist to a marked degree and in pre- only one study (Hayes et al., 1997) that
dictable ways (Singer & Luborsky, 1977). connected CT behavior to treatment out-
Two studies found surprisingly similar comes beyond proximate outcomes, and
results in their surveys of psychotherapists’ the results of that study only partially sup-
perceptions of stressful client behavior ported the link of CT to outcome.
(Deutsch, 1984; Farber, 1983). In both Furthermore, there are a few ways in
studies, suicidal statements were the most which our three meta-analyses fall short of the
stressful type of client behavior. Others ideal. First, the number studies for each was
were aggression and hostility, severe depres- small, especially for the CT-management-
sion, apathy or lack of motivation, and to-outcome analysis, which limits the
agitated anxiety. From a different perspec- extent to which one can address heteroge-
tive, the APA Insurance Trust has identi- neity of effect sizes and potential modera-
fied types of patients who are more likely tors. Second, there might be researcher
to result in unfavorable outcomes and bias within the findings because many of
increased probability of malpractice suits. the studies were conducted by the same
These include borderline and narcissistic set of scholars, or by researchers who were
personality disorders, dissociative disorder, trained at the same university, or trained
PTSD, suicidal and violent patients, and by faculty from the same university. These
abuse victims. These patients, who typically connections among researchers may have
have poor prognoses, are likely to prompt been responsible for the similarity in mea-
CT. Therefore, the inverse association sures in most studies. Such similarity is
helpful in creating consistency between knowledge valuably. Still, there is a great
studies but could also affect external valid- need for more controlled, quantitative field
ity and inflate the relations found. research in the CT area.
As far as limitations of research meth-
ods, the enormous complexity of CT led
early studies not “to work very well with
The meta-analytic evidence points to the
the more subtle, yet substantial, aspects of
likely conclusion that the acting out of
countertransference. Rather . . . studies have
CT is harmful and that CT management
been limited to more simplified and super-
is helpful. Several therapeutic practices
ficial problems, and restricted in terms
follow directly from these research-driven
of what could be measured” (Singer and
conclusions.
Luborsky, 1977, p. 448). Have we pro-
gressed beyond this point? Our evalua- • The effective psychotherapist can
tion leads to a “yes” and “no” answer. On work at preventing such acting out and
the affirmative side, although many of must manage internal CT reactions in a
the studies reviewed in this chapter were way that benefits the work.
laboratory experiments, these investiga- • Several therapist behaviors appear
tions have moved forward in terms of to be a useful part of this process.
both realism and complexity. For example, Using self-insight and self-integration
videotapes have replaced audiotapes as as examples, the therapist’s struggle to
client stimuli, and rather than choose from gain self-understanding and work on
predetermined written responses, therapist his or her own psychological health,
participants usually generate their own including boundary issues with patients,
verbal responses to the client stimulus. are fundamental to managing and
Also, in recent laboratory studies, CT effectively using one’s internal reactions.
tends to be operationalized multidimen- These two factors allow the therapist to
sionally along affective, cognitive, and pay attention to client behaviors that are
behavioral lines. A further advance has been affecting the therapist in particular ways,
the development of a line of research and why. As the therapist seeks to
on CT management. Such work moves understand what internal conflicts are
the field toward studying the ways in being stirred by the patient’s material,
which CT may be beneficially used (Jacobs, the therapist also considers how this
1991). process may relate to the patient’s life
On the negative side, laboratory studies outside the consulting room. Then the
possess inherent limitations. Experimental therapist may be in a good position to
manipulations are inevitably highly simpli- devise responses that will be helpful to
fied. Clinical meaning and external valid- the patient.
ity are sacrificed for experimental control. • One aspect of CT management,
However, we would offer that such simpli- self-integration, underscores the
fications are a reasonable and helpful way importance of the therapist resolving his
to proceed, so long as a sufficient number or her major conflicts, which in turn
of field studies are done to complement points to the potential value of personal
the analogues and to allow for methodolog- therapy for the psychotherapist. Personal
ical triangulation. Field studies have begun therapy for the therapist seems especially
to accumulate, as have qualitative studies, important when dealing with chronic
both of which have served to advance CT problems. Although we believe the
∗
evidence supports the pervasive nature of Cutler, R. L. (1958). Countertransference effects
CT, it seems obvious that chronic CT in psychotherapy. Journal of Consulting Psychology,
22, 349–356.
problems need to be dealt with by the
Deutsch, C. J. (1984). Self-reported sources of stress
therapist, and that personal treatment is a among psychotherapists. Professional Psychology:
likely vehicle for such resolutions. Clinical Research and Practice, 15, 833–845.
supervision, for experienced therapists as Diener, M., Hilsenroth, M., & Weinberger, J. (2008).
well as trainees, is another key factor in A primer on meta-analysis of correlation coeffi-
understanding and managing CT and in cients: The relationship between patient-reported
therapeutic alliance and adult attachment style as
using it to benefit the work of therapy.
an illustration. Psychotherapy Research, 18, 1–9.
• The psychotherapist can deal with Epstein, L., & Feiner, A. H. (1988). Counter-
CT that has already been acted out in the transference: The therapist’s contribution to
session. In addition to the need for the treatment. In B. Wolstein (Ed.), Essential papers
therapist to understand that indeed he on countertransference (pp. 282–303). New York:
or she was acting out personal conflicts, New York University Press.
Fauth, J. (2006). Toward more (and better) counter-
some research (Hill et al., 1996) points
transference research. Psychotherapy, 43, 16–31.
to the value to therapists’ admission that Fauth, J, & Hayes, J. A. (2006). Counselors’ stress
a mistake was made and that it was the appraisals as predictors of countertransference
therapists’ conflicts that were the source. behavior with male clients. Journal of Counseling
Therapists need not go into detail about and Development, 84, 430–439.
∗
the specific nature of those problems, for Fauth, J., & Williams, E. N. (2005). The in-session
self-awareness of therapist- trainees: Hindering
we suggest that doing so often serves the
or helpful? Journal of Counseling Psychology, 52,
therapist’s needs more than the patient’s 443–447.
(Myers & Hayes, 2006; Yeh & Hayes, Fiedler, F. E. (1951). A method of objective quanti-
2010). fication of certain countertransference attitudes.
• If CT is to be managed and used Journal of Clinical Psychology, 7, 101–107.
∗
for therapeutic gain, having and using Forester, C. (2001). Body awareness: An aspect
of countertransference management that moder-
a theory are not enough. Theory alone
ates vicarious traumatization. Dissertation Abstracts
may be used defensively, but the evidence International, 61 (5561).
suggests that theory in conjunction Freud, S. (1910/1957). Future prospects of psycho-
with personal awareness is a key to the analytic therapy. In J. Strachey (Ed.), The standard
therapeutic use of CT. edition of the complete works of Sigmund Freud
(Vol. 11, pp. 139–151). London: Hogarth Press.
∗
Friedman, S. C., & Gelso, C. J. (2000). The devel-
References opment of the Inventory of Countertransference
An asterisk (∗) indicates studies included in Behavior. Journal of Clinical Psychology, 56,
the meta-analysis. 1221–1235.
∗
Baehr, A. (2005). Wounded healers and relational Gelso, C. J., Fassinger, R. E., Gomez, M. J., &
experts: A grounded theory of experienced psy- Latts, M. G. (1995). Countertransference
chotherapists’ management and use of counter- reactions to lesbian clients: The role of homopho-
transference. Dissertation completed at Penn bia, counselor gender, and countertransference
State University. management. Journal of Counseling Psychology,
∗
Bandura, A., Lipsher, D. H., & Miller, P. E. (1960). 42, 356–364.
Psychotherapists’ approach-avoidance reactions Gelso, C. J., & Hayes, J. A. (1998). The psychother-
to patients’ expressions of hostility. Journal of apy relationship: Theory, research, and practice.
Consulting Psychology, 24, 1–8. New York: Wiley.
Chatzisarantis, N. L. D., & Stoica, A. (2009). Gelso, C. J., & Hayes, J. A. (2007). Countertrans-
A primer on the understanding of meta-analysis. ference and the inner world of the psychotherapist:
Psychology of Sport and Exercise, 10, 498–501. Perils and possibilities. Mahwah, NJ: Erlbaum.
Gelso, C. J., Hill, C. E., Mohr, J. J., Rochlen, A. B., ∗
Hofsess, C. D., & Tracey, T. J. G. (2010).
& Zack, J. (1999). Describing the face of trans- Countertransference as a prototype: The devel-
ference: Psychodynamic therapists’ recollections opment of a measure. Journal of Counseling
about transference in cases of successful long- Psychology, 57, 52–67.
term therapy. Journal of Counseling Psychology, Jacobs, T. J. (1991). The use of the self: Counter-
46, 257–267. transference and communication in the analytic
∗
Gelso, C. J., Latts, M., Gomez, M., & Fassinger, R. E. situation. Madison, CT: International Universities
(2002). Countertransference management and Press.
therapy outcome: An initial evaluation. Journal ∗
Kholooci, H. (2007). An examination of the
of Clinical Psychology, 58, 861–867. relationship between countertransference and
Greenberg, L. S., & Pinsof, W. M. (1986). Process mindfulness and its potential role in limit-
research: current trends and future perspectives. ing therapist abuse. Dissertation Abstracts
In L. S. Greenberg & W. M. Pinsof (Eds.), The International, 68 (9-B), 6312.
psychotherapeutic process: A research handbook ∗
Latts, M. G. (1996). A revision and validation of the
(pp. 3–20). New York: Guilford. Countertransference Factors Inventory. Unpublished
Hayes, J. A. (2004). The inner world of the psy- doctoral dissertation, University of Maryland,
chotherapist: A program of research on counter- College Park.
transference. Psychotherapy Research, 14, 21–36. ∗
Latts, M. G., & Gelso, C. J. (1995). Counter-
Hayes, J. A. & Gelso, C. J. (1991). Effects of transference behavior and management with
therapist-trainees’ anxiety and empathy on coun- survivors of sexual assault. Psychotherapy, 32,
tertransference behavior. Journal of Clinical 405–415.
Psychology, 47, 284–290. Levenson, H. (1995). Time-limited dynamic psycho-
Hayes, J. A. & Gelso, C. J. (1993). Male counselors’ therapy. New York: Basic Books.
discomfort with gay and HIV-infected clients. ∗
Ligiéro, D., & Gelso, C. (2002). Countertrans-
Journal of Counseling Psychology, 40, 86–93. ference, attachment, and the working alliance: The
Hayes, J. A. & Gelso, C. J. (2001). Clinical implica- therapist’s contribution. Psychotherapy, 39, 3–11.
tions of research on countertransference: Science ∗
Little, M. (1951). Countertransference and the
informing practice. Journal of Clinical Psychology, patient’s response to it. International Journal of
57, 1041–1051. Psychoanalysis, 32, 32–40.
Hayes, J. A., Gelso, C. J., VanWagoner, S. L., & Mitchell, S. A. (1993). Hope and dread in psycho-
Diemer, R. (1991). Managing countertransfer- analysis. New York: Basic Books.
ence: What the experts think. Psychological ∗
Mohr, J. J., Gelso, C. J., & Hill, C. E. (2005).
Reports, 69, 139–148. Client and counselor trainee attachment as
Hayes, J. A., McCracken, J. E., McClanahan, M. K., predictors of session evaluation and counter-
Hill, C. E., Harp, J. S., & Carozzoni, P. (1998). transference behavior in first counseling sessions.
Therapist perspectives on countertransference: Journal of Counseling Psychology, 53, 298–309.
Qualitative date in search of a theory. Journal of ∗
Myers, D., & Hayes, J. A. (2006). Effects of thera-
Counseling Psychology, 45, 468–482. pist general self-disclosure and countertransfer-
∗
Hayes, J. A., Riker, J. B. & Ingram, K. M. (1997). ence disclosure on ratings of the therapist and
Countertransference behavior and management session. Psychotherapy, 43, 173–185.
in brief counseling: A field study. Psychotherapy ∗
Peabody, S. A., & Gelso, C. J. (1982). Counter-
Research, 7, 145–154. transference and empathy: The complex rela-
∗
Hayes, J. A, Yeh, Y., & Eisenberg, A. (2007). Good tionship between two divergent concepts in
grief and not-so-good grief: Countertransference counseling. Journal of Counseling Psychology, 29,
in bereavement therapy. Journal of Clinical 240–245.
Psychology, 63, 345–356. Racker, H. (1957). The meanings and uses of coun-
Heimann, P. (1950). Countertransference. British tertransference. Psychoanalytic Quarterly, 26,
Journal of Medical Psychology, 33, 9–15. 303–357.
Hill, C. E., Nutt-Williams, E., Heaton, K. J., Reich, A. (1960). Further remarks on countertrans-
Thompson, G. B. J., & Rhodes, R. H. (1996). ference. International Journal of Psychoanalysis.
Therapist retrospective recall of impasses in long- 41, 389–395.
term psychotherapy: A qualitative analysis. ∗
Robbins, S. B. & Jolkovski, M. P. (1987). Manag-
Journal of Counseling Psychology, 43, 201–217. ing countertransference feelings: An interactional
model using awareness of feeling and theoretical ∗
VanWagoner, S. L., Gelso, C. J., Hayes, J. A., &
framework. Journal of Counseling Psychology, 34, Diemer, R. (1991). Countertransference and the
276–282. reputedly excellent psychotherapist. Psychotherapy,
Rosenberger, E. W., & Hayes, J. A. (2002a). 28, 411–421.
Therapist as subject: A review of the empirical ∗
Williams, E. N., & Fauth, J. (2005). A psycho-
countertransference literature. Journal of therapy process study of therapist in session
Counseling and Development, 80, 264–270. self-awareness. Psychotherapy Research, 15,
∗
Rosenberger, E. W., & Hayes, J. A. (2002b). 374–381.
Origins, consequences, and management of ∗
Williams, E. N., Hurley, K., O’Brien, K., &
countertransference: A case study. Journal of Degregorio, A. (2003). Development and vali-
Counseling Psychology, 49, 221–232. dation of the Self-Awareness and Management
Rosenthal, R., & DiMatteo, M. R. (2001). Meta- Strategies (SAMS) scale for therapists. Psycho-
analysis: Recent developments in quantitative therapy, 40, 278–288.
methods for literature reviews. Annual Review of Williams, E. N., Judge, A. B., Hill, C. E., &
Psychology, 52, 59–82. Hoffman, M. A. (1997). Experiences of novice
Singer, B. A., & Luborsky, L. (1977). Counter- therapists in prepracticum: Trainees,’ clients,’
transference: The status of clinical versus quanti- and supervisors’ perceptions of therapists’
tative research. In A. S. Gurman & A. M. Razin personal reactions and management strateg-
(Eds.), Effective psychotherapy: Handbook of ies. Journal of Counseling Psychology, 44,
research (pp. 433–451). New York: Pergamon. 390–399.
Teyber, E. (1997). Interpersonal process in psycho- ∗
Yeh, Y., & Hayes, J. A. (in press). How does
therapy: A relational approach. Pacific Grove, CA: disclosing countertransference affect perceptions
Brooks/Cole. of the therapist and the session? Psychotherapy.
PART
3
Tailoring the Therapy
Relationship to the
Individual Patient:
What Works in
Particular
C HA P TER
13 Reactance/Resistance Level
Larry E. Beutler, T. Mark Harwood, Aaron Michelson, Xiaoxia Song, and John Holman
While it may be conceptualized differently, merely a product of the patient’s poor

psychotherapists from all professions and motivation—it is also a failure of the thera-
perspectives struggle to overcome patients’ pist to fit the treatment to the patient. It
apparent resistance to change. This resis- is in consideration of this viewpoint that
tance is seen in more than simply a failure we have included a discussion of reactance
to improve; it is often seen in the patient’s in this chapter and tend to use this term
behaving in ways that are directly contrary interchangeably with the more usual term,
to the recommendations of the therapist resistance. By either term, our explicit objec-
and/or to the health of the patient. This tive in this chapter is to consider how a
great paradox of psychotherapy suggests patient’s failure to thrive may be a reflec-
that even the most well-intentioned patients tion of the therapy rather than of the patient
may possess ambivalence about making alone. We will address the notion that, by
changes that will help them and thus may looking beyond the patient to the demands
fail to act in their own self-interest. of the therapeutic environment, we can
Patients who fail to comply with therapy identify and target those processes that can
procedures, even when they believe that facilitate patient cooperation and improve
doing so will be helpful, are usually described outcomes.
as resistant. But such a label implies that The literature on patient resistance and
the problem is simply and solely that of the reactance has arisen from two simple obser-
patient and would disappear if the patient vations: First, in every form of psychotherapy,
were more committed. Such an unwar- some individuals don’t change, no matter
ranted assumption offers little to help the how skilled or knowledgeable the therapist
therapist improve the outcomes of work (Howard, Krause, & Lyons, 1993), and
with such patients. It is more accurate, we second, in the end, most psychotherapies
believe, to talk about the failure of one to seem to achieve similar amounts of change
respond in his or her own best interest as a to one another, a phenomenon characterized
problem of reactance rather than a problem as the dodo bird verdict (e.g., Beutler, 2009;
of resistance. Budd & Hughes, 2009; Wampold, 2001;
Reactance implies that the psychother- Wampold et al., 1997).
apy environment itself is a partner in induc- Since the mid-1970s, the preferred
ing noncompliance, and by extension, that methodology of comparative psychother-
the therapist has some control over the fail- apy research has been the randomized clini-
ure of therapy. The failure to change is not cal trial (RCT). This methodology has been
261
considered, in most circles, to be the “gold and overcome patient resistance. Cognitive
standard” for identifying research-supported therapy (or any other treatment studied),
treatments (variously called research- the training of cognitive therapists, and
informed, evidence-based, and empirically- the diagnosis of patients (and any other
supported; c.f. Norcross, Beutler, & Levant, homogenizing condition) are implicitly
2007). This methodology compares a man- assumed to accurately and sufficiently
ualized treatment with another manualized describe patients, therapists, treatments,
treatment or with a delayed or no-treatment and contexts. Limiting study to what the
control group. In actuality, these “treat- similarly trained therapists do in common,
ments” are not homogeneous but are repre- to similarly diagnosed patients, rather than
sented by one or more clusters of interventions including either the variations that exist
that may differ widely in what they are among commonly trained therapists, the
intended to do. Typically, the discrete inter- variability among patients within diagnos-
ventions within any “treatment” are aimed tic groups, or the nature of the therapy’s
at multiple goals, reflecting efforts both context may obscure important relational,
to create a therapeutic environment or pro- patient, and therapist contributions to psy-
cess and to affect the end point of treatment chotherapy outcome (Beutler, 2009).
itself. These multiple objectives represent In response to the foregoing concerns,
the goals that are valued by the theoretical contemporary research has begun to look for
model and underlie the particular treatment aptitude by treatment interactions (ATIs).
approach. These treatments (or more accu- This research largely has investigated how
rately, clusters of interventions bound to a different classes of treatment methods and
given theory) are applied by specially trained the pattern of their use (rather than the
therapists to selected groups of patients, clusters of interventions valued by broad
all of whom share a common diagnosis and theories of psychotherapy) interact with spe-
who are randomly assigned to treatments. cifically defined (and often extradiagnostic)
The specific training provided ensures that characteristics of patients (e.g., Beutler,
the therapists deliver the treatment in a Clarkin, & Bongar, 2000; Castonguay &
manner that is as similar as possible. Those Beutler, 2007). That is, it is increasingly
who depart too far from the ideal treatment thought that the fit or match between
behavior are subject to being dropped or specific interventions that share common
retrained to ensure that similarity among characteristics (rather than either the broader
therapists is maintained in how treatment is treatment or a specific technique) and the
delivered. personal (usually non-diagnostic) attributes
The effects of the therapist variability, of particular subgroups of patients is what
the influence of the treatment relationship, primarily instigates and maintains change.
and all other factors that are thought to This chapter examines the value of this
be extraneous to the specific “treatment” matching approach as applied to patient
studied are ostensibly controlled through resistance or reactance. Specifically, we
training, randomization, or the application will report the results from an original
of statistical controls. Unfortunately, this meta-analysis on matching a nondiagnostic
effort to eliminate variance associated with patient variable (resistance) to a group of
therapists, relationships, and contexts by therapy interventions that share a common
RCT methodologies inadvertently elimi- level of therapist directiveness. Our review
nates from study the very aspects of psycho- will assess the prevailing hypothesis that
therapy that would allow us to understand treatment outcomes are enhanced by a
262 ta i lo r i n g t he t he r a p y re l at i o n s hi p to t h e i n d i v i d ua l pat i e n t
good (in this case, an inverse) fit between a consideration of the evoking environ-
the patient’s level of trait-like resistance and ment whereas resistance implies a problem
the level of therapist directiveness. contained within the skin of the person.
Beyond this important difference, resistance
Definitions and Measures implies both a state and a traitlike quality
Patient Resistance/Reactance associated with psychopathology, while
The concept of resistance began to take reactance is confined to statelike behavior
shape in psychology with the development that occurs in normal personality expres-
of psychoanalytic theory. Classical psy- sion. Finally, reactance, by definition, is
choanalytic theory characterized resistance expressed as directly oppositional behavior
as the patient’s unconscious avoidance of while resistance can be identified by failure
or distraction from the analytic work. to act—stubbornness—as well as obstruc-
Resistance was an inherent, unconscious tionism and rebellion.
striving to avoid, repress, or control con- In other words, reactance seems to be a
flicted thoughts and feelings that threat- special expression of resistance that occurs
ened to become expressed toward a therapist in the form of rebellion and that is situa-
or family member. For example, a patient tionally induced (Brehm, 1976). As such,
with significant past trauma may feel threat- reactance may be affected by one’s tolerance
ened by an inquisitive therapist and pro- for events that limit freedom—it is respon-
tectively divert attention away from the sive to traitlike sensitivities as well as state-
threatening material through unconscious like properties of the environment. Whether
processes, or consciously attempt to with- these oppositional behaviors are manifest
hold, falsify, or even refuse disclosure of in any particular situation depends both on
relevant information. the level of traitlike sensitivity (i.e., “resis-
This concept of resistance has been incor- tance”) present to stimulate “reactance” and
porated into much of contemporary litera- the forcefulness of the external demand to
ture, movies, and even common parlance. comply. An adequate external demand is
Take, for example, the frequently confron- one that limits a person’s options; this stim-
tational phrase, “you are in denial” to alert ulates oppositional behavior (J. W. Brehm,
a person to his or her resistance to admit- 1976). Clients rarely attribute active oppo-
ting an addiction or other problem. sition as a characteristic of their own behav-
Outside of psychotherapy and particu- ior in psychotherapy (Kirmayer, 1990).
larly outside of psychoanalytic thought, the Most ascribe their oppositional response to
concept of resistance has achieved its greatest the effect of being a victim of circumstance,
recognition within the field of social psy- of disease, of others’ malevolence or of the
chology, under the label, reactance. In 1966, therapy itself.
J. W. Brehm proposed a theory of psy- Once activated by the environment, resis-
chological reactance, defining this term as a tance propensities can escalate to become
“. . . state of mind aroused by a threat to reactant—oppositional, noncompliant, and
one’s perceived legitimate freedom, moti- rigid (Tennen et al., 1981). An adolescent
vating the individual to restore the thwarted may be quick to perceive his or her free-
freedom” (Brehm & Brehm, 1981, p. 4). dom as being threatened by a parent and
In spite of the similarity in the defini- may resist in reactant ways—rebellious and
tions of resistance and reactance, there are oppositional behaviors. An adolescent with
several distinguishing features. We have high traitlike resistance may be particularly
already mentioned that reactance invokes sensitive to threats to freedom (e.g., being
b e u t l e r, ha rwo o d , mi c h e l s o n , s o n g , h o l m a n 263
disciplined by a parent) and consequently than directions, to use reflections versus
may exhibit a reactant oppositional behavior. interpretations, to use the application of
Resistance, as applied to a client’s behav- self-guided as opposed to therapist-guided
ior, implies that both the static refusal to homework assignments, and to use para-
cooperate or change and manifest rebellion doxical interventions (Beutler & Harwood,
are active processes driven by a common 2000; Malik et al., 2003) are all indica-
need to escape the therapist’s effort to limit tive of low levels of therapist directiveness.
his or her behavior, whether through direct Early research indicated that effective ther-
suggestion or via the inherent demands for apeutic change is greatest when the level of
change within the therapy process. It fol- a given therapist’s directiveness corresponds
lows that a therapist may elicit resistant inversely to a given patient’s level of resis-
behavior from a client by assuming more tance (e.g., Beutler & Harwood, 2000).
control of the patient’s behavior within
and outside of the therapy sessions than Measuring Resistance and
the recipient of these efforts can tolerate, Therapist Directiveness
by using confrontational techniques, or by For it to be useful in psychotherapy plan-
creating or failing to mend alliance rup- ning, resistance/reactance must occur with
tures, among other things. Thus, as we look some degree of consistency. We must start
for aspects of the therapeutic environment with the awareness that resistance, as a trait,
that may evoke resistance, therapist direc- is something that all individuals have.
tiveness has become the major contender Variability among people is a matter both of
(e.g., Beutler, 1983; Rohrbaugh, Tennen, degree (extremeness of reaction) and sensi-
Press, & White, 1981; Shoham-Solomon tivity (the likelihood of activating reactance).
& Hannah, 1991). Thus, measures that identify a patient’s
traitlike qualities of resistance are preferred
Therapist Directiveness for treatment planning. Unfortunately, there
Therapist directiveness refers to the extent are no current measures that reliably pre-
to which a therapist dictates the pace and dict when one’s predisposition to resistance
direction of therapy and implicates a direc- will become oppositional/reactant; how-
tion of needed change, as well as the overall ever, we can measure the strength of resis-
predominance of interventions chosen by tance by assessing the likelihood of its being
the therapist to elicit change, insight, or observed in different situations.
well-being. In other words, directiveness That is, resistant traits in psychotherapy
refers to the degree to which the therapist are identified by assessing the sensitivity of
becomes the primary agent of therapeutic a patient to external demands that reduce
process or change, whether through the his or her choices. A highly trait-resistant
selection of specific techniques or the adop- individual is easily stimulated to behave
tion of a specific interpersonal demeanor. with opposition to a situation in which
High therapist directiveness is illustrated external restraints to freedom are placed
by use of instruction, guidance, interpreta- on him or her. High traitlike resistance
tions, experiential procedures, and in some may or may not lead to broader psychopa-
cases, more confrontational therapeutic thology; however, easy arousal of resis-
styles; however, directiveness should not be tance behaviors is likely to be disruptive to
equated with a particular kind of treatment. relationships and social activities of many
Any therapist decision to vary usage of self- varieties. At this point, it is important to
help resources, to use suggestions rather remember that resistance is not isomorphic
with any particular kind of psychopathol- of the therapy, anger at the therapist’s
ogy or set of diagnoses. In other words, interventions, and the like as instances of
resistance is a normal process evoked within resistance. Therapist behaviors that include
most people under the right circumstances; directions, homework assignments, teach-
however, some individuals are more easily ing activities, and guidance are coded as
aroused to resistant behaviors than others. examples of directiveness.
Resistance is not the prerogative of any Alternatively, some studies have employed
diagnostic group or groups; it is best consid- paper and pencil tests to tap patient trait-
ered a nonpathological process rather than like, resistant qualities (e.g., Beutler et al.,
a pathology characteristic of one’s uncon- 2003; Karno, Beutler, & Harwood, 2002;
scious urges and impulses. Thus, resistance Piper, Joyce, McCallum, & Azim, 1998);
will be a potentially observable event when these tests ask the patient to rate their
any person has been subject to persuasion own tendencies to get in arguments, the
and social influence to change his or her frequency of past conflicts with authorities,
behavior, thoughts, and feelings. Resistance attitudes toward psychotherapy, and rated
occurs when a patient perceives his or her ease of changing behaviors that are offen-
options as being arbitrarily limited by this sive to others.
influence (e.g., Beutler, 1983; J.W. Brehm, For example, The Systematic Treatment
1966; S. S. Brehm, 1976). Selection-Clinician Rating Form (STS-CRF;
Given the variability of individual sensi- Beutler, Clarkin, & Bongar, 2000) illus-
tivity and thresholds that comprise resis- trates an observer-based rating of patient
tance, it is not likely to be reliably measured traitlike behavior while the use of the
as a grouping variable embedded within MMPI Dominance or Treatment Readiness
diagnoses. Neither are the efforts to cir- (TRT) Scales to measure resistant tenden-
cumvent resistance best measured at the cies represents examples of a self-report
level of treatment brand. Resistance is most rating system.
sensitively measured by individual observa- Not all patient measures of resistance
tions of a patient’s tendency to act against are aimed at general traitlike qualities. Both
or in opposition to those persuasive acts of observer-based ratings and self-report meth-
others that are designed to induce change. ods are also available as measures that are
Relatedly, intervention is best measured specific to identifying resistance in psy-
at the level of individual therapists and chotherapy. For example, the STS-CRF
interventions, with judgments based on the requires the clinician to rate the patient’s
degree to which the interventions limit previous responses to mental health treat-
choice. ment as a means of predicting a particular
Direct observer ratings of patient behav- patient’s resistance to treatment. From a
ior and of therapists evoking behavior, self-report perspective, the TRT (Negative
within the therapy session, have been used Treatment Indicators) is a content scale
in several research studies (e.g., Karno & included in and extrapolated from the
Longabaugh, 2005a, 2005b; Shoham- MMPI-2 (Butcher, Dahlstrom, Graham,
Salomon, Avner, & Neeman, 1989). In Tellegen, & Kaemmer, 1989) to assess resis-
observational methods, a rater typically tance to the specific case of mental health
codes such patient behaviors as failure to treatment. Another example of a state-
keep appointments, failure to complete based self-report measure is the Patient
homework assignments or other contractual Resistance Inventory (PRI) developed by
agreements, anger at the therapist, criticism Dowd and colleagues (Dowd, Milne, &
Wise, 1991; Dowd, Wallbrown, Sanders, introduces insensitivity into the measure-
& Yesenosky, 1994) as a specific measure ment of what are individual patient and
of resistance in psychotherapy. All of these treatment qualities.
measures are examples of direct and indi- In the rare instances in which the shared
vidualized methods of assessing each indi- characteristics of groups of individuals could
vidual patient’s level of resistance. be used to infer the presence of individual
Corresponding measures of the thera- trait-like levels of resistance traits, these
pist’s actions are even less frequently used shared characteristics (primarily diagnoses)
than measures of individual patient behav- were used to indirectly reflect in session
ior. Treatment type, embodied in a manual, levels of resistance levels. Such a leap from a
is usually considered a proxy for all thera- grouping variable to an estimate of an indi-
pist behaviors; however, it is not very likely vidual trait of resistance was easiest for
that one can accurately infer the therapist’s patients who were diagnosed with border-
proclivities for directiveness from a treat- line personality disorder (BPD), and then
ment manual. Certainly, directiveness cuts only when the samples were sufficiently
across therapeutic schools broadly (e.g., described and documented as having behav-
Malik et al., 2003) and dilutes estimates ioral characteristics of resistance. Likewise,
of directiveness on the basis of school or when a treatment that dictated directive
manual adherence. Thus, most compari- interventions was compared with one that
sons of different treatments do not include clearly represented a less directive approach
an individualized, direct assessment of (e.g., behavior therapy vs. nondirective ther-
either resistance or therapy directiveness. apy, as per Beutler, Mohr, et al., 1991; Karno
In the current review, for example, we & Longabaugh, 2005b; or interpretive vs.
found that relatively few of the many stud- insight-supportive therapy as per Piper et al,.
ies on psychotherapy had utilized a direct 1999), one might infer a difference in direc-
measure of patient resistance and that even tiveness sufficient to permit analyses. In
fewer used a direct measure of therapist the current review, such decisions were
directiveness. The absence of appropriate based on a nomination and review process
measures was so pervasive that we had to described further in the following sections.
resort to including a few studies that used
group-based measures of both patient and Clinical Examples
therapy qualities. While it is possible, under There are many examples of resistance in
ideal conditions, to estimate patient resis- psychotherapy: The patient who consis-
tance by identifying certain group-level tently fails to complete homework assign-
characteristics of the patients in the sample ments, the chronically late patient, the
and, perhaps, certain levels of directive- patient who agrees and then disagrees (“yes,
ness based on the comparative treatments but . . .”), and the patient who becomes
studied, we reserved the use of this pro- angry and verbally attacks the therapist’s
cedure to fewer than 50% of the studies skill or interventions. While any patient
reported and to exceptional circumstances may show some of these signs when the
where the homogeneity of the group or therapist moves too fast or makes a tactical
the disparity of the comparative treat- error, patients who show consistent, cross-
ments was sufficiently documented to war- situational resistant behaviors may be
rant equating a group-level quality with an spoken validly of as a “resistant patient.”
individual assessment. It should be noted Lisa was a 37-year-old European-
that such a group classification procedure American female in her third marriage.
She sought psychotherapy because of mild In this latter tactic, the therapist attempts
depressive symptoms. She presented with to gain the patient’s trust by acknowledg-
a matter-of-fact and assertive style. She ing, agreeing with, or even encouraging the
revealed that she had marital problems avoidance, with such assertions as, “Make
and described escalating arguments that sure you don’t reveal more than you want
lasted several hours. She indicated that her to” and “it is not wise to rush into change.”
primary goal was learning how to commu- At other times, the therapist provides an
nicate in a more effective way with her hus- interpretation to accompany the paradoxi-
band. The client admitted that her husband cal assertion. This interpretation is framed
was the one who told her to come to ther- with a twist in which the encouragement of
apy, although he was unwilling to engage the patient’s defensiveness is given a nega-
in couples therapy himself. tive valence. For example, it may be inter-
Lisa defended her decision to undertake preted as a natural process that will pass when
psychotherapy by describing the history of sufficient maturity or strength is achieved.
her symptoms in detail and reporting her The implication is that the patient is not
background. She opened the third treat- yet strong, and until he is, he should resist
ment session by asking, “So, what do we do and avoid anything that might frighten him.
now?” This was the client’s first therapy This interpretation is designed to mobilize
experience, and she declared she wanted the patient’s oppositional tendencies against
to move through the process and find a his or her own resistance.
solution as quickly as possible. She expected For example, the therapist suggested to
to be through with therapy in 6 months. Ray that he had failed to yet develop the
This form of resistance may simply result strength to face aspects of his relation-
from misunderstanding the nature of psy- ships and that until he developed this level
chotherapy and the demands and time of experience and maturity, the therapist
requirements associated with change. Such would continue to encourage him to avoid
resistance can often be countered by provid- discussing anything personal about him-
ing education about the treatment process. self. His resistance to being classified as
By contrast, Ray was a 34-year-old cocaine weak helped move him toward greater self-
abuser who was sent to treatment by his disclosure.
lawyer. He openly expressed a lack of inter- Another variety of resistance comes in
est in participating and spent the first two the form of periodic cooperation inter-
sessions sitting quietly but sullenly. He spersed with periods of anger at, rebellion
failed to complete homework assignments against, or distrust of the therapist. Barbara,
or performed them in an obviously incor- for example, seemed very interested in
rect and antagonistic way. This “reactant” finding ways to better develop relation-
behavior exemplifies the conditions in which ships with others. She had a long history of
fear of losing face, control, or freedom, and beginning relationships, becoming intensely
a resulting open distrust of the process, can involved, and then being abandoned and
drive oppositionalism and avoidance. rejected. She could not keep a relationship
Working with this magnitude and type of going for longer than a few months, and
resistance requires either a very slow and the end of the relationship was always dif-
nondirective treatment in which trust is ficult, filled with ambivalence and hostility.
developed gradually and painfully, or the She vacillated between being angry with and
use of paradoxical strategies in which resis- being dependent on others. Her relationship
tance is not only tolerated but prescribed. with the therapist was similar. She quickly
came to depend on her therapist’s sugges- the impact of “fit” and “misfit” between
tions, even to the point of calling in several patient resistance and therapist directive-
times a week to make sure that her deci- ness on outcome. They concluded that,
sions were appropriate. But on those occa- in both cases, more than 80% of studies
sions when her decisions did not meet her cited confirmed one of two relationships.
therapist’s ready approval or did not pro- The first relation found was that patient
duce the desirable results, Barbara became resistance was consistently and negatively
angry. Initially she expressed the anger in related to the achievement of therapeutic
a heated fashion and then immediately gains. The second relation that predicted
apologized and pleaded for understanding; treatment benefit was a consistent but
however, as time passed, she became more inverse correlation between patient resis-
overt with the expression of her anger and tance and the level of therapist directive-
began to distrust the therapist’s motives. ness. That is, when this inverse relationship
She accused her therapist of lying to her, was present, the likelihood and magnitude
asserted that therapy was just a way of of positive change increased.
manipulating her, and frequently became Based on these two findings, the authors
distraught if the therapist did not give concluded that measures of resistance/
direct answers to her many questions. reactance could be used to tailor thera-
Barbara represents a particularly difficult peutic interventions and optimize gains.
pattern in which her early behavior tended Advantageous outcomes would most likely
to seduce the therapist into a pattern of occur if therapists could selectively fit treat-
answering questions and giving direction. ments that vary in the level of applied con-
But, as soon as this pattern became estab- trol, structure, and directiveness in order to
lished, the patient then reversed her behav- optimally fit both the high- and the low-
ior and became angry at the therapist’s resistance patient.
control—the very behaviors that she ini-
tially desired and solicited from the thera- Meta-Analytic Review
pist. The situation calls for the therapist to Inclusion Criteria and Search Strategy
back away from providing direction, to The overriding objective of the current
rethink the processes by which help is being review was to investigate the hypothesis
offered, and to give space to the patient to that an inverse fit between patient level
make her own choices. In these instances, of resistance and therapist directiveness is
the therapist has a difficult task of letting conducive to enhanced treatment benefit.
the patient struggle to regain control. As the In order to ensure an optimal and reliable
therapist adopts a less directive approach, test of this hypothesis, we began with the
dropping her agenda of trying to change the description of an ideal prototypic study
patient, and becomes reflective and more that could best address the research ques-
able to listen, the patient is likely to open up tion of fit. The qualities of this prototypic
and express more feelings about her fears. study included the following:
Previous Reviews 1. A wide breadth of reliably applied

In the first edition of this volume, Beutler, therapeutic approaches and trained
Harwood, Alimohamed, and Malik (2002) therapists to ensure variance on the
identified nearly 30 studies that either dimension of directiveness
addressed the independent role of patient 2. A similarly wide range of moderately
resistance on psychotherapy outcome or impaired patients in order to ensure
variability on the dimension of traitlike patient diagnosis. Likewise, directiveness
resistance was assessed through patient-level ratings
3. Individualized, direct measures of therapist actions using external raters
of both patient resistance and therapy/ applying a pretested scale (the Therapy
therapist directiveness in order to avoid Process Rating Scale).
equating directiveness with a particular 4. Patients were randomly assigned
brand of treatment and resistance with to treatment type and then randomly
a particular diagnosis assigned to therapist within treatment
4. Random assignment of patients to type.
therapists within treatments in order to 5. Directiveness of treatment was
ensure equivalent dispersal of patients to monitored in early and late sessions, and
directive and nondirective interventions resistance was monitored after every five
5. Systematic monitoring both of sessions to ensure constancy of both
treatment variability/consistency on treatment and fit.
the dimension of directiveness and of 6. Outcome was assessed using standard
patient resistance. scales for depression and drug abuse,
6. Objective and uniform outcome including biological tests of use. Fit of
measurement that included analysis of fit treatment and patient was systematically
between patient resistance and therapy/ measured and assessed against patient,
therapist directiveness. treatment, and relationship contributors
to outcome at the end of treatment and
We eliminated from consideration stud-
6 months later.
ies that looked only at statelike qualities
of resistance since such measures would
The foregoing study represents a good fit
have little implication for pretreatment
with our ideal prototype but was not used
selection and confounded resistance with
as one of the studies that were subjected to
intervention.
meta-analysis because the measure of fit
To aid our selection further, we identi-
was a composite measure that included
fied a study that represented this model
the fit of resistance and directiveness along
most closely, which then served as a tem-
with two other measures of fit. While a
plate for evaluating other studies that we
significant finding was obtained, the fit
identified. The study we used as a template
of resistance and directives could not be
(Beutler et al., 2003) had the following
teased out of the composite score from the
methodological features:
published data.
1. Three manualized treatments to The next step in our procedure was to
ensure treatment breadth. cognitive, identify studies that most closely approxi-
narrative, and prescriptive therapies were mated our ideal and to retain these for the
designed to ensure variability of therapy meta-analysis. Our scope included studies
actions across therapies. published within a 20-year span from 1988
2. Patients were those with comorbid to 2008. We began by collecting studies
conditions of mild-to-moderate depression that had addressed patient resistance as a
and substance abuse disorder. mediator between treatment directiveness
3. Patient resistance was measured before and outcome from among the studies of
therapy using a self-report measure (the resistance × directiveness that were reviewed
MMPI-2 TRT scale). This avoided the in the first edition. This resulted in a list of
tendency to equate level of resistance with 20 studies, and we added to this list by
searching the PSYCInfo database using without reference to the findings. This deci-
search words associated with resistance/ sion provided some confidence that the
reactance, therapy directiveness, and the body of studies we included were of a high
like. The final step was to hand-search the enough quality to estimate the effects of
past year’s volumes of the most widely cited treatment by patient fit.
journals that emerged from our search. Our meta-analysis was based on this
These included the Journal of Consulting carefully selected sample of studies, all of
and Clinical Psychology, the Journal of which maintained a relatively uniform meth-
Counseling Psychology, Clinical Psychology: odology and adequate description to ensure
Science and Practice, and the Journal of consistency in the calculation of effect sizes.
Clinical Psychology. All but one of the selected studies employed
Few studies were found that met all a manual-driven and randomized assign-
six criteria in our methodological ideal. ment (RCT) of therapy. The one partial
The search yielded a total of 27 studies that exception (Calvert, Beutler, & Crago, 1988)
complied with three or more of our six did not employ a manualized psychother-
inclusion criteria and only five that com- apy but otherwise met criteria for an RCT
plied with all six of the criteria sufficiently design. It also had the advantages of a large
that we could extract effect size estimates. sample and was the only study that applied
The penultimate sample was comprised of the concepts to psychiatric inpatients. Two
10 studies that met five or more of the six studies reported initial and follow-up data
prototypic criteria. The most usual depar- on the same sample (Piper et al., 1998 and
ture from our ideal was the failure to use Piper et al., 1999), but only one provided
individual/direct measures of either resis- a sufficiently convincing measure of the
tance or directiveness or the failure to ade- patient resistance variable to include.
quately describe the treatment(s) or patients The 12 studies, involving 1,103 psy-
in sufficient detail as to ensure compli- chotherapy patients, are summarized in
ance with the first two criteria. Failure to Table 13.1. We discuss the result of the
report necessary statistics to ensure that meta-analysis in the following paragraphs,
level of fit was assessed against a reliable both in terms of main effects of resistance
outcome measure was a close second reason and directiveness as well as the mediational
for rejecting studies. effects of matching patient resistance levels
At this point, we elected to expand the with therapist directiveness.
data set to include two more studies in
order to reach a sample of 12 studies for Calculation of Effect Sizes
our analysis. We invited each of the coau- Effect sizes (ESs) associated with the fit
thors to nominate studies that they thought of treatment directiveness and patient
came closest to the ideal prototype, though resistance levels were calculated by the use
not in compliance with all six selection of several different formulas, each being
criteria. We accepted nominations of studies selected to best fit the characteristics of the
that departed from the criteria but included data presented in an individual study.
only as many as would ensure that the Cohen’s d was calculated in all cases and,
preponderance of the final data set were when ESs were presented as correlations
in substantial compliance with the quality or regressions, transformations were con-
criteria that we identified. The two senior ducted (using Borenstein, Hedges, Higgins,
authors made final decisions by a review of and Rothstein, 2009, as our general guide;
the methodology of each nominated study Lipsey & Wilson, 2001, and Hunter &
Table 13.1 Resistance and Psychotherapy Directiveness
Study name N Design Measure Measure N M ES M ES M ES 95% CI
resistance directiveness ES/study (direct) (resist) (fit)
Calvert et al. 108 RCT D (FIRO-B) D (TOQ) 1 0.52 0.42–0.61

(1988)
Beutler, Engle,, 62 RCT D (MMPI) I (CBT v FEP 3 0.34 0.88 0.79–0.96
et al. (1991) v S/Sd)
Beutler, Mohr, 63 RCT D (9 scales) I (BEH Vs 9 0.62 0.50–0.73
et al. (1991) ND)
Beutler et al. 46 RCT D (MMPI) D (TPRS) 1 0.33 1.40 1.18–1.61
(1993)
Piper et al. 98 RCT D (QOR) I (Interp v 4 0.31 0.43 0.64 0.54–0.58
(1999) insight)
Karno et al. 47 RCT D (MMPI) I (FST v CBT) 1 0.46 0.42 0.65 0.51–0.78
(2002)
Karno & Longabaugh
(2004) 140 RCT D (anger) D (TPRS) 3 1.16 1.04–1.33
138 RCT D (anger) I (MET v CBT) 2 0.43 0.33–0.50
(2005a) 169 RCT D (obs) D (obs) 4 1.21 1.12–1.29
(2005b) 139 RCT D (self-re) I (TPRS) 6 1.12 1.05–1.18
Clarkin et al. 62 RCT I (BPD) I (DBT v Pdyn v 3 0.14 0.10–0.17
(2007) Support)
Gregory et al. 30 RCT I (BPD) I (Pdyn v TAU) 4 0.52 0.34–0.69
(2008)
Total N 1102
Summary 0.38 0.82
weighted ESs
95%CIs for 0.32–0.43 0.81–0.86
summary
weighted ESs =
Note: For ease of interpretation, all effect sizes have been reported as positive values if they support the specified hypotheses. Negative values
indicate a failure to support the hypothesized relationships.
Key: Design = RCT (randomized clinical trial)
Measures of resistance and directiveness = Measure of resistance and directiveness are either directly measured (D) or indirectly measured (I).
Specifically, D indicates the use of direct observational ratings of directiveness (obs) or a standardized trait measure (e.g., the MMPI,
QOR-Quality of Object Relationships, FIRO-B, or STS-Clinician Rating Form) applied to each individual. I indicates that an indirect
measure of resistance was used based upon a grouping variable such as patient diagnosis—e.g., borderline personality disorder (BPD) or
substance abuse disorder (SAD) to indicate resistant groups.
Among measures of Directiveness, D indicates the use of a direct rating of therapist acts in treatment—e.g., using an observational rating like
the Therapy Process Rating Form (TPRS), or a simple observational rating (obs). I indicates the use of an indirect measure of directiveness,
based on the general directiveness of the treatment model used. Below are identifiers of the direct and indirect measures of the directive and
nondirective treatments employed.
TOQ = Therapist Orientation Questionnaire—a measure of therapist directiveness
Pdyn = psychodynamic treatment—moderately directive; TAU = treatment as usual, nondirective; BEH = Behavioral Tx, directive;
ND = nondirective or reflective, nondirective; CBT = cognitive therapy, directive;
FEP = focused expressive therapy, low directive; Interp = interpretive, highly directive; FST = family systems; NT = narrative therapy;
MET = motivational interviewing, nondirective; DBT = dialectic behavior therapy, directive
Support = supportive therapy, nondirective
N ES/Study = Number of effect sizes calculated for this study
M ES (Direct) = the mean effect size attributable to the directiveness of the treatment—combining all treatments
M ES (Resist) = the mean effect size attributable to the resistance variable—combining all varieties
M ES (Fit) = the mean difference between Effect sizes for “good” and “poor” fit, estimated in MR/Nat studies from correlational data.
All ESs are expressed as d.
Total effect size is weighted by the sample sizes of all studies.
271
Schmidt, 2004, were used as supplemen- All of the studies in our meta-analysis
tal sources). In all cases, the signs of the addressed the role of therapist directive-
effect sizes were changed where necessary ness and the differential or mediating effects
in Table 13.1 to ensure that positive signs of resistance on different treatments. Four
indicated support for one of three hypoth- analyses in this series were conducted on
eses: (1) positive outcomes are associated a single sample from Project Match and
with low levels of resistance, (2) positive three associated publications of these
outcomes are associated with high levels of data (Karno & Longabaugh, 2004, 2005a,
directiveness, and (3) good outcomes are 2005b). These analyses were included
associated with an inverse fit between patient separately because they entailed different
resistance and therapist directiveness. measures of both resistance and directive-
In calculating the mean effect sizes, we ness and varied in the samples used; how-
weighted each mean ES estimate by the ever, the presence of some redundancy
N of the study in which it occurred. This among these studies led us to later summa-
ensured that the vagaries of small samples rize the data with and without this
would not overly influence the conclu- replication.
sions. Finally, we calculated 95% confi- All studies in this series involved random
dence intervals for all d values, utilizing the assignment of patients to therapists, and
procedures outlined by Smithson (2003). all but one included randomization to a
manualized treatment as well. In 10 of the
Meta-Analytic Results 12 studies in our analyses, we evaluated
Resistant patients are assumed to experi- the fit of directiveness to patient resistance
ence less benefit and are more prone to pre- through individual, direct measure of the
maturely terminate from treatment than patient’s resistance, the therapist’s direc-
those who are cooperative. Unfortunately, tiveness, or both. This assessment at the
while the preponderance of studies avail- level of the person and session avoided
able in the literature is supportive of this equating treatment type with directiveness
claim, the reliability of the findings is less or diagnosis with patient resistance, and it
than optimal. In our sample of 12 con- assures independence of measurement.
trolled studies, for example, only two pro- Addressed in this way, we were able to dis-
vided reliable data on which to calculate an entangle therapist intervention from treat-
effect size attributable to patient resistance ment. Thus, it was possible to extract an
(Karno, Beutler, & Harwood, 2002; Piper effect size estimate for the use of directive
et al, 1999). These effect sizes were −.43 procedures, independently of patient resis-
and −.42 (note: the signs are changed in tance levels, based on four of the studies.
Table 13.1 to preserve consistency), sug- The weighted effect size of this analysis
gesting that high resistance was related to was 0.38. This ES is considered a signifi-
low outcomes. Though meager, this finding cant, moderate effect and provides support
is consistent with the evidence reported that directive treatments tend to be rela-
in our earlier review published in the first tively powerful compared with nondirec-
edition of this book based on a box score tive ones.
count of the preponderance of findings. As noted in Table 13.1, the mean ESs
Thus, we tentatively reaffirm that our ear- (d ) associated with matching effects, sum-
lier recommendation that psychothera- ming across different measures across these
pists avoid inciting patient resistance may studies, earned a large mean, weighted d
be valid. of 0.82. The effect size of 0.82 suggests
that approximately 15% of the variance in Representative Studies
outcome may be reflective of the fit of To give further flavor of the ways in which
directiveness and patient resistance; how- these indications of treatment–patient fit
ever, the range was relatively wide with were studied, a review of several studies and
0.14 (Clarkin et al., 2007) being the low their findings may prove helpful. Piper and
value and 1.40 (Beutler et al., 1993) mark- colleagues (1999), for example, compared
ing the high value. Such variation suggests interpretative (directive) and supportive
that the fit of treatment and patient is (nondirective) therapies among patients
important, but that additional mediators who varied on their interpersonal recep-
also are at work and are not accounted for tivity and attachment patterns (Quality of
in the data. Object Relations). In our analysis, we used
One yet unpublished (and hence, not the latter measure as a proxy measure of
included) study (Johannsen, reported in patient resistant tendencies. Piper et al. found
Beutler, 2009), comparing U.S. and Argen- that directive interventions evoked higher
tine samples, suggested that one of these rates of dropout than supportive ones, and
additional moderating variables might that among patients with high receptivity,
include variations among cultures. How- the application of directive interventions
ever, the one non–North American sample resulted in positive effects, while patient with
included in the current meta-analysis poor receptivity/object relations responded
(Beutler, Mohr, et al., 1991) was based on a poorly to directive interventions.
Swiss sample and reported results that were For another example, Beutler, Engle,
comparable to others in the set (see Table and colleagues (1991) demonstrated that
13.1). Moreover, when only the first of the directive therapies (cognitive therapy) and
four analyses of Project Match data by nondirective interventions (self-directed
Karno and Longabaugh (2004) was therapy) were differentially effective for
included in deriving the effect size, the reducing depressive symptoms, depending
result (d = 0.76) continued to be high and on the patient’s level of resistance. Among
supportive of the mediational hypothesis. very resistant patients, a self-directed ther-
It bears noting that Karno and apy regimen surpassed a directive one
Longabaugh (2004) conducted two analy- in affecting therapeutic gain. Conversely,
ses of the same patients, one in which indi- patients who were low on resistance did
vidualized measures were taken of both best with directive, cognitive therapy proce-
patient resistance (anger levels) and of dures. These results were cross-validated at
therapist directiveness (rater observations) 1-year follow-up (Beutler, Machado, Engle,
and the other that substituted a group mea- & Mohr, 1993) and independently repli-
sure of treatment directiveness based on cated in a cross-cultural sample of depressed
the global treatments compared. While patients treated with behavioral and non-
the individualized measurement yielded directive therapies (Beutler, Mohr, Grawe,
an effect size (d ) of 1.16, a high value, the Engle, & McDonald, 1991).
grouping method earned only an effect size Reactance theorists have suggested that
of (d ) 0.43. The magnitude of this differ- paradoxical interventions may be differ-
ence suggests the degree of dilution that entially effective among highly reactant
can be obtained by the use of group meth- patients since they capitalize on the patient’s
ods of clustering treatments (and patients) tendency to respond in oppositional ways.
as opposed to the more sensitive, individu- These studies were not included in the cur-
alized ones. rent analysis because they usually involved
subject analog designs. For example, Limitations of the Research
Shoham-Saloman, Avner, and Neemen Because resistance cannot be randomly
(1989) examined the mechanisms of change assigned to patients, they are not subject to
under paradoxical interventions (defined experimental designs that require direct
as therapeutic directives whose common random assignment. Randomized controlled
denominator is to attempt to induce change trials are possible by randomly assigning
by discouraging it). Although conducted patients (who vary in resistance) to treat-
with a nonclinical population, their study ments (which vary in the amount of direc-
revealed that, under paradoxical interven- tiveness), and to therapists within
tions, subjects who were high on resis- treatments (whose differential proclivities
tance benefitted more than those low on to adopt directive interventions can be
resistance. measured). Our meta-analysis relied heav-
ily on such evidence and excluded studies
Patient Contribution that did not utilize randomized procedures
The negative relationship between patient for assigning patients to treatments and
resistance and therapy outcome may evoke therapists. Many naturalistic and quasi-
some criticism among those who believe experimental studies failed to include suf-
that patient resistance represents a core ficient controls to meet our standards for
conflict that must be exposed and excised calculating meaningful effect size estimates.
in order for benefit to be achieved. One Based on 12 studies that we believe
of the implications of the current find- are representative of the best available, we
ings is that it may be advisable to reframe found that the evidence supports the
our conventional wisdom about the role hypotheses posed; however, we recognize
of interpreting and inciting resistance that there is a particular weakness within
in psychotherapy. Rather than being a this body of studies. More specifically, they
patient process that must be expunged in are not equally inclusive of other potential
order to promote change, resistance may mediators. For example, the study that we
best be viewed as a signal to the therapist used to exemplify a prototypic methodol-
that ineffective methods are being used. ogy model (Beutler et al., 2003) investi-
Namely, the therapist may better view gated the joint relationship of three pairs
some forms of resistance as a problem of mediating variables. They observed effect
of inadequately or inappropriately select- sizes that accounted for over 40% of the
ing interventions that evoke resistance. outcome variance. Such effects are higher
Although some patients may enter treat- than any of those reported here for the
ment with a certain level of resistance, reac- single resistance–directiveness dimension
tance, or suspicion of the motives of the and suggest that there are multiple dimen-
therapist, in general, resistance is typi- sions on which matching can be beneficial,
cally evoked through poor selection of each of which may add independent vari-
interventions; therefore, resistance is best ance to the predictive equation. The role
viewed as a problem of therapy, not of of patient coping style, stage of change,
the patient, and as such, it becomes a prob- cultural beliefs, and symptom severity are
lem for the therapist to solve. The therapist all cases in point where patient and treat-
can then find the appropriate therapeu- ment factors probably interact.
tic means that both stimulate move- Another limitation in studying patient
ment and reduce fear of losing control or resistance is the absence of consensually
freedom. accepted measures of traitlike resistance.
There is a certain circularity to the defini- Therapeutic Practices
tion of resistance that almost, by definition, Collectively, the foregoing results provide
determines that those patients who have it strong evidence that, other things being
will not do as well in psychotherapy as equal, low traitlike resistance serve as an
those who do not. At least this is true if we indicator for patients who respond to direc-
retain the view that resistance is a charac- tive interventions. Conversely, high resis-
teristic of the patient rather than a signal to tance-like traits are markers for identifying
or a characteristic of the therapist’s inter- patients who may respond poorly to inter-
ventions. Escaping the circularity would ventions that are authoritative and directive,
be assisted if there were generally accepted evoking states of resistance that interfere
measures of resistance traits. Numerous with progress, increase the likelihood of drop-
measures have been developed, but they out, and reduce effectiveness of treatment.
suffer from low or inconsistent inter- In practice, we recommend the follow-
correlations. The presence of stable pre- ing on the basis of these results:
dictive measures would greatly add to the
draw that this area has on contemporary • Psychotherapists can learn to
researchers. recognize the manifestations of resistance
Another concern is the role played by both as a state and as a trait. Cues for
different theories of psychotherapy in set- statelike manifestations of resistance
ting the level of therapist directiveness. include expressed anger at the treatment
Much of the research initially reviewed (but or therapist, ranging from simple
rejected in this analysis) assumed that dif- dissatisfaction with therapeutic progress
ferent treatment types are distinguished by to overt expressions of resentment
definably different levels of directiveness. and anger. Three responses to these
To the degree that this assumption is expressions of resistant states entail:
accurate (and that has not been adequately (1) acknowledgment and reflection
demonstrated), it follows that different of the patient’s concerns and anger,
treatments may be more or less appropriate (2) discussion of the therapeutic
for patients, depending on the level of resis- relationship, and (3) renegotiation of the
tance that these patients evince. Therapies therapeutic contract regarding goals and
that are thought to be directive (behavioral therapeutic roles. These responses are
and cognitive-behavioral) or nondirective designed to defuse the immediate
(self-directed or evocative) are presumed to consequences of resistance and to infuse
be advantageous for different patients, the patient with some sense of control,
though directiveness alone seems to offer a as suggested in formulations of reactance
better prospect of treatment outcome than theory (Beutler & Harwood, 2000; also
if it is nondirective. Indeed, cognitive and see Chapter 11 by Safran, Muran, and
behavioral therapies have been found to be Eubanks-Carter, this volume).
most useful for patients who are relatively • Anticipate these reactions by initially
low in resistance, whereas self-directed and assessing the level of patient reactance.
client-centered therapies have been found Patterns are assessed either by standardized
to be of most value for those who are highly psychological tests that tap interpersonal
resistant. Of course, such demonstrations suspiciousness and distrust or by attending
are only interpretable if it can be assured to the historical patterns that have
that different psychotherapies actually characterized the patient’s responses to
differ in level of therapist direction. authority. Patients with high resistance
traits typically manifest a history of decades of study on psychotherapy
difficulty taking direction, a tendency outcomes among depressed adults, which
toward stubbornness and obstructiveness, found a modest effect favoring directive
and difficulty working cooperatively in models and dis-favoring nondirective ones
groups. (Cuijpers, van Straten, Andersson, & van
• Match therapist directiveness to Oppen, 2008). To the degree that specific
patient reactance. High reactance indicates treatments are found to differ in efficacy,
a treatment that will deemphasize it may well be by virtue of how they differ
therapist authority and guidance, employ in the use of such common treatment
tasks that are designed to bolster patient characteristics as level of directiveness
control and self-direction, and rather than in how accurately they
deemphasize the use of rigid homework conceptualize psychopathology or
assignments. Among high-reactance psychotherapy processes.
patients, homework assignments can be • View some manifestations of client
constructed as experiments that require resistance as a signal that ineffective
minimal overt action on the part of methods are being used. That is, resistance
the patient, in order to avoid failure is best characterized as a problem of
and to reduce the opportunities for therapy delivery (not of the patient) and
oppositional behavior. The relative as such becomes a problem for the
amount of listening versus talking should therapist to solve. The therapist can find
shift more toward the patient, and fewer a means that both stimulates movement
instructions should be used. Self-directed and reduces fear of losing control or
assignments and reading might replace freedom.
the usual instructional activities of the
therapist. References
• Beware to avoid matching the level of ∗
= Studies included in the meta-analysis.
therapist directiveness to the therapist’s Arlow, J. A. (2000). Psychoanalysis. In R. J. Corsini
reactance level. This surfaces as a common & D. Wedding (Eds.), Current Psychotherapies
occurrence among neophyte therapists (6th ed.),. Itasca, IL: Peacock Press.
who unwittingly project their own Beutler, L. E. (1981). Convergence in counseling
and psychotherapy: A current look. Clinical
personality structure onto their clients. Psychology Review, 1, 79–101.
It is the patient’s, not the therapist’s, level Beutler, L. E. (1983). Eclectic psychotherapy: A sys-
of reactance that provides the optimal fit. tematic approach. New York: Pergamon Press.
• Avoid stimulating the patient’s level Beutler, L. E. (2009). Making science matter in clini-
of resistance. Based on the current review, cal practice: Redefining psychotherapy. Clinical
we conclude that there is strong and Psychology: Science and Practice, 16, 301–317.
Beutler, L. E., & Clarkin, J. (1990). Systematic
consistent support for a negative treatment selection: Toward targeted therapeutic
relationship between raising patient interventions. New York: Brunner/Mazel.
resistance and therapeutic outcome. Beutler, L. E., Clarkin, J. F., & Bongar, B. (2000).
While a causal chain cannot be certain, Guidelines for the systematic treatment of the
the consistency of the correlational depressed patient. New York: Oxford University
evidence is persuasive. Press.
∗
Beutler, L. E., Engle, D., Mohr, D., Daldrup, R. J.,
• Consider directive interventions as Bergan, J., Meredith, K., et al. (1991). Predictors
a class to be modestly more effective of differential and self-directed psychotherapeu-
than nondirective ones. This conclusion tic procedures. Journal of Consulting and Clinical
supports a recent meta-analyis of three Psychology, 59, 333–340.
Beutler, L. E., Goodrich, G., Fisher, D., & ∗
Calvert, S. J., Beutler, L. E., & Crago, M. (1988).
Williams, O. B. (1999). Use of psychological Psychotherapy outcome as a function of
tests/instruments for treatment planning. In therapist-patient matching on selected vari-
M. E. Maruish (Ed.), The use of psychological ables. Journal of Social and Clinical Psychology, 6,
tests for treatment planning and outcome assessment 104–117.
(2nd ed., pp. 81–113). Hillsdale, NJ: Lawrence Castonguay, L. G., & Beutler, L. E. (Eds.) (2006).
Erlbaum. Principles of therapeutic change that work:
Beutler, L. E., & Harwood, M. T. (2000). Prescriptive Integrating relationship, treatment, client, and
therapy: A practical guide to systematic treatment therapist factors. New York: Oxford University
selection. New York: Oxford University Press. Press.
Beutler, L. E., Harwood, T. M., Alimohamed, S., & ∗
Clarkin, J. F., Levy, K. N., Lezenweger, M. F.,
Malik, M. (2002). Functional impairment and & Kernberg, O. F. (2007). Evaluating three
coping style. In J. Norcross (Ed), Psychotherapy treatments of borderline personality disorder:
relationships that work (pp. 145–70). New York: A multiwave study. American Journal of
Oxford University Press. Psychiatry, 164, 922–28.
∗
Beutler, L. E., Machado, P. P., Engle, D., & Cortina, J. M., & Nouri, H. (2000). Effect size
Mohr, D. (1993). Differential patient X treat- for ANOVA designs. Thousand Oaks, CA: Sage
ment maintenance of treatment effects among Publications.
cognitive, experiential, and self-directed psycho- Cuijpers, P., van Straten, A., Andersson, G., &
therapies. Journal of Psychotherapy Integration, van Oppen, P. (2008). Psychotherapy for depres-
3,15–32. sion in adults: A meta-analysis of compara-
∗
Beutler, L. E., Mohr, D. C., Grawe, K., Engle, D., tive outcome studies. Journal of Consulting and
& McDonald, R. (1991). Looking for differential Clinical Psychology, 76, 909–22.
effects: Cross-cultural predictors of differential Dowd, E. T., Milne, C. R., & Wise, S. L. (1991).
psychotherapy efficacy. Journal of Psychotherapy The therapeutic reactance scale: A measure of
Integration, 1, 121–42. psychological reactance. Journal of Counseling
Beutler, L. E., Moleiro, C., Malik, M., Harwood, and Development, 69, 541–45.
T. M., Romanelli, R., Gallagher-Thompson, D., Dowd, E. T., Wallbrown, F., Sanders, D., &
et al. (2003). A comparison of the Dodo, EST, Yesenosky, Y. (1994). Psychological reactance
and ATI indicators among co-morbid stimulant and its relationship to normal personality variables.
dependent, depressed patients. Clinical Psychology Cognitive Therapy and Research, 18, 601–613.
& Psychotherapy, 10, 69–85. ∗
Gregory, R. J., Chlebowski, S., Kang, D.,
Borenstein, M., Hedges, L., Higgins, J., & Remen, A. L., Soderberg, M. G., Stepkovitch, J.,
Rothstein, H. (2009). Introduction to meta- et al. (2008). A controlled trial of psychody-
analysis. Chichester, UK: Wiley. namic psychotherapy for co-occurring border-
Brehm, J. W. (1966). A theory of psychological reac- line personality disorder and alcohol use disorder.
tance. New York: Academic Press. Psychotherapy, 45(1), 28–41.
Brehm, S. S. (1976). The application of social psy- Howard, K. I., Krause, M. S., & Lyons, J. (1993).
chology to clinical practice. Washington, DC: When clinical trials fail: A guide for disaggre-
Hemisphere. gation. In L. S. Onken & J. D. Blaine (Eds.),
Brehm, S. S. & Brehm, J.W. (1981). Psychological Behavioral treatments for drug abuse and depen-
reactance: A theory of freedom and control. dence (pp. 291–302) (NIDA Research Mono-
New York: Wiley Press. graph No. 137). Washington, DC: National
Budd, R., & Hughes, I. (2009). The dodo bird Institute of Drug Abuse.
verdict—controversial, inevitable and important: Hunter, J. E., & Schmidt, F. L. (2004). Methods of
A commentary on 30 years of meta-analyses. meta-analysis: Correcting error and bias in research
Clinical Psychology and Psychotherapy, 16, findings, 2nd edition. Thousand Oaks, CA: Sage
510–22. Publications.
Butcher J. N., Dahlstrom W. G., Graham J. R., ∗
Karno, M. P, Beutler, L. E., & Harwood, M.
Tellegen, A., & Kaemmer, B. (1989). Minnesota (2002). Interactions between psychotherapy
Multiphasic Personality Inventory-2: Manual for process and patient attributes that predict alco-
administration and scoring. Minneapolis, MN: hol treatment effectiveness: A preliminary report.
National Computer Systems. Journal of Alcohol Studies, 27, 779–97.
∗ forms of psychotherapy and patient personality
Karno, P. M., & Longabaugh, R. (2004). What
do we know? Process analysis and the search for variables. Journal of Consulting and Clinical
a better understanding of Project Match’s anger- Psychology, 66(3), 558–67.
∗
by-treatment matching effect. Journal of Studies Piper, W. E., McCallum, M., Joyce, A. S., Azim,
in Alcohol, 65, 501–512. H. F., & Ogrodniczuk, J. S. (1999). Follow-up
∗ findings for interpretive and supportive forms of
Karno, P. M., & Longabaugh, R. (2005a). Less
directiveness by therapists improves drinking psychotherapy and patient personality variables.
outcomes of reactance client in alcoholism Journal of Consulting and Clinical Psychology, 67,
treatment. Journal of Consulting and Clinical 267–273.
Psychology, 73, 262–67. Rohrbaugh, M., Tennen, H., Press, S., & White, L.
∗ (1981). Compliance, defiance, and therapeu-
Karno, P. M., Longabaugh, R. (2005b). An exami-
nation of how therapist directiveness interacts tic paradox: Guidelines for strategic use of
with patient anger to predict alcohol use. Journal paradoxical interventions. American Journal of
of Studies on Alcohol, 66, 825–32. Orthopsychiatry, 51, 454–67.
Kirmayer, L. J. (1990). Resistance, reactance, and Shoham-Salomon, V., Avner, R., & Neemen, R.
reluctance to change: A cognitive attributional (1989). You’re changed if you do and changed if
approach to strategic interventions. Journal you don’t: Mechanisms underlying paradoxical
of Cognitive Psychotherapy: An International interventions. Journal of Consulting and Clinical
Quarterly, 4, 83–103. Psychology, 57(5), 590–98.
Kirsch, I., & Sapirstein, G. (1998). Listening Shoham-Salomon, V., & Hannah, M. T. (1991).
to Prozac by hearing placebo: A meta-analysis Client-treatment interaction in the study of dif-
of antidepressant medications. Treatment and ferential change processes. Journal of Consulting
Prevention, 1(1), Article 0001c, at http://journals. and Clinical Psychology, 59, 217–25.
apa.org/prevention/volume 1/pre 0010001c.html. Smithson, M. (2003). Confidence intervals. Series:
November, 1998. Quantitative applications in the social sciences,
Lipsey, M. W., & Wilson, D. B. (2001) Practical No. 140. Thousand Oaks, CA: Sage Publications.
meta-analysis. Applied social research methods Tennen, H., Rohrbaugh, M., Press, S., & White, L.
series, Volume 49. Thousand Oaks, CA: Sage (1981). Reactance theory and therapeutic para-
Publications. dox: A compliance defiance model. Psychotherapy,
Malik, M. L., Beutler, L. E., Gallagher-Thompson, D., 18(1), 14–22.
Thompson, L., & Alimohamed, S. (2003). Are Thalheimer, W., & Cook, S. (2002). How to calcu-
all cognitive therapies alike? A comparison of late effect sizes from published research articles:
cognitive and non-cognitive therapy process A simplified methodology. Retrieved March 15,
and implications for the application of empiri- 2009 from http://work-learning.com/effect_
cally supported treatments (ESTs). Journal of sizes.htm.
Consulting and Clinical Psychology, 71, 150–58. Wampold, B. E. (2001). The great psychotherapy
Norcross, J., Beutler, L. E., & Levant, R. (Eds.) debate: Models, methods, and findings. Hillsdale,
(2006). Evidence based practices in mental health: NJ: L. Erlbaum Associates.
Debate and dialogue on the fundamental questions. Wampold, B. E., Mondin, G. W., Moody, M.,
Washington, DC: American Psychological Stich, F., Benson, K., & Ahn, H. (1997). A meta-
Association. analysis of outcome studies comparing bona
Piper, W. E., Joyce, A. S., McCallum, M., & fide psychotherapies: Empirically, “All must have
Azim, H. F. (1998). Interpretative and supporive prizes.” Psychological Bulletin, 122, 203–215.
C HA P TER
14 Stages of Change
John C. Norcross, Paul M. Krebs, and James O. Prochaska
In the transtheoretical model, behavior represents a period of time as well as a set of

change is conceptualized as a process that tasks needed for movement to the next stage.
unfolds over time and involves progres- Although the time an individual spends in
sion through a series of five stages: precon- each stage may vary, the tasks to be accom-
templation, contemplation, preparation, plished are assumed to be invariant.
action, and maintenance. At each stage of Precontemplation is the stage at which
change, we propose that different change there is no intention to change behavior
processes and relational stances produce in the foreseeable future. Most patients in
optimal progress. Matching psychotherapy this stage are unaware or under-aware of
to the individual patient thus requires their problems. Families, friends, neigh-
matching the processes of change and the bors, or employees, however, are often well
therapeutic relationship to his/her stage of aware that the precontemplators have prob-
change. Furthermore, as clients progress lems. When precontemplators present for
from one stage to the next, the therapeutic psychotherapy, they often do so because
relationship also progresses. of pressure from others. Usually they feel
In this chapter, we review the volumi- coerced into changing by a spouse who
nous research evidence on the stages of threatens to leave, an employer who threat-
change as it applies to psychotherapy. Our ens to dismiss them, parents who threaten
meta-analysis is intended to address two to disown them, or courts that threaten to
specific aims: First, to assess the ability of punish them. Resistance to recognizing or
stages of change and related readiness mea- modifying a problem is the hallmark of
sures to predict psychotherapy outcomes; precontemplation.
and second, to assess the outcomes from Contemplation is the stage in which
psychotherapy studies that matched treat- patients are aware that a problem exists and
ment to specific stages or readiness levels are seriously thinking about overcoming it
of change. We illustrate the meta-analytic but have not yet made a commitment to
results with examples from select studies take action. Contemplators struggle with
using the stages of change. their positive evaluations of their dysfunc-
tional behavior and the amount of effort,
Definitions and Measures energy, and loss it will cost to overcome it.
Stages of Change People can remain stuck in the contempla-
Following are brief descriptions of each tion stage for long periods. In one study, we
of the five stages of change. Each stage followed a group of 200 smokers in the
279
contemplation stage for 2 years. The modal consistently engaging in a new incompati-
response of this group was to remain in the ble behavior for more than 6 months are
contemplation stage for the entire 2 years of the criteria for considering someone to be
the study without ever moving to significant in the maintenance stage. Stabilizing behav-
action (Prochaska & DiClemente, 1983). ior change and avoiding relapse are the
Serious consideration of problem resolution hallmarks of maintenance.
is the central element of contemplation.
Preparation is a stage that combines Measures of Stages and Readiness
intention and behavioral criteria. Indivi- Multiple assessment devices have been
duals in this stage are intending to take developed over the years to assess a person’s
action in the next month and have unsuc- stage of change or “readiness to change.” The
cessfully taken action in the past year. As a measures vary in format—questionnaires,
group, patients prepared for action report algorithms, ladders, and interviews—as well
some small behavioral changes—“baby as in specificity—generic measures for vari-
steps,” so to speak. While they have made ous problems and disorder-specific mea-
some reductions in their problem behav- sures (see the Measures link at www.uri.
iors, patients in the preparation stage have edu/research/cprc/).
not yet reached a criterion for effective The most frequent measure in psycho-
action, such as abstinence from smoking, therapy research studies is the University of
alcohol abuse, or heroin use. They are Rhode Island Change Assessment (URICA;
intending, however, to take such action in McConnaughy et al., 1983, 1989). This
the immediate future. 32-item questionnaire yields separate
Action is the stage in which individuals scores on four continuous scales: Precon-
modify their behavior, experiences, and/or tempation, Contemplation, Action, and
environment in order to overcome their Maintenance (people in preparation score
problems. Action involves the most overt high on both the contemplation and action
behavioral changes and requires consider- scales). Items that are used to identify pre-
able commitment of time and energy. contemplation include “As far as I’m con-
Modifications of the problem made in the cerned, I don’t have any problems that need
action stage tend to be most visible and changing” and “I guess I have faults but
receive the greatest external recognition. there’s nothing that I really need to change.”
Individuals are classified in the action stage Contemplators endorse such items as
if they have successfully altered the dys- “I have a problem and I really think I should
functional behavior for a period from 1 day work on it” and “I’ve been thinking that
to 6 months. Modification of the target I might want to change something about
behavior to an acceptable criterion and sig- myself.” Patients in the action stage endorse
nificant overt efforts to change are the hall- statements like, “I am really working hard
marks of action. to change” and “Anyone can talk about
Maintenance is the stage in which people changing; I am actually doing something
work to prevent relapse and consolidate the about it.” Representative maintenance
gains attained during action. For addictive items are, “I may need a boost right now to
behaviors this stage extends from 6 months help me maintain the changes I’ve already
to an indeterminate period past the initial made” and “I’m here to prevent myself
action. For some behaviors, maintenance from having a relapse of my problem.”
can be considered to last a lifetime. Remain- Other researchers have constructed
ing free of the problem behavior and/or additional measures of readiness to change.
The Stages of Change Readiness and next month are in the preparation stage.
Treatment Eagerness Scale (SOCRATES) Clients who state that they are currently
was developed for measuring readiness for changing their problem are in the action
change with regard to problem drinking stage.
as an alternate measure to the URICA
(Miller & Tonigan, 1996). This 19-item Processes of Change
measure produces three continuous scales: The stages of change represent when people
Ambivalence, Recognition, and Taking change; the processes of change entail how
Steps, which are considered to represent people change. The processes of change
continuously distributed motivational pro- represent an intermediate level of abstrac-
cesses. The SOCRATES has been found to tion between metatheoretical assumptions
be related to quit attempts for smoking ces- and specific techniques spawned by those
sation (DiClemente et al., 1991), alcohol theories. While there are 400-plus ostensi-
use (Isenhart, 1997; Zhang, Harmon, bly different psychotherapies, we have been
Werkner, & McCormick, 2004), and drug able to identify only 8 to 10 different pro-
use (Henderson, Saules, & Galen, 2004). cesses of change based on principal compo-
In fewer research studies but more fre- nents analysis. We prefer to conceptualize
quently in clinical practice, the stages are change in terms of processes or principles,
measured using a series of questions that not in terms of specific techniques.
result in a discrete categorization. We ask if Change processes are overt and covert
the individual is seriously intending to activities that individuals engage in when
change the problem in the near future, typ- they attempt to modify problem behaviors.
ically within the next 6 months. If not, they Each process is a broad category encom-
are classified as precontemplators. Clients passing multiple techniques, methods, and
who state that they are seriously consider- relationship stances traditionally associated
ing changing the problem behavior in the with disparate theoretical orientations.
next 6 months are classified as contempla- Table 14.1 presents the processes receiv-
tors. Those intending to take action in the ing the most theoretical and empirical
Table 14.1 Definitions and Representative Interventions of the Processes of Change

Process Definition: Interventions
Consciousness Increasing information about self and problem: observations; confrontations; interpretations;
raising awareness exercises; bibliotherapy
Self-reevaluation Assessing how one feels and thinks about oneself with respect to a problem: value clarification;
imagery; corrective emotional experience
Dramatic relief Experiencing and expressing feelings about one’s problems and solutions: psychodrama;
(emotional arousal) cathartic work; grieving losses; role-playing
Self-liberation Choosing and commitment to act or believe in ability to change: decision-making methods;
motivational interviewing; commitment-enhancing techniques
Counterconditioning Substituting alternative or incompatible behaviors for problem: relaxation; desensitization;
assertion; cognitive restructuring; behavioral activation.
Stimulus control Avoiding or controlling stimuli that elicit problem behaviors: restructuring one’s environment;
avoiding high-risk cues; fading techniques; altering relationships
Reinforcement Rewarding one’s self or being rewarded by others for making changes: contingency contracts;
overt and covert reinforcement; self-reward
n o rc ro s s , k re b s , p ro c h a s k a 281
support in our work along with their relating that encourages others to take care
definitions and representative interven- of us, or a sense of self-worth based on suf-
tions. A common and finite set of change fering and self-sacrifice. Dramatic relief can
processes has been repeatedly identified also include facing the fear, guilt or regret
across diverse disorders (Prochaska & that would come from not changing. If a
DiClemente, 1985). patient clings tenaciously on to safe and
secure patterns that are also self-defeating
Stages × Processes and self-destructive, how will he/she feel in
The transtheoretical model posits that differ- the future?
ent processes of change are differentially effec- As people progress from precontempla-
tive in certain stages of change. In general tion to contemplation, they rely more on
terms, change processes traditionally associ- the process of self-reevaluation. “How do I
ated with the experiential, cognitive, and think and feel about myself as a couch
psychoanalytic persuasions are most useful potato or a passive person? How will I think
during the earlier precontemplation and con- and feel about myself as a more active
templation stages. Change processes tradi- or proactive person?” Many couch potatoes
tionally associated with the existential and perceive joggers as road hazards, public
behavioral traditions, by contrast, are most nuisances. Why would they want to become
useful during action and maintenance. one of them? The lesson learned here is that
Consciousness raising will help clients psychotherapy can help people find posi-
progress from precontemplation to con- tive images that can draw them into a
templation. In particular, patients need to healthier future, just as the tobacco indus-
increase their awareness of the advantages try provides attractive images that draw
of changing and the multiple benefits of young people into an unhealthy future.
psychotherapy. They also typically benefit As patients progress into the preparation
from enhanced awareness of themselves, stage, they rely more on the process of self-
their disorders, and their defenses. liberation. This is the belief that they have
Contemplation can be a safe haven for the ability to change their behavior and the
clients and therapists alike. Clients are commitment and recommitment to act on
intending to make major changes, but not that belief. This process is what the public
right now. First they need to increase con- calls willpower. People often overrely on
sciousness more and more and more. this process. And when they relapse because
Reflecting, feeling and reevaluating how of overreliance on one change process, they
they have been and how they might become attribute their failure to lack of willpower.
can be hard work at times. But it can also Self-liberation can be enhanced via many
be very meaningful and even fun. And such routes. As with each change process, we try
sharing builds a therapeutic bond that can to provide expert guidance as to whether
be hard to let go. Who wants to give up clients are overutilizing, underutilizing, or
such a close relationship? How can you fail appropriately utilizing willpower compared
as a therapist by having such a good thera- with their peers who have been most suc-
peutic relationship? By allowing your client cessful in progressing from preparation to
to stay stuck in contemplation. action. Such feedback requires scientific
The process of dramatic relief (emotional assessments with adequate reliability and
arousal) can include anticipatory grieving, validity. Another way to enhance self-
the sadness and loss of letting go of a liberation is to give clients choices. If we
best friend in a bottle, of a childlike way of only give them one choice (go to AA), they
won’t be as committed as they would be if stance a transtheoretical therapist (Prochaska)
we gave them two choices (AA or motiva- would adopt with a patient in the precon-
tional interviewing). Two choices won’t templation stage. The client is a 32-year-
enhance willpower as much as three choices old stockbroker in precontemplation for
(AA or motivational interviewing or cogni- chronic cocaine abuse. The stage of change
tive-behavioral therapy). was briefly outlined, and then the client,
During action, clients receiving adequate Donald, was given feedback that his assess-
reinforcement for their efforts secure better ment indicated he was in the precon-
treatment outcomes. One problem is they templation stage. Did he concur? “Yeh,
expect to be reinforced by others much more probably.”
than others will reinforce them. Average
acquaintances are good for one or two rein- Therapist : We know that individuals in
forcements before they start to take the change the precontemplation stage often feel
for granted. Thus, clients need to be prepared coerced into entering therapy rather
to rely more on self rather than social rein- than being there by choice. What
forcements, including from the therapist. pressures were there on you to seek
Clients will learn and practice counter- psychotherapy?
conditioning (response substitution) as they Client : Lots of people have been on my
replace healthier and happier behaviors back. My girlfriend, my mother. My
for their problem behaviors. This process job may be in jeopardy. They all
includes the classic reciprocal inhibition think it’s caused by cocaine. But I’ve
methods: assertion to counter passivity; been using it for years, and it’s never
relaxation to replace anxiety; cognitive sub- been a problem.
stitutions instead of negative thinking; Therapist : How do you react when
exposure to counter avoidance. people pressure you to quit cocaine
As clients progress into the maintenance when you’re not ready?
stage, they do not have to work as hard, but Client : I get angry. I tell them to mind
they have to apply change processes to pre- their own business.
vent relapse. They particularly have to be Therapist : You get defensive.
prepared for the situations that are most Client : Sure, wouldn’t you? Nobody
likely to induce relapse. likes to be told what to do, to be
But for all, psychotherapy will probably treated like a kid.
terminate before the problem is terminated. Therapist : How would you react if
This is one reason why therapists and cli- I told you to quit cocaine?
ents alike can feel anxious about termina- Client : I would get angry. I would tell
tion. They both know that under certain myself you’re just like all the others—
conditions the risk of relapse is real. Of think you know better than me how
course, clients can return for brief therapy to run my life.
if they lapse or relapse. They can analyze Therapist : Would you want to drop out
what they did right, what mistakes they of therapy?
made, and what they need to do differently Client : Probably. I don’t react well to
to keep moving ahead. being controlled.
Therapist : I appreciate you sharing your
Clinical Examples reactions with me. Let me share my
The following exchange from a psychother- main concern. I am concerned that
apy session demonstrates the relational you might drop out of therapy before
I have a chance to make a significant Client : Sure.
difference in your life. I don’t want to Therapist : Would that improve their
coerce or control you. I do want to relationships?
help you to be freer to do what is Client : It should.
best for your life. So will you let me Therapist : And be more open and less
know if I am pressuring you or defensive.
parenting you? Client : I can see that.
Therapist : And do better in their job
Client : You’ll know. and make more money.
Client : I don’t know about that.
Historically, confrontation was one of the
Therapist : It’s true. How about we make
recommended ways of relating to defensive
a deal. If your income goes up 10%,
and resistant clients. By consistently con-
my fee goes up 10%?
fronting patients’ defenses and resistance,
Client : That would be worth it.
therapists expected to be able to break
Therapist : You might not believe this,
through their denial and other defenses.
but there’s only one other thing you
Research has shown, however, that a con-
could do for an hour a week that
frontational style of relating drives many
would give you more benefits than
patients away and increases premature ter-
therapy.
mination (Miller & Rollnick, 2002; Miller,
Client : What’s that?
Wilbourne, & Hettema, 2003).
Therapist : I’m not going to tell you
Later, in the same session, the therapist
because you might invest in that
adopts an affirming, Socratic style and relies
instead.
primarily on consciousness-raising strategies
(Donald laughs!)
that the research evidence suggests will assist
a patient to progress from precontemplation
The psychotherapist’s relational stance at
to contemplation. This entails increasing
different stages can be characterized as fol-
awareness of the pros of changing and the
lows. With patients in precontemplation,
multiple benefits of sticking with treatment.
often the role is like that of a nurturing
Therapist : We know people are likely to parent joining with a resistant youngster
complete therapy if they appreciate who is both drawn to and repelled by
its many benefits. Donald, how do the prospects of becoming more indepen-
you think people benefit from dent. With clients in contemplation,
therapy? the role is akin to a Socratic teacher who
Client : It makes the therapist encourages clients to achieve their own
better off. insights into their condition. With clients
Therapist : That’s good! And how about who are in the preparation stage, the stance
the client? is more like that of an experienced coach
Client : I expect it helps them solve their who has been through many crucial matches
problems. and can provide a fine game plan or can
Therapist : That’s true. And would that review the participant’s own plan. With
help them to feel better about clients who are progressing into action
themselves? and maintenance, the psychotherapist
Client : Yeah, it should. becomes more of a consultant who is avail-
Therapist : And would that improve able to provide expert advice and support
their moods? when action is not progressing as smoothly
as expected. As termination approaches in Tailoring Treatments to Stages
lengthier treatment, the therapist is con- A large number of psychosocial treatments
sulted less and less often as the client have been tailored to stage of change or
experiences greater autonomy and ability readiness for change. These have primarily
to live a life freer from previously disabling been population-based studies delivered via
problems. computer, mail, or phone, with a focus on
health behavior change. Such interventions
have assessed and provided specific feed-
Previous Meta-Analyses
back by stage of change and other con-
Empirical research on the stages of change
structs, such as self-efficacy. Results of these
has taken a number of tacks over the past
studies clearly show the effectiveness of tai-
30 years (for reviews, see Prochaska et al.,
loring or matching to the patient’s stage of
2001; Prochaska & Norcross, 2010), result-
change.
ing in a vast literature. In this section, we
We conducted a meta-analysis on 87
review the results of earlier meta-analyses
prospective, tailored interventions deliv-
on the integration of the stages and pro-
ered via computer or mail across smok-
cesses of change and on the ability of the
ing cessation, physical activity, healthy diet,
stages of change to predict the outcomes of
and mammography screening (Krebs,
behavior change.
Prochaska, & Rossi, in press). The mean
effect size of d = 0.18 (95% CI = 0.16– 0.20)
Stages × Processes represents a 39% increase (OR = 1.39) over
Years of research in behavioral medicine the assessment or minimal care conditions
and psychotherapy converge in showing with which the interventions were com-
that different processes of change are dif- pared and indicates a medium-size effect
ferentially effective in certain stages of for population-based interventions (Rossi,
change. Rosen (2000) performed a meta- 2002). The subset of studies that intervened
analysis of 47 cross-sectional studies exam- on smoking cessation, for instance, resulted
ining the relations of the stages and the in an absolute increase of 6% in quit rates,
processes of change. The studies involved a rate comparable to that observed with
smoking, substance abuse, exercise, diet, 4 to 8 individual in-person counseling
and psychotherapy. The mean effect sizes sessions (Fiore et al., 2008).
(d ) were approximately 0.70 for variation Although supportive of matching or tai-
in cognitive-affective processes by stage loring to the patient’s stages of change,
and 0.80 for variation in behavioral pro- these studies did not include face-to-face
cesses by stage, both moderate-to-large psychotherapy, nor did they address the
effects. Effect sizes for stages by processes disorders most commonly treated by mental
did not vary significantly by the problem health professionals. Thus, we undertook a
treated. For the five studies that examined new meta-analysis specifically focusing on
the change processes in psychotherapy, the stages of change in psychotherapy.
behavioral processes peaked in action while
cognitive-affective processes peaked in con- Meta-Analytic Review: Stages
templation or preparation. Of particular Predicting Outcome
interest was the finding that “use of help- Here, we present the results of an original
ing relationships was strongly related to meta-analysis conducted with aim of gaug-
stages in studies of psychotherapy” (Rosen, ing the ability of the stages of change to
2000, p. 601). predict psychotherapy outcomes.
Search Strategy and Criteria using the standardized mean difference
A combination of search methods was used (Cohen’s d ). Results reported as correla-
to locate all published and in-press studies tions (r), mean differences (F or t), or tests
that matched psychotherapy to stage of of variance (X 2) were transformed to d (per
change or that employed a measure of read- Lipsey & Wilson, 2001). Each obtained
iness for change to predict outcomes after a effect size estimate was weighted by the
course of treatment. The electronic data- inverse of the variance of the estimate,
bases PsycINFO and PubMed were searched which gives greater weight to studies with
for studies indicating reference to psycho- better estimates (for the most part, studies
therapy and stages of change, readiness, with larger sample sizes). If insufficient
and motivation as well as for instru- information was reported for effect size
ments used to measure these constructs calculation, the study was excluded, but if
(e.g., URICA, SOCRATES, Contemplation the study indicated that the effect was
Ladder). To locate studies that may have simply “nonsignificant,” it was included
employed similar techniques, we also con- with the effect size entered as zero, a con-
ducted a forward search for articles that servative strategy that nevertheless preserves
cited identified studies, examined reference some data.
lists from published studies, and searched We employed a random-effects variance
for articles published by authors of studies estimation model. This model assumes
deemed suitable for inclusion. both study-level error and, as well, variabil-
Studies selected for analysis met the fol- ity among studies due to sampling of stud-
lowing criteria, which were consistent with ies from a population of studies. This
inclusion criteria for other meta-analytic enables generalization to a population of
reviews in this volume: (a) studies reported studies. Variability of the random-effects
results of behavioral/psychological face-to- variance component was tested with the Q
face treatment; (b) treatment was provided by test, the significance of which indicates that
mental health professionals; (c) patients had a there is variability among the effect size of
DSM-III or IV diagnosis; (d) treatments con- the sampled studies and suggests that there
sisted of at least three group or individual are factors (i.e., moderators) that could
sessions; (e) readiness to change measured explain this variability.
prior to treatment was used to predict treat- Publication bias, the tendency for signifi-
ment outcome; and (f) sufficient statistical cant study results to get reported more often
information was available to calculate an than nonsignificant results, can upwardly
effect size. skew effect size estimates in meta-analysis.
The search yielded 1,686 references, the Mean effects were assessed for degree of
abstracts of which were reviewed for possi- publication bias using two techniques: fail-
ble inclusion. Of these, 113 papers were safe N, and trim and fill. Fail-safe N calcu-
chosen for full text review, and 39 studies lates the number of unpublished studies
met inclusion criteria and were included in with a null effect size needed to reduce the
the present analysis. overall effect to nonsignificance. Trim and
fill (Duval & Tweedie, 2000) assesses the
Methodological Decisions symmetry of a plot of effect size by sample
The primary database was created, and the size (funnel plot) under the assumption
results were analyzed using the Compre- that when publication bias exists, a dispro-
hensive Meta-Analysis software package portionate number of studies will fall to the
(Biostat, 2006). Effect sizes were calculated bottom right of the plot. This technique
thus determines the number of asymmetri- independence of outcomes, where studies
cal outcomes, imputes their counterparts reported more than one outcome, an over-
to the left, and estimates a corrected mean all mean effect size per study was included
effect size. for calculating the overall mean effect.
The 39 analyzed studies represented
a variety of diagnoses and outcome mea- The Studies
sures with some studies reporting more Table 14.2 summarizes the attributes of the
than one outcome (e.g., substance use and 39 studies, encompassing 8,238 psycho-
treatment dropout). To ensure statistical therapy patients. All studies reported data
Table 14.2 Summary of Studies and Samples (k = 39) included in the Meta-Analysis
Characteristic k %
Country
United States 25 64%
Canada 7 18%
Australia 2 5%
United Kingdom 2 5%
Spain 2 5%
Germany 1 3%
Study design
Single group pre-post 24 62%
Randomized controlled trial 15 38%
Patient age
Adult (18+) 33 85%
Adolescent (13–17) 6 15%
Patient race/ethnicity
White (>60% of sample) 26 67%
Mix (none greater than 60% of sample) 6 15%
African American (>60% of sample) 4 10%
Data not reported 3 8%
Treatment setting
Outpatient 25 64%
Inpatient 14 36%
Treatment manual used 12 31%
Number of treatment sessions
<10 4 10%
10–19 13 33%
20+ 4 10%
Treatment orientation
Cognitive-behavioral 19 49%
12-step 4 10%
Other 5 13%
Readiness Measure
University of Rhode Island Change Assessment 27 69%
Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES) 5 13%
Anorexia Stages of Change Questionnaire 2 5%
Other 5 13%
only from final follow-ups, which were stages of change reliably predict outcomes
mostly conducted immediately upon treat- in psychotherapy. That is, the amount
ment completion. Thirteen studies were of progress clients make during treat-
randomized controlled trials while the ment tends to be a function of their pre-
remainder used a one-group pre-post treatment stage of change. For example, an
design. Six studies concerned treatments intensive action- and maintenance-oriented
for adolescents (ages 13–17), while the smoking cessation program for cardiac
others focused on adults (18+). Sample patients achieved success for 22% of pre-
sizes ranged from N = 42 to N = 1,075, contemplators, 43% of the contemplators,
with an average of 211 participants at and 76% of those in action or prepared
recruitment and a 77% retention rate at for action 6 months later (Ockene et al.,
follow-up. Most samples (k = 26) were 1992).
comprised of primarily white participants If patients progress from one stage to
(>60%), four with primarily African- the next during the first month of treat-
American participants (>60%), and six ment, they can double their chances
studies recruited a racially mixed sample. of taking action in the next 6 months. Of
(Note that k denotes the number of stud- the precontemplators who were still in pre-
ies, in contrast to N, which refers to the contemplation at 1-month follow-up, only
number of participants in a study.) Samples 3% took action by 6 months. For the pre-
on average were 38% female (and ranged contemplators who progressed to contem-
0%–100%). Fourteen studies conducted plation at 1 month, 7% took action by
interventions in an inpatient setting, and 6 months. Similarly, of the contempla-
the number of treatment sessions ranged tors who remained in contemplation at
from 4 to 28 with 12 being the modal 1 month, only 20% took action by 6 months.
number. Twelve studies reported using a At 1 month, 41% of the contemplators
treatment manual, with cognitive-behavioral who progressed to the preparation stage
treatment the most common theoretical attempted to quit by 6 months (Prochaska,
orientation guiding treatment (k = 19). The DiClemente, Velicer, et al. 1985). Such
most common readiness measures were the data indicate that treatments designed
University of Rhode Island Change to help patients progress just one stage
Assessment (URICA; k = 27) and the Stages in a month can double the chances of
of Change Readiness and Treatment participants taking action in the near
Eagerness Scale (SOCRATES; k = 5). future.
Effect Size Effect Size by Outcome

The 39 studies reported 71 separate out- We analyzed the effect size for the stages of
comes. Results of the individual studies are change for two particular outcomes of
summarized in Table 14.3. The mean effect interest: enhancement of the working alli-
size was d = .46 with a 95% confidence ance (k = 4) and adherence to treatment/
interval of .35 to .58 (range −.20 to 2.7), premature dropout (k = 24). Three studies
Q(38) = 186.05, p < 0.001. Analysis of included working alliance as an outcome
publication bias suggested a fail-safe N with one reporting outcomes from two
of 2,554. samples (Connors et al., 2000). The mean
By convention, a d of .46 indicates a effect size for these four outcomes was
medium effect, demonstrating that the d = .61 (95% CI = .36–.86, p < 0.001).
Table 14.3 Effect Sizes by Study
Study Primary diagnosis Readiness measure N d SE 95% CI
Lower Upper
Alexander & Morris, 2008 Domestic abuse URICA 210 0.44 0.19 0.08 0.81
Ametller et al., 2005 Eating disorder Anorexia Stage 70 0.34 0.12 0.10 0.58
of Change
Blanchard et al., 2003 Substance abuse URICA 252 0.16 0.13 −0.10 0.42
Brodeur et al., 2008 Domestic abuse URICA-DV 302 0.11 0.12 −0.12 0.34
Callaghan et al., 2005 Substance abuse URICA 130 0.74 0.19 0.37 1.11
Callaghan et al., 2008 Substance abuse URICA 60 0.37 0.41 −0.44 1.18
(Budney, 2000)
Callaghan et al., 2008 Substance abuse URICA 90 0.62 0.29 0.05 1.20
(Budney, 2006)
Carpenter et al., 2002 Substance abuse URICA 174 0.49 0.22 0.05 0.92
Chung & Maisto, 2009 Substance abuse Contemplation Ladder 142 0.03 0.23 −0.43 0.49
Connors et al., 1998a Alcohol abuse URICA 682 0.49 0.08 0.34 0.65
Connors et al., 1998b Alcohol abuse URICA 465 0.52 0.10 0.33 0.70
Demmel et al., 2004 Alcohol abuse SOCRATES 350 0.58 0.13 0.33 0.83
Derisley et al., 2000 General therapy URICA 60 1.30 0.32 0.68 1.92
Dozois et al., 2004 Anxiety URICA 81 0.34 0.24 −0.12 0.80
Eckhardt et al., 2008 Domestic abuse URICA-DV 199 0.52 0.16 0.21 0.84
Geller et al., 2004 Eating disorder RMI 60 0.78 0.35 0.10 1.47
Gossop et al., 2006 Substance abuse SOCRATES 1,075 0.23 0.08 0.08 0.38
Haller et al., 2004 Substance abuse URICA 75 0.87 0.26 0.36 1.38
Henderson et al., 2004 Substance abuse URICA 96 0.63 0.22 0.20 1.06
Hewes & Janikowski, 1998 Alcohol abuse SOCRATES 58 2.49 0.60 1.31 3.68
Hunt et al., 2006 PTSD URICA 42 0.68 0.35 0.00 1.36
Isenhart 1997 Alcohol abuse SOCRATES 125 0.69 0.19 0.32 1.07
Kerns et al., 2000 Pain management Pain Stages of Change 68 0.24 0.10 0.05 0.44
Kinnaman et al., 2007 Alcohol abuse URICA 120 −0.02 0.19 −0.39 0.34
Lewis et al., 2009 Depression Stage of Change Q 332 0.30 0.12 0.08 0.53
Mitchell, 2006 Substance abuse SOCRATES 357 0.71 0.11 0.49 0.93
Pantalon et al., 2002 Substance abuse URICA 117 0.14 0.20 −0.25 0.52
Pantalon et al., 2003 Psychiatric URICA 120 −0.20 0.09 −0.38 −0.02
inpatients
Petry et al., 2005 Gambling disorder URICA 234 0.70 0.16 0.38 1.01
Project Match Group, 1999 Alcohol abuse URICA 806 0.28 0.07 0.14 0.42
(Continued)
289
Table 14.3 Continued
Study Primary diagnosis Readiness measure N d SE 95% CI
Lower Upper
Rooney et al., 2007 PTSD URICA 50 0.63 0.31 0.03 1.23
Scott & Wolfe, 2003 Domestic abuse URICA 194 0.63 0.21 0.23 1.04
Smith et al., 1995 General therapy URICA 74 1.84 0.33 1.20 2.48
Soler et al., 2008 Borderline PD URICA 60 0.54 0.61 −0.67 1.74
Stotts et al., 2003 Alcohol abuse URICA 115 0.49 0.24 0.03 0.96
Tambling & Johnson, 2008 Relational problem URICA 469 −0.09 0.13 −0.34 0.15
Treasure et al., 1999 Eating disorder URICA 125 0.70 0.47 −0.22 1.61
Wade et al., 2009 Eating disorder Anorexia Stage of 47 2.67 0.50 1.68 3.65
Change
Willoughby et al., 1996 Alcohol abuse URICA 152 −0.15 0.17 −0.49 0.18
Overall Effect Size 0.46 0.06 0.35 0.58
URICA = University of Rhode Island Change Assessment; SOCRATES = Stages of Change Readiness and Treatment Eagerness Scale;
RMI = Readiness and Motivation Interview.
For the 24 studies that reported client Potential Moderators

adherence to suggested treatment or pre- Categorical moderators were examined
mature dropout from treatment, the mean using a statistical test for meta-analysis that
effect size was d = .42 (95% CI = .24–.60). employs weighted data and compares
The stage of change reliably predicts within- and between-group heterogene-
psychotherapy dropout, which is an impor- ity using the Q statistic as employed by
tant finding given that a review of 126 the Comprehensive Meta-analysis software
studies found that about 50% of patients package (Biostat, 2006). A sample size of
will leave treatment prematurely (Pekarik 10 or more studies is necessary to pro-
& Wierzbicki, 1986). While stage of change vide sufficient statistical power for detect-
alone is an important indicator of treating differences between groups (Lipsey &
ment dropout, we found in a separate study Wilson, 2001). Continuous moderators
that assessing both stage and processes of were examined using meta-regression tech-
change predicted psychotherapy dropout niques, which correct variance estimates
with 90% accuracy among clients with a for sample size.
variety of mental health problems (Brogan, The significant Q test for our meta-analysis
Prochaska, & Prochaska, 1999). The 40% indicated that there was sufficient variabil-
of the patients who terminated quickly ity among the effect sizes of the studies to
(less than three sessions) and prematurely, look for moderators that could explain this
as judged by their therapists, had a group variability. We conducted moderator analy-
profile representing the precontemplation ses for patient characteristics, treatment
stage. The 20% of patients who terminated features, and diagnostic categories. We
quickly but appropriately had a group pro- could not search for potential moderators
file representing action, while the 40% who of assessment time or rater perspective as all
remained in psychotherapy had a stage pro- stage measures were completed by patients
file similar to contemplation. at the beginning of intake or treatment.
Nor was there sufficient variability in distress, which were deemed sufficiently
these 39 studies in the measures used similar to group together to increase reli-
to assess stages to explore moderators; ability of the estimate. Outcome measures
more than 30 studies employed the University included the State-Trait Anxiety Inventory,
of Rhode Island Change Assessment Beck Depression Inventory, Children’s
(URICA). Depression Rating Scale, and the Outcome
For patient characteristics, we found no Questionnaire 45. The mean effect size was
statistically significant difference between d = .45 (95% CI = .19–.71, p < .001).
adolescent and adult populations, nor by Across studies, readiness to change was
race/ethnicity (all ps > .10). However, effect moderately to strongly related to progress
size was positively correlated with having a in psychotherapy for various DSM-IV dis-
larger number of female participants orders. For instance, low motivation as
(p = .02). For treatment features, we found indicated on the Anorexia Nervosa Stages
no differences in effect size between inpa- of Change Questionnaire predicted hospi-
tient and outpatient treatment settings, talization in adolescent patients (Ametller
between treatments that used a manual or et al., 2005) as well as improvement
those that did not, nor by number of treat- in problem eating (Wade et al., 2009).
ment sessions. However, for studies report- Changes in stages predicted PTSD symp-
ing primary theoretical orientation, 12-step tom severity in a population of veterans
programs had the highest effect size (k = 4, at treatment follow-up 3 months later
d = .73) as compared to cognitive-behav- (Rooney et al., 2007). Improvement in
ioral treatment (k = 19, d = .39) or other Action scores during psychotherapy was
orientations (k = 5, d = .24; p = .001). positively related to decreases on the
We also analyzed the effect size of the Children’s Depression Rating Scale after
stages of change for particular diagnos- 12 weeks of treatment (Lewis et al., 2009).
tic categories: addictions, eating disorders,
and mood disorders. Fourteen studies Meta-Analytic Review: Stage-
predicted addiction outcomes using base- Matched Treatments
line readiness to change. The most fre- Our second aim was to conduct a meta-
quently used outcome measures were the analysis that assessed the outcomes from
Addiction Severity Index, Severity of psychotherapy studies that matched treat-
Dependence Scale, Timeline Followback, ment to specific stages or readiness levels
and the Alcohol Use Questionnaire. The of change. We were interested in learning
mean effect for the 14 studies was d = .37 whether stage-matching clients in psycho-
(95% CI = .23–.52, p < .001). Four studies therapy produced the superior results found
assessed the relationship between baseline in behavioral medicine and population-
readiness to change and prediction of eating based studies reviewed earlier. Unfortu-
disorder outcomes. Two studies employed nately, we located no controlled group
the Eating Disorders Inventory, one a mea- studies meeting our inclusion criteria that
sure from the European COST Action B6 matched psychotherapy to client stage or
Project, and one a count of relapse to assess readiness. As a result, we could not perform
outcomes. The mean effect size was d = .99 a meta-analysis.
(95% CI = 0.24–1.74, p < .001). Seven A number of studies did use in-person
studies assessed the relationship between sessions and delivered treatment based on
baseline readiness to change and prediction stage or readiness to change but other-
of mood disorder symptoms or relational wise did not meet inclusion criteria in that
treatment either was a single session, pro- In the computer condition, participants
vided by medical staff, or focused on health completed by mail or telephone 40 ques-
behaviors such as smoking, physical activ- tions that were entered into computers that
ity, or diabetes management (Champion generated feedback reports. These reports
et al., 2003; Chouinard & Robichaud- informed participants about their stage of
Ekstrand, 2007; Clark, Hampson, Avery, change, their pros and cons of changing,
& Simpson, 2004; Patten, et al., 2008; Van and their use of change processes appropri-
Sluijs, Van Poppel, Twisk, Brug, & Van ate to their stages. At baseline, partici-
Mechelen, 2005; Wiggers et al., 2005). The pants were given positive feedback on what
one study that intervened on psychiatric they were doing correctly and guidance on
and substance use diagnoses was not indi- which principles and processes they needed
vidually stage tailored (James et al., 2004). to apply more in order to progress. In two
All of the studies we did locate reported progress reports delivered over the next
findings in support of stage-matching 6 months, participants also received posi-
treatments. tive feedback on any improvement they
The failure to locate stage-matching made on any of the variables relevant to
studies in psychotherapy reflects, first, the progressing.
obvious dearth of such studies, and second, In the personalized condition, smokers
the limited reach of conventional psycho- received four proactive counselor calls over
therapy. Psychotherapy has traditionally the 6-month intervention period. Three of
taken a passive and narrow approach to the calls were based on the computer
health care—passively waiting for individ- reports. Counselors reported much more
uals suffering from mental disorders in the difficulty in interacting with participants
contemplation or preparation stages to without any progress data. Without scien-
contact their offices. When psychotherapy tific assessments, it was harder for both cli-
proactively reaches out to individuals and ents and counselors to tell whether any
populations alike, suffering from all behav- significant progress had occurred since their
ioral health conditions, in all stages of last interaction.
change, then we will achieve a transforma- Abstinence rates were compared for
tion in psychotherapy. each of the four treatment groups over
To illustrate, several of our studies inves- 18 months with treatment ending at
tigated the results of reaching out to patient 6 months. The two self-help manual condi-
populations. A series of clinical trials apply- tions paralleled each other for 12 months.
ing stage-matched interventions for health At 18 months, the stage-matched manuals
behavior change have been conducted. moved ahead (18% vs. 11% abstinent).
In our first large-scale clinical trial, we This is an example of a delayed action effect,
compared four treatments: a home-based which we often observe with stage-matched
action-oriented tobacco cessation pro- programs specifically and which others
gram (standardized); stage-matched manu- have observed with self-help programs gen-
als (individualized); expert system computer erally. It takes time for participants in early
reports plus manuals (interactive); and stages to progress all the way to action.
counselors plus computers and manuals Therefore, some treatment effects as mea-
(personalized). We randomly assigned by sured by action will be observed only after
stage 739 smokers to one of the four treat- considerable delay.
ments (Prochaska, DiClemente, Velicer, & The results of the computer alone
Rossi, 1993). and computer plus counselor conditions
paralleled each other for 12 months. Then, the expert system proved 34% more effec-
the effects of the counselor condition flat- tive than assessment alone; in the second
tened out (18%) while the computer con- population study, it was 31% more effec-
dition effects continued to increase (25% tive (23.2% abstinent vs. 17.5%). Working
abstinent). We can only speculate as to the with populations, we were able to produce
delayed differences between these two con- the outcomes normally found in intense
ditions. Participants in the personalized clinic-based programs with low participa-
condition may have become somewhat tion rates of much more selected samples of
dependent on the social support and social smokers, namely, about 25% abstinence at
control of the counselor calling. The last long-term follow-up. The research to date
call was after the 6 months assessment and indicates that proactive, stage-matched
benefits would be observed at 12 months. treatments emerge as a powerful and inclu-
Termination of the counselors could result sive approach to behavior change.
in no further progress because of the loss
of social support. The classic pattern for Limitations of the Research
therapies for all addictions is rapid relapse Although more than 1,500 research stud-
beginning as soon as the treatment is termi- ies have been conducted on the stages of
nated. Some of this rapid relapse could well change, none have directly and prospec-
be due to the sudden loss of social support tively matched and mismatched psycho-
or social control provided by the counselors therapy to the patient’s stage of change.
and other participants in therapy Rather, the available research concerns the
programs. predictive utility of the stages of change in
The next test was to demonstrate the effi- terms of outcomes and dropouts, the dif-
cacy of the expert system when applied to ferential use of the processes of change at
an entire population recruited proactively. various stages of change, and the relative
With over 80% of 5,170 smokers partici- efficacy of diverse forms of service delivery.
pating and fewer than 20% in the prepara- Further, the majority of published research
tion stage, we demonstrated significant concerns health behaviors and addictive
benefit of the expert system at each 6-month disorders, as contrasted to the wide range
follow-up (Prochaska et al., 2005). The of neurotic disorders.
point prevalence abstinence rates for expert
stage-matched systems versus assessment
alone were: 9.7% vs. 7.4%; 18.0% vs.
Three decades of clinical research on the
14.5%; 21.7% vs. 16.6%; and 25.6% vs.
stages of change, including the meta-analyses
19.7% at 6, 12, 18, and 24 months, respec-
reviewed in this chapter, have identified a
tively. The advantages over proactive assess-
number of therapist behaviors that will
ment alone increased at each follow-up for
improve psychotherapy outcomes.
the full 2 years assessed. The implications
here are that stage-matched interventions • Assess the client’s stage of change.
in a population can continue to demon- Probably the most obvious and direct
strate benefits long after the intervention implication is to assess the stage of a
has ended. client’s readiness for change and to tailor
The system’s efficacy was replicated in treatment accordingly. In clinical practice,
an HMO population of 4,000 smokers assessing stage of change typically entails
with 85% participation (Prochaska et al., a straightforward question: “Would you
2001). In the first population-based study, say you are not ready to change in the next
6 months (precontemplation), are thinking underestimate the pros of changing,
about changing in the next 6 months overestimate the cons, and are not
(contemplation), are thinking about particularly conscious that they are
changing in the next month (preparation), making such mistakes (Prochaska, 1994).
or have already made some progress Compared with their peers in other stages,
(action)?” The stages are problem specific, precontemplators rate the cons of
so the question will probably be asked changing—and of psychotherapy—as
several times for multidisordered patients. higher than the pros (Hall & Rossi, 2008).
• Beware treating all patients as though No wonder they are at a high risk for
they are in action. Professionals frequently dropping out. If psychotherapists try to
design excellent action-oriented treatments impose action on these patients, they are
but then are disappointed when only a likely to drive them away, consequently
small percentage of clients seek that blaming the clients for being resistant,
therapy or remain in therapy. The vast unmotivated, noncompliant, or not ready
majority of patients are not in the action for therapy. Historically, it has been
stage. Aggregating across studies and therapists who were not ready or
populations (Velicer et al., 1995), we motivated to match their relationship and
estimate that 40% are in interactions to the clients’ needs, and who
precontemplation, 40% in contemplation, were resistant to trying new approaches to
and only 20% prepared for action. Thus, retaining more clients. Motivational
professionals offering only action-oriented Interviewing (Miller & Rollnick, 2002)
programs are likely to underserve or has brilliantly incorporated these lessons
misserve the majority of their target into its philosophical spirit and its
population. The therapeutic treatment methods.
recommendation is to move from an • Tailor the processes to the stages.
action paradigm to a stage paradigm. The research reliably demonstrates that
• Set realistic goals by moving one stage patients optimally progress from
at a time. A goal for many patients, precontemplation and contemplation into
particularly in a time-limited managed preparation by use of consciousness
care environment, is to set realistic goals, raising, self-liberation, and dramatic relief/
such as helping patients progress from emotional arousal. Patients progress best
precontemplation to contemplation. Such from preparation to action and
progress means that patients are changing maintenance by use of
if we view change as a process that unfolds counterconditioning, stimulus control,
over time, through a series of stages. and reinforcement management. To
Helping patients break out of the chronic, simplify: change processes traditionally
stuck phase of precontemplation is a associated with the insight or awareness
therapeutic success, since it almost doubles therapies for the early stages, and change
the chances that patients will take effective processes associated with the action
action in the next 6 months. If we can therapies for the later stages.
help them progress two stages with brief • Avoid mismatching stages and processes.
therapy, we triple the chances they will A person’s stage of change provides
take effective action. proscriptive as well as prescriptive
• Treat precontemplators gingerly. We information on treatments of choice.
know that, across every disorder that has Action-oriented therapies may be quite
been studied, people in precontemplation effective with individuals who are in the
preparation or action stages. These same a precontemplator, a Socratic teacher role
programs tend to be ineffective or with contemplator, an experienced coach
detrimental, however, with individuals in with a patient in action, and then a consul-
precontemplation or contemplation. tant once into maintenance.
We have observed two frequent
mismatches (Prochaska, Norcross, & • Practice integratively. Psychotherapists
DiClemente, 1995). First, some therapists moving with their patients through the
rely primarily on change processes most stages of change over the course of
indicated for the contemplation stage— treatment will probably employ relational
consciousness raising, self-reevaluation— stances and change processes traditionally
while they are moving into the action stage. emphasized by disparate systems of
They try to modify behaviors by becoming psychotherapy. That is, they will practice
more aware, a common criticism of classi- integratively (Norcross & Goldfried,
cal psychoanalysis: insight alone does not 2005). Competing systems of
necessarily bring about behavior change. psychotherapy have promulgated
Second, other therapists rely primarily on purportedly rival processes of change.
change processes most indicated for the However, ostensibly contradictory
action stage—reinforcement management, processes become complementary when
stimulus control, counterconditioning— embedded in the stages of change. While
without the requisite awareness, decision some psychotherapists insist that such
making, and readiness provided in the con- theoretical integration is philosophically
templation and preparation stages. They impossible, our research has consistently
try to modify behavior without awareness, documented that psychotherapists in their
a common criticism of radical behaviorism: consultation rooms can be remarkably
overt action without insight is likely to lead effective in synthesizing powerful change
to temporary change. processes across the stages (Valasquez,
Maurer, Crouch, & DiClemente, 2001).
• Prescribe stage-matched “relationships • Anticipate recycling: Most
of choice” as well as “treatments of choice.” psychotherapy patients will recycle several
We conceptualize this practice, paralleling times through the stages before achieving
the notion of “treatments of choice” in long-term maintenance. Accordingly,
terms of treatment methods, as offering professionals and programs expecting
“therapeutic relationships of choice” in people to progress linearly through the
terms of interpersonal stances (Norcross & stages of change are likely to gather
Beutler, 1997). Once you know a patient’s disappointing results. Be prepared to
stage of change, then you will know which include relapse prevention in treatment,
relationship stances to apply in order to anticipate the probability of recycling
help him/her progress to the next stage patients, and try to minimize therapist
and eventually maintenance. Rather guilt and patient shame over recycling
than apply therapy relationships in a (Prochaska, Norcross, & DiClemente,
haphazard or trial-and-error manner, 2005).
practitioners can use them in a more • Shift to an expanded view of
systematic style across the course of psychotherapy as proactive, population-based
psychotherapy. health care. Psychotherapists need not
These relational matches, as reviewed discard effective means of assisting
earlier, entail a nurturing parent stance with individuals suffering from mental
disorders. Instead, we can add to these S. Rollnick (Eds.), Motivational interviewing:
invaluable services by providing proactive Preparing people for change. New York: Guilford.
DiClemente, C. C., Prochaska, J. O., Fairhurst, S. K.,
recruitment and treatment of entire
Velicer, W. F., Velasquez, M. M., & Rossi, J. S.
populations suffering from chronic (1991). The process of smoking cessation:
biobehavioral conditions. Such an An analysis of precontemplation, contempla-
expansion could produce unprecedented tion and preparation stages of change. Journal
impacts on the health and happiness of of Consulting and Clinical Psychology, 59,
the populace. 295–304.
Duval, S., & Tweedie, R. (2000). Trim and fill:
A simple funnel-plot-based method of testing and
References adjusting for publication bias in meta-analysis.
Biostat. (2006). Comprehensive meta-analysis. Biometrics, 56(2), 455–63.
Englewood, NJ: Author. Fiore, M., Jaen, C. R., Baker, T. B., et al.
Brogan, M. M., Prochaska, J. O., & Prochaska, J. M. (2008). Treating tobacco use and dependence:
(1999). Predicting termination and continua- 2008 Update. Rockville, MD: U.S. Department
tion status in psychotherapy using the transtheo- of Health & Human Services. Public Health
retical model. Psychotherapy, 36, 105–113. Service.
Budney, A. J., Moore, B. A., Rocha, H. L., & Hall, K. L., & Rossi, J. S. (2008). Meta-analytic
Higgins, S. T. (2006). Clinical trial of abstinence- examination of the strong and weak principles
based vouchers and cognitive-behavioral therapy across 48 health behaviors. Preventive Medicine,
for cannabis dependence. Journal of Consulting 46(3), 266–74.
and Clinical Psychology, 74, 307–316. Henderson, M. J., Saules, K. K., & Galen, L. W.
Budney, A. J., Higgins, S. T., Radonovich, K. J., (2004). The predictive validity of the University
& Novy, P. L. (2000). Adding voucher-based of Rhode Island Change Assessment question-
incentives to coping skills and motiva- naire in a heroin-addicted polysubstance abuse
tional enhancement improves outcomes during sample. Psychology of Addictive Behaviors, 18(2),
treatment for marijuana dependence. Journal 106–112.
of Consulting and Clinical Psychology, 68, Isenhart, C. E. (1997). Pretreatment readiness for
1051–1061. change in male alcohol dependent subjects:
Champion, V., Maraj, M., Hui, S., Perkins, A. J., Predictors of one-year follow-up status. Journal
Tierney, W., Menon, U., et al. (2003). of Studies on Alcohol, 58(4), 351–57.
Comparison of tailored interventions to increase James, W., Preston, N. J., Koh, G., Spencer, C.,
mammography screening in nonadherent older Kisely, S. R., & Castle, D. J. (2004). A group
women. Preventive Medicine, 36(2), 150–58. intervention which assists patients with dual
Chouinard, M. C., & Robichaud-Ekstrand, S. diagnosis reduce their drug use: A randomized
(2007). Predictive value of the transtheoreti- controlled trial. Psychological Medicine, 34(6),
cal model to smoking cessation in hospitalized 983–90.
patients with cardiovascular disease. European Krebs, P., Prochaska, J. O., & Rossi, J. S. (in press).
Journal of Cardiovascular Prevention and Rehabili- Defining what works in tailoring: A meta-
tation, 14(1), 51–58. analysis of computer-tailored interventions for
Clark, M., Hampson, S. E., Avery, L., & Simpson, R. cancer-preventive behavior change. Preventive
(2004). Effects of a tailored lifestyle self-manage- Medicine.
ment intervention in patients with Type 2 diabe- Lipsey, M., & Wilson, D. B. (2001). Practical meta-
tes. British Journal of Health Psychology, 9(3), analysis. Thousand Oaks, CA: Sage.
365–79. McConnaughy, E. A., DiClemente, C. C.,
Connors, G. J., DiClemente, C. C., Dermen, K. H., Prochaska, J. O., & Velicer, W. F. (1989). Stages
Kadden, R., Carroll, K. M., & Frone, M. R. of change in psychotherapy: A follow-up report.
(2000). Predicting the therapeutic alliance in Psychotherapy, 26, 494–503.
alcoholism treatment. Journal of Studies on McConnaughy, E. A., Prochaska, J. O., & Velicer,
Alcohol, 61(1), 139–49. W. F. (1983). Stages of change in psychotherapy:
DiClemente, C. C. (1991). Motivational interview- Measurement and sample profiles. Psychotherapy,
ing and the stages of change. In W. R. Miller & 20, 368–75.
Miller, W. R., & Rollnick, S. (2002). (Eds.), Prochaska, J. O., DiClemente, C. C., Velicer, W. F.,
Motivational interviewing: Preparing people for Ginpil, S., & Norcross, J. C. (1985). Predicting
change (2nd ed.). New York: Guilford. change in smoking status for self-changers.
Miller, W. R., & Tonigan, J. S. (1996). Assessing Addictive Behaviors, 10, 395–406.
drinkers’ motivation for change: The Stages of Prochaska, J. O., DiClemente, C. C., Velicer, W. F.,
Change Readiness and Treatment Eagerness & Rossi, J. S. (1993). Standardized, individual-
Scale (SOCRATES). Psychology of Addictive ized, interactive, and personalized self-help pro-
Behaviors, 10(2), 81–89. grams for smoking cessation. Health Psychology,
Miller, W. R., Wilbourne, P. L., & Hettema, J. E. 12(5), 399–405.
(2003). What works? A summary of alcohol Prochaska, J. O., & Norcross, J. C. (2010). Systems
treatment outcome research. In R. K. Hester & of psychotherapy: A transtheoretical analysis (7th
W. R. Miller (Eds.), Handbook of alcoholism treat- edition). Pacific Grove, CA: Brooks/Cole.
ment approaches: Effective alternatives (3rd ed., Prochaska, J. O., Norcross, J. C., & DiClemente, C. C.
pp. 13–63). Boston: Allyn & Bacon. (1995). Changing for good. New York: Avon.
Norcross, J. C., & Beutler, L. E. (1997). Determining Prochaska, J. O., Norcross, J. C., & DiClemente, C. C.
the therapeutic relationship of choice in brief (2005). Stages of change: Prescriptive guidelines.
therapy. In J. N. Butcher (Ed.), Objective psycho- In G. P. Koocher, J. C. Norcross, & S. S. Hill.
logical assessment in managed health care: A practi- (Eds.), Psychologists’ desk reference (2nd edition).
tioner’s guide. New York: Oxford University New York: Oxford University Press.
Press. Prochaska, J. O., Norcross, J. C., Fowler, J, Follick, M.,
Norcross, J. C., & Goldfried, M. R. (Eds.). (2005). & Abrams, D. B. (1992). Attendance and out-
Handbook of psychotherapy integration (2nd ed.). come in a work-site weight control program:
New York: Oxford University Press. Processes and stages of change as process
Ockene, J., Kristellar, J., Ockene, I., & Goldberg, R. and predictor variables. Addictive Behaviors, 17,
(1992). Smoking cessation and severity of illness. 35–45.
Health Psychology, 11, 119–26. Prochaska, J. O., Redding, C. A., & Evers, K.
Patten, C. A., Decker, P. A., Dornelas, E. A., (2001). The transtheoretical model and stages of
Barbagallo, J., et al. (2008). Changes in readiness change. In K. Glanz, F. M. Lewis, & B. K. Rimer
to quit and self-efficacy among adolescents (Eds.), Health behavior and health education (3rd
receiving a brief office intervention for smoking Edition). San Francisco: Jossey-Bass.
cessation. Psychology, Health and Medicine, 13(3), Prochaska, J. O., Velicer, W. F., Fava, J. L.,
326–36. Ruggiero, L., Laforge, R. G., Rossi, J. S., et al.
Pekarik, G., & Wierzbicki, M. (1986). The relation- (2001). Counselor and stimulus control enhance-
ship between clients’ expected and actual treatments of a stage-matched expert system inter-
ment duration. Psychotherapy, 23(4), 532–34. vention for smokers in a managed care setting.
Prochaska, J. O. (1994). Strong and weak principles Preventive Medicine, 32(1), 23–32.
for progressing from precontemplation to action Prochaska, J. O., Velicer, W. F., Fava, J. L., Rossi, J. S.,
on the basis of twelve problem behaviors. Health & Tsoh, J. Y. (2001). Evaluating a population-
Psychology, 13(1), 47–51. based recruitment approach and a stage-based
Prochaska, J. O., & DiClemente, C. C. (1983). expert system intervention for smoking cessa-
Stages and processes of self-change in smoking: tion. Addictive Behaviors, 26, 583–602.
Toward an integrative model of change. Journal Prochaska, J. O., Velicer W. F., Redding, C. A.,
of Consulting and Clinical Psychology, 5, 390–95. Rossi, J. S., Goldstein, M., DePue, J., et al.
Prochaska, J. O., & DiClemente, C. C. (1985). (2005). Stage-based expert systems to guide a pop-
Common processes of self-change in smoking, ulation of primary care patients to quit smoking,
weight control, and psychological distress. In eat healthier, prevent skin cancer and receive
S. Shiffman & T. Wills (Eds.), Coping and sub- regular mammograms. Preventive Medicine, 41,
stance abuse: A conceptual framework. New York: 406–416.
Academic Press (pp. 345–63). Project MATCH Research Group. (1999). Matching
Prochaska, J. O., DiClemente, C. C., & Norcross, J. C. alcoholism treatments to client heterogeneity:
(1992). In search of how people change: Applica- Project MATCH posttreatment drinking
tions to addictive behaviors. American Psychologist, outcomes. Journal of Studies on Alcohol,
47, 1102–1114. 58, 7–29.
Rosen, C. S. (2000). Is the sequencing of change Blanchard, K. A., Morgenstern, J., Morgan, T. J.,
processes by stage consistent across health prob- Labouvie, E., & Bux, D. A. (2003). Motivational
lems? A meta-analysis. Health Psychology, 19(6), subtypes and continuous measures of readiness
593–604. for change: Concurrent and predictive validity.
Rossi, J. S. (2002). Comparison of the use of signifi- Psychology of Addictive Behaviors, 17(1),
cance testing and effect sizes in theory-based health 56–65.
promotion research. Paper presented at the 43rd Brodeur, N., Rondeau, G., Brochu, S., Lindsay, J.,
Annual Meeting of the Society for Multivariate & Phelps, J. (2008). Does the transtheoretical
Experimental Psychology. model predict attrition in domestic violence
Valasquez, M. M., Maurer, G., Crouch, C., & treatment programs? Violence and Victims, 23(4),
DiClemente, C. C. (2001). Group treatment for 493–507.
substance abuse: A stages-of-change therapy manual. Callaghan, R. C., Hathaway, A., Cunningham, J. A.,
New York: Guilford. Vettese, L. C., Wyatt, S., & Taylor, L. (2005).
Van Sluijs, E. M. F., Van Poppel, M. N. M., Does stage-of-change predict dropout in a cul-
Twisk, J. W. R., Brug, J., & Van Mechelen, W. turally diverse sample of adolescents admitted to
(2005). The positive effect on determinants of inpatient substance-abuse treatment? A test of
physical activity of a tailored, general practice- the transtheoretical model. Addictive Behaviors,
based physical activity intervention. Health 30(9), 1834–47.
Education Research, 20(3), 345–56. Callaghan, R. C., Taylor, L., Moore, B. A.,
Velicer, W. F., Fava, J. L., Prochaska, J. O., Jungerman, F. S., Vilela, F. A. D. B., &
Abrams, D. B., Emmons, K. M., & Pierce, J. P. Budney, A. J. (2008). Recovery and URICA
(1995). Distribution of smokers by stage in three stage-of-change scores in three marijuana treat-
representative samples. Preventive Medicine, ment studies. Journal of Substance Abuse
24(4), 401–411. Treatment, 35(4), 419–26.
Wiggers, L. C. W., Oort, F. J., Dijkstra, A., Carpenter, K. M., Miele, G. M., & Hasin, D. S.
De Haes, J. C. J. M., Legemate, D. A., & (2002). Does motivation to change mediate the
Smets, E. M. A. (2005). Cognitive changes in effect of DSM-IV substance use disorders on
cardiovascular patients following a tailored treatment utilization and substance use? Addictive
behavioral smoking cessation intervention. Behaviors, 27(2), 207–25.
Preventive Medicine, 40(6), 812–21. Chung, T., & Maisto, S. A. (2009). “What I got
Zhang, A. Y., Harmon, J. A., Werkner, J., & from treatment”: Predictors of treatment content
McCormick, R. A. (2004). Impacts of motiva- received and association of treatment content
tion for change on the severity of alcohol use by with 6-month outcomes in adolescents. Journal
patients with severe and persistent mental illness. of Substance Abuse Treatment, 37(2), 171–81.
Journal of Studies on Alcohol, 65(3), 392–97. Demmel, R., Beck, B., Richter, D., & Reker, T.
(2004). Readiness to change in a clinical sample
of problem drinkers: Relation to alcohol use,
Studies Included in the Meta-Analysis self-efficacy, and treatment outcome. European
Alexander, P. C., & Morris, E. (2008). Stages of Addiction Research, 10(3), 133–38.
change in batterers and their response to treat- Derisley, J., & Reynolds, S. (2000). The transtheo-
ment. Violence and Victims, 23(4), 476–92. retical stages of change as a predictor of premature
Allen, J., Anton, R. F., Babor, T. F., Carbonari, J., termination, attendance and alliance in psycho-
Carroll, K. M., Connors, G. J., et al. (1998). therapy. British Journal of Clinical Psychology,
Matching alcoholism treatments to client het- 39(4), 371–82.
erogeneity: Project MATCH three-year drinking Dozois, D. J. A., Westra, H. A., Collins, K. A.,
outcomes. Alcoholism: Clinical and Experimental Fung, T. S., & Garry, J. K. F. (2004). Stages
Research, 22(6), 1300–1311. of change in anxiety: Psychometric properties
Ametller, L., Castro, J., Serrano, E., Martínez, E., of the University of Rhode Island Change
& Toro, J. (2005). Readiness to recover in Assessment (URICA) scale. Behaviour Research
adolescent anorexia nervosa: Prediction of and Therapy, 42(6), 711–29.
hospital admission. Journal of Child Psychology Eckhardt, C., Holtzworth-Munroe, A., Norlander, B.,
and Psychiatry and Allied Disciplines, 46(4), Sibley, A., & Cahill, M. (2008). Readiness
394–400. to change, partner violence subtypes, and
treatment outcomes among men in treatment Mitchell, D., & Angelone, D. J. (2006). Assessing
for partner assault. Violence and Victims, 23(4), the validity of the Stages of Change Readiness
446–75. and Treatment Eagerness Scale with treatment-
Geller, J., Drab-Hudson, D. L., Whisenhunt, B. L., seeking military service members. Military
& Srikameswaran, S. (2004). Readiness to Medicine, 171(9), 900–904.
change dietary restriction predicts outcomes in Pantalon, M. V., Nich, C., Frankforter, T., &
the eating disorders. Eating Disorders: The Journal Carroll, K. M. (2002). The URICA as a measure
of Treatment & Prevention, 12(3), 209–24. of motivation to change among treatment-seek-
Gossop, M., Stewart, D., & Marsden, J. (2007). ing individuals with concurrent alcohol and
Readiness for change and drug use outcomes cocaine problems. Psychology of Addictive
after treatment. Addiction, 102(2), 301–308. Behaviors, 16(4), 299–307.
Haller, D. L., Miles, D. R., & Cropsey, K. L. (2004). Pantalon, M. V., & Swanson, A. J. (2003). Use of
Smoking stage of change is associated with reten- the University of Rhode Island change assess-
tion in a smoke-free residential drug treatment ment to measure motivational readiness to
program for women. Addictive Behaviors, 29(6), change in psychiatric and dually diagnosed indi-
1265–70. viduals. Psychology of Addictive Behaviors, 17(2),
Henderson, M. J., Saules, K. K., & Galen, L. W. 91–97.
(2004). The predictive validity of the University Petry, N. M. (2005). Stages of change in treatment-
of Rhode Island Change Assessment question- seeking pathological gamblers. Journal of Consulting
naire in a heroin-addicted polysubstance abuse and Clinical Psychology, 73(2), 312–22.
sample. Psychology of Addictive Behaviors, 18(2), Rooney, K., Hunt, C., Humphreys, L., Harding, D.,
106–112. Mullen, M., & Kearney, J. (2005). A test of
Hewes, R. L., & Janikowski, T. P. (1998). Readiness the assumptions of the transtheoretical model
for change and treatment outcome among indi- in a post-traumatic stress disorder population.
viduals with alcohol dependency. Rehabilitation Clinical Psychology and Psychotherapy, 12(2),
Counseling Bulletin, 42(1), 76–93. 97–111.
Hunt, Y. M., Kyle, T. L., Coffey, S. F., Stasiewicz, P. R., Scott, K. L., & Wolfe, D. A. (2003). Readiness to
& Schumacher, J. A. (2006). University of Rhode change as a predictor of outcome in batterer
Island Change Assessment - Trauma: Preliminary treatment. Journal of Consulting and Clinical
psychometric properties in an alcohol-dependent Psychology, 71(5), 879–89.
PTSD sample. Journal of Traumatic Stress, 19(6), Smith, K. J., Subich, L. M., & Kalodner, C. (1995).
915–21. The transtheoretical model’s stages and processes
Isenhart, C. E. (1997). Pretreatment readiness for of change and their relation to premature termi-
change in male alcohol dependent subjects: nation. Journal of Counseling Psychology, 42(1),
Predictors of one-year follow-up status. Journal 34–39.
of Studies on Alcohol, 58(4), 351–57. Soler, J., Trujols, J., Pascual, J. C., Portella, M. J.,
Kerns, R. D., Wagner, J., Rosenberg, R., Barrachina, J., Campins, J., et al. (2008). Stages
Haythornthwaite, J., & Caudill-Slosberg, M. of change in dialectical behaviour therapy for
(2005). Identification of subgroups of persons borderline personality disorder. British Journal of
with chronic pain based on profiles on the pain Clinical Psychology, 47(4), 417–26.
stages of change questionnaire. Pain, 116(3), Stotts, A. L., Schmitz, J. M., & Grabowski, J.
302–310. (2003). Concurrent treatment for alcohol and
Kinnaman, J. E. S., Bellack, A. S., Brown, C. H., tobacco dependence: Are patients ready to quit
& Yang, Y. (2007). Assessment of motivation both? Drug and Alcohol Dependence, 69(1), 1–7.
to change substance use in dually-diagnosed Tambling, R. B., & Johnson, L. N. (2008).
schizophrenia patients. Addictive Behaviors, The relationship between stages of change and
32(9), 1798–1813. outcome in couple therapy. American Journal of
Lewis, C. C., Simons, A. D., Silva, S. G., Rohde, P., Family Therapy, 36(3), 229–41.
Small, D. M., Murakami, J. L., et al. (2009). Treasure, J. L., Katzman, M., Schmidt, U., Troop, N.,
The role of readiness to change in response to Todd, G., & De Silva, P. (1999). Engagement
treatment of adolescent depression. Journal and outcome in the treatment of bulimia ner-
of Consulting and Clinical Psychology, 77(3), vosa: First phase of a sequential design compar-
422–28. ing motivation enhancement therapy and
cognitive behavioural therapy. Behaviour Research and New Zealand Journal of Psychiatry, 43(3),
and Therapy, 37(5), 405–418. 235–43.
Wade, T. D., Frayne, A., Edwards, S. A., Robertson, T., Willoughby, F. W., & Edens, J. F. (1996). Construct
& Gilchrist, P. (2009). Motivational change validity and predictive utility of the stages of
in an inpatient anorexia nervosa population change scale for alcoholics. Journal of Substance
and implications for treatment. Australian Abuse, 8(3), 275–91.
C HA P TER
15 Preferences
Joshua K. Swift, Jennifer L. Callahan, and Barbara M. Vollmer
In recent years, health care professions have edition of this chapter, Arnkoff, Glass, and
emphasized the inclusion of patient prefer- Shapiro (2002) reviewed 10 studies examin-
ences as an essential part of best prac- ing the relation between therapy outcomes
tice standards (e.g., American Psychological and matching clients to a preferred treatment.
Association, 2006; Institute of Medicine, Results from their review were inconclusive,
2001). In psychology, client preferences with only 2 of the 10 studies finding a signifi-
have been identified as one of the three cant positive relationship between treatment
key components of evidence-based prac- preference matching and outcome.
tice, along with the best available research Since the 2002 review, there has been
and clinical expertise. In particular, APA’s increased interest in studying the prefer-
(2006) evidence-based practice policy states ence effect; compared with the 10 studies
that treatment decisions should be made found in 2002, Swift and Callahan (2009)
in collaboration with the patient, with the identified 28 studies that tested this effect.
central goal to maximize patient choice. Their meta-analysis found that clients who
Involving clients in the decision-making received their preferred treatments were
process when providing psychological treat- significantly less likely to drop out from
ments is important not only because it therapy prematurely and were significantly
allows them the freedom to direct their more likely to show improved outcomes
own lives and determine their care, but also compared with clients whose preferences
because it might provide them with pre- were either not considered or not matched.
ferred services thought to lead to improved Unfortunately, Swift and Callahan’s review
therapy outcomes. only examined preferences for treatment
The impact of client preferences on ther- type. Thus, in order to further our under-
apy outcomes has been studied empirically standing of the influence of client prefer-
for at least 40 years. In perhaps the earliest ences on therapy, an updated meta-analysis
review of the topic, Rosen (1967) surveyed a of the preference effect for all types of client
number of studies that examined preferences, preferences is needed.
but discussed only one study that actually In this chapter, we review the empiri-
looked at the influence preferences exert on cal evidence supporting the accommodation
treatment outcomes. Based on this early of patient preferences when providing psy-
review, Rosen concluded that preferences chological treatments. Specifically, we exam-
“might” have an effect on a number of ine whether providing patients with their
outcome-related variables. In the previous preferred therapy conditions influences rates
301
of premature termination and overall ther- preferring to have a therapist that has a
apy outcomes. We begin by defining and empathetic personality style). Finally, treat-
providing clinical examples of preference ment preferences involve specific desires
matching, then provide a summary of our for the type of intervention that will be
meta-analysis of the research looking at out- used (e.g., preferring a psychodynamic
come and dropout effects, and conclude with approach versus a behavioral approach,
recommendations for therapeutic practices. preferring psychotherapy compared to
pharmacotherapy).
Definitions and Measures Various ways of measuring patient pref-
In the previous edition of this chapter, erences can be found in the literature.
client preferences were defined as the Perhaps the most popular measure has been
behaviors or attributes of the therapist or to directly ask patients what condition they
therapy that clients value or desire (Arnkoff would prefer to receive: for example, asking
et al., 2002). In other words, client prefer- patients if they would prefer medication,
ences represent what clients would want psychotherapy, or a combination treatment
the therapy encounter to be like if the (Kocsis et al., 2009), or asking patients
choice were left to them. This definition of if they would prefer a male or a female
preferences based on desires and values therapist (Zlotnick, Elkin, & Shea, 1998).
should be contrasted to definitions of the In a variation of this type of measure, a
similar concept of client expectations, few studies have provided patients with
which focus more on what the client descriptions and/or demonstrations of their
actually believes should or will happen in options prior to asking them to state a pref-
therapy (see Chapter 18 for a review of erence. For example, some researchers
expectations). Studies have indicated that have played audiotapes of therapists pro-
although these two constructs are corre- viding descriptions of themselves and their
lated, client preferences and expectations approaches to therapy and then asked
are distinct phenomena that can influence patients to indicate which therapist they
therapy in different ways (Proctor & Rosen, would prefer to work with (Manthei,
1981; Tracey & Dundon, 1988). Vitalo, & Ivey, 1982). Other researchers
Three main types of client prefer- have had clients briefly discuss therapy
ences have been identified in the literature: options with a psychotherapist or physi-
role preferences, therapist preferences, and cian prior to being asked to state a prefer-
treatment type preferences. Role preferences ence for one treatment or another (Adamson,
involve the behaviors and activities that Sellman, & Dore, 2005; Calsyn, Winter, &
clients desire themselves and their thera- Morse, 2000). The Treatment Preference
pists to engage in while in therapy (e.g., Interview (Vollmer, Grote, Lange, & Walker,
preferring the therapist to take an active 2009) is one example of a discussion-based
advice-giving role versus a listening role, measure that allows clients to express
preferring that cognitive-behavioral treat- preferences for each of the three main pref-
ment be administered in a group format erence domains: roles, therapists, and
rather than an individual format). Therapist treatments.
preferences entail characteristics that clients In contrast to directly asking patients for
hope their therapists will possess (e.g., pre- their preferences, some researchers have
ferring the therapist to have had many years employed questionnaires or rating scales
of clinical experience, preferring the thera- that assess preferences as well as their degree
pist to have a similar ethnic background, or strength. Assessing preference strength is
of value because one might expect that stron- into initial treatment planning as well
ger preferences, compared with slighter pref- as ongoing therapy decision making. In both
erences, would have a greater influence on cases, the client’s preferences were assessed
treatment outcomes. For example, research- during an intake appointment using the
ers have not only invited depressed patients Treatment Preference Interview (Vollmer
to indicate if they preferred interper- et al., 2009; Table 15.1). This interview
sonal psychotherapy or pharmacotherapy, was again administered at every third
but they also asked them to rate on a five- session, allowing these clients to indicate
point, Likert-type scale how strongly they whether their preferences had changed
wanted their preferred treatment (Raue, and whether therapy was accommodating
Schulberg, Heo, Klimstra, & Bruce, 2009). their preferences.
Similarly, the Treatment Preferences and
Experiences Questionnaire (Berg, Sandahl, Case Example 1
& Clinton, 2008) was developed to allow “Linda,” a 55-year-old divorced, Caucasian
patients to rate their preferences according woman, contacted the clinic because she
to four intervention and behavior domains: felt that her gambling was out of control.
outward orientation (concrete and direc- While she thought that her work perfor-
tive problem-solving interventions), inward mance had not suffered, other aspects of
orientation (interventions focusing on her life had. Linda’s financial problems had
reflection and inner mental processes), sup- worsened: she had to sell her home, was
port (wanting advice, encouragement, and thousands of dollars in debt, and was con-
sympathy from the therapist), and catharsis stantly being called by collection agencies.
(focusing on expressive interventions). Her family members, who were her only
support system since she no longer had
Clinical Examples friends, had lost respect for her. She reported
The following case examples demonstrate feeling depressed about the impact gam-
how client preferences can be incorporated bling had on her life. At the beginning of
Table 15.1 Treatment Preference Interview

Preference factor Question content and examples
Therapist’s Strong preferences for counselor’s: gender, age, ethnicity or race, language, sexual orientation,
characteristics religion, or other?
Role preferences Prior therapy or experience being helped: What was most helpful? What was the worst a therapist
could do?
Preferences for the counselor’s approach: Preference for a therapist who takes charge, is active/
talkative and expressive/warm, or client taking charge, and the therapist is more quiet and
reserved?
Preferences for treatment modality: Individual, couple, group, or family sessions?
Preferences for therapy tasks: Try new things between sessions, reading self-help books, watching
self-help movies, going online for information
Type of therapy Beliefs about the causes of the problem: Will of God, unlucky experiences, biological makeup,
unmet emotional needs, unrealistic expectations, relationship conflicts, lack of self-knowledge,
lifestyle, or lack of will power?
Preferences for type of therapy: solution-focused, cognitive-behavioral, or psychodynamic
therapy? (Therapy descriptions were also provided, including typical goals, therapist–client
relationship, and tasks.)
Preferences for who decides about the type of therapy: Client makes the decision, client and
therapist collaborate, or therapist makes the decision?
s w i f t, c a l l a h a n , vo l l me r 303
her therapy, Linda scored 68 on the the twelve-step approach nor the individu-
Outcome Questionnaire 45.2 (OQ-45.2; als who attended meetings, as they did not
Lambert et al., 1996), in the clinical range seem to have problems with Internet gam-
of disturbance, and she met diagnostic cri- bling. Linda stated that her treatment goals
teria for pathological gambling. were to regain her “sense of empowerment,”
Linda responded positively to the possi- to earn the perception of being a “strong
bility of being asked about her preferences. woman” from her family members, to learn
She stated that preferences were important how to take better care of her health, and to
to her “in order to individualize her treat- change her gambling habit.
ment, since no one treatment is best for every For the most part, the therapist adhered
person.” During the Treatment Preference to Linda’s preferences. Throughout their
Interview, Linda was first asked about her sessions, the therapist focused on Linda’s
preferences for therapist characteristics. feelings and the meanings attached to them.
Linda expressed a strong preference for a The therapist noted that Linda found it
female therapist, partially due to a positive much easier to express her thoughts rather
past experience with a female counselor than her emotions. By the fourth session,
and partially due to the difficulty she had Linda provided written feedback that she
in trusting men after an abusive marriage. wanted therapy to continue to focus on
In addition, she indicated that she preferred building awareness of her feelings, and that
a therapist “who is warm, caring, and shows she found the realization of a connection
emotions” as well as empathy. In terms of between her emotions and actions to be
role preferences, Linda strongly preferred enlightening.
individual therapy sessions compared with Although the therapist largely adhered
family or group therapy. Additionally, to Linda’s preferences for roles and type
Linda did not want her therapist to “take a of therapy, the therapist noted that she
more directive and active approach . . . by breached them during the 12th session
giving opinions and making suggestions.” when she introduced the principles of cog-
She indicated that she desired to have a nitive-behavioral therapy (CBT) and the
“collaborative” relationship with her thera- idea of a thought log as a homework exer-
pist, “having specific goals to guide our cise. The therapist quickly noticed that
work together.” Linda became disengaged when discussing
In terms of treatment preferences, Linda CBT and concluded that the introduction
responded positively to two therapy descrip- of CBT and homework was a mistake
tions. Her first choice was psychodynamic, because it had not accommodated Linda’s
particularly because she was interested preferences. The therapist’s observation was
in identifying repetitive patterns in her life corroborated in Linda’s ratings of the work-
and relationships. Her second choice was ing alliance for that session; the item asking
motivational enhancement. This option about whether “my therapist and I agree on
appealed to Linda because of the collabora- ways to achieve my goals” was rated lower.
tive nature of the relationship and ele- The therapist and client mutually termi-
ments that drew upon her strengths while nated after 16 sessions. By this time Linda’s
still acknowledging her weaknesses. Linda OQ-45.2 score had dropped into the
strongly disliked twelve-step facilitation as normal range, 47, indicating a clinically
a result of her experience at Gamblers significant improvement. Her rating of the
Anonymous meetings. She did not identify working alliance had also improved; going
with the spiritual principles prevalent in from an initial score of 9 to a score of 14.5
on the revised short version of the Working homework assignments between sessions.
Alliance Inventory (WAI-S; Hatcher & In terms of treatment preferences, Ruth
Gillaspy, 2006). Overall, the client reported strongly desired a CBT approach to her
that working on family issues had been therapy. Ruth’s treatment goals were to
helpful, that her urge to gamble had dissi- improve communication with her boy-
pated, and that she was taking better care friend, friends, and coworkers; to adjust
of herself. At her termination session, better to her partner’s reemergence of alco-
Linda’s written response to the question hol abuse; and to learn better problem-
“Do you feel that your therapist had a good solving and decision-making skills.
understanding of your goal(s) for therapy?” Ruth started therapy by “no-showing”
was as follows: “Yes—She took my lead for the second session. When her therapist
and went with it . . . I believe it was a good called to encourage her to continue treat-
match. I respect and admire her.” ment, she responded by returning and
regularly attending after that point. Her
Case Example 2 therapist attributed Ruth’s “no-show” as a
“Ruth,” a 39-year-old pregnant Caucasian test of whether the therapist would be non-
woman, contacted the clinic for assistance judgmental and willing to work with her.
with relationship problems with her boy- Therapy primarily focused on Ruth’s auto-
friend, indicating that she had become matic thoughts of not feeling worthy to
increasingly concerned about his problems have a good life and her need to look to
with alcohol. Ruth reported that he became others for validation. Ruth indicated that
emotionally abusive when he was drinking, therapy helped her learn to recognize and
and this was not the type of relationship challenge her maladaptive thinking pat-
she was hoping for when starting a family. terns and develop more assertive behaviors
Ruth wanted to be able to enjoy her preg- in her social and work relationships. When
nancy and look forward to becoming a assessed at different time points during
parent. Her friends urged her to leave her therapy, Ruth continued to express a pref-
boyfriend; however, she hoped that he erence for CBT. For example, when her
would change and become a good father therapist introduced thought logs, she
for his child. At the beginning of her ther- wrote on her Session Feedback Survey that
apy, Ruth scored 82 on the OQ-45.2, she “liked CBT.”
falling in the clinical range. Throughout therapy, Ruth continued to
During the Treatment Preference Inter- express hope that her boyfriend would
view, Ruth indicated that she had no strong change his behaviors. As might have been
preferences for her therapist’s age, gender, expected from her original goals, Ruth
or sexual orientation. However, she did responded more positively to discussions
state that she wanted a therapist whom she concerning ways to manage her conflicts
could “connect” to, who would actively with her boyfriend, rather than to ques-
work with her on an equal basis, who would tions about the evidence that he would
not be too confronting or challenging, and change. However, by the end of therapy
who would not “judge her too harshly” she was able to look at her relationship in
for being pregnant and unmarried. When terms of her assumptions and beliefs and was
asked about role preferences, Ruth expressed able to state “If my partner behaves poorly,
a desire for self-help books and self- it is not a reflection on me,” and “People
help movies to be incorporated into her are responsible for their own behavior.”
counseling, and openness to completing Interestingly, on the occasions when her
therapist was challenging, such as suggest- or outcome. Using these terms, 3,895 cita-
ing her need to set better boundaries, Ruth tions were identified. Several journals were
rated the working alliance as lower for those also hand searched for relevant studies.
sessions. Further search strategies included pull-
Ruth attended a total of 20 sessions. ing citations from the reference lists of
After the 11th session, she took a break relevant articles and exploring all studies
from therapy when she was about to give in PsycINFO that cited a relevant study.
birth. She later returned to therapy for an All abstracts from the resulting citations
additional nine sessions. This return to were reviewed. Based on the abstracts, 134
therapy included a switch to a new male potentially relevant articles were further
therapist as compared with the previous evaluated to determine if they met inclu-
female one. Perhaps due to the lack of pref- sion criteria.
erence concerning therapist demographic
characteristics, which was expressed prior Inclusion Criteria
to therapy, Ruth transitioned easily to her All published studies in the English lan-
second therapist. Ruth, in her written com- guage that assessed client preferences prior
ments about both therapists, noted several to treatment and examined the effect (on
times that they were nonjudgmental and therapy dropout or outcome) of matching
that she found it helpful to feel that she clients to their preferred therapy conditions
could talk about personal concerns in con- were included in this meta-analysis. Studies
fidence. By the end of treatment, Ruth’s were excluded if they used a nonclini-
OQ-45.2 score had dropped to the low 60s, cal sample (e.g., students participating for
indicating clinically significant improve- course credit), studied a variable not related
ment. Her WAI-S score had improved to to a clinical problem (e.g., speed reading),
12.25 out of 15 points. did not involve matching of at least part of
the sample to their preferred therapy con-
Meta-Analytic Review dition, did not involve the administration
The preceding case examples illustrate the of a psychological treatment (e.g., use of
probable influence of patient preferences on only medication groups, use of interview-
treatment progress, but here we examine only interventions), or did not include a
more systematically the relation of patient measure of therapy dropout or outcome
preferences to psychotherapy outcomes. We (e.g., examined the preference effect for
summarize the results of our meta-analysis treatment satisfaction). Where multiple
of studies comparing dropout rates and/or studies analyzed the data from the same
outcomes between preference-matched and group of clients, the study with the most
preference non-matched patients. recent follow-up period or with the largest
sample was used in the analysis. After fur-
Search Strategy ther review, a total of 38 studies were deemed
We began with an initial search of PsycINFO eligible for inclusion in the meta-analysis.
for articles published between 1967 (Rosen’s
review of client preferences) and September Study Coding
2009. The electronic search was conducted These 38 studies were coded by two inde-
using the following terms: preference or choice, pendent evaluators to assess a number of
in combination with therapy or psychother- variables: the type of preference (role, ther-
apy or treatment or therapist or counselor or apist, or treatment), problem treated, treat-
therapeutic alliance or role, and matching ments that were provided (e.g., CBT, IPT,
pharmacotherapy, psychodynamic), method moderators, estimating the percentage of
of allocation to preference conditions (par- variability due to true differences among the
tially randomized preference trial, random- studies. Calculations were completed using
ized/assigned to treatment conditions, or Comprehensive Meta-Analysis, Version 2
randomized/assigned to preference match (Borenstein, Hedges, Higgins, & Rothstein,
conditions), and primary outcome (identi- 2005).
fied through statements/hypotheses made
by the original authors). The two indepen- Effect on Dropout
dent coders showed a high level of agree- Eighteen of the 35 studies compared drop-
ment (97.78%) across all variables. Where out rates between clients who received their
a discrepancy was found, a third coder was preferred therapy conditions and those who
asked to code the relevant variable. did not. An odds ratio effect size was calcu-
lated for each of the studies, which repre-
Methodological Decisions sents the ratio of dropouts versus completers
We were interested in measuring outcome between the compared groups. While an
and dropout differences between those odds ratio of 1 indicates that an equal
clients who were matched and those clients number of clients dropped out of each
who were not matched to preferred ther- group, in our analyses an odds ratio less
apy conditions. The results from each of than 1 indicates fewer clients dropped out
the studies were summarized using odds of the preference-matched groups, and an
ratios when examining therapy dropout odds ratio greater than 1 indicates fewer
and Cohen’s d when examining therapy clients dropped out of the preference non-
outcome. Of the 38 studies deemed eligible matched groups. A forest plot of the odds
for inclusion, 3 did not contain sufficient ratio effect sizes for each study and the
outcome or dropout data to include their aggregate effect can be viewed in Figure 15.1.
results in either analysis; thus, 35 studies The overall effect on dropout was signifi-
were included in the remaining analyses. cant (OR = 0.59, CI.95: 0.44 to 0.78,
Where outcome results were reported for p < 0.001), indicating that clients who
multiple measures within a single study, received their preferred conditions were
only one primary outcome measure (see between a half and a third less likely to drop
study coding above) from each study was out of therapy prematurely compared with
used in our analyses. clients who did not receive their preferred
Effect sizes and confidence intervals for therapy conditions, or for every 5 non-
each of the studies were calculated, follow- matched clients who dropped out prema-
ing which an aggregate effect size was then turely, only 3 matched clients dropped out.
calculated across studies using a random- Heterogeneity between studies was not
effects model. A fail-safe N was calculated, found [Q(17) = 22.46, p = 0.17, I 2 = 24.31].
representing the number of nonsignificant, Calculation of the fail-safe N indicated that
nonpublished studies that would be needed 89 unpublished studies with nonsignificant
to dilute the results of the meta-analysis. results would be required to reduce the
Moderators were next tested using the results of the preference effect on therapy
Q-statistic and a random-effects model. dropout to a nonsignificant level.
A significant Q-statistic between groups
indicates a difference that is greater than Effect on Outcome
expected by chance. In addition, the I 2 Thirty-three of the 35 studies included an
statistic was calculated for each group of outcome comparison between clients who
Pref. type Study name Odds ratio Odds ratio and 95% C.I.
role Ersner-Hershfield et al. (1979) 0.34

Kludt and Perlmuter (1999) 0.58
Macias et al. (2005) 0.23
McKay et al. (1995) 0.76
McKay et al. (1998) 1.13
Renjilian et al. (2001) 1.43
Sterling et al. (1997) 0.83
therapist Manthei et al. (1982) 0.46
Proctor & Rosen (1981) 0.36
Zlotnick et al. (1998) 0.63
treatment Bakker et al. (2000) 0.84
Elkin et al. (1999) 0.19
Fuller (1988) 0.45
Kocsis et al. (2009) 0.80
Leykin et al. (2007) 0.62
Raue et al. (2009) 0.05
Rokke et al. (1999) 0.08
Van et al. (2009) 0.63
Total 0.59
0.1 0.2 0.5 1 2 5 10
Favors matched Favors non-matched
clients being less clients being less
likely to drop out likely to drop out
Fig. 15.1 Dropout effect sizes (odd ratios) for preference match vs. nonmatch groups.
did and did not receive a preferred therapy considered—role, therapist, or treatment.
condition. Cohen’s d was calculated for Preference type was tested as a moderating
each of these studies, and a forest plot of variable for the preference effect on therapy
the effect sizes can be viewed in Figure 15.2. dropout and therapy outcome. A total
The overall effect size was d = 0.31 (CI.95: of 11 studies examined role preferences,
0.20 to 0.43), indicating a small but sig- 3 examined therapist preference, and 21
nificant (z = 5.39, p < 0.001) outcome effect examined treatment preferences. Of the 18
in favor of those clients who received studies that reported dropout rates, the dif-
their preferred therapy conditions. Hetero- ference in effect size estimates between these
geneity between the 33 studies was found groups was not significant [Q(2) = 1.59,
[Q(32) = 57.78, p < 0.01, I 2 = 44.63], indi- p = 0.45], indicating that therapy dropout
cating that the studies did differ signifi- was similarly influenced by matching cli-
cantly in their outcome effect size estimates. ents to any of the three preferred therapy
Calculation of the fail-safe N indicated that conditions. In terms of treatment outcome,
427 unpublished studies with nonsignifi- the difference between preference type
cant results would be required to reduce groups was also not significant [Q(2) = 0.10,
the results of the preference effect on treat- p = 0.88], indicating that treatment out-
ment outcome to a nonsignificant level. come was also similarly influenced by
matching clients to their preferred therapy
Moderators roles, therapists, or types of treatment.
Preference Characteristics Preference for or against Pharmacotherapy.
Preference Type. The effect of receiving or not Several studies specifically examined the
receiving a preferred therapy condition may preference effect when psychotherapy was
be moderated by what type of preference is compared with pharmacotherapy. Of the
Pref. type Study name Effect size d Effect size and 95% C.I.
role Al-Otaiba et al. (2008) 0.52

Cooper (1980a) 0.69
Cooper (1980b) 0.54
Gossop et al. (1986) 0.51
Kludt and Perlmuter (1999) 0.14
Macias et al. (2005) 0.08
McKay et al. (1995) 0.09
McKay et al. (1998) 0.25
Renjilian et al. (2001) −0.14
Sterling et al. (1997) 0.47
therapist Manthei et al. (1982) 0.23
Zlotnick et al. (1998) 0.37
treatment Adamson et al. (2005) 0.52
Bakker et al. (2000) 0.31
Berg et al. (2008) 0.65
Brown et al. (2002) 0.37
Calsyn et al. (2000) 0.10
Chilvers et al. (2001) 0.33
Devine & Fernald (1973) 1.19
Dyck & Spinhoven(1997) 0.10
Elkin et al. (1999) 1.15
Fuller (1988) −0.27
Gum et al. (2006) 0.07
Iacoviello et al. (2007) 1.15
Kadish (1999) 0.22
Kocsis et al. (2009) 1.01
Leykin et al. (2007) 0.31
Lin et al. (2005) 0.23
Raue et al. (2009) −0.18
Rokke et al. (1999) 0.37
Van et al. (2009) 0.10
Wallach (1988) 0.44
Ward et al. (2000) −0.08
Total 0.31
−2.00 −1.00 0.00 1.00 2.00
Outcome favors Outcome favors
non-matched clients matched clients
Fig. 15.2 Outcome effect sizes (d) for preference match vs. nonmatch groups.
studies that reported dropout rates, 7 exam- (k = 21) was d = 0.21 (CI.95: 0.10 to 0.31).
ined the preference effect for psychotherapy This difference showed a trend toward sig-
versus pharmacotherapy and 11 examined nificance [Q(1) = 3.49, p = 0.06], indicating
the preference effect for one form of psy- that preferences for psychotherapy versus
chotherapy versus another. The difference pharmacotherapy may have a greater influ-
between these groups was not significant ence on treatment outcome than prefer-
[Q(1) = 0.37, p = 0.55]. In terms of treat- ences between two forms of psychotherapy.
ment outcome, the average effect size for
studies comparing preferences for psycho- Client Characteristics
therapy versus pharmacotherapy (k = 12) The only client characteristic that was reli-
was d = 0.36 (CI.95: 0.24 to 0.49), while the ably reported across studies and that could
average effect size for studies comparing be compared between studies was client
preference for one form of psychother- diagnosis/problem (e.g., anxiety, depres-
apy versus another form of psychotherapy sion, substance abuse). In terms of therapy
dropout, average odds ratios between the treatment dropout and outcome. In con-
three compared groups (depression, sub- trast, the other two types of designs make
stance abuse, and obesity) were not signifi- comparisons between clients who are given
cantly different [Q(2) = 3.04, p = 0.22], a therapy that matches their preferences and
indicating that matching clients to their clients who are given a therapy that directly
preferred therapy conditions had a similar opposes their preferences, thus maximizing
effect on therapy dropout rates regardless the differences between groups.
of the problem being treated. In terms Type of study design was tested as a
of therapy outcome, studies of anxiety moderator for the preference effect for both
(k = 6) found an average preference effect therapy dropout and outcome. In terms of
of d = 0.49 (CI.95: 0.19 to 0.79), studies of therapy dropout, the difference between
depression (k = 12) found an average pref- the three design groups was not significant
erence effect of d = 0.35 (CI.95: 0.13 to [Q(2) = 3.33, p = 0.19], indicating that
0.57), studies of a health concern (k = 3) study design did not moderate the effect
found an average preference effect of preferences had on therapy dropout. In
d = −.07 (CI.95: −.43 to 0.29), studies of terms of therapy outcome, the three types
serious mental illness (k = 2) found an aver- of study design [PRPTs (d = 0.16, CI.95:
age preference effect of d = 0.09 (CI.95: −.22 0.00 to 0.32), studies allocating to prefer-
to 0.40), and studies of substance abuse ence conditions (d = 0.24, CI.95: 0.01 to
(k = 8) found an average preference effect 0.46), and studies allocating to treatment
of d = 0.34 (CI.95: 0.18 to 0.51). Preference conditions (d = 0.45, CI.95: 0.28 to 0.62)]
effect differences between these groups were found to be significantly different in
showed a trend toward significance [Q(4) = their estimates [Q(2) = 6.17, p = 0.02].
7.71, p = 0.10]. While matching client While the largest outcome difference
preferences did positively influence treat- between matched and nonmatched clients
ment outcomes for anxiety, depression, and was found in studies that randomized or
substance abuse, preference matching assigned clients to a treatment condition,
showed little benefit in the treatment of the smallest outcome difference between
health concerns and serious mental illness. preference-matched and nonmatched cli-
ents was found in the group of PRPTs, as
Design Characteristics predicted.
Study Design. The studies included in this Dropout/Outcome Measurement Type. The
meta-analysis varied in the designs used to studies included in this meta-analysis dif-
examine the preference effect. These designs fered both in how therapy dropout was
included partially randomized preference defined (for those that included an assess-
trials (PRPTs), studies that randomized or ment of dropout) and in how treatment
assigned clients to a treatment condition, outcome was measured. Thus, measure-
and studies that randomized or assigned ment type was also tested as a design mod-
clients to a preference condition. The dif- erator for the overall preference effects.
ferences between studies in terms of design Regarding outcomes, 7 studies measured
may have influenced the magnitude of the outcome by ratings from an independent
preference effect. Because PRPTs actually rater (e.g., HRSD, SCID), 6 studies used
only compare clients who express prefer- an objective measurement (e.g., BMI, urine
ences with clients who do not express strong analysis), and 19 studies measured out-
preferences, these studies may underesti- come by patient self-report. A significant
mate the influence preferences have on difference in effect size estimates between
these groups was not found [Q(3) = 0.10, time of outcome measurement was not a
p = 0.99]. moderating variable.
In terms of the preference dropout effect
by assessment type, the studies (k = 11) that Patient Contributions
defined dropout as not completing a full In our meta-analysis we saw that client
treatment protocol found an average prefer- preferences influenced both who dropped
ence effect size of OR = 0.73 (CI.95: 0.56 to out of therapy prematurely and who showed
0.94), while the studies (k = 2) that assessed greater improvements while in therapy,
dropout by therapist rating found an aver- thus illustrating the importance of accom-
age preference effect size of OR = 0.38 (CI.95: modating client preferences in the therapy
0.12 to 1.20) and the studies (k = 5) that encounter. Patient preferences can be
defined dropout as having attended less viewed as a variable that patients contrib-
than a set number of sessions found an ute to the therapy relationship because
average preference effect size of OR = 0.34 most patients enter therapy with specific
(CI.95: 0.19 to 0.60). The effect size esti- desires or hopes concerning what treatment
mates from these three groups showed will be like. Patient preferences have been
a trend toward significance [Q(2) = 5.63, found to be influenced by a number of
p = 0.06]. When dropout was defined other variables, such as demographic char-
by completion of a treatment protocol, acteristics, beliefs about the nature of their
smaller differences (compared with differ- problems, level of symptom severity, previ-
ences found by the other two definitions ous experience with therapy, expectations
of dropout) between those who received a for therapy, and other life experiences (e.g.,
preferred therapy condition and those Bedi et al., 2000; Churchill et al., 2000;
who received a nonpreferred therapy condi- Ertl & McNamara, 2000; Gum et al., 2006;
tion were observed. Perhaps clients who Riedel-Heller, Matschinger, & Angermeyer,
received a nonpreferred therapy were will- 2005; Vincent & LeBow, 1995; Wanigaratne
ing to “stick it out” through a treatment & Barker, 1995; Wong, Kim, Zane, Kim,
that had a defined number of sessions in & Huang, 2003). Given the number of
its protocol, but these nonmatched clients variables that could possibly influence pref-
prematurely terminated when a predeter- erences for therapy, each client should be
mined number of treatment sessions had viewed as a unique individual with differ-
not been set. ent hopes or desires for what therapy will
Time of Outcome Measurement. Time of be like. It may well be labeled as the thera-
outcome measurement was also assessed as pist’s responsibility to elicit these individual
a design characteristic that may have had preferences and then make treatment deci-
an influence on the overall preference effect. sions in conjunction with their clients.
In testing this variable, we were examin- In turn, it could be considered a patient
ing whether receiving a preferred therapy responsibility to be forthcoming with ther-
condition resulted in improved outcomes apists concerning treatment preferences. If
(over clients who did not receive a preferred therapists are unaware of their patients’
condition) equally early on in therapy, preferences, they will not likely accommo-
immediately after the completion of ther- date them. However, some clients may be
apy, and at follow-up time points. There hesitant about expressing their preferences
was not a significant difference in the due to a number of factors. For example,
effect size estimates between these groups clients who specifically enter treatment to
[Q(2) = 0.41, p = 0.82], indicating that work on an addictive behavior may indicate
a preference to work on another problem been identified as the gold standard for
or to use a treatment that is not as directed measuring treatment effects, patients in
at their addiction because they are not these trials are not randomized into prefer-
yet ready to change. Some clients may also ence (matched versus not-matched) condi-
be hesitant about expressing preferences tions; thus, there is no guarantee that
because they do not know that it is appro- patients in the preference conditions are
priate to do so or because they are unaware similar or even comparable. RCTs may
that different treatment options even exist. also fail to properly account for client pref-
Additionally, patients who see their thera- erences because many clients who hold
pists as authority figures or who have yet to strong preferences (the group where one
develop trust in their therapists may be might expect to see the largest preference
hesitant because they think the therapists effect) refuse randomization into treatment
know best, or they worry that their prefer- groups. In response to this limitation, the
ences will be ignored or not taken seriously. PRPT has been developed. Although cli-
In each of these situations, therapists should ents with strong preferences are more likely
seek to overcome the barriers that prevent to be included in studies using this design,
clients from expressing their preferences. in PRPTs the preference effect is likely to
At the same time, some patients may be be attenuated because no clients actually
hesitant about airing their preferences receive a nonpreferred treatment. PRPTs
because they do not want their preferences only compare clients who have stronger
to be taken into account when treatment preferences with clients who hold weaker
decisions are made. In one psychology or no preferences. We believe the most
department clinic, 42% of the clients pre- clinically appropriate and methodologically
ferred therapists to make treatment deci- sound design to measure the preference
sions, compared with clients making the effect is to randomize clients into prefer-
decision, or clients and therapists collabo- ence conditions. Regrettably, only about
rating to make treatment decisions together a third of the preference effect studies
(Grote, Lange, Walker, & Vollmer, 2009). that have been conducted have used this
Yet, a client’s desire to not be involved design.
in the treatment decision-making process An additional limitation of the research
remains a preference that therapists can reviewed is that most studies have failed to
elicit and address. examine how other client variables influ-
ence the preference effect. None of the stud-
Limitations of the Research ies included in this review examined whether
A number of limitations exist with the cur- patient gender, age, or race/ethnicity affected
rent body of research. Although this area of whether preferences impacted therapy
research has a history dating back over 40 dropout or treatment outcomes. One might
years, the number of published studies is hypothesize that just as client diagnosis
still relatively small. This limitation is par- moderates the preference effect, other
ticularly evident for studies examining cli- demographic variables could also moderate
ents’ preferences for their therapists, with the magnitude of the overall effect.
only three studies being found. In addition, A final concern is that very few of the
research has primarily examined the prefer- studies identified employed a measure of
ence effect post hoc in studies that were preference strength; only one study in this
designed to study treatment effects. Although meta-analysis assessed whether preference
randomized controlled trials (RCTs) have strength influenced the preference effect.
One might expect that receiving or not preferences are addressed, fewer clients
receiving a preferred therapy condition drop out of therapy prematurely, and
would exert large dropout and outcome clients show greater improvements in
effects for clients who strongly desire a therapy outcomes.
given condition. On the other hand, receiv- • When a therapist believes that a
ing a preferred therapy condition may make client’s preferences for therapy are not in
little difference to patients who only slightly the client’s best interest, share these
prefer one condition over another. However, concerns with the client so that treatment
given the lack of assessment of prefer- decisions can still be made collaboratively.
ence strength in the current research, this
hypothesis could not be tested. References
An asterisk (∗) indicates studies included in
Therapeutic Practices the meta-analysis.
Based on the existing research, we can con- *Adamson, S. J., Sellman, J. D., & Dore, G. M.
(2005). Therapy preference and treatment
clude that client preferences exert an influ- outcome in clients with mild to moderate alco-
ence on therapy dropout and treatment hol dependence. Drug and Alcohol Review, 24,
outcomes. Specifically, clients who receive 209–216.
a treatment that matches or considers their *Al-Otaiba, Z., Worden, B. L., McCrady, B. S., &
preferences, compared with clients who Epstein, E. E. (2008). Accounting for self-
receive non-preferred conditions or clients selected drinking goals in the assessment of treat-
ment outcome. Psychology of Addictive Behaviors,
whose preferences are ignored, are about 22, 439–43.
one-half to one-third less likely to drop out American Psychological Association Presidential
of treatment prematurely and are more likely Task Force on Evidence-Based Practice (2006).
to show improved therapy outcomes. Given Evidence-based practice in psychology. American
this significant preference effect, we offer Psychologist, 61, 271–85.
the following clinical recommendations: Arnkoff, D. B., Glass, C. R., & Shapiro, S. J. (2002).
Expectations and preferences. In J. C. Norcross
• Assess clients’ preferences prior to the (Ed.), Psychotherapy relationships that work:
start of treatment. This assessment can Therapist contributions and responsiveness to
address preferences for therapy roles, patients. (pp. 335–56). New York: Oxford
University Press.
therapist characteristics, and treatment *Bakker, A., Spinhove, P., Van Balkom, A. J. L. M.,
types. Vleugel, L., & Van Dyck, R. (2000). Cognitive
• Seek to overcome barriers that might therapy by allocation versus cognitive therapy by
prevent clients from expressing their preference in the treatment of panic disorder.
preferences, such as paucity of information Psychotherapy and Psychosomatics, 69, 240–43.
about therapy options, lack of trust in the Bedi, N., Chilvers, C., Churchill, R., Dewey, M.,
Duggan, C., Fielding, K., et al. (2000). Assessing
therapist, or low readiness to change. effectiveness of treatment of depression in pri-
• Address client preferences throughout mary care: Partially randomized preference trial.
the therapy process. Clients may change British Journal of Psychiatry, 177, 312–318.
their preferences after starting treatment, *Berg, A. L., Sandahl, C., & Clinton, D. (2008).
or clients may feel as if their preferences The relationship of treatment preferences and
are not being addressed despite therapists’ experiences to outcome in generalized anxiety
disorder. Psychology and Psychotherapy: Theory,
attempts to do so. Research, and Practice, 81, 247–259.
• Accommodate client preferences Borenstein, M., Hedges, L. V., Higgins, J. P. T., &
whenever possible. The findings from this Rothstein, H. R. (2005). Comprehensive meta-
meta-analysis illustrate that when client analysis, Version 2. Englewood, NY: Biostat.
*Brown, T. G., Seraganian, P., Tremblay, J., & treatment outcomes. Presentation at the American
Annis, H. (2002). Matching substance abuse Psychological Convention, Toronto, Canada.
aftercare treatments to client characteristics. *Gum, A. M., Arean, P. A., Hunkeler, E., Tang, L.,
Addictive Behaviors, 27, 585–604. Katon, M., Hitchcock, P., et al. (2006).
*Calsyn, R. J., Winter, J. P., & Morse, G. A. (2000). Depression treatment preferences in older
Do consumers who have a choice of treatment primary care patients. The Gerontologist, 46,
have better outcomes? Community Mental Health 14–22.
Journal, 36, 149–60. Hatcher, R. L., & Gillaspy, J. A. (2006).
*Chilvers, C., Dewey, M., Fielding, K., Gretton, V., Development and validation of a revised short
Millwer, P., Palmer, B., et al. (2001). Anti- version of the Working Alliance Inventory.
depressant drugs and generic counseling for Psychotherapy Research, 16, 12–25.
treatment of major depression in primary care: *Iacoviello, B. M., McCarthy, K. S., Barrett, M. S.,
Randomized trial with patient preference arms. Rynn, M., Gallop, R., & Barber, J. P. (2007).
British Medical Journal, 322, 1–5. Treatment preferences affect the therapeutic alli-
Churchill, R., Khaira, M., Gretton, V., Chilvers, C., ance: Implications for randomized controlled
Dewey, M., Duggan, C., et al. (2000). Treating trials. Journal of Consulting and Clinical
depression in general practice: Factors affecting Psychology, 75, 194–98.
patients’ treatment preferences. British Journal of Institute of Medicine (2001). Crossing the quality
General Practice, 50, 905–906. chasm: A new health system for the 21st century.
*Cooper, J. (1980). Reducing fears and increasing Washington, DC: Institute of Medicine.
assertiveness: The role of dissonance reduction. *Kadish, D. A. (1999). Psychological mindedness
Journal of Experimental Social Psychology, 16, and psychotherapy orientation preference as pre-
199–213. dictors of treatment outcome for social phobia.
*Devine, D. A., & Fernald, P. S. (1973). Outcome Dissertation Abstracts International, 60, 832.
effects of receiving a preferred, randomly assigned, *Kludt, C. J., & Perlmuter, L. (1999). Effects of
or nonpreferred therapy. Journal of Consulting control and motivation on treatment outcome.
and Clinical Psychology, 41, 104–107. Journal of Psychoactive Drugs, 31, 405–414.
*Dyck, V. R., & Spinhoven, P. (1997). Does prefer- *Kocsis, J. H., Leon, A. C., Markowitz, J. C.,
ence for type of treatment matter?: A study of Manber, R., Arnow, B., Klein, D. N., & et al.
exposure in vivo with or without hypnosis in the (2009). Patient preference as a moderator of out-
treatment of panic disorder with agoraphobia. come for chronic forms of major depressive dis-
Behavior Modification, 21, 172–86. order treated with Nefazodone, Cognitive
*Elkin, I., Yamaguchi, J. L., Arnkoff, D. B., Glass, C. R., Behavioral Analysis System of Psychotherapy, or
Sotsky, S. M., & Krupnick, J. L. (1999). “Patient- their combination. Journal of Clinical Psychiatry,
treatment fit” and early engagement in therapy. 70, 354–61.
Psychotherapy Research, 9, 437–51. Lambert, M. J., Hansen, N. B., Umpress, V.,
*Ersner-Hershfield, S., Abramowitz, S. I., & Baren, J. Lunnen, K., Okiishi, J., & Burlingame, G. M.
(1979). Incentive effects of choosing a therapist. (1996). Administration and scoring manual for
Journal of Clinical Psychology, 35, 404–406. the OQ-45.2. Stevenson, MD: American
Ertl, M. A., & McNamara, J. R. (2000). Predicting Professional Credentialing Services LLC.
potential client treatment preferences. Psycho- *Leykin, Y., DeRubeis, J., Gallop, R., Amsterdam, J. D.,
therapy, 37, 219–27. Shelton, R. C., & Hollon, S. D. (2007). The
*Fuller, T. C. (1988). The role of patient preference relation of patients’ treatment preferences to out-
for treatment type in the modification of weight come in a randomized clinical trial. Behavior
loss behavior. Dissertation Abstracts International, Therapy, 38, 209–217.
49, 2932. *Lin, P., Campbell, D. G., Chaney, E. F., Liu, C.,
*Gossop, M., Johns, A., & Green, L. (1986). Opiate Heagerty, P., Felker, B. L., et al. (2005). The
withdrawal: Inpatient versus outpatient pro- influence of patient preference on depression
grammes and preferred versus random assign- treatment in primary care. Annals of Behavioral
ment to treatment. British Medical Journal, 293, Medicine, 30, 164–73.
103–104. *Macias, C., Barreira, P., Hargreaves, W., Bickman, L.,
Grote, J., Lange, R., Walker, C., & Vollmer, B. Fisher, W., & Aronson, E. (2005). Impact of
(2009). The effect of client choice on referral source and study applicants’ preference
for randomly assigned service on research enroll- substance abuse treatment program. The
ment, service engagement, and evaluative out- American Journal on Addictions, 6, 168–76.
comes. American Journal of Psychiatry, 162, Swift, J. K., & Callahan, J. L. (2009). The impact
781–87. of client treatment preferences on outcome:
*Manthei, R. J., Vitalo, R. L., & Ivey, A. E. (1982). A meta-analysis. Journal of Clinical Psychology,
The effect of client choice of therapist on therapy 65, 368–81.
outcome. Community Mental Health Journal, 18, Tracey, T. J., & Dundon, M. (1988). Role antici-
220–29. pations and preferences over the course of coun-
*McKay, J. R., Alterman, A. I., McLellan, A. T., seling. Journal of Counseling Psychology, 35,
Boardman, C. R., Mulvaney, F. D., & O’Brien, C. P. 3–14.
(1998). Random versus nonrandom assignment *Van, H. L., Dekker, J., Koelen, J., Kool, S.,
in the evaluation of treatment for cocaine abus- Aalst, G. V., Hendriksen, M., et al. (2009).
ers. Journal of Consulting and Clinical Psychology, Patient preference compared with random allo-
66, 697–701. cation in short-term psychodynamic supportive
*McKay, J. R., Alterman, A. I., McLellan, A. T., psychotherapy with indicated addition of phar-
Snider, E. C., & O’Brien, C. P. (1995). Effect of macotherapy for depression. Psychotherapy
random versus nonrandom assignment in a com- Research, 19, 205–212.
parison of inpatient and day hospital rehabilita- Vincent, N., & LeBow, M. (1995). Treatment pref-
tion for male alcoholics. Journal of Consulting erence and acceptability: Epistemology and locus
and Clinical Psychology, 63, 70–78. of control. Journal of Constructivist Psychology, 8,
*Proctor, E. K., & Rosen, A. (1981). Expectations 81–96.
and preferences for counselor race and their rela- Vollmer, B., Grote, J., Lange, R., & Walker, C.
tion to intermediate treatment outcomes. Journal (2009). A Therapy Preferences Interview:
of Counseling Psychology, 28, 40–46. Empowering clients by offering choices.
*Raue, P. J., Schulberg, H. C., Heo, M., Klimstra, S., Psychotherapy Bulletin, 44, 33–37.
& Bruce, M. L. (2009). Patients’ depression *Wallach, H. S. (1988). Clients’ expectations
treatment preferences and initiation, adherence, and results of psychological therapy for dysmen-
and outcome: A randomized primary care study. orrheal. Dissertation Abstracts International,
Psychiatric Services, 60, 337–43. 49, 1961.
*Renjilian, D. A., Nezu, A. M., Shermer, R. L., Wanigaratne, S., & Barker, C. (1995). Clients’ pref-
Perri, M. G., McKelvey, W. G., & Anton, S. D. erences for styles of therapy. British Journal of
(2001). Individual versus group therapy for obe- Clinical Psychology, 34, 215–22.
sity: Effects of matching participants to their *Ward, E., King, M., Lloyd, M., Bower, P.,
treatment preferences. Journal of Consulting and Sibbald, B., Farrelly, S., et al. (2000). Randomized
Clinical Psychology, 69, 717–21. controlled trial of non-directive counseling,
Riedel-Heller, S. G., Matschinger, H., & cognitive-behavior therapy, and usual general
Angermeyer, M. C. (2005). Mental disorders— practitioner care for patients with depression I:
Who and what might help? Help-seeking Clinical effectiveness. British Medical Journal,
and treatment preferences of the lay public. 321, 1383–88.
Social Psychiatry and Psychiatric Epidemiology, Wong, E. C., Kim, B. S. K., Zane, N. W. S.,
40, 167–74. Kim, I. J., & Huang, J. S. (2003). Examining
*Rokke, P. D., Tomhave, J. A., & Jocic, Z. (1999). culturally based variables associated with ethnic-
The role of client choice and target selection in ity: Influences on credibility perceptions of
self-management therapy for depression in older empirically supported interventions. Cultural
adults. Psychology and Aging, 14, 155–69. Diversity & Ethnic Minority Psychology, 9,
Rosen, A. (1967). Client preferences: An overview 88–96.
of the literature. The Personnel and Guidance *Zlotnick, C., Elkin, I., & Shea, M. T. (1998). Does
Journal, 45, 785–89. the gender of a patient or the gender of a thera-
*Sterling, R. C., Gottheil, E., Glassman, S. D., pist affect the treatment of patients with major
Weinstein, S. P., & Serota, R. D. (1997). Patient depression? Journal of Consulting and Clinical
treatment choice and compliance: Data from a Psychology, 66, 655–59.
C HA P TER
16 Culture
Timothy B. Smith, Melanie M. Domenech Rodríguez, and Guillermo Bernal
The therapist–client relationship is highly with considering culture in a “systematic

dependent on context. Factors such as the fashion” in broad areas of practice. The
therapy format (e.g., family, individual more recent Guidelines on Multicultural
therapy), clinical setting (e.g., group home, Education, Training, Research, Practice,
wilderness retreat), and personal characteris- and Organizational Change for Psychologists
tics of the participants (e.g., age, gender, (APA, 2003) specify that psychologists
culture) influence the content and process of apply culturally appropriate skills in psycho-
therapy. Psychotherapy can be adapted across logical practice, taking cultural context into
nearly infinite therapist–client combina- account at all times. In short, recognizing
tions to achieve positive client outcomes, as and aligning with client culture is not only
evidenced across the other chapters in this best practice, it is ethical practice (APA,
volume. 2002; Bernal, Jiménez-Chafey, & Domenech
In this chapter, we focus on the context Rodríguez, 2009; Smith, 2010).
of client culture. We situate this discussion Despite the clear professional mandates
in the context of evidence-based practice to account for client culture, the implemen-
(EBP), defined by the American Psycho- tation of these standards appears limited.
logical Association Presidential Task Force Engagement into mental health services for
on Evidence-Based Practice (APA, 2006) as ethnic minorities has been low (U.S.
“the integration of the best available research Surgeon General, 2001) and continues to
with clinical expertise in the context of be so (Gonzalez et al., 2010). Some scholars
patient characteristics, culture [emphasis have argued that this low engagement is a
added] and preferences.” (p. 273). Client result of incongruous therapy–client match
culture is an essential context with which (Dumas, Moreland, Gitter, Pearl, &
therapy should align. Nordstrom, 2008) and low relevance of
Professional standards and guidelines available treatments to ethnic minorities
across the mental health professions recog- (Miranda, Azocar, Organista, Muñoz, &
nize the centrality of cultural contexts. The Lieberman, 1996). Other evidence points
Guidelines for Providers of Psychological to language, economic, and structural
Services to Ethnic, Linguistic, and Culturally barriers, such as a lack of mental health
Diverse Populations (APA, 1993), for one clinics in ethnic neighborhoods (Alegría
prominent example, unequivocally state et al., 2002).
that culture and language impact psycho- Disproportionately low rates of utilization
logical services. Psychotherapists are tasked and retention among ethnic minorities may
316
also be related to practitioner demographics. compatible with the client’s cultural
In the United States, the vast majority of patterns, meanings, and values” (Bernal
treatment professionals are white/European et al., 2009, p. 362).
American, primarily English-speaking A less structured conceptualization of
(APA, 2009; NSF, 2009). In a survey of cultural adaptation considers mental health
psychologists, only 12% of respondents treatments tailored to clients’ cultural beliefs
reported speaking a language other than and values, provided in a setting considered
English well enough to provide services in “safe” by the client and conducted in the
that language, and 9% reported actually clients’ preferred language (Miranda,
providing services in another language Nakamura, & Bernal, 2003; Whaley &
(APA, 2010). Meanwhile nearly 20% of the Davis, 2007). For instance, mental health
U.S. population speaks a language other clinics provide culturally adapted services
than English in the home (Shin & Kominski, when they regularly consult with cultural
2010). Ethnic minorities represent roughly group representatives, provide language-
25% of the population in the United States appropriate resources, or modify their intake
and are expected to surpass 50% between procedures to help orient clients unfamiliar
2040 and 2050 (Ortman & Guarnieri, with psychotherapy (Muñoz, 1982).
2009). While neither therapist ethnicity Guidelines for adapting therapy to
nor non–English language fluency imply clients’ cultures have emerged (Barrera &
cultural competence or lack thereof González Castro, 2006; Bernal, Bonilla, &
(Schwartz et al., 2010), the demographic Bellido, 1995; Castro, Barrera, & Martínez,
mismatch between therapists and clients 2004; Hwang, 2006, 2009; Lau, 2006;
may present challenges to client engage- Leong, 1996; Leong & Lee, 2006;
ment in therapy (e.g., Ridley, 1984). Whitbeck, 2006), built on several decades
In this chapter, we consider adaptations of scholarship (Pedersen, 1999). A synthesis
to psychotherapy based on client culture of the work of several international scholars
and present relevant clinical examples. We with expertise in cross-cultural psycho-
then present an original meta-analysis of therapy identified common themes regard-
culturally adapted treatment in mental ing cultural adaptation:
health. We conclude with probable mod-
erators, limitations of the research reviewed, • Therapists must practice flexibly.
and recommended therapeutic practices • Therapists must remain open to what
based on the research evidence. clients bring to therapy.
• Services must be meaningful within
Definitions and Measures the cultural context that they are
Although sometimes broadly considered delivered.
culture relevant or culture sensitive (Atkinson, • Assessments should be conducted
Bui, & Mori, 2001; Hall, 2001; LaRoche prior to implementing treatment.
& Christopher, 2009; Tanaka-Matsumi, • Traditional treatments should not be
2008), the term culturally adapted treat- summarily dismissed but rather used as an
ments has been used frequently in the existing resource.
literature. A precise definition of cultural • Therapists must experience and
adaptation is “the systematic modification communicate empathy with the client
of an evidence-based treatment (EBT) or in a culturally appropriate manner.
intervention protocol to consider language, • Observations of therapy across
culture, and context in such a way that it is cultures provide an opportunity to
s m i t h , d o m e n e c h ro d r í g u e z , b e r n a l 317
learn more about important cultural can help the psychotherapist align treatment
features. with the client rather than presume that
• Therapists must proceed with caution the client will accommodate to the psycho-
in interpreting cultural differences as therapy.
deficits (Draguns, 2008). In addition to considering when and
how to culturally adapt a treatment, psycho-
These common themes can be clinically therapists may want to consider the tension
actualized by eight elements of culturally between population fit and treatment fidel-
adapted treatments: language, persons, met- ity. If a traditional intervention such as
aphors, content, concepts, goals, methods, cognitive therapy is adapted in content and
and context (Bernal & Sáez-Santiago, 2006). format with an Asian American client by
A language-appropriate service refers to not infusing the Buddhist principle of mind-
only the use of clients’ preferred language fulness, for example, there comes a point at
but also to understanding the meaning of which the causal explanations of cognitive
particular uses of language by different therapy may no longer predominate in the
groups such as adolescents. Person factors adapted treatment (e.g., therapy may facili-
include characteristics such as race and tate meditative relaxation/awareness over
ethnicity. Studies of racial, ethnic, and the explicit refutation of irrational thoughts).
language match generally show that clients Research contains a broad spectrum of
both prefer therapists matched to themselves opinions about maintaining traditional
and typically remain in treatment longer treatment fidelity when working with
compared with those who are not matched ethnic minority clients. Some scholars call
(e.g., Coleman, Wampold, & Casali, 1995). for the creation of new therapies specific to
Infusion of cultural metaphors, symbols, and each cultural group that are explicitly
overarching cultural concepts can align ther- aligned with their beliefs, values, and prac-
apy with existing client heuristics. For tices (Comas-Diaz, 2006; Gone, 2009), yet
instance, cultural sayings can be used in others propose implementing traditional
therapy to more clearly convey meaning or EBTs with minimal or no alterations
insight. Attending to the cultural content of (Chambless & Ollendick, 2000). Many
a mental health treatment can enhance scholars, however, seem to opt for an inte-
alignment with client worldviews. For exam- grated or hybrid model of cultural adapta-
ple, some groups are more collectivistic than tion that takes into account both fidelity
others, so notions such as individuation, dif- and fit (Castro et al., 2004; Domenech
ferentiation, and dependence may need to Rodríguez & Wieling, 2004; Hwang, 2006,
be contextualized so as to not pathologize 2009; Lau, 2006; Whitbeck, 2006).
clients with a collectivistic worldview. The These scholars recommend adaptation of
categories of goals and methods imply the existing evidence-based therapies for cultural
consideration of customs and cultural values fit while retaining the original mechanisms
in setting treatment goals and establishing of behavioral change or symptom reduction.
suitable procedures to reach those goals. For example, a Parent Management Training–
And finally, by the consideration of context, Oregon (PMTO) intervention (Domenech
broader issues come into focus such as the Rodríguez, Baumann, & Schwartz, 2011)
social and economic realities that may with Spanish-speaking Latino families main-
include acculturative stress, migration, avail- tained behavioral therapy principles, such
ability of social supports, and so on. In brief, as applying immediate contingencies for
explicit consideration of these eight elements desired behaviors, but the specific behaviors
thought to be desirable, the specific contin- practices, one clinician supported linkages
gencies used, the context in which they to medical practitioners when a family’s
are presented and delivered, the frames or inability to communicate in English inter-
metaphors used to explain the concepts to fered with their ability to secure urgent
caregivers, and the therapeutic process are all medical care for a child (Domenech
changeable or decentrable (for a description Rodríguez, McNeal, & Cauce, 2008). The
of the concept of decentering, see Domenech therapist’s actions went beyond tradi-
Rodríguez & Wieling, 2004). To facilitate tional services by making contact with the
precise descriptions and evaluations of these clients at home in their preferred language
types of adaptations, an observational (Spanish), conducting an evaluation that
measure of cultural adaptation is being devel- went beyond presenting symptoms (in this
oped by the PMTO team. case, a child’s sleep disturbance turned out
to be caused by persistent stomach pain) to
Clinical Examples include cultural and contextual informa-
In a broad sense, all mental health treat- tion, respecting the father’s role in the
ments are informed by cultural contexts. family and working within the cultural
What have been termed “traditional” worldview of the parents, alleviating
Western treatments are inextricably inter- parents’ fears about seeking medical care
woven with European/European-American that were based on their undocumented
culture, so much so as to render the cultural immigration status, arranging for payment
influences nearly invisible (Smith, Richards, of medical services through a public health
Granley, & Obiakor, 2004). Yet, in an program, accompanying the parents to the
increasingly multicultural society, culture medical office visit, and linking the family
cannot remain invisible. with a Hispanic community liaison who
There are a number of ways in which could provide subsequent assistance.
cultural centering of mental health interven- Oppositely, in a recent and poignant
tions can be achieved (Barrera & González negative example, Dr. Guerda Nicholas, a
Castro, 2006; Bernal et al., 1995; Castro well-known Haitian psychotherapist, had
et al., 2004; Domenech Rodríguez & Wieling, sharp words for practitioners wishing to
2004; Hwang, 2006, 2009; Lau, 2006; engage in relief work in Haiti following
Leong, 1996; Leong & Lee, 2006; Whitbeck, the January 12, 2010 earthquake in Haiti
2006). Still other ways to adapt therapy to that claimed hundreds of thousands of
better serve ethnic minority clients include: lives. Dr. Nicolas was quoted as saying
providing additional or ancillary services (e.g., “Please stay away—unless you’ve really,
child care, home visits, referrals for legal or really done the homework” (Marcus, 2010).
medical assistance), supporting consultation/ Among her examples, Dr. Nicolas shared a
collaboration with community/family (e.g., situation in which a psychologist from the
religious clergy and indigenous healers such United States was speaking with a Haitian
as curanderos or santeros), and providing out- woman who had lost her child, her home,
reach services that move beyond the tradi- and a leg. The woman was most upset about
tional patient–therapist office visit to facilitate losing her leg, but the psychologist, appar-
access to services by disadvantaged popula- ently believing that the woman was avoid-
tions (e.g., Alberta & Wood, 2009; Miranda, ing a sensitive topic, insisted on discussing
2006; Pedersen, 2000; Sue & Sue, 2008). the child’s death. Dr. Nicolas lamented that
For instance, in the course of psycho- this psychologist had added one more task
therapy focused on improving parenting to the work of local therapists in Haiti: that
of ensuring that the woman and other research) example of collaboratively access-
Haitians understood that not all therapists ing experiential phenomena and then
would respond in the same unhelpful directly applying that understanding to the
manner. In this example it is evident that treatment rendered.
the psychotherapist privileged his or her
own understanding of trauma over the Previous Meta-Analyses
client’s experience, failing to connect with Evidence has slowly accumulated regarding
the client. Rather than assume that clients the efficacy and effectiveness of culturally
will adapt their ways to fit Western psycho- adapted treatments. In a comprehensive
logical theories, psychotherapists need to analysis of a decade of the randomized
make concerted efforts to align their prac- clinical trials (RCT) conducted with NIH
tices with clients’ lived experiences. funds, less than 50% of the studies reported
This principle is demonstrated in an any data specific to client culture, and all
innovative study conducted in Australia groups except white/European Americans
with Aboriginal people suffering from and African Americans were underrepre-
chronic mental illness (Nagel, Robinson, sented (Mak, Law, Alvidrez, & Pérez-Stable,
Condon, & Trauer, 2009). The study had 2007). Previous reviews have indicated
several notable features. First, it employed that psychotherapy with ethnic minority
a mixed-methods design that entailed a clients is equally effective as that with
12-month qualitative phase followed by a white/European Americans (Hall, 2001;
nested randomized controlled trial. The Miranda et al., 2005; Sue, 1988; Zane
treatment development invited both et al., 2004), but these reviews cite a limited
Aboriginal mental health workers and number of RCTs.
recovered patients as key informants to Two meta-analyses of culturally adapted
understand indigenous views of mental interventions, one specific to children and
illness. Group and individual in-depth youth (Huey & Polo, 2008) and another
interviews were conducted as well as field with clients of all ages (Griner & Smith,
observations. The themes that emerged 2006), found average effect sizes of moder-
were the importance of the family, strength ate magnitude (d = 0.44 and d = 0.45,
derived from cultural traditions, and the respectively), although the results of both
value of storytelling to share information. meta-analyses were moderated by several
These themes were used to inform the factors. The overall positive meta-analytic
process and content of the assessment, findings have been somewhat surprising,
intervention, and ancillary materials. The given the lack of direct measurement of
resulting culturally adapted treatment was cultural adaptation and sparse information
subsequently compared with treatment as available on how cultural adaptations were
usual. In all, 49 patients were randomly implemented.
assigned, and outcomes were evaluated at
baseline, 6-, 12-, and 18-month follow- Meta-Analytic Review
ups. The culturally adapted intervention Methods
produced better outcomes in well-being, Inclusion and Exclusion Criteria. We included
health, and substance dependence with in our meta-analysis those studies that pro-
changes maintained over time. Conducted vided quantitative data regarding clients’
in a remote indigenous area of Australia experiences in mental health treatments that
with a historically underserved population, explicitly accounted for clients’ culture,
this study is an excellent clinical (and ethnicity, or race. We included treatments
for mental illness, emotional distress/well- had appeared from January 2004 to July
being, family problems, and problem behav- 2009 using several electronic databases:
iors (such as physical aggression but not Academic Search Premier, Dissertation
pregnancy or sexual behavior). Substance Abstracts, Mental Health Abstracts, and
abuse prevention and treatment programs PsycINFO. Search terms included a list of
were excluded unless they also targeted root words relevant to psychotherapy
psychological variables (e.g., depression, (clinic, counsel, intervention, psychotherapy,
self-esteem). We excluded studies that service, therapy, and treatment) that were
accounted for generic contextual/ecological crossed with combinations of the root
factors (such as poverty or family systems) terms culture/cultural, ethnic, multicultural,
or other client characteristics (such as and race/racial that were crossed with
gender) unless they explicitly accounted for root terms adapt, appropriate, consonant,
culture, ethnicity, or race (e.g., Latina compatible, competent, congruent, focused,
women). Selection of clients from a particu- informed, relevant, responsive, sensitive, skill,
lar group or assignment of clients to thera- and specific. Three undergraduate research
pists of the same ethnic group or native assistants sequentially reviewed retrieved
language (ethnic or language matching) titles, then abstracts and full texts of appar-
were insufficient criteria for inclusion; some ently relevant reports. One of these assis-
aspect of the content, format, or delivery of tants manually examined the reference
the intervention had to be purposefully sections of past reviews and of studies
changed to align with clients’ culture, meeting the inclusion criteria to locate
ethnicity, or race. We extracted effect size articles not identified in the database
data from psychological and behavioral searches. Finally, we sent personal e-mail
outcomes but not educational, substance requests to several colleagues and posted
use/abuse, or physical health outcomes if general solicitations on several professional
reported. listservs: APA Division 12 Section VI:
A previous meta-analysis (Griner & Clinical Psychology of Ethnic Minorities;
Smith, 2006) aggregated studies using APA Division 45; Association of Black
disparate research designs. This procedure Psychologists; National Latino/a Psycho-
is problematic because correlational designs, logical Association; and the Society of
single-group pre- to posttest designs, and Indian Psychologists.
experimental designs provide distinct data Coding Procedures. Coders were six
that also typically differ in terms of effect undergraduate and four graduate students
size magnitude. Moreover, potential threats with prior experience and training in meta-
to internal validity plague single-group analytic coding. To increase the accuracy of
designs (Campbell & Stanley, 1966). We coding and data entry, two team members
therefore restricted the present meta- coded each article. Subsequently, two
analytic review to quasi-experimental and different team members coded the same
experimental designs. article. Coders extracted several objectively
Search Strategies. We included studies verifiable characteristics of the studies,
identified in prior meta-analyses and including participants’ age, gender, and
reviews (Griner & Smith, 2006; Hall, race; the outcome evaluated; and compo-
2001; Huey & Polo, 2008; Miranda et al., nents of the research design and interven-
2005; Smith, 2010; Zane et al., 2004). tion. Discrepancies across coding pairs were
We subsequently searched for additional resolved through further scrutiny of the
published and unpublished studies that manuscript to the point of consensus.
Statistical Methods. Data within studies 0.46 represents a medium effect size, indi-
were transformed to the metric of Cohen’s d. cating that patients receiving culturally
Across all studies we assigned positive d adapted treatments typically experienced
values to indicate beneficial results and neg- superior outcomes to those of patients in
ative d values to comparatively worse results control groups. Substantial heterogeneity
for the culturally adapted intervention. characterized the effect sizes (range = −.97
When multiple effect sizes were reported to 2.80), with 74% of the variability in effect
within a study (e.g., across different mea- sizes due to true between-study variability
sures of outcome), we averaged the several (I 2 = 74; Q (64) = 247, p < 0.001). No extreme
values (weighted by N) to avoid violat- outliers were observed.
ing the assumption of independent sam- We conducted several analyses to deter-
ples. Aggregate effect sizes were calculated mine if the meta-analytic results may have
using random effects models following been influenced by publication bias (the
confirmation of heterogeneity. A random exclusion of studies with negative or nonsig-
effects approach produces results that best nificant results because they tend to be
generalize beyond the sample of studies unpublished and difficult to locate).
reviewed. The assumptions made in this Calculation of Orwin’s fail-safe N indi-
meta-analysis clearly warrant this method: cated that there would need to be at least
The belief that different kinds of modifica- 103 studies averaging d = 0 that were “miss-
tions to mental health treatments and client ing” from our literature search for the
characteristics moderate the effectiveness overall results to be reduced to a trivial
of psychotherapy implies that the studies magnitude (d < 0.10). Although unlikely,
reviewed will estimate different popula- it was possible that many studies with
tion effect sizes. Random effects models nonsignificant findings remained unac-
take such between-studies variation into counted for over a 30-year period, leaving
account. open the possibility of publication bias.
Egger’s regression test reached statistical
Results significance (p < 0.001), and our examina-
Statistically nonredundant effect sizes were tion of the funnel plot of the effect sizes by
extracted from 65 studies that evaluated their standard error indicated approxi-
culturally adapted interventions using quasi- mately 15 “missing” studies on the left side
experimental or experimental designs. These of the distribution, where statistically non-
studies and their ESs are summarized in significant results would be located in the
Table 16.1. Data were reported from 8,620 expected funnel-shaped distribution. When
participants, with an average age of 24.4 years we reestimated the average weighted effect
(range = 5 to 73; SD = 16); 55% of the size using “trim and fill” methodology (Duval
participants within studies were female. Of & Tweedie, 2000), the recalculated value
the total, 39% were Asian American, 32% was d = 0.27 (95% CI = 0.16 to 0.38).
were Hispanic/Latino(a), 20% were African
American, 4% were Native American, 1% Moderators and Mediators
were white/European American, and 4% Given the substantial heterogeneity in the
indicated “other” affiliations including ethnic omnibus effect size estimate, we evaluated
groups outside North America. what factors may have accounted for the
Across all 65 studies, the weighted average variation across the 65 studies. Analyses of
effect size was d = 0.46 (95% CI = 0.36– effect size moderation were conducted
0.56). By conventional benchmarks, a d of using random effects weighted correlations
Table 16.1 Studies Included in the Meta-Analysis
Study N Mean age Effect size Lower limit Upper limit
Acosta, Yamamoto, Evans, & Skilbeck (1983) 151 31 0.42 0.08 0.76
Banks, Hogue, Timberlake, & Liddle (1998) 64 12 0.59 0.07 1.11
Belgrave (2002) 49 0.61 0.02 1.20
Botvin, Schinke, & Diaz (1994) 304 13 0.11 −0.09 0.30
Cardemil, Reivich, Beevers, 168 11 0.15 −0.13 0.43
Seligman, & James (2007)
Costantino, Malgady, & Rogler (1994) 90 11 0.22 −0.17 0.61
Crespo (2006) 36 38 1.10 0.37 1.84
Dai et al. (1999) 30 73 0.96 0.08 1.83
Domenech Rodríguez & Crowley (2008) 195 23 0.29 0.01 0.56
Falconer (2002) 25 20 −0.20 −0.99 0.58
Gallagher-Thompson, Arean, 70 52 0.51 0.03 0.99
Rivera, & Thompson (2001)
Garza (2004) 29 8 0.21 −0.53 0.94
Gilchrist, Schinke, Trimble, & Cvetkovich (1987) 97 11 0.04 −0.35 0.43
Ginsburg & Drake (2002) 9 16 0.87 −0.50 2.24
Gonzalez (2003) 57 10 0.27 −0.25 0.79
Grodnitzky (1993) 28 14 0.31 −0.45 1.07
Gutierrez & Ortega (1991) 73 19 0.60 0.09 1.12
Hammond & Yung (1991) 19 14 0.80 −0.16 1.76
Heppner, Neville, Smith, Kivlighan, 41 20 0.25 −0.37 0.87
& Gershuny (1999)
Hinton et al. (2005) 40 52 2.80 1.92 3.68
Hinton, Hofmann, Pollack, & Otto (2009) 24 50 2.26 1.24 3.27
Hogue, Liddle, Becker, & Johnson-Leckrone (2002) 114 13 0.53 0.14 0.92
Huey & Rank (1984) 48 0.95 0.36 1.53
Huey & Pan (2006) 15 24 0.82 −0.25 1.90
Jackson (1997) 14 17 −0.05 −1.10 1.01
Johnson, & Breckenridge (1982) 128 5 0.20 −0.14 0.54
Jones (2008) 10 42 1.58 0.58 2.58
Kataoka et al. (2003) 198 11 0.46 0.12 0.80
Kim, Omizo, & D’Andrea (1998) 48 16 0.91 0.33 1.50
(Continued)
323
King (1999) 80 13 0.17 −0.27 0.60
Kohn, Oden, Muñoz, Robinson, & Leavitt (2002) 18 47 0.81 −0.15 1.77
Kopelowicz, Zárate, Smith, Mintz, 162 38 0.85 0.51 1.19
& Liberman (2003)
LaFromboise, & Howard-Pitney (1995) 62 16 0.33 −0.19 0.85
Lau & Zane (2000) 317 37 0.14 0.07 0.21
Malgady, Rogler, & Constantino (1990a) 80 14 0.46 −0.02 0.94
Malgady, Rogler, & Constantino (1990b) 210 8 0.44 0.16 0.72
Malgady, Rogler, & Constantino (1990b) 90 14 0.27 −0.17 0.71
Martinez & Eddy (2005) 52 13 0.46 −0.10 1.01
Matos, Bauermeister, & Bernal (2009) 32 5 1.81 0.97 2.64
Mausbach, Bucardo, McKibbin, 59 49 0.70 0.15 1.26
Cardenas, & Barrio (2008)
Mickens-English (1996) 60 35 −0.09 −0.61 0.43
Mokuau, Braun, Wong, Higuchi, & Gotay (2008) 10 57 1.21 −0.17 2.58
Moran (1999) 85 11 −0.23 −0.67 0.21
Myers et al. (1992) 1 92 31 0.38 −0.06 0.82
1
Myers et al. (1992) 81 33 0.78 0.34 1.22
Nagel, Robinson, Condon, & Trauer (2009) 49 33 0.67 0.08 1.26
Nyamathi, Leake, Flaskerud, 858 33 −0.11 −0.24 0.03
Lewis, & Bennett (1993)
Ochoa (1981) 21 15 0.58 −0.32 1.48
Pantin et al. (2003) 167 12 0.35 0.07 0.62
Parker (1990) 23 17 0.22 −0.61 1.05
Rosselló & Bernal (1999) 59 15 0.81 0.23 1.40
Rosselló, Bernal, & Rivera-Medina (2008) 112 15 1.21 0.82 1.60
Rowland et al. (1995) 31 15 0.40 −0.31 1.11
Royce (1998) 55 14 0.04 −0.48 0.56
Santisteban et al. (2003) 85 16 0.46 0.03 0.90
Schwarz (1989) 72 38 0.18 −0.30 0.66
Shin (2004) 47 66 1.56 0.91 2.21
Shin & Lukens (2002) 47 37 1.07 0.45 1.69
(Continued)
324
Sobol (2000) 89 −0.16 −0.60 0.28
Szapocznik et al. (1986) 31 15 0.38 −0.32 1.09
Szapocznik et al. (1989) 76 9.4 0.30 −0.14 0.74
Telles et al. (1995) 40 30 −0.97 −1.62 −0.32
Timberlake (2000) 74 0.19 −0.29 0.67
Xiong et al. (1994) 62 31 0.51 −0.01 1.03
Zhang et al. (2002) 97 35 0.54 0.15 0.93
Note: Effect sizes and 95% confidence intervals are expressed in the metric of Cohen’s d.
1
The publication by Myers et al. (1992) contained two studies, both included in our analyses.
for continuous-level variables and random clinical status (p = 0.46), moderated effect
effects weighted analyses of variance for size magnitude.
categorical variables. Differences were observed between
studies using participants of different races
Participant Characteristics (Q (3, 48) = 12.8, p = 0.005). Specifically,
We evaluated the association between effect 7 studies with Asian American participants
sizes and the following characteristics of (d = 1.18, 95% CI = 0.79 to 1.60) had an
study participants: gender composition average effect size of more than twice that
(percentage of females), average age, mental of 14 studies of African American partici-
health status (normal community members, pants (d = 0.47, 95% CI = 0.19 to 0.76),
at-risk group members, clients in clinical 26 studies of Hispanic/Latino(a) partici-
settings), and racial composition. Of these, pants (d = 0.47, 95% CI = 0.28 to 0.65),
participants’ average age was significantly and 5 studies of Native American partici-
associated (r = 0.39 p < 0.001) with the pants (d = 0.22, 95% CI = −.20 to 0.64).
magnitude of effect sizes within studies. Differences were also found between stud-
Investigation of the associated scatter plot ies using culturally homogeneous samples
(funnel plot) revealed that studies with (i.e., all participants were of the same
adult participants over age 35 tended to culture) and culturally heterogeneous sam-
have effect sizes of larger magnitude than ples (Q(1, 63) = 5.2, p = 0.02). Interventions
studies with children, adolescents, and delivered to a specific cultural group were
young adults. Further investigation of the much more effective (d = 0.51, 95%
data revealed that there was substantial CI = 0.40 to 0.63) than interventions deliv-
overlap between participant age and the ered to mixed groups (d = 0.18, 95%
clinical status of the population investi- CI = −.08 to 0.44).
gated: normal community samples had an
average age of 20 years; at-risk groups had Study Design Variables
an average age of 21 years; and clinical We next evaluated the association between
populations had an average age of 32 years. effect sizes and several characteristics of
Nevertheless, we confirmed through study design: random assignment, control
random effects weighted multiple regres- group condition, type of outcome evalu-
sion that participant age (p = 0.01), not ated, and the time of outcome assessment
administration (number of sessions com- to psychotherapy process and outcome (Zane
pleted at posttest). Of these, the only statis- et al., 2004). Yet, due in part to the limited
tically significant difference observed was outcome research with ethnic minorities, the
across the source of the outcome evaluation available research on client characteristics
(Q(2, 108) = 6.7, p = 0.04); outcome evalua- relies on analog studies, and much of the
tions provided by therapists tended to be work has focused on establishing efficacy
associated with effect sizes of much lower (Miranda et al., 2005; Zane et al., 2004).
magnitude (d = 0.09) than those provided Furthermore, the high degree of heterogene-
by the clients (d = 0.45) or external observ- ity across the major ethnocultural groups—
ers (d = 0.45). Native Americans, Asian Americans, African
We next evaluated whether authors Americans, and Latinos—and diversity
included descriptions of treatment compo- within those groups calls into question any
nents that aligned with the eight points of generalization that can be made in linking
Bernal’s model (Bernal et al., 1995; Bernal client characteristics to the therapy relation-
& Sáez-Santiago, 2006). Each of the eight ship and even to outcome.
components was assigned a binary value With this caveat in mind, we turn to the
(yes = 1, no = 0), which we summed to patient’s contribution to the relationship in
obtain a total number of culturally adapted culturally adapted psychotherapies and the
components described within each study. distinctive perspective he/she brings to the
This total value was positively associated interaction. In a comprehensive review of
with effect sizes (r = 0.28, p = 0.007), indi- the research on psychotherapy with diverse
cating that studies describing treatments populations, Zane and colleagues (2004)
with relatively more cultural adaptations examined several cultural groups and
tended to be more effective than studies discussed the salient client variables that
describing treatments with fewer cultural include: preference for a therapist of the
adaptations. To ascertain the amount of vari- same ethnicity and language, valuing inter-
ance in effect sizes explained by the cultural personal over instrumental orientation, the
adaptations described within studies, we role of the experience of discrimination and
simultaneously entered into a random effects prejudice, preference for therapists who
weighted multiple regression of the eight evoke positive attitudes and trustworthiness,
binary variables of language matching, acculturation, causal attributions on the
ethnic matching, metaphors, content, con- nature of illness and symptoms, and cultur-
ceptualization, goals, methods, and context ally specific symptoms in some popula-
(described previously). The resulting model tions. The authors found that the most
explained 20% of the variance in effect sizes salient commonality across the four groups
(p = 0.03); the two variables that reached examined was that a substantial number of
statistical significance were descriptions of ethnic minorities prefer therapists of their
therapeutic goals that explicitly matched own ethnicity. Subsequent research has
clients’ goals (b = 0.29, p = 0.02) and descrip- confirmed this finding of client preference
tions of using metaphors/symbols in therapy but has failed to find evidence that treat-
that matched client cultural worldviews ment outcomes improve as a result of ethnic
(b = 0.37, p = 0.02). matching (Cabral & Smith, 2010).
Much work remains to be done to under-
Patient Contributions stand the impact of culture-specific patient
There is a small but growing literature on characteristics on treatment impact. In much
client characteristics and their contribution of treatment outcome research, cultural
values are implied rather than measured therapies (Luborsky et al., 2006). Even
(LaRoche & Christopher, 2009); for exam- though our analyses indicated that ratings
ple, positive results may be the result of a of client outcomes provided by therapists
cultural value such as personalismo, but per- were of lower magnitude than those provided
sonalismo is not directly measured. When by clients or external observers, researcher
clients are asked about these, some under- allegiance to culturally adapted interven-
standing of cultural values map onto clients’ tions may nevertheless be associated with
but others do not (Bermúdez, Kirkpatrick, outcomes of the studies included in our
Hecker, & Torres-Robles, 2010). A note- review. Adjusting the results for apparently
worthy example to the contrary measured “missing” nonsignificant findings (due to
Asian American and European American possible publication bias) reduced the
cultural values and found that both related magnitude of the omnibus effect size. Until
to differential outcomes for Asian American additional unpublished reports appear or
clients (Kim, Ng, & Ahn, 2005). until studies explicitly control for researcher
allegiance, the adjusted value of d = 0.27
Limitations of the Research represents a lower estimate of the compara-
Psychotherapy outcome research has accu- tive benefit of culturally adapted interven-
mulated over several decades, with now tions than the omnibus value of d = 0.46.
thousands of research reports and hun- A third limitation of the research concerns
dreds of meta-analyses. By comparison, the the heterogeneity of the adapted treatments.
research investigating culturally adapted Studies included in this review used a variety
treatments is miniscule. The amount and of means to align mental health interven-
pace of clinical outcome research specific to tions with clients’ cultures, with an average
clients’ cultural backgrounds has remained of four of the eight components (Bernal
consistently low. The studies included in et al., 1995; Bernal & Sáez-Santiago, 2006)
this review appeared at a steady rate of about being explicitly described by authors within
2.3 per year since 1981. Because the long- studies. Specifically, 74% described provid-
term success of any initiative depends on ing therapy in the clients’ preferred language,
the consistent replication of supportive 53% matched clients with therapists of
findings, the single greatest limitation of the similar ethnic/racial backgrounds, 42%
research specific to culturally adapted utilized metaphors/objects from client
mental health interventions is that more cultures, 77% included explicit mention of
evidence needs to accumulate. cultural content/values, 37% adhered to the
Across the history of psychotherapy, there client’s conceptualization of the presenting
have been multiple cycles wherein a new problem, 14% solicited outcome goals from
theory or treatment attains popularity fol- the client, 43% modified the methods of
lowing initial research support, but then delivering therapy based on cultural
enthusiasm and implementation decline considerations, and 55% addressed clients’
when subsequent research fails to replicate contextual issues. A regression model includ-
the initial positive results. This failure to ing all of these variables explained 20% of
replicate results can be partially attributable effect size variation, and treatments that
to increased empirical rigor and the identifi- included greater numbers of these adapta-
cation of potential confounds omitted in tions tended to be more effective than treat-
previous studies. Researcher allegiance ments with fewer cultural adaptations.
effects, in particular, have been identified Another limitation of the research base
as a confound in comparisons of specific was the lack of systematic measurement of
cultural adaptation within studies. Few superior to those that do not explicitly
studies confirmed the fidelity of the treat- incorporate cultural considerations.
ments provided. All interventions in this Thus, we advance the following research-
review were developed by Western-trained supported therapeutic practices:
professionals, with mention of consultation
with indigenous healers or cultural experts • Clients will tend to benefit when
in 30 of 65 studies (46%). Many authors psychotherapists make attempts to align
adapted traditional (Western) mental health treatment with clients’ cultural
interventions, but evaluation of the fidelity backgrounds.
to the causal mechanisms assumed by the • Asian American clients and
traditional intervention was rare. In short, adult clients tended to benefit most
existing clinical outcome research of cultural from culturally adapted treatments
adaptations has inconsistently achieved relative to clients of other groups and
high levels of methodological rigor. younger ages. Nevertheless, because
Finally, we observed that much of the both age and Asian American culture
research describes preventative interven- are likely mediating factors of accultur-
tions with at-risk populations. Although ation status (integration with mainstream
preventative interventions are essential in Western society vs. maintaining ancestral
at-risk communities, the fact that only 20 cultural worldviews), therapists should
of the 65 studies were conducted in clinical particularly aware of how client age and
settings with mental health clients must be acculturation interact with their
acknowledged as a gap in coverage. Although treatments.
we have no reason to suspect that the • Treatments explicitly aligned with
benefit of culturally adapted interventions clients’ outcome goals will tend to be
would differ between clinical populations more effective than other treatments.
and at-risk populations (and no differences • Treatments involving cultural
were observed in the meta-analytic results), metaphors and modes of expression
researchers have come to rely on greater will tend to be more effective than
aggregate numbers of clinical studies than other treatments. Whenever feasible,
present coverage allows. psychotherapy should be conducted in
the client’s preferred language.
Therapeutic Practices • Different combinations of the
Mental health treatments typically yield eight components of culturally adapted
patient outcomes of similar magnitude, interventions (Bernal et al., 1995) proved
irrespective of differences in content effective across studies. Rather than exert
(Lambert, 1999). Bona fide comparisons treatment-specific effects, it is possible
of client outcomes across different thera- that cultural adaptations to treatment
pies usually average between a d of 0 and influence common factors, such as
0.20 (Wampold et al., 1997). By compari- the therapeutic alliance and patient
son, the omnibus effect size obtained in preferences. The specific procedures taken
this meta-analysis (d = 0.46) exceeds those to align therapy with client culture may
expected values. Even if we interpret the matter less than the fact that therapists
omnibus effect size adjusted for possi- attempt to make the alignment by
ble publication bias (d = 0.27), these results using several methods (Smith, 2010).
remain important. Culturally adapted Treatments that include multiple
mental health therapies are moderately cultural adaptations will tend
to be more effective than treatments Alberta, A. J., & Wood, A. H. (2009). A practical
with only a few cultural adaptations. skills model for effectively engaging clients in
multicultural settings. The Counseling Psychologist,
• Consistent components identified in
37(4), 564–79.
culturally adapted therapies include the Alegría, M., Canino, G., Rios, R., Vera, M.,
following: Calderon, J., Rusch, D., et al. (2002). Inequa-
• Work to establish a strong lities in use of specialty mental health services
therapeutic alliance. Desire to among Latinos, African Americans, and non-Latino
understand the client. Demonstrate that Whites. Psychiatric Services, 53(12), 1547–55.
American Psychological Association (1993).
you hold the client in high regard.
Guidelines for providers of psychological services
• Confirm client expression and to ethnic, linguistic, and culturally diverse popu-
reception, optimally in the client’s lations. American Psychologist, 48(1), 45–48.
preferred language. American Psychological Association (2002). Ethical
• Verify clients’ expectations and principles of psychologists and code of conduct.
conceptualizations of optimal mental American Psychologist, 57, 1060–73.
American Psychological Association (2003).
health; align treatment goals accordingly.
Guidelines on multicultural education, training,
• Use therapeutic methods that are research, practice, and organizational change for
compatible with the clients’ values and psychologists. American Psychologist, 58, 377–402.
conceptualization of improvement. American Psychological Association (2006).
• Maintain a feedback loop whereby Evidence-based practice in psychology: APA presi-
clients express progress and dential task force on evidence-based practice.
American Psychologist, 61, 271–85.
expectations. For clients preferring
American Psychological Association (2009). 2007
directive therapeutic approaches, doctorate employment survey. Retrieved online
completion of rating scales would be 5/17/2010 from http://www.apa.org/workforce/
preferable to interpersonal dialogue publications/07-doc-empl/index.aspx
about process issues. American Psychological Association (2010). 2008
• Respond immediately to client APA Survey of Psychology Health Service Providers.
Special Analysis. Washington, DC: American
feedback; verify the congruity of your
Psychological Association.
response. Atkinson, D. R., Bui, U., & Mori, S. (2001).
• Culturally adapted treatments Multiculturally sensitive empirically supported
were much more beneficial when they treatments—an oxymoron? In C. M. Alexander
were specific to clients of a given ethnic (Ed.), Handbook of multicultural counseling (2nd
group than when they were provided ed.). (pp. 542–74). Thousand Oaks, CA: Sage
Publications, Inc.
to a conglomerate of clients from many ∗
Banks, R., Hogue, A., Timerlake, T., & Liddle, H.
ethnocultural groups. The more culturally (1998). An Afrocentric approach to group social
focused and specific the treatment, the skills training with inner-city African American
more effective it will probably prove. adolescents. Journal of Negro Education, 65(4),
414–23.
Barrera, M., & González Castro, F. (2006).
References A heuristic framework for the cultural adapta-
Citations marked with an asterisk were tion of interventions. Clinical Psychology: Science
included in the meta-analysis. & Practice, 13(4), 311–316.
∗ ∗
Acosta, F. X., Yamamoto, J., Evans, L. E., & Belgrave, F. Z. (2002). Relational theory and
Skilbeck, W. M. (1983). Preparing low-income cultural enhancement interventions for African
Hispanic, Black, and White patients for psycho- American adolescent girls. Public Health Reports,
therapy: Evaluation of a new orientation 117(1), 76–81.
program. Journal of Clinical Psychology, 39(6), Bernal, G., Bonilla, J., & Bellido, C. (1995).
872–77. Ecological validity and cultural sensitivity for
outcome research: Issues for cultural adaptation ∗
Costantino, G., Malgady, R. G., & Rogler, L. H.
and development of psychosocial treatments (1994). Storytelling through pictures: Culturally
with Hispanics. Journal of Abnormal Child sensitive psychotherapy for Hispanic children
Psychology, 23, 67–82. and adolescents. Journal of Clinical Child
Bernal, G., Jimenez-Chafey, M. I., & Domenech Psychology, 23(1), 13–20.
Rodriguez, M. M. (2009). Cultural adaptation Comas-Diaz, L. (2006). Latino healing: The
of treatments: A resource for considering culture integration of ethnic psychology into psycho-
in evidence-based practice. Professional Psychology: therapy. Psychotherapy: Theory, Research, Practice,
Research and Practice, 40(4), 361–68. and Training, 4, 436–453.
Bernal, G., & Sáez-Santiago, E. (2006). Culturally ∗
Crespo, M. M. (2006). Effects of culturally
centered psychosocial interventions. Journal of specific dynamically oriented group art therapy with
Community Psychology, 34, 121–32. immigrant Latinas. Unpublished doctoral disser-
Bernal, G., & Scharrón-del-Río, M. (2001). Are tation, California Institute of Integral Studies,
empirically supported treatments valid for ethnic San Francisco, CA.
minorities? Toward an alternative approach for ∗
Dai, Y., Zhang, S., Yamamoto, J., Ao, M.,
treatment research. Cultural Diversity and Ethnic Belin, T. R., Cheung, F., & Hifumi, S. S. (1999).
Minority Psychology, 7, 328–42. Cognitive behavioral therapy of minor depres-
Bermúdez, J. M., Kirkpatrick, D. R., Hecker, L., sive symptoms in elderly Chinese Americans:
Torres-Robles, C. (2010). Describing Latino A pilot study. Community Mental Health Journal,
families and their help-seeking attitudes: Chal- 35(6), 537–42.
lenging the family therapy literature. Contem- Domenech Rodríguez, M. M., Baumann, A., &
porary Family Therapy, 32, 155–72. Schwartz, A. (2011). Cultural adaptation of an
∗
Botvin, G. J., Schinke, S. P., Epstein, J. A., & Diaz, T. empirically supported intervention: From theory
(1994). Effectiveness of culturally focused and to practice in a Latino/a community context.
generic skills training approaches to alcohol and American Journal of Community Psychology 47,
drug abuse prevention among minority youths. 170–186.
Psychology of Addictive Behavior, 8(2), 116–27. ∗
Domenech Rodríguez, M. M., & Crowley, S. L.
Cabral, R. C., & Smith, T. B. (manuscript (2008, August). Criando con amor: Promoviendo
submitted). Racial/ethnic matching of clients armonia y superacion: Findings from a randomized
and therapists in mental health services: A meta- trial of a culturally adapted PMTO intervention.
analytic review of preferences, perceptions, and Paper presented at the annual meeting of the
outcomes. American Psychological Association, Boston.
Campbell, D. T., & Stanley, J. C. (1966). Domenech Rodríguez, M., McNeal, C. T., & Cauce,
Experimental and quasi-experimental designs for A. M. (2008). Counseling the marginalized. In
research. Chicago: Rand McNally. P. B. Pedersen, J. G. Daguns, W. J. Lonner, &
∗
Cardemil, E. V., Reivich, K. J., Beevers, C. G., J. E. Trimble (Eds.), Counseling across cultures
Seligman, M. E. P., & James, J. (2007). The (6th ed.) (pp. 223–38). Thousand Oaks, CA: Sage.
prevention of depressive symptoms in low- Domenech Rodríguez, M., & Wieling, E. (2004).
income, minority children: Two-year follow-up. Developing culturally appropriate, evidence-
Behaviour Research and Therapy, 45(2), 313–27. based treatments for interventions with ethnic
Castro, F. G., Barrera, M., & Martinez, C. R. minority populations. I. In M. Rastogin &
(2004). The cultural adaptation of prevention E. Wieling (Eds.), Voices of color: First person
interventions: Resolving tensions between accounts of ethnic minority therapists (pp. 313–33).
fidelity and fit. Prevention Science, 5(1), 41–45. Thousand Oaks, CA: Sage.
Chambless, D., & Ollendick, T. (2001). Empirically Draguns, J. G. (2008). What have we learned about
supported psychological interventions: Contro- the interplay of culture with counseling and
versies and evidence. Annual Review of Psychology, psychotherapy? In U. P. Gielen, J. G. Draguns &
52, 685–716. J. M. Fish (Eds.), Principles of multicultural
Coleman, H. L. K., Wampold, B. E., & Casali, S. L. counseling and therapy. (pp. 393–417). New York:
(1995). Ethnic minorities’ ratings of ethnically Routledge/Taylor & Francis Group.
similar and European American counselors: Dumas, J. E., Moreland, A. D., Gitter, A. H.,
A meta-analysis. Journal of Counseling Psychology, Pearl, A. M., & Nordstrom, A. H. (2008). Engaging
42(1), 55–64. parents in preventive parenting groups: Do ethnic,
socioeconomic, and belief match between par- ∗
Grodnitzky, G. R. (1993). Hero modeling versus
ents and group leaders matter? Health Education non-hero modeling as interventions for Puerto-
& Behavior, 35(5), 619–33. Rican and Anglo adolescents exhibiting behavior
Duval, S., & Tweedie, R. (2000). Trim and fill: problems. Unpublished doctoral dissertation,
A simple funnel-plot-based method of testing Hofstra University, Hempstead, NY.
and adjusting for publication bias in meta- ∗
Gutierrez, L. M., & Ortega, R. (1991). Developing
analysis. Biometrics, 56(2), 455–63. methods to empower Latinos: The importance
∗
Falconer, J. (2002). The effectiveness of a culturally of groups. Social Work with Groups, 14(2),
relevant eating disorder prevention intervention 23–43.
with African American college women. Unpub- Hall, G. C. N. (2001). Psychotherapy research
lished doctoral dissertation, University of and ethnic minorities: Empirical, ethical, and
Missouri-Columbia. conceptual issues. Journal of Counseling and
∗
Gallagher-Thompson, D., Arean, P., Rivera, P., & Clinical Psychology, 69, 502–510.
Thompson, L. W. (2001). A psychoeducational ∗
Hammond, W. R., & Yung, B. R. (1991).
intervention to reduce distress in Hispanic family Preventing violence in at-risk African-American
caregivers: Results of a pilot study. Clinical youth. Journal of Health Care for the Poor and
Gerontologist, 23(1/2), 17–32. Underserved, 2(3), 359–73.
∗ ∗
Garza, Y. (2004). Effects of culturally responsive child- Heppner, M. J., Neville, H. A., Smith, K.,
centered play therapy compared to curriculum-based Kivlighan, D. M., & Gershuny, B. S. (1999).
small group counseling with elementary-age Hispanic Examining immediate and long-term efficacy of
children experiencing externalizing and internal- race prevention programming with racially
izing behavior problems: A preliminary study. diverse college men. Journal of Counseling
Unpublished doctoral dissertation, University of Psychology, 46(1), 16–26.
North Texas, Denton. ∗
Hinton, D., Chhean, D., Pich, V., Safren, S.,
∗
Gilchrist, L. D., Schinke, S. P., Trimble, J. E., & Hofmann, S., & Pollack, M. (2005).
Cvetkovich, G. T. (1987). Skills enhancement to A randomized controlled trial of cognitive-
prevent substance abuse among American Indian behavior therapy for cambodian refugees with
adolescents. The International Journal of the treatment-resistant PTSD and panic attacks:
Addictions, 22(9), 869–79. A cross-over design. Journal of Traumatic Stress,
∗
Ginsburg, G. S., & Drake, K. L. (2002). School- 18(6), 617–29.
based treatment for anxious African-American ∗
Hinton, D. E., Hofmann, S. G., Pollack, M. H., &
adolescents: A controlled pilot study. Journal of Otto, M. W. (2009). Mechanisms of efficacy of
the American Academy of Child & Adolescent CBT for Cambodian refugees with PTSD:
Psychiatry, 41(7), 768–75. Improvement in emotion regulation and orthos-
Gone, J. P. (2009). A community-based treatment tatic blood pressure response. CNS Neuroscience
for Native American historical trauma: Prospects & Therapeutics, 15(3), 255–63.
for evidence-based practice. Journal of Consulting ∗
Hogue, A., Liddle, H. A., Becker, D., &
and Clinical Psychology, 77(4), 751–62. Johnson-Leckrone, J. (2002). Family-based
∗
Gonzalez, M. (2003). The effectiveness of a culturally prevention counseling for high-risk young
based social work activity group in the promotion of adolescents: Immediate outcomes. Journal of
ethnic identity and adaptive behaviors: A study of Community Psychology, 30(1), 1–22.
Puerto Rican children. Unpublished doctoral ∗
Huey, S. J., & Pan, D. (2006). Culture-responsive
dissertation, Catholic University of America, one-session treatment for phobic Asian
Washington, DC. Americans: A pilot study. Psychotherapy: Theory,
Gonzalez, H. M., Vega, W. A., Williams, D. R., Research, Practice, Training, 43(4), 549–54.
Wassim Tarraf, W., West, B. T., & Huey, S., & Polo, A. (2008). Evidence-based
Neighbors, H. W. (2010). Depression care in the psychosocial treatments for ethnic minority
United States: Too little for too few. Archives of youth. Journal of Clinical Child and Adolescent
General Psychiatry, 67, 37–46. Psychology, 37(1), 262–301.
Griner, D., & Smith, T. B. (2006). Culturally ∗
Huey, W. C., & Rank, R. C. (1984). Effects of
adapted mental health intervention: A meta- counselor and peer-led group assertive training
analytic review. Psychotherapy: Theory, Research, on Black adolescent aggression. Journal of
Practice, Training, 43(4), 531–48. Counseling Psychology, 31(1), 95–98.
Hwang, W. (2006). The psychotherapy adaptation ∗
Kopelowicz, A., Zarate, R., Gonzalez Smith, V.,
and modification framework: Application to Mintz, J., & Liberman, R. (2003). Disease
Asian Americans. American Psychologist, 61(7), management in Latinos with schizophrenia:
702–715. a family-assisted, skills training approach.
Hwang, W. (2009). The formative method for Schizophrenia Bulletin, 29(2), 211–27.
adapting psychotherapy (FMAP): A community- Lambert, M. J. (1999). Are differential treatment
based developmental approach to culturally effects inflated by researcher therapy allegiance?
adapting therapy. Professional Psychology: Research could clever Hans count? Clinical Psychology:
and Practice, 40(4), 369–77. Science and Practice, 6(1), 127–30.
∗ ∗
Jackson, P. A. (1997). The effect of exposure to LaFromboise, T., & Howard-Pitney, B. (1995).
culturally relevant/historically based material on The Zuni life skills development curriculum:
level of frustration tolerance, level of depression, Description and evaluation of a suicide
and mediation of anger in African-American young prevention program. Journal of Counseling
males. Unpublished doctoral dissertation, The Psychology, 42(4), 479–86.
California School of Professional Psychology, La Roche, M., & Christopher, M. (2009). Changing
Alameda. paradigms from empirically supported treatment
∗
Johnson, D. L. & Breckenridge, J. N. (1982). The to evidence-based practice: A cultural perspective.
Houston Parent-Child Development Center and Professional Psychology: Research and Practice,
the primary prevention of behavior problems in 40(4), 396–402.
young children. American Journal of Community Lau, A. S. (2006). Making the case for selective and
Psychology, 10(3), 305–316. directed cultural adaptations of evidence-based
∗
Jones, L. V. (2008). Preventing depression: treatments: Examples from parent training.
Culturally relevant group work with black Clinical Psychology: Science & Practice, 13(4),
women. Research on Social Work Practice, 18(6), 295–310.
626–34. ∗
Lau, A., & Zane, N. (2000). Examining the effects
∗
Kataoka, S. H., Stein, B. D., Jaycox, L. H., of ethnic-specific services: An analysis of cost-
Wong, M., Escudero, P., Tu, W., et al. (2003). utilization and treatment outcome for Asian
A school-based mental health program for trau- American clients. Journal of Community
matized Latino immigrant children. Journal of Psychology, 28(1), 63–77.
the American Academy of Child & Adolescent Leong, F. T. L. (1996). Toward an integrative model
Psychiatry, 42(3), 311–318. for cross-cultural counseling and psychotherapy.
Kim, B. S. K., Ng, G. F., & Ahn, A. J. (2005). Applied & Preventive Psychology, 5(4), 189–209.
Effects of client expectation for counseling Leong, F. T. L., & Lee, S. H. (2006). A cultural
success, client-counselor worldview match, and accommodation model for cross-cultural psycho-
client adherence to Asian and European American therapy: Illustrated with the case of Asian
cultural values on counseling process with Asian Americans. Psychotherapy: Theory, Research,
Americans. Journal of Counseling Psychology, Practice, Training, 43, 410–423.
52(1), 67–76. Luborsky, L. B., Barrett, M. S., Antonuccio, D. O.,
∗
Kim, B. S. K., Omizo, M. M., & D’Andrea, M. J. Shoenberger, D., & Stricker, G. (2006). What
(1998) The effects of culturally consonant group else materially influences what is represented and
counseling on the self-esteem and internal locus published as evidence? In R. F. Levant (Ed.),
of control orientation among Native American Evidence-based practices in mental health: Debate
adolescents. Journal of Specialists in Group Work, and dialogue on the fundamental questions
23(2), 145–16. (pp. 257–98). Washington, DC: American
∗
King, B. S. (1999). The effect of a cultural-based life Psychological Association.
skills curriculum on American Indian adolescent self- Mak, W., Law, R., Alvidrez, J., & Pérez-Stable, E.
esteem and locus of control. Unpublished doctoral (2007). Gender and ethnic diversity in NIMH-
dissertation, University of Arkansas, Fayetteville. funded clinical trials: Review of a decade of
∗
Kohn, L. P., Oden, T., Munoz, R. F., Robinson, A., published research. Administration and Policy in
& Leavitt, D. (2002). Adapted cognitive Mental Health and Mental Health Services
behavioral group therapy for depressed low- Research, 34(6), 497–503.
income African American women. Community ∗
Malgady, R. G., Rogler, L. H., & Costantino, G.
Mental Health Journal, 38, 497–504. (1990a). Hero/heroine modeling for Puerto
Rican adolescents: A preventive mental health ∗
Mokuau, N., Braun, K. L., Wong, L. K.,
intervention. Journal of Counseling Psychology, Higuchi, P., & Gotay, C. C. (2008). Development
58, 469–74. of a family intervention for native Hawaiian
∗
Malgady, R. G., Rogler, L. H., & Costantino, G. women with cancer: A pilot study. National
(1990b). Culturally sensitive psychotherapy For Association of Social Workers. 53(1), 9–19.
Puerto Rican children and adolescents: A pro- ∗
Moran, J. R. (1999). Preventing alcohol use among
gram of treatment outcome research. Journal of urban American Indian youth: The seventh
Consulting and Clinical Psychology, 58(6), 704–712. generation program. Journal of Human Behavior
Marcus, E. (2010). PTSD manifests differently in in the Social Environment, 2, 51–67.
Haitian patients, says researcher. Huffington Muñoz, R. E (1982). The Spanish speaking consumer
Post. Retrieved online May 31, 2010 from http:// and the community mental health center. In
www.huffingtonpost.com/erin-marcus/ptsd- E. E. Jones & S. Korchin (Eds.), Minority mental
manifests-differentl_b_580825.html health (pp. 362–98). New York: Praeger.
∗ ∗
Martinez C. R., Jr., & Eddy, J. M. (2005). Effects Myers, H. F., Alvy, K. T., Arrington, A.,
of culturally adapted parent management Richardson, M. A., Marigna, M., Huff, R., et al.
training on Latino youth behavioral health (1992). The impact of a parent training program
outcomes. Journal of Consulting and Clinical on inner-city African-American families. Journal
Psychology, 73(4), 841–51. of Community Psychology, 20, 132–47.
∗ ∗
Matos, M., Bauermeister, J. J., & Bernal, G. Nagel, T., Robinson, G., Condon, J., & Trauer, T.
(2009). Parent-Child Interaction Therapy for (2009). Approach to treatment of mental illness
Puerto Rican preschool children with ADHD and substance dependence in remote indigenous
and behavior problems: A pilot efficacy study. communities: Results of a mixed methods study.
Family Process, 48, 232–52. The Australian Journal of Rural Health, 17(4),
∗
Mausbach, B. T., Bucardo, J., McKibbin, C. L., 174–82.
Cardenas, V., & Barrio, C. (2008). Evaluation of a National Science Foundation (2009). Doctorate
culturally tailored skills intervention for Latinos recipients from US universities: Summary report
with persistent psychotic disorders. American 2007–2008. Retrieved online May 17, 2010
Journal of Psychiatric Rehabilitation, 11(1), 61–75. from http://www.nsf.gov/statistics/nsf10309/pdf/
∗
Mickens-English, P. (1996). The efficacy of an nsf10309.pdf
Afrocentric/holistic group psychotherapy approach ∗
Nyamathi, A. M., Leake, B., Flaskerud, J.,
for black women. Unpublished doctoral disserta- Lewis, C., & Bennett, C. (1993). Outcomes of
tion, Kent State University, Ohio. specialized and traditional AIDS counseling
Miranda, J. (2006). Improving services and outreach programs for impoverished women of color.
for women with depression. In C. M. Mazure & Research in Nursing & Health, 16, 11–21.
G. P. Keita (Eds). Understanding depression in ∗
Ochoa, M. L., (1981). Group counseling Chicana
women: Applying empirical research to practice and troubled youth: An exploratory group counseling
policy. (pp. 113–35). Washington, DC: American project. Unpublished doctoral dissertation,
Psychological Association. University of Massachusetts, Amherst.
Miranda, J., Azocar, F., Organista, K. C., Muñoz, R. F., Ortman, J. M., & Guarneri, C. E. (2009). United
& Lieberman, A. (1996). Recruiting and retain- States population projections: 2000 to 2050.
ing low-income Latinos in psychotherapy research. U.S. Census Bureau. Retrieved online June 1,
Journal of Consulting and Clinical Psychology, 2010 from http://www.census.gov/population/
64(5), 868–74. www/projections/analytical-document09.pdf
Miranda, J., Bernal, G., Lau, A., Kohn, L., ∗
Pantin, H., Coatsworth, J. D., Feaster, D. J.,
Hwang, W., & La Framboise, T. (2005). State of Newman, F. L., Briones, E., Prado, G., et al.
the science on psychosocial interventions for (2003). Familias Unidas: The efficacy of an
ethnic minorities. Annual Review of Clinical intervention to promote parental investment in
Psychology, 1, 113–42. Hispanic immigrant families. Prevention Science,
Miranda, J., Nakamura, R., & Bernal, G. (2003). 4(3), 189–201.
Including ethnic minorities in mental health ∗
Parker, L. (1990). The role of cultural traditions
intervention research: A practical approach to a in alcohol and drug abuse prevention: A Native
long-standing problem. Culture, Medicine and American study. Unpublished doctoral disserta-
Psychiatry, 27(4), 467–86. tion, Brown University, Providence, RI.
Pedersen, P. B. (1999). Multiculturalism as a fourth ∗
Shin, S. K., & Lukens, E. P. (2002). Effects of
force. Philadelphia: Brunner/Mazel. psychoeducation for Korean Americans with
Pedersen, P. B. (2000). A handbook for developing chronic mental illness. Psychiatric Services, 53(9),
multicultural awareness (3rd ed.). Alexandria, 1125–31.
VA: American Counseling Association. ∗
Shin, S. K. (2004). Effects of culturally relevant
Rey, G., & Sainz, M. (2007). Tailoring an interven- psychoeducation for Korean American families
tion model to help indigenous families cope of persons with chronic mental illness. Research
with excessive drinking in central Mexico. Salud on Social Work Practice, 14(4), 231–39.
Mental, 30(6), 32–42. Smith, T. B. (2010). Culturally congruent practices
Ridley, C. R. (1984). Clinical treatment of the in counseling and psychotherapy: A review of
nondisclosing Black client: A therapeutic research. In J. G. Ponterotto, J. M. Casas, L. A.
paradox. American Psychologist, 39, 1234–44. Suzuki, & C. M. Alexander (Eds.), Handbook of
∗
Rosselló, J. & Bernal, G. (1999). The efficacy of multicultural counseling. (3rd ed., pp. 439–50).
cognitive behavioral and interpersonal treatments Thousand Oaks, CA: Sage.
for depressed Puerto Rican adolescents. Journal of Smith, T. B., Richards, P. S., Granley, M., &
Consulting and Clinical Psychology, 67, 734–45. Obiakor, F. (2004). Practicing multiculturalism:
∗
Rosselló, J., Bernal, G., & Rivera-Medina, C. An introduction. In T. B. Smith (Ed.), Practicing
(2008). Individual and group CBT and IPT multiculturalism: Affirming diversity in counseling
Puerto Rican adolescent with depressive symp- and psychology (pp. 3–16). Boston: Allyn &
toms. Cultural Diversity and Ethnic Minority Bacon.
Psychology, 14, 234–45. ∗
Sobol, D. A. (2000). An adolescent-parent conflict
∗
Rowland, M. D., Halliday-Boykins, C. A., resolution training program for ethnically diverse
Henggeler, S. W., Cunningham, P. B., Lee, T. G., families. Unpublished doctoral dissertation,
Kruesi, M. J. P., et al. (2005). A randomized trial University of Southern California, Los Angeles.
of multisystemic therapy with Hawaii’s Felix Sue, S. (1988). Psychotherapeutic services for ethnic
Class youths. Journal of Emotional and Behavioral minorities: Two decades of research findings.
Disorders, 13(1), 13–23. American Psychologist, 43(4), 301–308.
∗
Royce, D. (1998) Mentoring high-risk minority Sue, D. W., & Sue, D. (2008). Counseling the
youth: Evaluation of the Brothers project. culturally diverse: Theory and practice (5th ed.).
Adolescence, 33(129), 145–58. New York: John Wiley.
∗ ∗
Santisteban, D. A., Coatsworth, J. D., Szapocznik, J., Rio, A., Perez-Vidal, A.,
Perez-Vidal, A., Kurtines, W. M., Schwartz, S. J., Kurtines, W., Hervis, O., & Santisteban, D.
LaPerriere, A., et al. (2003). Efficacy of Brief (1986). Bicultural Effectiveness Training (BET):
Strategic Family Therapy in modifying Hispanic An experimental test of an intervention modality
adolescent behavior problems and substance use. for families experiencing intergenerational/inter-
Journal of Family Psychology, 17(1), 121–33. cultural conflict. Hispanic Journal of Behavioral
Schwartz, A. L., Domenech Rodríguez, M. M., Sciences, 8(4), 303–30.
Santiago-Rivera, A., Arredondo, P., & Field, L. D. ∗Szapocznik, J., Santisteban, D., Rio, A.,
(2010). Cultural and linguistic competence: Perez-Vidal, A., Santisteban, D., &
Welcome challenges from successful diversifica- Kurtines, W. M. (1989). Family Effectiveness
tion. Professional Psychology: Research and Practice, Training: An intervention to prevent drug
41, 210–20. abuse and problem behaviors in Hispanic
∗
Schwarz, D. A. (1989). The effect of a Spanish adolescents. Hispanic Journal of Behavioral
pre-therapy orientation videotape on Puerto Rican Sciences, 11(1), 4–27.
clients’ knowledge about psychotherapy, improvement Tanaka-Matsumi, J. (2008). Functional approaches
in therapy, attendance patterns and satisfaction to evidence-based practice in multicultural coun-
with services. Unpublished doctoral dissertation, seling and therapy. In P. A. Hays & G. Y. Iwamasa
Temple University, Philadelphia. (Eds.), Principles of multicultural counseling and
Shin, H. B., & Kominski, R. A. (2010). Language therapy. (pp. 169–98). New York: Routledge/
use in the United States: 2007. U.S. Census Taylor & Francis Group.
Bureau. Retrieved on line June 1, 2010 from ∗
Telles, C., Karno, M., Mintz, J., Paz, G.,
http://www.census.gov/population/www/socdemo/ Arias, M., Tucker, D., et al. (1995). Immigrant
language/ACS-12.pdf families coping with schizophrenia: Behavioural
family intervention v. case management with a Whitbeck, L. B. (2006). Some guiding assumptions
low-income Spanish-speaking population. British and a theoretical model for developing culturally
Journal of Psychiatry, 167(4), 473–79. specific preventions with Native American
∗Timberlake, T. L. (2000). A comprehensive people. Journal of Community Psychology, 34(2),
approach to social skills training with urban African 183–92.
∗
American adolescents. Unpublished doctoral Xiong, W., Phillips, M. R., Hu, X., Wang, R.,
dissertation, Temple University, Philadelphia. Dai, Q., Kleinman, J., et al. (1994). Family-
U.S. Surgeon General (2001). Mental health: based intervention for schizophrenic patients in
Culture, race, and ethnicity. A supplement to China: A randomized controlled trial. British
mental health: A report of the surgeon general. Journal of Psychiatry, 165(2), 239–47.
Rockville, MD: U.S. Department of Health and Zane, N., Hall, G. N., Sue, S., Young K., &
Human Services. Nuñez (2004). Research on psychotherapy with
Wampold, B. E., Mondin, G. W., Moody, M., culturally diverse populations. In M. Lambert
Stich, F., Benson, K., & Ahn, H. (1997). A meta- (Ed.), Bergin and Garfield’s Handbook of
analysis of outcome studies comparing bona fide Psychotherapy and Behavior Change, 5th Edition
psychotherapies: Empirically, “all must have (767–804). New York: Wiley.
∗
prizes.” Psychological Bulletin, 122(3), 203–215. Zhang, Y., Young, D., Lee, S., Li, L., Zhang, H.,
Whaley, A. L., & Davis, K. E. (2007). Cultural Xiao, Z., et al. (2002). Chinese Taoist cognitive
competence and evidence-based practice in mental psychotherapy in the treatment of generalized
health services: A complementary perspective. anxiety disorder in contemporary China.
American Psychologist, 62(6), 563–74. Transcultural Psychology, 39(1), 115–29.
C HA P TER
17 Coping Style
Larry E. Beutler, T. Mark Harwood, Satoko Kimpara, David Verdirame, and Kathy Blau
It is important that children, early on, and fear. They viewed both the occurrence
acquire the ability both to engage in self- and anticipation of external events as intru-
reflection and to appraise the behavior of sions, as threats that upset their internal
others. As children begin to look both inter- experiences and produced avoidance and
nally and externally, they learn to integrate seclusion. In later life, these children were
and compare the information obtained from observed frequently to develop substantial
each without becoming overwhelmed with amounts of anxiety and to become over-
either. The integration between internal whelmed by their fears and avoidant in their
sensitivity and external judgment, between behaviors. They often became socially with-
the subjective and the objective, requires drawn, self-critical, phobic, and intolerant
that humans maintain a complex but mod- of emotional experience or environmental
ulated response to both sources of informa- change. They turned to internal experience,
tion and rely on a flexible system of values fantasy, and obsessive reconstruction of
by which to appraise both the impact of events to achieve stability.
others on self and of self on others. By contrast, Kagan observed that other
A perfect balance is unlikely and, not infants were less reactive to these events
infrequently, an individual will develop a and, instead, preferred attending to external
preference for, or sensitivity to, either internal happenings while ignoring internal experi-
experiences or external events. This preferences. Those with this temperament of low
ence results in governing temperaments of reactivity or low sensitivity were relatively
infants and, later in life, distinctive coping more tolerant of novelty and change; they
styles. were observed to seek, rather than avoid,
Kagan (1998) observed that some infants stimulation from their environment, to
were, by nature, behaviorally highly reactive— take action to engage and change their envi-
very responsive—to internal events, resulting ronments, and to be gregarious and outgo-
in a degree of emotionality that contributed ing in their relationships with others. When
to behavioral instability. He concluded that they did develop problems, the problems
hyperreactive children were easily over- frequently expressed themselves as intrusive
whelmed and distressed by sudden or novel behaviors, insensitivity to other’s feelings
stimuli in their environments. Their responses and needs, lack of empathy, and with overt
were characterized by high arousal, distress, signs of anger and rage.
336
Patterns like those observed by Kagan In their search for factors that mediate
occur within all age groups. Introversion– the effects of psychotherapy, researchers
extroversion (Eysenck, 1960), internal- have been drawn to examine how patient
ization–externalization (Welsh, 1952), and attributes may determine one’s response to
a bimodal array of similarly descriptive different therapeutic interventions. One of
terms have characterized these distinctions these specific patient attributes is coping
among the experiences that people prefer style, a patient dimension that is both
and the way they adapt to change. These reminiscent of the temperament described
and related terms constitute psychometri- by Kagan (1998) and matched with the
cally valid and clinically useful descriptors degree to which effective change is moder-
identifying a continuum of ways that ated by insight. Early research discovered a
people respond to novelty and change. At relation between patient coping style and
one end of this continuum are individuals the differential use of psychological treat-
who protect themselves from stimulation ments that either sought to change skills
by being self-critical, avoiding change, and and behaviors directly or that focused on
withdrawing in the face of anticipated the indirect processes of achieving insight
change or discord. These individuals are and internal awareness (Beutler & Clarkin,
sensitized and overreactive to change and 1990). Specifically, among patients whose
are prone to be overwhelmed by fear. They characteristic coping styles were identified
seek stability and safety in a focus on as internalizing (or hyperreactive) outcomes
internal experiences rather than on the were positively associated with the use of
instability and uncertainty of external insight- or awareness-oriented therapies.
events. At the other end of the continuum The latter interventions include those
are individuals who prefer to embrace focused on improved emotional awareness
novelty and change with activity and asser- or interpersonal sensitivity. Conversely,
tion (e.g., Beutler, Moos, & Lane, 2003; among patients whose characteristic coping
Beutler, Clarkin, & Bongar, 2000). They styles were identified as externalizing, posi-
seek contact with others, enjoy change, and tive outcomes were associated with thera-
are gregarious in their interactions with pies that rely largely on enhancing skill
their world. development and encouraging direct symp-
In virtually all cultures, individuals with tom change.
a highly reactive temperament are described The chapter (Beutler et al., 2002) in the
as internalizing, avoidant, restrained, or intro- earlier edition of this volume provided a
verted. Those with a low reactive tempera- box score on the association of patient
ment, in contrast, have been described as coping style with treatment type in predict-
externalizing, gregarious, and extroverted. ing psychotherapy outcome. Fifteen of 19
Across cultures, there are preferences for studies confirmed the expected pattern
one or another of these temperamental between the goodness of patient–treatment
styles; Western cultures tend to foster the fit and outcome; however, the inclusion
development of external, assertive, and criteria for those 19 studies were somewhat
individualistic styles of adjusting to change, lenient, and the box score did not yield an
while those living in Eastern cultures prefer index of the magnitude of the association.
more avoidant, self-inspection, and inter- In this chapter, we delve deeper into the
nalizing styles, even sharing attachments rationale for patient coping style as an indi-
across the communal group (Kawai, 1993, cator for differential psychotherapy and will
1996). subject the hypotheses to a meta-analysis.
b e u t l e r, ha rwo o d , k i m pa r a , ve rd i r a m e , b l au 337
This will allow an assessment both of the through increasing insight or personal
statistical significance of the findings and awareness. These distinctions will become
the strength of this matching dimension clear as we inspect some of the ways that
as a contributor to outcome. Our meta- these concepts have been defined and
analysis will also permit us to analyze studies measured in the past.
comprised mainly of patients with one type
of coping style (external or internal), and a Coping Style
separate estimate can be derived for each of Coping style has been described by different
these styles and the differences can be personality and psychopathology theorists
compared. via a collection of often unrelated-sounding
but conceptually similar terms. Two con-
Definitions and Measures ceptual aspects of coping have proven con-
In order to determine the effectiveness of troversial. First, some theorists define
matching a patient’s coping style to the coping style in terms of how one deals
focus of psychotherapeutic interventions, with environmental novelty and change
both the patient’s coping style and the under normal conditions (e.g., Lazarus &
nature of treatments ranging from insight/ Folkman, 1984; McKay et al., 1998),
awareness- to symptom-focused must be whereas others emphasize the adaptability
defined in operational terms. While numer- of one’s of coping efforts when faced with
ous instruments have been developed to stressful situations or unusual environ-
assess this distinction among treatments, mental changes (e.g., Eysenck, 1960; Latack
few address the level of observed therapist & Havlovic, 1992). Second, some empha-
behavior in these terms. One exception to size the role of traitlike aspects of coping
this rule is an instrument developed by (e.g., Endler, Parker, & Butcher, 1993), a
Beutler and colleagues to assess dimensions position that is in contrast to those that
both of patients and treatments. This concentrate on state or situational qualities
instrument conceptualizes both coping of coping (e.g., Ouimette, Ahrens, Moos,
style and therapy focus as existing along a & Finney, 1997).
continuum, with the nature of the effective We have incorporated these varying
interaction assumed to vary as a function of theoretical points within a broad, statistical
the intersect between each continuum for a definition (Beutler & Clarkin, 1990). We
given patient and therapist. In measuring define coping style as the pattern of behav-
coping style, for example, ratings of exter- ior that is predominantly employed when
nalization and internalization are ordered one faces a new or unusual situation. This
along a continuum based on the prepon- definition combines both state- and trait
derance of actions that occur under condi- aspects of one’s response and removes the
tions of environmental change (Beutler, requirement that coping styles only be
Moos, & Lane, 2003). Likewise, measures observed during and following stressful situ-
of treatment focus consider it to be best ations. Thereby, the definition effectively
described as a continuum that extends from eliminates the need to judge the level of
being insight/awareness-oriented to symp- stress experienced or the generalizability of
tom/skill-oriented. This latter designation the situation in which it has occurred.
is based both on a rating of the objectives of From this broader perspective, coping
the treatment and the degree to which the styles are recurrent patterns of behavior
efforts to induce change are aimed directly that characterize the individual when con-
at symptomatic behaviors or indirectly fronting new or problematic situations.
This style identifies one’s vulnerability to Multiphasic Personality Inventory (MMPI-2;
change and one’s predominant tendency to Butcher, 1990; Butcher & Beutler, 2003),
respond to novelty. Thus, coping styles are supplemented by reviewing the patient’s
not discrete behaviors but are a cluster of past and present reactions to problems. The
related behaviors that are distinguished internalization ratio (IR) formula, extracted
because they are repetitive, durable across from the MMPI-2, has been used frequently
similar events, and observable when by our own research group to capture the
problems or unexpected events are being interactive nature of coping style and
addressed. Descriptively, the specific behav- treatment focus (e.g., Beutler, Engle, et al.,
iors that form the clusters include both 1991; Beutler, Moliero, et al., 2003). In
repetitive situational responses such as our modification of a formula originally
impulsivity, discrete acting-out behaviors, proposed by Welsh (1952), eight MMPI-2
escape and direct avoidance, and general subscale scores are entered as a standard T
temperaments. Unlike more narrow defini- scores:
tions of coping style, definitions like ours Hy + Pd + Pa + Ma
are based upon correlated clusters of behav- IR =
iors and are not explanatory concepts. Hs + D Pt Si
Given the diversity of measurements used An IR that favors the numerator suggests
to study coping styles, we will adopt this that a patient is disposed to use externaliz-
broad definition in our meta-analysis. ing coping behaviors. These individuals
Following the description of Eysenck blame others for their feelings (Pa); they
(1957) and Kagan (1998), the quality that display active, dependent behaviors (Hy),
distinguishes internalizing traits and dispo- high levels of unfocused energy (Ma), are
sitions from other coping styles is that they impulsive, and frequently have social
are governed by the forces of inhibition and adjustment problems (Pd).
excitation. Internalizers/introverts are more These individualized, patient-level meth-
easily overwhelmed by change and tend to ods of measurement serve the broad defini-
become shy, withdrawn, and self-inspective, tion used in this literature somewhat better
while externalizers/extroverts are more than more indirect measures that cluster
likely to act out, to seek stimulation and groups of individuals under a categorical
change, to directly escape or withdraw from classification based on either coping (e.g.,
conflict, and to be confrontational and gre- Lazarus & Folkman, 1984) or diagnosis.
garious in expressing problems. Animal Using diagnosis as a proxy for coping style,
behaviorists have extended these qualities for example, treats all patients within a
to proactive versus reactive behaviors diagnostic group as if they were identical on
(Koolhaas et al., 1999), and others have this dimension, occluding the many varia-
incorporated similar concepts into the Big tions that exist within diagnostic groups;
Five personality factors (Costa & McCrae, however, when using archival data, direct,
1985). individual-level measures of coping style
For research purposes in psychotherapy, are often not available. In such instances, a
patient coping styles are typically measured categorical definition of the patient’s domi-
objectively through individualized observa- nant coping style must be inferred through
tions and ratings (e.g., Beutler, Clarkin, & indirect observation of shared behaviors,
Bongar, 2000) or through standardized, using what information is available.
self-report, omnibus personality and psycho- Diagnostic problems that are characterized
pathology measures such as the Minnesota by intense distress, ruminations, and social
withdrawal are usually indirectly identified instruction in the use of problem-solving
as related to internalizing patterns of coping. strategies, or efforts to enhance patient
Thus, Axis I diagnoses within the spectrum self-monitoring in order to identify the
of anxiety disorders as well as Axis II predominant procedures used to evoke
avoidant personality disorders can usually changes in symptoms and overt problems
be assumed to be internalizing conditions, as well as to stimulate the resolution of
while antisocial personality, substance inferred problems or causes. Where possible,
abuse, and paranoid personality disorder the use of direct measures is advantageous
can be seen as more highly dominated in either case. The measures are reliable,
by externalizing patterns. Unfortunately, easily tested for interrater validity, and can
because they are diverse and do not reliably be used to rate a wide array of discrete
map onto individual coping style descrip- techniques that share a common set of
tions, reliance on diagnosis or other cate- objectives.
gorical definitions of personalities as a proxy Unfortunately, as with coping style, there
for a patient’s dominant coping style must are many instances when direct observa-
be undertaken with considerable caution. tions of therapy interactions are not
This consideration led us to arbitrarily limit possible. When using archival data or when
the proportion of studies in our meta-anal- working from published reports, the focus
ysis that used such proxy measures to no of the treatment often must be inferred
more the 25% of the studies included in from the theoretical rationale underlying
our analysis. the therapy used. Usually, it is most reliable
simply to categorize the treatment model
Treatment Focus in terms of purity as a prototypic insight/
The therapist’s treatment methods are awareness-oriented procedure or a symptom/
also measured in two ways—through direct skill-focused procedure. In this bifurcation,
observations of each individual therapist’s interpersonal, experiential, and psycho-
behaviors or by indirect measures based on dynamic therapies are usually classed as
the system of psychotherapy used. There is insight-focused procedures, and cogni-
little doubt that the most sensitive measure tive, cognitive-behavioral, and behavioral
of treatment focus is to observe and calibrate models are identified as symptom-focused
in-therapy actions and intentions of the interventions. However, it is an oversim-
therapist. Using individual, direct measur- plification to think of the distinction
ing methods, rating the use of various tech- between direct, symptom change–focused
niques such as interpretation, transference and indirect change–focused interventions
analysis, dream analysis, interpersonal anal- as discrete and finite. More accurately,
ysis, and the like can identify procedures psychological treatments are ordered along
that are most frequently associated with a continuum that ranges from the degree to
the effort to evoke insight and awareness which they address mediating variables
of previously cathected, unconscious, and to the degree to which they focus on the
symbolized material (e.g., Beutler, Moleiro, symptoms themselves (Beutler, Moleiro,
et al., 2003). et al., 2003). For example, in the purest
Direct observation such as the foregoing form of symptom-focused interventions,
can yield numerical data on the frequency behavior therapy directly addresses changes
of any treatment methods. One can count in symptoms and skills while eschewing
the use of symptom reports, techniques the presence of “underlying” problems.
based on reinforcement paradigms, therapist These therapies take each symptom that is
disruptive to the patient’s adjustment or measures, he/she may observe the patient’s
happiness at face value, working sequen- response to life crises by withdrawal and self-
tially to eradicate it. At the other end of the blame (internalizing) or by becoming angry,
treatment focus dimension, psychodynamic blaming, and avoidant (externalizing).
procedures make an indirect or mediated L.C. was a 42-year-old, married man
change on expressed problems and symp- who was referred for psychotherapy by his
toms. These methods take little note of the physician, who he also described as his best
symptom, itself, seeing it as merely a sym- friend. The patient’s presenting problems
bolized expression of some unseen and more were many, including substance abuse,
important underlying conflict. That which depression, impaired work performance,
is not directly seen, but which can be inferred and deficits in interpersonal functioning.
from the theoretical model used, is then The patient recalled being “very depressed”
assumed as the point of focus for the change since the age of 12 and described a family
effort. Treatments that emphasize uncon- history of abuse, alcohol dependence, and
scious processes are examples of these finally, abandonment. He was on his own at
indirect interventions. age 16, and what had begun 3 years before
The difficulty with this categorical as recreational marijuana use rapidly devel-
classification can be seen when the thera- oped into extensive cocaine, metham-
pies studied are those that attempt to phetamine, and heroin abuse. He held
achieve symptom change through indirect several jobs between the ages of 16 and 40,
means. Interpersonal psychotherapy (IPT), losing most because of behaviors related
for example, is explicitly not focused on to chemical dependency. At age 29 he got
improving historical insight in the same married and was divorced by age 35. At
way that psychodynamic therapy is. Instead, age 40 he began his own Internet business
it deals with improving one’s awareness of in an effort to escape the rigid rules that had
interpersonal and emotional forces that frequently led to his termination from other
affect one’s behavior. Thus, it is focused on jobs. His progress had been uneven and
the enhancement of social and emotional slow; he maintained a marginal existence
awareness but not on insight, per se, and its on the income that he could produce.
status lies somewhere between the direct Direct and indirect measures of L.C.’s
symptom focus of cognitive therapy and coping style indicated a mixed but predom-
the indirect focus of psychodynamic ther- inantly internalizing style. Indirect measure,
apy. While not as sensitive as therapist- based on diagnosis, reveals a mix of inter-
level measures, a classification of treatment nalizing depression and some substance
based on relative position along the con- acting out. A direct measure (the MMPI-2,
tinuum from insight/awareness to symptom/ IR) of L.C.’s coping pattern revealed that
skill focus can be assessed for reliability. he had a mixture of both internalizing and
Descriptions of efforts to ensure treatment externalizing coping patterns, with an over-
fidelity can be used in research practice to all balance favoring the use of internalizing
provide some cross-validation of one’s strategies. Hypochondriasis, Depression,
classifications. anxiety (Psychasthenia), and Social Intro-
version scales averaged 7 points higher than
Clinical Examples the corresponding externalizing scales. His
There are many examples of how patient internalizing style of coping was further
coping styles manifest in psychotherapy. Even illustrated and observed in how he concep-
if the therapist does not have self-report tualized the cause and the consequences
of his drug use. He expressed the belief that likely to be much more consistently inter-
his drug abuse began because he was weak nalizing, with a lot of attention given to self-
and defective—an introtensive injunction. evaluation and criticism. This means that
He indicated that his problem had contin- one would probably move quite quickly to a
ued because he was not strong enough theme- or insight-focused intervention.
to follow his conscience—a self-critical There are equivalent examples of the
injunction. While not a religious man, he differential treatment of individuals who
expressed strong guilt for having “enticed” prefer externalizing coping styles. Patterns
his wife into a marriage in which he was of consistent acting out and conflict with
unable to take care of her. authorities are examples of individuals who
In contrast, R.W. was a 43-year-old cope in externalized ways. The identifica-
woman with a history of social avoidance tion of a preference for these externalized
and shyness. In her 20s and 30s, the prob- patterns may be inferred from diagnoses
lems had become so bad that she had to like antisocial personality disorder, border-
quit her job as a secondary school teacher line personality disorder, substance abuse or
because she could not face her class. At that dependency, and varieties of impulse disor-
time she was diagnosed with social phobia ders. While these categorical, indirect mea-
and with avoidant personality disorder. The sures of coping style are useful, they lack
MMPI-2 provided a direct confirmation the sensitivity that a continuous measure
for an internalizing coping style. The might provide.
Internalization ratio showed dramatic eleva- R.G., for example, was a 21-year-old
tions on Psychasthenia (Scale 7) and Social woman referred for psychotherapy from
Introversion (Scale 10), with a secondary her psychiatrist because of a long-standing
elevation on Depression (Scale 2), relative pattern of explosive outbursts. In recent
to the externalizing scales. years, she had begun abusing alcohol and
L.C. and R.W. would differ with respect had been arrested for driving under the
to the symptoms that would be of primary influence on two occasions. She had expe-
focus during the early phase of treatment rienced problems in school because of her
and in the theme that guided the insight- failure to control her temper and had been
oriented work. For L.C. the initial symp- a chronic problem to her parents because of
tom focus would probably be on behaviors similar behavior. She had been in and out
that indicated risk for drug abuse and of treatment since age 8, but with little
self-harm, with a secondary focus on social help. Except for her first experience with
functioning. In contrast, the initial focus behavior therapy, her treatment had focused
for R.W. would probably be on social with- on allowing expression of her feelings,
drawal and depression with secondary trying to uncover the source of her rage,
attention given to any issues of self-harm and developing self-awareness and insight.
that emerge. R.G.’s direct measure of coping style, using
R.W.’s theme is likely to be quite different the Systematic Treatment Selection-Clinician
than L.C.’s. R.W. may represent the hyper- Rating Form (STS-CRF, Fisher, Beutler, &
reactive temperament described by Kagan Williams, 1999) and the MMPI-2, confirmed
(1998) and, thus, be an early developmental the dominance of impulsivity and confronta-
phenomenon; therefore, hypervigilance, tional coping behaviors over rational self-
chronic fear, and a dread of appraisal from control. She evidenced poor insight, high
others would probably dominate the theme. levels of poorly directed energy, and a strong
Compared with L.C., the coping style is sense of persecution. Accordingly, treatment
focused, not on self-expression and “unload- meta-analytic review of matching treatment
ing,” but on control and tolerance for the directiveness to patient resistance/reactance.
discomfort associated with anger and envi- This procedure began with identifying a set
ronmental stimulation. of six criteria that would characterize an
Psychotherapy began by identifying ideal study:
specific situations in which problematic
1. A wide breadth of reliably applied
behaviors and symptoms occurred. R.G.
therapeutic approaches and trained
was taught to self-monitor her arousal and
therapists to ensure variance on the
to identify risk-provoking situations. She
dimension of treatment focus
then was engaged in learning stress toler-
2. A similarly wide range of moderately
ance, where negative emotions were selec-
impaired patients in order to ensure
tively evoked by visual imagery and role
variability on the dimension of coping
playing. Instruction in prosocial behavior
style
accompanied all of these interventions. For
3. Individualized (i.e., direct) measures
example, behavioral rehearsal was used to
of both patient coping style and treatment
engage her in communication training and
focus in order to avoid equating focus
to develop useful skills in impulse control,
with a particular brand of treatment and
self-appraisal, and tolerance for novelty and
coping style with a particular diagnosis
change.
4. Random assignment of patients to
therapists within treatments in order to
Meta-Analytic Review
ensure equivalent dispersal of patients
Literature Search and
to insight/awareness- and symptom/
Inclusion Criteria
skill-focused interventions
In undertaking our meta-analysis, our focus
5. Systematic monitoring both of
was to identify studies in which interaction
treatment variability/consistency on the
of coping style and treatment focus could
dimension of treatment focus and of
be assessed and effect sizes could be calcu-
patient coping style
lated. That is, they addressed the moderat-
6. Objective and uniform outcome
ing role that patient coping style exerted on
measurement that included analysis of fit
the effectiveness of a particular treatment
between patient coping style and
focus (direct behavior change or indirect
treatment focus.
insight change). In addition, like the corre-
sponding chapter in the earlier volume of The second step was then to identify a
this book, we also wanted to assess the model study, an investigation whose meth-
independent effects of patient coping style, ods represented these criteria most closely.
if any, on outcome. That is, we wanted to This model study then served as a template
know the main effects of patient coping for evaluating other studies that we identi-
style. This latter, or main effect analysis, fied. The study (Beutler, Moleiro, et al.,
addresses a prognostic question while mod- 2003) had all of the requisite methodo-
erating studies address a treatment planning logical features for inclusion in our review
question: What treatment is best for what except that it utilized a composite algorithm
patient? to fit treatment to patient. The composite
In identifying relevant literature, we included the fit of coping style to treatment
followed the general outline that was used focus but incorporated two other matching
by Beutler, Harwood, Michelson, Song, and factors as well. Because it did not permit a
Holman (this volume, Chapter 13) in their pure test of the coping style by treatment
focus “fit,” it was not included in the investigator to conduct necessary analyses
current review as a primary study. or report statistics. In other cases, the prob-
The next step in our procedure was to lem was that their statistical procedures
identify studies that most closely complied provided data that were appropriate, but
with the criteria established for inclusion we were unable to reliably calculate effect
in the meta-analysis. Our scope included sizes. Some examples of excluded studies
studies published within a 19-year span may make the decision process more clear.
from 1990 to 2009. We began by collecting One excluded study (Beutler & Mitchell,
studies that had addressed patient coping 1981) reported the treatment outcomes of
style explicitly as a mediator between treat- 40 patients. Patient coping style (internalizer
ment focus and outcome in the prior or externalizer) was assessed using the MMPI.
edition of this volume. This resulted in a The results revealed systematic patient apti-
list of 19 studies, and we added to this list tude (coping style) × treatment interaction
by searching the PsychINFO database effects independent of diagnoses. External-
using search words associated with coping izing patients were found to achieve greater
style, personality, introversion, extraver- benefit from experiential treatment than
sion, etc. The final step was to hand-search from analytic-based therapy; however, among
the 2009 volumes of the most widely cited internalizing patients, insight-oriented (ana-
journals that emerged from our search. lytic) treatment achieved its greatest effects
These included the Journal of Consulting and, correspondingly, the behavioral thera-
and Clinical Psychology, the Journal of pies had the least beneficial impact. Unfor-
Counseling Psychology, and the Journal of tunately, these results were based on a box
Clinical Psychology. score tabulation of studies that were indica-
We excluded studies that had major tive of a relation between therapy–patient fit
methodological weaknesses and those and outcome. The lack of more precise statis-
whose results did not allow the calculation tics rendered this study inappropriate for
of an effect size (ES). Methodological weak- inclusion in the meta-analysis.
nesses included the failure to use direct A study by Barber and Muenz (1996) was
measures of coping style, absence of blind included as part of the meta-analysis in an
or masked outcome measures, indefinite early draft of this chapter but excluded in the
forms of treatment in which the focus could final sample. The exclusion occurred because
not be defined with relative certainty, and the two senior authors could not uniformly
inaccurate interpretations or calculations. identify a difference in coping styles as being
Applying both indirect measures of characteristic of the two subsamples studied.
coping style and direct ones, we initially Obsessive and “avoidant” patients were both
identified 26 studies that had addressed the judged to be representative of predominantly
roles of coping style (or a reasonable proxy internalizing coping styles.
measure) either as a main effect or as a Three widely regarded and large-sample
mediator of treatment outcome. From the studies also were excluded from our final
pool of 26 studies, 12 met at least four of analysis and deserve special attention here
the six criteria and permitted an analysis of because of this. For example, Beutler, Clarkin,
ESs. The main reason studies were pared and Bongar (2000) compared several treat-
from the initial set of 26 studies was that ment fit dimensions in a large-scale study
they did not report data from which effect of 284 patients, nine treatments, and over
sizes could be computed. In many cases 30 therapists. This study included specific
this was simply a failure on the part of the and direct measures of both internal and
external coping styles from the MMPI, as We used the calculation procedure and
well as observational and direct measures of formula that best fit the characteristics of
treatment focus. Unfortunately, the study the data presented in an individual study.
also applied a complex Structural Equation We used Cohen’s d in all cases, but ESs
Modeling (SEM) analysis to assess the find- were often presented as correlation coeffi-
ings, and we could find no reliable method cients or even regression coefficients. In
to extract an effect size estimate for the inter- these cases, we transferred all estimates of
action effects of the specific matching dimen- ES to a d coefficient. Several sources were
sions associated with coping style. consulted in making this transformation.
The second large-sample study, Project When there was a difference between for-
MATCH (Project Match Research Group, mulas, and no single one was consensually
1997), is the largest randomized clinical accepted as the one of choice, which was
trial (RCT) of matching variables conducted often the case, we used the formula that
to date. In this study, 952 outpatients and was most consistent with other sources
774 inpatients diagnosed as alcohol depen- and that provided what appeared to be the
dent were assigned to one of three 12-week, most unbiased estimate of ES. We fre-
manual-guided treatments (cognitive- quently calculated and recalculated formu-
behavioral coping skills therapy = CBT, las two or three times to ensure accuracy. If
motivational enhancement therapy = MET, means, sample sizes, and SDs were avail-
or 12-step facilitation therapy = TSF). All able, we always employed the same for-
three of the treatments were symptom mula across studies; however, when data
focused. None of the treatments could were incomplete or reports did not con-
reliably be judged by our raters as insight or tain some important information, we relied
awareness oriented. This precluded a reli- on accepted alternative procedures (e.g.,
able test of the fit between treatment focus Borenstein, Hedges, Higgins, & Rothstein,
and patient coping style. 2009; Lipsey & Wilson, 2001; Hunter &
The third large-sample study that was Schmidt, 2004). We also calculated an
excluded was the United Kingdom Alcohol overall mean ES estimate across studies,
Treatment Trial (UKATT Research Team, weighting the individual study ds with the
2007). Over 420 alcoholic patients were number of patients. Our source for the cal-
treated with one of two structured interven- culation of 95% confidence intervals was
tions. The treatments consisted of either Smithson (2003).
motivational enhancement therapy (MET)
or social behavior and network Therapy. Results: Main Effects of Coping Style
Once again, our raters could not distinguish and Treatment Focus
between the two treatments. Both were The 12 studies and their results are presented
identified as being symptom and skill in Table 17.1. Only four of these studies
focused. This and the absence of specific provided information from which we
outcome and follow-up data precluded a test could extract an ES estimate on the predic-
of the treatment-fitting hypothesis regarding tive value of coping style. Only one of
coping style and treatment focus. these was on an internalizing group of
patients (Knekt et al., 2008), and three
Coding Studies and Calculating ESs (Beutler, Moliero, et al., 2003; Karno et al,
Effect sizes (ESs) associated with the fit of 2002; Longabaugh et al, 1994) were on
treatment focus and patient coping style were externalizing groups. Thus, the number of
calculated as suggested by several sources. studies was insufficient to calculate a reliable
346
Table 17.1 Coping Style and Therapy Focus

Study N Design Measure TX Focus Sample N ESs/ M ES (focus) M ES M ES “fit” 95% CI
coping style Study (CS)
Beutler, Engle, et al. (1991) 63 RCT I (FEP/ins vs. SSD/sym) D (Int-Ext) 3 1.63 0.75 0.64–0.86
Litt, Babor, et al. (1992) 79 RCT I (CST/Sym vs. Interact/Ins) D (Ext) 2 0.63 0.52–0.74
Beutler, et al. (1993) 46 RCT I (CT/Sym FEP/Ins & SSD/Ins) D (Int-Ext) 4 1.16 1.64 1.34–1.94
Longabaugh, Rubin, et al. (1994) 140 RCT I (CBT/Sym vs. ECBT/Ins) I (Ext) 14 0.12 0.68 0.37 0.29–0.45
Calvert, Beutler, & Crago (1988) 108 MR/Q-E D (TOQ) (Sym-Ins) D (Int-Ext) 3 0.81 0.73-0.88
Kadden, et al., (1989) 96 Nat I (CBT/sym vs. IPT/Ins) D (Ext) 2 0.60 0.50-0.70
Karno et al. (2002) 47 RCT I (CT/Sym vs. FST/Ins) I (Ext) 4 0.02 0.30 0.50 0.36–0.64
Beutler, Moliero, et al. (2003) 40 RCT/MR I (CT/Sym vs. NT/ins) I/D (Ext) 4 1.01 0.99 0.71 0.57–0.85
Milrod, Leon, et al. (2007) 49 RCT I (PFP/Ins vs. ART/Sym) I (Int) 4 0.92 0.71 0.58–0.84
Knekt, Lindfors, et al. (2008) 326 RCT I (SFT/Sym vs. STD/Ins & LTD/Ins) D (Int) 1 0.94 0.94 0.17 0.13–0.21
Kimpara (in Beutler, 2009) 121 Nat D (SFT/Ins vs. Sym) D (Int) 1 1.17 0.76 0.68–0.84
Johannsen (in Beutler, 2009) 92 Q-E/MR D (TPRS/Ins vs. Sym) D (Int-Ext) 1 0.61 0.51–0.71
Total N 1,291
Summary weighted ESs 0.85∗ 0.55∗
95% CIs for summary weighted ESs 0.82–0.88 0.52–0.58
N = Participants in study
Design = RCT (randomized clinical trial), MR (correlational), Nat (naturalistic), Q-E (quasi-experimental)
Measure Tx focus = either direct (designated as D) or indirect (designated as I). Indirect measures are based on the model of treatment used and identified as either either symptom- (Sym) or insight-
(Ins) focused; direct measures are based on a individual measure of the use of insight or symptom change procedures.
Direct measures include: TOQ (Therapist Orientation Questionnaire), TPRS (Therapy Process Rating Scale).
Indirect measures of Tx Focus are based on the model of treatment studied:
CT = cognitive therapy; FEP = focused expressive therapy; SSD = supportive self-directed therapy;
CST = coping skills training; Interact = Interactive; CBT = cognitive behavioral therapy; ECBT = relationship enhanced CT;
IPT = interpersonal therapy; Interp = interpretive; Supp =supportive; FST = family systems; NT = narrative therapy;
PT = prescriptive therapy; PFP = panic-focused psychodynamic; ART = applied relaxation;
SFT = solution-focused therapy; STD = short-term dynamic therapy; LTD = long-term dynamic therapy
Sample/coping Style = coping style type. CS is measured either directly (designated as D) or indirectly (designated as I). Direct measures are an individual personality scale (unspecified here).
Indirect measures are based on the type or diagnosis of the patient group as either internalizing (Int) or externalizing (Ext) or both (Int-Ext)
N ESs = Number of effect sizes calculated for this study
M ES (focus) = The mean effect size attributable to the treatment focus variable—combining all treatments
M ES (CS) = The mean effect size attributable to the CS variable—combining all varieties
M ES “fit” = The mean difference between effect sizes for “good” and “poor” fit, estimated in MR/Nat studies from correlational data
All ESs are expressed as d.
∗
designates a weighted mean effect size across studies.

difference among the effect sizes repre- size (d ) was computed for each study, based
sented by the two coping styles. Thus, we on all dependent variables. The size of the
are unable to conclude whether there was a mean of means, then, indicated the role
substantial effect in favor of one or the of treatment fit. A good fit was taken as
other way of coping. being composed of either: (1) externalizing
Estimating the effect of the therapist’s patients and symptom-focused therapy or
treatment focus was an easier matter since (2) internalizing patients and insight-focused
all the treatments could be coded in the therapy. The overall mean of the estimated
same direction relative to their insight or ES reflecting level of “fit” was d = 0.55
symptomatic focus. The results of these (p < 0.05; CI = 0.52–0.58). This value indi-
analyses indicated d = 0.85 (p < 0.05; cates a medium effect size (Cohen, 1988)
CI = 0.82–0.88) favoring symptom-fo- associated with fitting patient coping style
cused over insight-focused interventions. to treatment focus. The average well-
This is a large effect, and clearly, at least matched treatment produced an 8% greater
among treatments comprising the majority effect than a randomly matched treatment—
of this data set, a direct symptomatic focus the average patient with a good fit was better
is superior to an indirect, insight focus of off than 58% of those with a poor match.
treatment; however, this conclusion must The findings were consistent across stud-
be considered with caution because of ies in demonstrating the selective efficacy
several factors: (1) the included studies of symptom/skill-building methods and
were selected because they allowed the insight-/interpersonally oriented methods
assessment of a matching or selective treat- as a function of patient coping style. All
ment effect, and many studies that included studies found results in the same direction;
variation in patient coping style without a interpersonal and insight-oriented therapies
corresponding measure of treatment focus are more effective among internalizing
may have been excluded; and (2) fully half patients, whereas symptom and skill-
of the studies included were conducted by building therapies are more effective among
members or former members of our own externalizing patients.
research group, leaving the conclusions This meta-analytic result supports the
subject to potential investigator bias. conclusions of the earlier review and adds
important information about the strength of
Results: Moderated Effects of Patient the effect. Moreover, given the correspon-
Coping Style dence among the two reviews, one an inclu-
The 12 studies in our final meta-analysis all sive review and this, a truncated review of
allowed a test of the proposition that coping only those studies that reported relevant
style could serve as a moderator of the effect statistics, the conclusion gains some veracity.
of differential treatment focus. Nine of the
12 studies used a direct measure of patient Patient Contributions
coping style. Only three used a direct Coping style is a relatively stable and endur-
measure of therapy focus. The individual ing patient quality; it is best conceptualized
studies had from 1 to 14 effect sizes as a personality trait (Beutler, Moos, &
comparing the level of fit to outcome. Lane, 2003). Clearly, coping style is an
The statistic of interest in these analyses aptitude that contributes to differential
was the variance accounted for by “fitting” treatment outcome when it interacts with
the patient’s coping style to the treatment treatment focus. Its independent effect is
focus. A weighted composite mean effect uncertain, as noted previously. In the earlier
edition of this volume, the review of coping use of symptom/skill-focused interventions.
style (Beutler et al., 2002) suggested that Both values reflect medium effect sizes, but
internalizing patients were better prognostic the difference between them (d = 0.16;
risks in psychotherapy than externalizing p < 0.10) was nonsignificant. Thus, we are
patients, but that finding does not achieve unable to conclude that those with one
the level of consistency required for clinical style of coping (e.g., internalizing) are more
application in the current analysis. In an likely to benefit from psychotherapy than
effort to partially salvage a reliable test of those with the other. Notably, this also
this hypothesis, we looked at the relative means that we did not find evidence for
effect sizes associated with treatment focus, Kagan’s (1998) assumption either that the
separating and comparing the values for fearful, hypersensitive internalizer would
patients with each of the two coping styles. be more of a prognostic risk than the under-
That is, we calculated the mean effect sizes responsive externalizer.
associated with insight/awareness-focused Judging from the current findings,
and symptom/skill-building-focused thera- patient’s coping styles are distributed
pies on samples of patients who were predom- broadly within the population at large and
inantly internalizing and compared them all along the coping style continuum.
with samples of patients who were predomi- Individuals with both internal and external
nantly externalizing. styles of coping are capable of benefitting
In our sample of 12 studies, there were from psychotherapy, assuming that the
three that were done on samples of patients nature of that treatment is appropriate to
whose diagnoses suggested that they may be their own coping style.
dominantly internalizing coping types.
These included patients diagnosed with Limitations of the Research
chronic shyness and avoidant personality There are limitations to any research analysis,
disorder (Kimpara—cited in Beutler 2009), including meta-analyses. Three major threats
those with chronic depression (Knekt et al., need to be considered in our meta-analytic
2008), and those with anxiety disorders review. First, several studies are excluded
(Milrod et al., 2007). The effect sizes of these because they do not include data that allows
studies indicate, generally, that the impact effect sizes to be constructed in a way that is
of the focus of treatment was moderate. The comparable across studies. That was certainly
weighted mean effect size (d) of these three a problem here where 12 studies found in
studies was 0.37 (p < 0.05; CI = 0.27–0.47), our review of the literature were not included
indicating a medium effect size favoring because of missing statistical information.
insight/awareness treatments over symptom/ Nonetheless, a tabulation of these studies
skill-focused treatments. confirmed that the direction of their find-
By comparison, 5 of the 12 studies in ings were consistent with the direction of
our sample were conducted on patients the effect sizes we computed.
whose diagnosis suggested a dominantly Second, it is of some concern that 4 of
externalizing coping pattern (i.e., substance the 12 studies in this meta-analysis utilized
abusers; Litt et al., 1992; Longabaugh et al., an indirect measure of patient coping style.
1994; Kadden et al., 1998; Karno et al., While nine of the studies employed a cate-
2002; Beutler, Moliero et al., 2003). These gorical measure based on the treatment
studies earned a mean, weighted effect models/manuals employed, this is a much
size of treatment focus (d ) of 0.53 less serious breach of the criteria of ade-
(p < 0.05; CI = 0.40–0.67), favoring the quacy than the use of a proxy measure for a
patient trait. A diagnosis of alcohol depen- Beutler and Harwood (2000) to
dence does not, logically, equate to a sensi- enable the clinician to make any
tive indication of one’s dominant coping necessary in-session treatment-matching
style of externalization. However, we lim- adjustments. These procedures combine
ited the use of such proxy measures, and self-report and clinician ratings to define
the limitation placed on the findings by characteristic ways that the patient
this categorical proxy of coping style augurs responds to change and novelty.
well to ensure the results are conservative. • Match the patient’s coping style
In follow-up research, however, we urge to the focus of treatment. Patients who
those who seek to understand ATI relation- manifest externalizing tendencies can be
ships to employ direct, individual mea- provided with treatments that are focused
sures of both the treatment variable and, on skill building and on symptom change.
especially, the patient variable under inves- In contrast, those who manifest patterns
tigation that reveals a masked relationship of self-criticism and emotional avoidance
because the other studies do not have such are more likely to benefit from an
a feature. interpersonally focused and insight-
Third, the pool of 12 studies analyzed oriented treatment.
here may prove restrictive in another sense. • Even with internalizing patients, the
This is a relatively small number of studies, research suggests that there is value in
the majority of which were conducted by beginning treatment with direct,
one senior researcher (Beutler) and various symptom-focused methods. As the coping
colleagues. The possibility of the file drawer style of the patient becomes clear, it may
phenomenon was not also considered in then be optimal to switch to a more
the analyses. Although obviously not a indirect, insight approach if that patient’s
concern in that the authors of this chapter coping style is weighted toward
would be aware of any unpublished studies internalizing patterns.
of their own, we do not know if there exist • More broadly, remember that the
other unpublished studies that were left in focus of treatment represents both an aspect
a file drawer because their results did not of one’s theoretical orientation and some
favor the predicted aptitude by treatment personal proclivity or preference. Effective
interaction. psychotherapists will recognize a patient’s
distinctive aptitude, such as coping style,
reactance level, stage of change, ethnic/
racial heritage, and other moderators, in
Patient coping style emerges in the research
order to modify treatment to fit the patient
as a moderator of the effects of treatment
and his/her unique circumstances.
focus on outcomes. We offer, in closing,
several practice recommendations based on
the research reviews: References
• Assess each patient’s predominant An asterisk (∗) indicates studies included in
coping style using a direct, individual the meta-analysis.
measure in the interest of treatment Badger, T. A., & Collins-Joyce, P. (2000). Depres-
sion, psychosocial resources, and functional
planning. Assessment of these patient ability in older adults. Clinical Nursing Research,
attributes need not be time consuming or 9(3), 238–55.
tedious; cues for the identification of a Barber, J. P., & Muenz, L. R. (1996). The role of
variety of patient attributes are included in avoidance and obsessiveness in matching patients
to cognitive and interpersonal psychotherapy: Barbara study of fitting therapy to patients: First
Empirical findings for the treatment for depres- results. Paper presented at the International
sion collaborative research program. Journal Society for Psychotherapy Research, Chicago.
∗
of Consulting and Clinical Psychology, 64, Beutler, L. E., Moleiro, C., Malik, M., Harwood, T. M.,
951–58. Romanelli, R., Gallagher-Thompson, D., et al.
Beutler, L. E., & Berren, M. R. (1995). Integrative (2003). A comparison of the Dodo, EST, and
assessment of adult personality. New York: ATI indicators among co-morbid stimulant
EGuilford Press. dependent, depressed patients. Clinical Psychology
Beutler, L. E., & Clarkin, J. F. (1990). Systematic & Psychotherapy, 10, 69–85.
treatment selection: Toward targeted therapeutic Beutler, L. E., Moos, R. H., & Lane, G. (2003).
interventions. New York: Brunner/Mazel. Coping, treatment planning, and treatment
Beutler, L. E., Clarkin, J. F., & Bongar, B. (2000). outcome: discussion. Journal of Clinical
Guidelines for the systematic treatment of the Psychology, 59, 1151–67.
depressed patient. New York: Oxford University Billings, A. G., & Moos, R. H. (1985). Life stressors
Press. and social resources affect posttreatment
∗
Beutler, L. E., Engle, D., Mohr, D., Daldrup, R. J., outcomes among depressed patients. Journal of
Bergan, J., Meredith, K., et al. (1991). Predictors Abnormal Psychology, 94, 140–53.
of differential response to cognitive, experiential, Borenstein, M., Hedges, L., Higgins, J., &
and self-directed psychotherapeutic techniques. Rothstein, H. (2009). Introduction to meta-
Journal of Consulting and Clinical Psychology, analysis. Chichester, UK: Wiley.
59, 333–340. Bussing, R., Zima, B. T., & Perwien, A. R. (2000).
Beutler, L. E., Goodrich, G., Fisher, D., & Self-esteem in special education children with
Williams, O. B. (1999). Use of psychological ADHD: Relationship to disorder characteristics
tests/instruments for treatment planning. In M. and medication use. Journal of the American
E. Maruish (Ed.), The use of psychological tests for Academy of Child & Adolescent Psychiatry, 39(10),
treatment planning and outcome assessment 1260–69.
(2nd ed., pp. 81–113). Hillsdale, NJ: Lawrence Butcher, J. N. (1990). The MMPI-2 in
Erlbaum. psychological treatment. New York: Oxford
Beutler, L. E., & Harwood, T. M. (2000). Prescriptive University Press.
∗
psychotherapy: A practical guide to systematic Calvert, S. J., Beutler, L. E., & Crago, M. (1988).
treatment selection. New York: Oxford University Psychotherapy outcomes as a function of
Press. therapist-patient matching on selected variables.
Beutler, L. E., Kim, E. J., Davison, E., Karno, M., Journal of Social and Clinical Psychology, 6,
& Fisher, D. (1996). Research contributions to 104–117.
improving managed health care outcomes. Cohen, J. (1988). Statistical power analysis for the
Psychotherapy, 33, 197–206. behavioral science (2nd ed.). Hillsdale, NJ:
∗
Beutler, L. E., Machado, P. P. P., Engle, D., & Lawrence Erlbaum Associates.
Mohr, D. (1993). Differential patient X Cortina, J. M., & Nouri, H. (2000). Effect size for
treatment maintenance aiming cognitive, experi- ANOVA designs. Thousand Oaks, CA: Sage
ential, and self-directed psychotherapies. Journal Publications.
of Psychotherapy Integration, 3, 15–31. Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO
Beutler, L. E., & Mitchell, R. (1981). Differential Personality Inventory manual. Odessa, FL:
psychotherapy outcome among depressed and Psychological Assessment Resources.
impulsive patients as a function of analytic and Ellicott, A., Hammen, C., Gitlin, M., Brown, G.,
experiential treatment procedures. Psychiatry, & Jamison, K. (1990). Life events and the course
44, 297–306. of bipolar disorder. American Journal of Psychiatry,
Beutler, L. E., Mohr, D. C., Grawe, K., Engle, D., 147, 1194–1198.
& MacDonaled, R. (1991). Looking for differ- Endler, N. S., Parker, J. D. A., & Butcher, J. M.
ential treatment effects: Cross-cultural predictors (1993). A factor analytic study of coping styles
of differential psychotherapy efficacy. Journal of and the MMPI-2 Content Scales. Journal of
Psychotherapy Integration, 1(121–41). Clinical Psychology, 49, 523–527.
Beutler, L. E., Moleiro, C., Malik, M., & Eysenck, H. J. (1957). The dynamics of anxiety and
Harwood, T. M. (2000, June). The UC Santa hysteria. New York: Praeger.
Fahy, T. A., & Russell, G. F. M. (1993). Outcome Koolhaas, J. M., Korte, S. M., De Boer, S. F.,
and prognostic variables in bulimia-nervosa. Van der Vegt, B. J., Van Reenan, C. G.,
International Journal of Eating Disorders, 14, Hopster, H., et al. (1999). Coping style in
135–45. animals: Current status in behavior and stress-
Fisher, D., Beutler, L. E., & Williams, O. B. (1999). physiology. Neuroscience and Biobehavioral
STS Clinician Rating Form: Patient assessment Reviews, 23, 925–35.
and treatment planning. Journal of Clinical Koran, L. M. (2000). Quality of life in obsessive-
Psychology, 55, 825–42. compulsive disorder. The Psychiatric Clinics of
Hoglend, P. (1993). Personality disorders and long- North America, 23, 509–517.
term outcome after brief dynamic psychotherapy. Latack, J. C., & Havlovic, S. J. (1992). Coping with
Journal of Personality Disorders, 7(2), 168–81. job stress: A conceptual evaluation framework
Hunter, J. E., & Schmidt, F. L. (2004). Methods of for coping measures. Journal of Organizational
meta-analysis: Correcting error and bias in research Behavior, 13, 479–508.
findings, 2nd edition. Thousand Oaks, CA: Sage Lazarus, A. A. (1981). The practice of multi-modal
Publications. therapy. New York: McGraw-Hill.
Judd, L. L., Paulus, M. P., Wells, K. B., & Lazarus, R. S., & Folkman, S. (1984). Stress,
Rapaport, M. H. (1996). Socioeconomic burden appraisal and coping. New York: Springer.
of subsyndromal depressive symptoms and major Lipsey, M. W., & Wilson, D. B (2001) Practical
depression in a sample of the general population. meta-analysis. Applied social research methods
American Journal of Psychiatry, 153(11), series, Volume 49. Thousand Oaks, CA: Sage
1411–1417. Publications.
∗ ∗
Kadden, R. M., Cooney, N. L., Getter, H., & Litt, M. D., Babor, T. F., DelBoca, F. K.,
Litt, M. D. (1989). Matching alcoholics to Kadden, R. M., & Cooney, N. L. (1992). Type
coping skills or interactional therapies: of alcoholics: II. Application of an empirically
Posttreatment results. Journal of Consulting and derived typology to treatment matching. Archives
Clinical Psychology, 57, 698–704. of General Psychiatry, 49, 609–614.
Kagan, J. (1998). Galen’s prophecy: temperament in Littlefield, C. H., Rodin, G. M., Murray, M. A., &
human nature. NY: Basic Books. Craven, J. L. (1990). Influence of functional
∗
Karno, M., Beutler, L. E., & Harwood, T. M. impairment and social support on depressive
(2002). Interactions between psychotherapy symptoms in persons with diabetes. Health
procedures and patient attributes that predict Psychology, 9(6), 737–49.
∗
alcohol treatment effectiveness: A preliminary Longabaugh, R., Rubin, G. M., Malloy, P.,
report. Addictive Behavior, 27, 779–797. Beattie, M., Clifford, P. R., & Noel, N. (1994).
Kawai, H. (1993). Monogatari to ningenno kagaku Drinking outcomes of alcohol abusers diagnosed
(Stories and human’s science). Tokyo: Iwanami as antisocial personality disorder. Alcoholism:
Books. Clinical and Experimental Research, 18, 778–85.
Kawai, H. (1996). The Japanese psyche: Major motifs Miller, W. R., & Joyce, M. A. (1979). Prediction
in the fairy tales of Japan. Dallas, TX: Spring. of abstinence, controlled drinking, and heavy
Keijsers, G. P. J., Hoogduin, C. A. L., & drinking outcomes following behavioral self-
Schaap, C. P. D. R. (1994). Predictors of control training. Journal of Consulting and
treatment outcome in the behavioral treatment Clinical Psychology, 47, 773–75.
∗
of obsessive-compulsive disorder. British Journal Milrod, B., Leon, A. C., Busch, F., Rudden, M.,
of Psychiatry, 165, 781–86. Schwalberg, M., Clarkin, J., et al. (2007).
∗
Knekt, P., Lindfors, O., Hrakanen, T., A randomized controlled clinical trial of psycho-
Valikoski, M., Virtala, E., & Laaksonen, M. A. analytic psychotherapy for panic disorder.
(2008). Randomized trial of the effectiveness of American Journal of Psychiatry, 164, 265–72.
long-term and short-term psychodynamic Ouimette, P., Finney, J., & Moos, R. (1999). Two
psychotherapy and solution-focused therapy on year posttreatment functioning and coping of
psychiatric symptoms during a 3-year follow up. patients with substance abuse and posttraumatic
Psychological Medicine, 38, 689–703. stress disorder. Psychology of Addictive Behaviors,
Koenig, H. (1998). Depression in hospitalized older 13, 105–114.
patients with congestive heart failure. General Prochaska, J. O., & DiClemente, C. C. (1992).
Hospital Psychiatry, 20(1), 29–43. The transtheoretical approach. In J. C. Norcross
& M. R. Goldried (Eds.) Handbook of psycho- depression and response to cognitive behavior
therapy integration. (pp. 300–34). New York: therapy. American Journal of Psychiatry, 148,
Basic Books. 784–89.
Project Match Research Group. (1997). Matching UKATT Research Team. (2007). UK Alcohol
alcoholism treatments to client heterogeneity: Treatment Trial: Client-treatment matching
Project Match posttreatment drinking outcomes. effects. Addiction, 103, 228–38.
Journal of Studies in Alcoholism, 58, 7–29. Veiel, H. O., Kuhner, C., Brill, G., & Ihle, W.
Shea, M. T., Elkin, I., Imber, S. D., Sotsky, S. M., (1992). Psychosocial correlates of clinical
Watkins, J. T., Collins, J. F., et al. (1992). Course of depression after psychiatric in-patient treatment:
depressive symptoms over followup: findings form Methodological issue and baseline differences
the National Institute of Mental Health Treatment between recovered and non-recovered patients.
of Depression Collaborative Research Program. Psychological Medicine, 22, 425–27.
Archives of General Psychiatry, 49, 782–87. Welsh, G. S. (1952). An anxiety index and an inter-
Thase, M. E., Simons, A. D., Cahalane, J., nalization ratio for the MMPI. Journal of
McGeary, J., & Harden, T. (1991). Severity of Consulting Psychology, 16, 65–72.
C HA P TER
18 Expectations
Michael J. Constantino, Carol R. Glass, Diane B. Arnkoff,

Rebecca M. Ametrano, and JuliAnna Z. Smith
Patients’ expectations have long been con- consequences of participating in treat-

sidered a key ingredient and common factor ment (outcome expectations) and expecta-
of successful psychotherapy (e.g., Frank, tions about the nature and process of
1961; Goldfried, 1980; Goldstein, 1960; treatment (treatment expectations). For out-
Rosensweig, 1936; Weinberger & Eig, 1999). come expectations, we include a compre-
Influenced by classic social psychological hensive meta-analysis of the association
findings that substantiated the influence between pre- or early-therapy expectations
of expectations on people’s perceptions, and posttreatment outcome. Given the
motivations, and actions (e.g., Asch, 1946; many types of treatment expectations
Kelley, 1950; Secord, 1958), researchers and the heterogeneity of research methods
and clinicians became interested in how used to study them, we did not conduct
expectations specifically affect psychother- a meta-analysis of their association with
apy. In his classic book, Persuasion and outcome. Instead, we include a substantive,
Healing, Frank (1961) argued that for any though not exhaustive, narrative review of
therapy to be effective there must be within that research. We also review (a) definitions
the patient a mobilization of hope for of expectations and similar constructs,
improvement. According to Frank, patients (b) expectancy measurement, (c) mediators
enter therapy because they are demoralized, and moderators of the expectation–outcome
and restoring their hope and positive expec- link, (d) patient factors related to expecta-
tation is a powerful change ingredient. tions, and (e) limitations of the extant
Others have since concurred with this per- research base. In concluding, we offer
spective (e.g., Kirsch, 1985; Shapiro, 1981), therapeutic practices based on the research
some going so far as to suggest that most results.
psychotherapies are inextricably linked with
the manipulation and revision of patients’ Definitions and Measures
expectations (Greenberg, Constantino, & Outcome Expectations
Bruce, 2006; Kirsch, 1990). Definitions. Outcome expectations reflect
This chapter will review the research patients’ prognostic beliefs about the conse-
evidence linking patient expectations quences of engaging in treatment (Arnkoff,
with treatment outcome across a variety Glass, & Shapiro, 2002). In psychotherapy,
of psychotherapies and clinical contexts. outcome expectations are typically assessed
We consider both expectations about the on a continuum of the potential benefits
354
of treatment, with rare consideration of Arnow, Blasey, & Agras, 2005; Safren,
plausible expected negative effects (Schulte, Heimberg, & Juster, 1997). On the other
2008). Outcome expectations come in dif- hand, pretreatment outcome expectations
ferent guises. For example, patients have often exist prior to having any substantial
beliefs about a treatment’s utility even before information about the forthcoming treat-
they have contact with a therapist or the ment. Credibility, though, is a perception
treatment. Patients also have malleable based on knowledge gained through direct
during-treatment expectations that are experience or observation (Schulte, 2008;
influenced by their own history, their Tinsely, Bowman, & Ray, 1988). From
interactions with the therapist, and their another perspective, credibility reflects
ongoing appraisal of the treatment’s course what a patient thinks will happen (a cog-
and efficacy (Schulte, 2008). nitive process), while expectations assess
Outcome expectations are differentiated what a patient feels will happen (an affec-
from constructs such as treatment motiva- tive process; Devilly & Borkovec, 2000).
tion and therapy preferences. Motivation, Thus, although conceptually related, expec-
which encompasses patients’ desire and tancy and credibly are likely distinct.
readiness for change, does not necessarily Measures. Historically, patient expecta-
correspond to positive prognostic expectations have been viewed as potential artifacts
tions (see Norcross, Krebs, & Prochaska, requiring control in experimental treat-
this volume, Chapter 14). Patients in dis- ment trials. Thus, as predictive factors and
tress might be highly motivated to engage potential change ingredients, expectations
in treatment yet have low expectation or have tended to be undervalued, with few
faith that therapy can actually help them studies providing a primary assessment of
(Rosenthal & Frank, 1956). Preferences expectations (Weinberger & Eig, 1999).
(see Swift, Callahan, & Vollmer, this Rather, expectations have often been
volume, Chapter 15) are distinguishable assessed secondarily as a manipulation
from expectations in that they reflect some- check, so that researchers can point to the
thing valued or desired, which might be comparability of expectancies engendered
distinct from what is expected (Arnkoff by different treatments, thus eliminating
et al., 2002). For example, a patient might expectancy effects as a rival hypothesis
have a preference for working with a same- to any between-group effects observed
sex psychotherapist yet expect that it would (Borkovec & Nau, 1972; Holt & Heimberg,
be more helpful to work with an other-sex 1990). Most expectation measures have
therapist. been brief (in many cases one item only;
Another related construct involves treat- e.g., Heine & Trosman, 1960) and often
ment credibility, or how plausible, suitable, study specific (and thus lacking in psycho-
and logical a treatment seems to the patient metric validation; e.g., Barrios & Karoly,
(Arnkoff et al., 2002). There is some debate 1983). In some cases, the measures have
over whether credibility and expectancy been confounded with another belief con-
are distinct constructs. On the one hand, struct (such as credibility; e.g., Hardy et al.,
outcome expectations might develop, at 1995) or even an outcome measure (e.g.,
least in part, from how credible a treat- Evans, Smith, Halar, & Kiolet, 1985).
ment seems (Hardy et al., 1995). Moderately Borkovec and Nau (1972) pioneered
significant correlations between expec- the use of a brief (4-item) questionnaire
tancy and credibility/suitability scales sup- to assess whether the rationale of placebo
port this perspective (e.g., Constantino, therapies generated equivalent ratings of
co n s ta n t i n o , gl a s s , a r n ko f f, a m e t r a n o , s m i t h 355
credibility and treatment outcome expec- Perhaps one of the purest measures of out-
tancy as did behavioral treatments for come expectancy (aside from 1-item mea-
public-speaking anxiety. Easily adaptable for sures) is the Patient Prognostic Expectancy
different conditions, this measure became the Inventory (PPEI; Martin & Sterne, 1975),
most frequently used credibility/expectancy although it has been used fairly infre-
measure in psychotherapy research. Using quently. The PPEI assesses, on a 4-point
trauma as an example, the measure includes response scale, patients’ expected improve-
three credibility items rated on a 9-point, ment as a result of hospital treatment across
Likert-type scale (“At this point, how logi- 15 domains (e.g., depression-sadness, feel-
cal does the therapy offered seem to you?” ing afraid, keeping a job).
“At this point, how successful do you think The Expectations About Counseling
this treatment will be in reducing your measure (EAC; Tinsely, Workman &
trauma symptoms?” “How confident would Kass, 1980) and its short form (Tinsely &
you be in recommending this treatment to Westcot, 1990) predominantly assess treat-
a friend who experiences similar prob- ment expectations but also contain a 3-item
lems?”), and one outcome expectancy item scale assessing treatment outcome expec-
rated from 0% to 100% (“By the end of the tancies. This scale has strong psychometric
therapy period, how much improvement in properties.
your trauma symptoms do you think will Finally, a promising scale for indepen-
occur?”). In subsequent psychometric anal- dently assessing patients’ outcome expecta-
yses (Devilly & Borkovec, 2000), the three tions and perceived treatment suitability
credibility items hung together, while the is the Patients’ Therapy Expectation and
expectancy item hung together with two Evaluation (PATHEV; Schulte, 2008). This
additional affectively anchored items (“At measure consists of three, factor analyti-
this point, how much do you really feel cally derived subscales: hope of improve-
that therapy will help you to reduce your ment (confidence in treatment efficacy),
trauma symptoms?” “By the end of the fear of change, and suitability. Perhaps what
therapy period, how much improve- is most promising about this measure is
ment in your trauma symptoms do you that it differentiates hope (e.g., “I believe
feel will occur?”). Devilly and Borkovec my problems can finally be solved”) and
named their update to Borkovec and fear (e.g., “Sometimes I am afraid that my
Nau’s measure the Credibility/Expectancy therapy will change me more than I want”),
Questionnaire (CEQ). both of which are components of expecta-
The Reaction to Treatment Questionnaire tions (Heckhausen & Leppmann, 1991).
(RTQ; Holt & Heimberg, 1990) has also The PATHEV also includes the assessment
been used in several studies. This measure of pessimistic expectations (e.g., “Actually,
is comprised of the four Borkovec and I am rather skeptical about whether treat-
Nau (1972) items (yielding a treatment ment can help me”).
“credibility score”) and nine items assess-
ing patients’ confidence that the treatment Treatment Expectations
would eliminate anxiety in specific social Definitions. Treatment expectations reflect
situations (a situationally based “confidence” beliefs about what will transpire during
or outcome expectancy scale). Scored in treatment. One form of treatment expecta-
this manner, though, the credibility scale tion reflects role expectations, or beliefs
lacks differentiation between credibility about how a person occupying a given posi-
and expectancy. tion should behave (Arnkoff et al., 2002).
Patients may have role expectations of both you expect to be concerned with how you
themselves (e.g., crying in session) and appear to your therapist?”), advice seeking
their psychotherapist (e.g., providing sup- (e.g., “How strongly do you expect to
port). Patients also have process expectations get definite advice from your therapist?”),
about the type of work that will transpire audience seeking (e.g., “How strongly do
and the duration of treatment (Greenberg you expect to feel like opening up without
et al., 2006). In this chapter we focus any help from your therapist?”), and rela-
predominantly on role expectations but tionship seeking (e.g., “How strongly do you
caution that in some cases researchers expect to behave in a spontaneous manner?”).
did not distinguish between role and pro- Subsequent analyses (Bleyen, Vertommen,
cess expectations, tending to use “role” as a Vander Steene, & Van Audenhove, 2001)
blanket term. found adequate support for this four-factor
Measures. Although many studies have structure, but a better fit for a five-factor
employed idiosyncratic measures of treat- model that split the first factor into approval
ment expectations, there are two widely seeking and impression (e.g., “How strongly
used and well-validated measures. The do you expect to be concerned with the
aforementioned EAC (Tinsely et al., 1980) impression you make on your therapist?”).
and its brief version (EAC-B; Tinsely &
Westcot, 1990) assess four empirically Clinical Examples
derived expectancy domains: patient atti- Outcome Expectations
tudes and behaviors (e.g., motivation, open- By definition, therapy outcome expecta-
ness), counselor attitudes and behaviors tions are cognitions regarding a probable
(e.g., acceptance, confrontation), counselor future resulting from treatment. Such expec-
characteristics (e.g., expertness, trustwor- tations can be positive (e.g., “I have faith
thiness), and counseling process and out- that I can do the work and feel better”),
come (e.g., immediacy, concreteness, and negative (e.g., “I can’t imagine ever feeling
the aforementioned outcome scale) (see better, even after this therapy”), or ambiva-
also Ægisdóttir, Gerstein, & Gridley, 2000 lent (e.g., “Well, I am willing to give it a
for a proposed three-factor solution). shot, but I’m just not sure this will work . . .
The Psychotherapy Expectancy Inventory’s I have been depressed for a long time”).
(PEI; Rickers-Ovsiankina, Geller, Berzins, Of course, patients’ hopes and expecta-
& Rogers, 1971) factor analytically derived tions may conflict. For example, a patient
scales correspond to Apfelbaum’s (1958) might have a desperate wish to feel emo-
three clusters of expected therapist roles: tional relief (e.g., “I hope to feel like my
nurturant (to be guided by an affiliative old self ”), yet have what he or she deems a
other), model (to be guided to help one- more reality-based expectation or predic-
self ), and critical (to receive guidance and tion (e.g., “I expect that therapy might not
correction). The authors added a fourth help me completely and that I will never
dimension, cooperative (to become auton- fully be what I used to be”).
omous and equal to the counselor), which Prognostic expectations are also affected
they purported comes about only toward by context, including perhaps most power-
treatment’s end. Subsequent reanalysis led to fully one’s own learning experiences. For
the Psychotherapy Expectancy Inventory- example, a male patient might have had
Revised (PEI-R; Berzins, 1971), with a positive therapy experience with an
renamed but conceptually consistent scales older female therapist in the past, which
of approval seeking (e.g., “How strongly do has led him to have greater faith in either
recommencing therapy with this same understanding, sympathetic, interested and
therapist or seeing a new therapist with competent person who would be unlikely
perceived salient similarities (e.g., gender, to engage in criticism, anger or ridicule.
age, theoretical orientation). They also want someone who will not
Outcome expectations and treatment be pessimistic about them, nor turn them
expectations probably interact. For exam- away, but who will at the same time
ple, a patient might generally have high not deny that the patient has difficulties”
outcome expectations prior to therapy and (p. 360).
also expect therapy to focus exclusively on But many patients have not formed pre-
early childhood (treatment expectation). treatment expectations for psychotherapy. In
Upon meeting with a well-regarded thera- a study of former Veterans Administration
pist who tends to work from a here-and- clinic patients (Kamin & Caughlan, 1963),
now, problem-oriented perspective, this the authors concluded that “. . . almost 75%
patient’s outcome expectations might take entered therapy with no clear concepts
a hit. However, psychotherapists can often of its modus operandi. They understood
frame their approach in accord with the neither their own role, nor that of the
patient’s treatment expectations, thereby therapist . . . repeatedly commented that
enhancing the patient’s outcome expecta- therapists were too passive, disinterested,
tions. For example, the therapist might cold, incomprehensible, enigmatic, even
say, “Actually, in discussing your current though polite, patient, and probably well
problems and relationships, we will likely meaning . . . therapists are analytically ori-
see traces of these same problems and pat- ented, but the patients are not” (p. 666).
terns from your earlier life. People learn Other surveys indicated that many
many things early on that have a lasting patients hold incongruent or unrealistic
influence on their present thoughts and expectations, almost having a “. . . naive,
feelings. Thus, although we might lean wishful, or magical view of counseling”
toward discussing the here-and-now, your (Tinsley, Bowman, & Barich, 1993, p. 50).
childhood will not be off limits, and I sus- Some patients expect the psychotherapist,
pect that we will learn something quite like their physician, will tell them what is
useful from connecting past to present. wrong and fix them. Others seeing a cogni-
How does this sound to you?” tive-behavioral psychotherapist may expect
to lay on a couch and talk about their child-
Treatment Expectations hood, while patients with psychodynamic
In one study (Garfield & Wolpin, 1963), or experiential psychotherapists may be
27% of surveyed clinic patients expected frustrated by their clinician’s limited input
that therapy would predominantly center (Walborn, 1996).
on their early life, and 47% thought that
the central focus would be on their life Research Review
just before therapy. Half of the patients In this section, we provide research reviews
indicated that the most important thing for both outcome and treatment expecta-
a therapist does is to help patients under- tions. For the former, we summarize a
stand themselves better, while 33% pointed previous box count review as well as pres-
to advice. Forty percent thought that ent a comprehensive and original meta-
the therapist could read their mind at analytic review of the association between
least moderately. The authors concluded: pre- or early-therapy outcome expectations
“Patients appear to be seeking a sincere, and treatment outcomes. For the latter, we
summarize a previous box count and offer Given these mostly positive but still
a selective review of relevant studies pub- mixed findings, it remains difficult to
lished since the previous version of this determine the consistency of the outcome
chapter (Arnkoff et al., 2002). Our reviews expectancy effect across various treatment
are limited to clinical samples receiving contexts, as well as its magnitude. The cur-
psychotherapy, with more specific inclu- rent meta-analysis attempts to shed addi-
sion/exclusion criteria for the meta-analysis tional light on these questions by focusing
discussed below. on the aggregated effect of outcome expec-
tations on posttreatment status. In addition
Outcome Expectations to examining the overall effect of outcome
Arnkoff and colleagues (2002) presented a expectations on treatment outcome, we
box count review of studies through the examined the potential moderating influ-
year 2000 that examined the association ence of several clinical variables: presenting
between patient outcome expectations and diagnosis, treatment orientation, treatment
psychotherapy outcomes (e.g., treatment modality, treatment setting, design type,
continuation, patient self-report, behav- and date of publication.
ior). They found that 12 studies revealed Search and Inclusion Procedures. We first
a significant positive association, 7 revealed conducted an extensive PsycINFO database
mixed findings, and 7 others demonstrated search for all references through December
no effect. 2009. We included the following 14
Since 2000, several other researchers searches (limited to published sources writ-
have summarized research on patient expec- ten in English): expecta∗ (any derivation
tations. A review of the psychiatric litera- of expectation) in combination with psy-
ture (Noble, Douglas, & Newman, 2001) chotherapy, treatment, therapy, counseling,
found that research prior to 1980 generally counselling, outcome, improvement, change,
suggested a curvilinear relationship between dropout, dropping out, premature termina-
outcome expectations and outcome. That tion, duration, patient, and client. This data-
is, patients with moderate outcome expec- base search yielded 39,250 citations. We
tations demonstrated better outcomes than then searched PubMed using the terms
those with extremely high or extremely expecta∗ and psychotherapy and expecta∗ and
low expectations (e.g., Goldstein, 1962). counseling, which yielded an additional
For the period from 1980 to 1999, several 15 citations. Finally, we hand-searched the
studies demonstrated a positive association reference lists of prior review articles, as
(Hansson & Berglund, 1987; Sotsky et al., well as the last four issues of 10 clinical
1991), while one study revealed no signi- journals (to ensure that we did not miss
ficant effects (Basoglu et al., 1994) and any citations because of a lag before appear-
another showed a negative association ing in PsycINFO or PubMed). These hand
(Lax, Basoglu, & Marks, 1992). A review searches revealed 13 additional citations for
of additional studies from 2000 to 2005 a total initial yield of 39,278 citations. We
(Greenberg et al., 2006) found several reviewed the titles and abstracts of all cita-
studies demonstrating a positive association tions and applied the inclusion/exclusion
between outcome expectations and either criteria in the next paragraph to create a
alliance quality (e.g., Constantino et al., candidate list.
2005) or posttreatment outcomes (e.g., To be included in the meta-analysis,
Joyce, Ogrodniczuk, Piper, & McCallum, studies had to (a) be correlational (this is
2003). reflective of the field in that virtually no
studies exist that experimentally mani- are reported, the actual parameters being
pulate outcome expectations in a way not estimated differ depending on which vari-
confounded by treatment expectations), ables are included. In addition, there was
(b) include a measure of patients’ own generally insufficient information avail-
outcome expectations at pretreatment or able to calculate standard errors of these
following Session 1, and (c) include a estimates and, consequently, to determine
posttreatment symptom outcome measure accurate inverse variance weights for the
not explicitly referenced as a follow-up meta-analysis. Thus, the total number of
occasion. Studies were excluded if they independent samples on which we con-
(a) examined a nonclinical sample (e.g., ducted our meta-analysis was 46.
students participating for course credit), In some studies, researchers assessed
(b) involved non-clinically-oriented out- outcome expectancies with more than
comes (e.g., well-being behaviors such as one measure and/or at both baseline and
exercise promotion programs), (c) focused postsession. Furthermore, in many stud-
solely on expectations other than for treat- ies, researchers assessed multiple treatment
ment outcomes/consequences, (d) did not outcomes. For studies that included multi-
involve a psychotherapist (e.g., self-help), ple outcomes, our goal was to identify
(e) only inferred outcome expectation and to code up to the three most psy-
through tests of placebo treatment effects, chometrically sound symptom measures.
(f ) assessed outcome expectations with However, for the few studies that included
a measure that was capturing a related, but more than three sound symptom measures,
distinct construct (e.g., treatment credibility, we coded more than three. In order to
motivation), (g) assessed outcome expecta- create a single effect size for each indepen-
tions retrospectively only, (h) employed dent sample, we averaged across the multi-
an experimental manipulation of outcome ple expectancy and/or outcome measures
expectancies, or (i) involved a treatment of and time points by creating a weighted
fewer than three sessions. average based on the sample size for each
Based on these criteria, 186 candidates effect reported.
were selected from the titles and abstracts Data Analyses. To estimate the direct
review. We fully read these candidates and effect of outcome expectations on outcome,
ruled out another 108 studies at this stage. we examined mean difference scores and
Thus, 78 studies were fully coded (by the bivariate associations. We also included
first four authors) for study characteristics effects between outcome expectations and
relevant to this review. In the case of articles treatment outcome that accounted for
that included multiple studies on separate pretreatment levels of symptomatology
samples, we coded these samples separately. (either with partial correlations, the use
For studies from separate articles that of researcher-derived change scores for
analyzed data from the same sample, we an outcome variable, or patient-reported
coded them as one sample. We excluded change on an outcome variable). We sum-
studies that reported only multivariate effects marized the averaged results from each
because of the difficulty obtaining accurate independent sample using the r statistic
estimates of comparable effects across stud- following the procedures outlined in Lipsey
ies (Lipsey & Wilson, 2001): the inclusion and Wilson (2001). In situations where
of other variables in the model equations the coefficient was unknown and reported
means that even when standardized effects only as nonsignificant, we used a conserva-
(such as standardized regression coefficients) tive approach of setting r to zero.
We next calculated an overall r across bias, we also calculated a fail-safe N to
samples. As sample sizes and, consequently, determine the number of nonsignificant
the precision of the effect size estimates file drawer studies that would be required
varied from study to study, the effect to attenuate the current results to an effect
sizes from the independent samples were less than r = 0.10 (i.e., less than a small
weighted by the inverse of their variance. effect in behavioral science research). The
As we desired to generalize to a popula- fail-safe N was 9 studies—that is, about
tion of studies, we used a random-effects 9 studies would need to have an average
model. r of 0.00 to bring the weighted mean
Finally, we grouped samples on the vari- r below 0.10. Thus, it seems reasonable to
ous moderator characteristics and compared suggest some caution in interpreting our
average weighted effect sizes using a mixed- meta-analytic findings.
effects model analogous to an ANOVA, We also examined the overall weighted
which tests whether the systematic vari- effect without an outlier based on sample
ance in r is a function of the categori- size (one study that contributed more than
cal variable included (Lipsey & Wilson, half of the total n to the meta-analysis).
2001). Without this study, the effect size and
Results. The meta-analysis included confidence intervals were essentially equiv-
8,016 patients across the 46 samples. In all alent. The same was the case when we
but one study, the patients were identified examined the overall effect removing a diff-
as predominantly (>80%) adult (age 18 erent outlier (i.e., a study that reported a
to 65). In all studies that reported race (13 moderate averaged negative effect). Because
of 46), the patients were predominantly the coefficients for many individual tests
white (>60%). In the studies reporting were not reported, and because we primar-
gender (41 of 46), 53.7% included pre- ily limited our coded outcome variables
dominantly (>60%) women, 9.8% pre- to three for any given study, we also tabu-
dominantly men, and 36.6% mixed (no lated the total number of tests conducted
predominant sex). and the proportion of those tests that
Table 18.1 includes averaged effect sizes had significantly positive or significantly
on outcome across all relevant analyses for negative effects (as researchers typically
each independent sample. We coded the reported significance and direction of
direction of the effect such that positive rs effects even in the absence of coefficients).
reflect positive associations between out- As indicated in Table 18.1, of 253 total
come expectations and favorable treat- tests (including studies that reported one
ment outcome, whereas negative rs reflect or more multivariate effects), 54 (21%)
a negative association. The overall weighted demonstrated a significantly positive asso-
effect size across samples was r = 0.12, ciation and 7 (3%) a significantly negative
p < 0.001 (CI.95 0.10 to 0.15), indicating association.
a small, but significant positive effect. Potential Moderators. There were no
Expressed as Cohen’s (1988) d, the effect statistically significant moderator effects
size was 0.24. of the expectancy–outcome association for
A test of homogeneity revealed signifi- any of the five potential moderators we
cant heterogeneity between studies, Q(45) = evaluated. However, there was a trend for
92.00, p < 0.001, which is not surprising design type. The direction of this trend,
given the different study designs, instruments, though, is influenced by the inclusion or
and eras. To address potential publication exclusion of the large sample outlier, thus
Table 18.1 Study Characteristics and Average Weighted Effects for Samples Included in the Meta-Analysis
Expectancy
Source Treatment Measure Time Primary outcomes ES Study Total Number
(Type/ (r) N tests significant
Modality) (+, −)
Abouguendia Mixed/ SS (multi) B General symptoms 0.27 107 3 2, 0
et al. (2004) Group Grief symptoms
Target objectives
Barrios & Mixed/ SS (multi) B Migraine ns 30 7 0, 0
Karoly Individual Total headache
(1983) Total disability
Basoglu et al. CBT/ SS (single) B Global improvement ns 154 2 0, 0
(1994) Individual Panic free
Bloch et al. NR/Group SS (multi) B Goal attainment 0.31 27 3 1, 0
(1976) Main problem change (therapist)
Main problem change (rater)
Borkovec & CBT/ CEQ S1 HAM-A 0.26 66 10 5, 0
Costello Individual Assessor GAD severity
(1993) PSWQ
Borkovec CBT/ CEQ S1 End state functioning ns 76 1 0, 0
et al. (2002) Individual
Buwalda & O/Group AS (single) S1 GIAS 0.07 140 1 0, 0
Bouman
(2008)
Calsyn et al. O/NR SS (multi) B Psychotic symptoms 0.14 65 1 0, 0
(2003) Program satisfaction
Chambless CBT/ CEQ S1 Anxious apprehension 0.17 64 5 5, 0
et al. (1997) Group Anxiety & skill (patient)
Anxiety & skill (rater)
Clark et al. CBT/ CEQ S1 Panic/Anxiety 0.50 43 1 1, 0
(1999) Individual
Collins & NR/NR SS (multi) S1 VETS global 0.09 4589 4 3, 0
Hyer (1986) PARS global
VETS improvement
Constantino Mixed/ SS (single) S1 Purge frequency 0.13 220 1 0, 0
Crits- PD/ SS (single) B HAM-A 0.10 68 6 2, 0
Christoph Individual SS (multi) B BAI
et al. (2004) PSWQ
Dearing et al. CBT/ SS (multi) B CSQ 0.00 208 3 0, 0
(2005) Mixed
Devilly & CBT/NR CEQ S1 HAM-A 0.34 67 8 2, 0
Borkovec PSWQ
(2000)– STAI
Study 2
(Continued )
362
Expectancy
Modality) (+, −)
Devilly & Mixed/ CEQ S1 STAI 0.06 22 18 1, 0
Borkovec Individual BDI
(2000)– SCL-90-R global
Study 3
Gaudiano & O/Group CEQ B Months in treatment 0.16 61 2 0, 0
Miller HAM-D
(2006) BRMS
Ghosh et al. CBT/ SS (single) B FQ phobic severity ns 84 2 0, 0
(1988) Individual Help received
Goldstein NR/ SS (multi) B Personality change ns 15 1 0, 0
(1960) Individual
Goossens Mixed/ CEQ B Motor behavior 0.12 171 4 2, 0
et al. (2005) Individual Pain coping & control
Negative effect
Greer (1980) NR/ PPEI B, S1 DGWBS −0.37 60 12 0, 7
Individual Social adjustment
General outcome
Hardy et al. Mixed/ CEQ B, S1 BDI 0.15 117 8 7, 0
(1995) Individual SCL-90-R
IIP
Joyce et al. Mixed/ SS (multi) B Disturbance (patient) 0.17 144 7 4, 0
(2003) Individual Disturbance (rater)
Disturbance (therapist)
Karzmark NR/NR SS (multi) S1 CSQ ns 110 1 0, 0
et al. (1983) GAS
Lax et al. CBT/ AS (NR) B Rituals 0.20 55 40 4, 0
(1992) Individual Obsessive thoughts
O-C checklist
Lipkin EXP/ SS (single) B TAT change 0.25 9 4 0, 0
(1954) Individual
Lorentzen & PD/Group SS (single) B GAF 0.16 69 5 1, 0
H glend SCL-90
(2004) IIP-C
Martin et al. NR/NR PPEI B PEQ adjustment 0.17 46 7 0, 0
(1976) PEQ improvement
Mathews CBT/ SS (multi) B Phobic severity 0.17 36 2 1, 0
et al. (1976) Individual SS (single) B
McConaghy CBT/ SS (single) S1 Anomalous urge 0.41 20 6 1, 0
et al. (1985) Individual Sexual urges
Sexual behavior
(Continued )
363
Expectancy
Modality) (+, −)
Meyer et al. Mixed/ AAE B BDI/HAM-D 0.22 151 1 1, 0
(2002) Individual (single)
Moene et al. O/ AS (single) B VRS motor conversion 0.27 24 2 0, 0
(2003) Individual Disability
O’Malley INT/ SS (single) B Change 0.36 35 3 0, 0
et al. (1988) Individual SAS
HAM-D
Persson & Mixed/ SS (single) B Global disorder 0.13 71 18 3, 0
Nordlund Individual Free anxiety
(1983) Ego restriction
Price et al. CBT/ CEQ S1 QATF 0.51 72 2 2, 0
(2008) Individual FFI
Richert NR/ AS (multi) B Self-satisfaction 0.11 26 3 1, 0
(1976) Individual Complexity
Permeability
Schoenberger CBT/ CEQ S1 PRCS 0.11 56 10 0, 0
et al. Group SS (multi) S1 FNE TBCL
(1997)
Shaw (1977) CBT/ SS (single) B FFQ 0.79 17 1 1, 0
Group
Spinhoven Mixed/ SS (single) B Pain reduction 0.27 165 1 1, 0
& ter Kuile Individual
(2000)
Stern & CBT/ SS (multi) B Main phobia ns 16 23 0, 0
Marks Individual Panic
(1973) Anxiety
ter Kuile CBT/ SS (single) B Headache 0.35 156 1 1, 0
Tollinton NR/NR SS (multi) B Distress 0.59 30 1 1, 0
(1973)
Van Minnen CBT/ CEQ S1 PTSD Symptom Scale 0.19 59 2 0, 0
et al. Individual
(2002)–
Sample 1
Van Minnen CBT/ CEQ S1 PTSD Symptom Scale 0.19 63 2 0, 0
et al. (2002)– Individual
Sample 2
Vannicelli & NR/ SS (single) B DASa 0.06 100 3 0, 0
Becker Combined SS (single) B DASb
(1981) RFQ
(Continued )
364
Expectancy
Modality) (+, −)
Wenzel et al. CBT/ AAE B HAM-D 0.23 32 5 1, 0
(2008) Individual (single) SSI
BDI-II
Note: (alphabetized within sections). Treatment type: CBT = predominantly cognitive and/or behavioral therapy; EXP = predominantly
humanistic/experiential therapy; INT = predominantly interpersonal/relational therapy; Mixed = different patients received different
treatments (none predominant, or >60%); NR = not reported; O = predominantly other therapy; PD = predominantly psychodynamic
therapy; Treatment modality: Combined = patients who received more than one treatment modality; Mixed = different patients received
different modalities (none predominant, or >60%); NR = not reported; Expectancy measure: AAE (single) = Attitudes and Expectations
Questionnaire single expectancy item; AS (multi) = author-specific expectancy measure with multiple items; AS (single) = author-specific
expectancy measure with single item; CEQ = Credibility/Expectancy Questionnaire or modified version (including Borkovec & Nau’s 1972
version); NR = not reported; PPEI = Patient Prognostic Expectancy Inventory; SS (multi) = study-specific expectancy measure with multiple
items; SS (single) = study-specific expectancy measure with single item; Expectancy assessment time: B = baseline; S1 = postsession 1; Primary
outcomes: BAI = Beck Anxiety Inventory; BDI = Beck Depression Inventory; BDI-II = Beck Depression Inventory–Second Edition; BRMS =
Bech-Rafaelson Mania Scale; CSQ = Client Satisfaction Questionnaire; DASa = Drinking Abstinence Scale; DASb = Drinking Adjustment
Scale; DGWBS = Dupuy General Well-Being Scale; FFI = Fear of Flying Inventory; FFQ = Flight Fear Questionnaire; FNE = Fear of
Negative Evaluation; FQ = Fear Questionnaire; GAD = generalized anxiety disorder; GAF = Global Assessment of Functioning; GAS =
Global Assessment Scale; GIAS = Groningen Illness Attitudes Scale; HAM-A = Hamilton Anxiety Rating Scale; HAM-D = Hamilton Rating
Scale for Depression; IIP = Inventory of Interpersonal Problems; IIP-C = Inventory of Interpersonal Problems–Circumplex; O-C =
obsessive-compulsive; PARS = Personal Adjustment and Role Skills Scale; PEQ = Psychotherapy Evaluation Questionnaire; PRCS = Personal
Report of Confidence as a Speaker; PSWQ = Penn State Worry Questionnaire; QATF = Questionnaire on Attitudes Toward Flying; RFQ =
Role-Functioning Questionnaire; SAS = Social Adjustment Scale; SCL-90-R Global = Symptom Checklist-90-Revised Global Distress; SSI =
Scale for Suicidal Ideation; STAI–State-Trait Anxiety Inventory; TAT = Thematic Apperception Test; TBC = Timed Behavior Checklist;
VETS Global = Veterans Adjustment Scale Global Adjustment Score; VETS Improvement = Veterans Adjustment Scale problem
improvement item; VRS = Video Rating Scale; ES = effect size (coefficients coded such that positive rs reflect positive associations between
outcome expectations and favorable outcome and negative rs a negative association); ns = nonsignificant; Study N: total initial study sample
size; Total tests: total number of tests of an expectancy–symptom outcome association reported in the study, including those for which no
coefficient was reported and/or a multivariate model was examined; Number significant: The number of significant positive and negative
associations between expectancy and symptom outcome, including those for which no coefficient was reported and/or a multivariate model
was examined.
rendering it problematic to interpret. The That is, the patient outcome expectation
specific results were: link to treatment outcome was fairly con-
• Presenting diagnosis [Q(3) = 2.00, sistent across each of these variables.
p = 0.57], coded as mood (n = 4), anxiety
(n = 17), substance abuse (n = 3), and Treatment Expectations
other (n = 8) Many who have written about treatment
• Treatment orientation [Q(1) = 1.21, expectations assume that they influence
p = 0.27], coded as cognitive-behavioral outcome; however, the research evidence
(n = 22) or other (n = 24) does not strongly support this assumption.
• Treatment modality [Q(2) = 2.31, In early reviews (e.g., Duckro et al., 1979)
p = 0.31], coded as individual (n = 30), and in our review of role expectation stud-
group (n = 7), or other (n = 3) ies through the year 2000 (Arnkoff et al.,
• Design type [Q(2) = 5.63, p = 0.06], 2002), the findings were equivocal (see
coded as comparative clinical trial also Noble et al., 2001). Arnkoff et al. iden-
(n = 23), open trial (n = 10), or tified 37 studies that addressed the rela-
naturalistic setting (n = 12) tion between role expectations and/or role
• Publication date [Q(1) = 0.13, expectation disconfirmation and an out-
p = 0.72], coded as before 2000 (n = 26) come measure, with disconfirmation being
or from 2000 to 2009 (n = 20). defined as a discrepancy between patient and
therapist role expectations and hypothesized likely to terminate therapy prematurely
to lead to a poor outcome (Goldstein, (Aubuchon-Endsley & Callahan, 2009).
1962). Nineteen studies demonstrated Another finding in the treatment expec-
some evidence for a significant, positive tations literature concerns the specific pro-
association between role expectations/ cess expectation about treatment duration
absence of disconfirmation and either (Clarkin & Levy, 2004). Several studies have
continuation in psychotherapy or patient, reported that the longer patients expect
therapist, or independent clinician ratings therapy to last, the longer they remain in
of psychotherapy outcome. Twelve studies, treatment (e.g., Jenkins, Fuqua, & Blum,
however, had mixed results, while eight 1986; Mueller & Pekarik, 2000). However,
revealed no significant relationship between several studies found that expectations for
role expectancy and outcome (note that duration were either unrelated to actual
studies with more than one type of out- duration (Hochberg, 1986) or showed sig-
come measure could be counted more nificant, but small associations (Pekarik &
than once). The 19 positive, 12 mixed, Wierzbicki, 1986).
and 8 nonsignificant results should be In sum, a dispassionate review of the
interpreted with caution in that many extant research finds mostly positive but
studies with positive results (especially the weak and mixed associations between
older ones) employed poor measurement treatment expectations and psychotherapy
of role expectations and/or outcome (e.g., outcomes. Insufficient numbers of well-
interviews with no quantification or con- controlled studies exist to either conduct
sensus analysis of qualitative data, drop- a meta-analysis (especially when carefully
out assessed by an arbitrary number of separating studies by specific type of treat-
sessions attended). Particularly when the ment expectation) or to render a more
quality of measurement of expectations definitive conclusion.
is taken into account, there was no out-
come measure for which the significant Mediators
findings outweighed the mixed and nega- Outcome Expectations
tive findings. Although the correlational data (includ-
Subsequent research on the association ing in our own meta-analysis) suggest that
of treatment expectations and psychother- outcome expectations show a small but
apy outcome has shown additional posi- significant association with treatment out-
tive findings. For example, Schneider and come, little is known about the specific
Klauer (2001) found that higher expecta- mechanisms through which they operate
tions of active involvement in psycho- (Arnkoff et al., 2002). Recently, however,
therapy were related to greater change several researchers have hypothesized that
in interpersonal functioning. In another the expectancy–outcome association is
study, treatment expectations in behavioral mediated by the patient–therapist alliance.
medicine treatment, specifically rejection Several studies have provided partial sup-
of the treatment rationale, predicted pro- port for this model in demonstrating that
gram dropout (Davis & Addis, 2002). patients’ outcome expectations are positively
A study examining the relationship between associated with alliance quality (a necessary
patients’ pretreatment role expectations step in demonstrating mediation) across
and attrition found that individuals who various treatments for various conditions
scored outside of the normative range on (e.g., Connolly Gibbons et al., 2003;
the PEI-R total score were seven times more Constantino et al., 2005).
We are aware of three studies that have (1998) found that patient expectations for
directly investigated the putative mediator the “typical session” were associated with
pathway. In research on patients with major better patient-rated alliance quality. In the
depressive disorder receiving short-term same study, better alliance quality was also
individual psychotherapy or pharmaco- associated with less discrepancy between
therapy (Meyer et al., 2002), patients with patients’ expectations for the typical session
mixed diagnoses receiving short-term indi- and their actual experience of session use-
vidual psychotherapy (Joyce et al., 2003), and fulness and comfort. In another study by
group counseling for grief (Abouguendia, the same investigators (Joyce, McCallum,
Joyce, Piper, & Ogrodniczuk, 2004), the Piper, & Ogrodniczuk, 2000), patients’ base-
therapeutic alliance was at least a partial line role behavior expectations interacted
mediator of patient expectancy effects on with their quality of object relations (QOR)
outcome, implicating the alliance as a robust to predict alliance quality. For patients
mechanism. Meyer and associates drew with higher QOR, higher expectations of
on goal theory (e.g., Austin & Vancouver, contributing to the treatment process was
1996) to explain this finding, suggesting associated with negative change in alliance
that people will only work toward a goal quality across short-term individual psy-
if they believe that they have a chance of chotherapy, suggesting that QOR may be
achieving it. Thus, patients who have posi- an important moderator of the expectancy–
tive outcome expectations (compared with alliance association. In a study of the asso-
more pessimistic beliefs) may be more likely ciation between patients’ pretreatment role
to engage in a collaborative working rela- expectations and early self-rated alliance
tionship with their therapist, which in turn quality, Patterson, Uhlin, and Anderson
may promote clinical improvement. (2008) found that role expectations acco-
Several investigators (e.g., Bootzin & unted for 31% of the variance in the goal
Lick, 1979; Higginbotham, 1977; Lick & dimension of the alliance, 30% in the
Bootzin, 1975) have postulated that outpatients’ bond with the therapist, and 24%
come expectations may produce change in the task dimension. Specific types of
through promoting greater patient adher- expectations were associated with the alli-
ence to the treatment regimen. One study ance; patients who were committed to
of cognitive-behavioral therapy for anxiety therapy and expected to take responsibility
found preliminary support for this perspec- for their work in therapy tended to report
tive (Westra, Dozois, & Marcus, 2007). stronger alliances.
Early homework compliance mediated the
association between baseline expectation of Patient Contributions
reducing one’s anxiety and early symptom Although the clinical importance of patient
change. outcome and treatment expectations has
been well documented, we have a paltry
Treatment Expectations understanding of factors that develop and
Similar to outcome expectations, formal maintain such beliefs. The available litera-
examination of mediators of the association ture suggests that diverse factors correlate
between treatment expectations and out- with or determine patients’ expectations.
come has been limited. However, as with
outcome expectations, there is some indi- Outcome Expectations
rect evidence that the alliance might be a A study of CBT for fibromyalgia and
mediator. For example, Joyce and Piper chronic low back pain found that less fear
of reinjury and active pain-coping strate- about therapist directiveness and in-session
gies were associated with higher pretreat- religious behavior (Belaire & Young, 2002;
ment outcome expectations (Goossens, Turton, 2004).
Vlaeyen, Hidding, Kole-Snijders, & Evers, Intrapsychic and historical variables
2005). These results are consistent with might also partially determine treatment
findings that less pain-related fear, more expectations. For example, in a sample of
internal control of pain, and lower depres- undergraduate students, adaptive perfec-
sion were associated with higher treatment tionism, also known as healthy, positive
outcome expectations for chronic low back striving, was associated with positive expec-
pain sufferers (Smeets et al., 2008). The tations toward both counseling process
association between more severe present- and outcome (Oliver, Hart, Ross, & Katz,
ing symptomatology and lower treatment 2001). For another example, group ther-
outcome expectations has also been found apy patients who had previously been
with socially phobic patients (Safren et al., in therapy had higher expectations about
1997). General hope might also be an the group treatment than patients with-
important determinant of outcome expec- out prior therapy experience (MacNair-
tations. For example, a study of students Semands, 2002).
seeking mental health counseling found
that patients who indicated more hopeless- Limitations of the Research
ness had lower expectations of improve- Outcome Expectations
ment (Goldfarb, 2002). Several limitations characterize our meta-
analysis on outcome expectations. First,
Treatment Expectations because we decided to retain all studies that
Several cultural and demographic variables met our a priori criteria (in the service of
have emerged as correlates of patient treat- comprehensiveness), the analysis contained
ment expectations. For example, Icelandic studies of varying quality. However, to the
students expected their psychotherapists extent that the more recent studies included
to have more expertise than did American improved measurement and methodology,
students (Ægisdóttir & Gerstein, 2000). it is interesting to note that publication
African-American and Latino/a college stu- year was not a moderator. Second, expec-
dents have reported higher multicultural tancy research has been plagued by poor
competence expectations of therapists com- measurement. In fact, of the 46 studies in
pared to Asian-American, white-American, our meta-analysis, we coded 31 (67.4%) as
and biracial students (Constantine & involving “poor” expectancy measurement.
Arorash, 2001). Problems included, but were not limited
Religion has also predicted treatment to, the use of 1-item scales, measures that
expectations. Highly religious married confounded expectancy and another con-
Christian couples (compared with low-to- struct, scales that confounded outcome
moderately religious participants) were more and treatment expectations, measures that
likely to believe that a Christian marital ther- used the same questions for both expected
apist would be more effective than a non- outcome and actual outcome, and the use
Christian therapist (Ripley, Worthington, & of projective measures to assess outcome
Berry, 2001). Other research has found dif- expectations. Third, the positive weighted
ferences between evangelical and nonevan- effect may have been inflated by the imbal-
gelical Christians, and highly and moderately anced reporting of coefficients from only
conservative Christians, in their expectations positive findings (i.e., a publication bias).
Finally, there could be a file drawer prob- Another limitation is that outcome
lem, especially with our excluding disserta- expectations have tended to be viewed as a
tions. However, because expectations were relatively static construct, often assessed at
often not related to a primary, hypothesis- baseline or early treatment only. However,
driven research question, we found that some studies have suggested that expec-
many authors openly reported negative tancies change as patients move beyond
or nonsignificant findings. This was sup- treatment’s early stages. For example,
ported by the symmetrical distribution of Holt and Heimberg (1990) found that
effect sizes across the studies included in patients rated treatments as less credible,
our meta-analysis. Nevertheless, as previ- and expectations for improvement were
ously noted, the relatively small fail-safe N lower when the RTQ was completed fol-
suggests caution in our results. lowing Session 4 compared with the end
In addition to these measurement and of Session 1 in cognitive-behavioral group
statistical limitations, other problems char- therapy. The authors concluded, “Credi-
acterize the outcome expectancy litera- bility and outcome expectancy erode when
ture. As reflected in our review, there are exposed to treatment reality” (p. 214), per-
few data to support a direct causal relation haps as patients become more cynical before
between outcome expectations and favor- experiencing much progress. Others have
able treatment outcomes. The most relevant suggested that prognostic expectaions might
experimental work relates to the use of be too high and unrealistic at treatment’s
pretreatment preparation to improve treat- start, thus requiring time to rework their
ment response. In some of the earliest unrealistic nature (Greer, 1980). Whatever
work, Frank and colleagues developed a the case, it appears that expectations are
pretreatment Role Induction Interview malleable, thus limiting the empirical and
(RII) that addressed (a) the treatment ratio- clinical utility of static assessments of this
nale, (b) the importance of attendance, construct.
(c) patient and therapist expectations for
role behavior, and (d) outcome expecta- Treatment Expectations
tions (Hoehn-Saric et al., 1964). In a con- Unlike outcome expectations, there has
trolled trial where patients either did or been more experimental work on treatment
did not engage in the RII prior to treat- expectations. Since the early RII work dis-
ment, RII patients achieved significantly cussed above, there have been a variety of
greater improvement (on both therapist- interventions used to manipulate patient
and patient-rated measures), had better treatment expectancies, and the research
attendance levels, and engaged in more suggests that they are malleable (Dew &
objectively coded favorable therapy behav- Bickman, 2005; Tinsley et al., 1988). In a
ior than the no-RII controls. However, comprehensive review of manipulation
most subsequent role induction stud- studies on adults involving audiotapes, vid-
ies have focused on socializing patients eotapes, verbal instructions, printed mate-
to treatment and on manipulating their rials, or counseling interviews, Tinsley and
expectations about how therapy will unfold colleagues (1988) found significant changes
and the role that they should expect to in treatment expectancies in about 50% of
play. Manipulation studies specifically the studies. One study found that present-
attempting to heighten patients’ prog- ing a credible treatment rationale helps to
nostic outcome expectations are virtually generate positive expectancies about the ther-
nonexistent. apy, and that more positive expectancies are
found when the therapy is based on scien- can be more general, such as “It makes
tific research, tested in clinical trials, and sense that you sought treatment for your
new in relation to other therapies (Kazdin problems” or “Your problems are exactly
& Krouse, 1983). There have also been the type for which this therapy can be
role induction studies using a pre-post con- of assistance” (Constantino, Klein, &
trol group design to try to enhance patient Greenberg, 2006). The therapist can also
engagement and to address misperceptions express confidence and competence in
about treatment, and many of these have such statements as, “I am confident that
been shown to improve retention and working together we can deal effectively
compliance (Dew & Bickman, 2005; Katz with your depression,” while maintaining
et al., 2004; Walitzer et al., 1999). a sense of understanding that the patient
However, many manipulation studies might not fully believe this statement at
are fraught with methodological problems, the outset.
including lack of random assignment, use • Personalize expectancy-enhancing
of analog participants in a laboratory set- statements based on patient experiences
ting, and the use of expectancy instruments or strengths. For example, a therapist can
with unknown or suspect psychometric state, “You have already conquered two
properties (Tinsley et al., 1988). Moreover, major hurdles in admitting to yourself that
most address treatment expectancies at one you have a problem and in seeking help,
time only (often pretreatment). which is not easy to do. This suggests a
motivation and desire to change, despite
Therapeutic Practices any questions you might have about
Outcome Expectations whether you can change.” Or a clinician
Although many therapies include elements might convey, “You strike me as someone
that address various expectations, such who can really accomplish things that
strategies are rarely emphasized or explicit you put your mind to.”
(Greenberg et al., 2006). We offer here sev- • Offer a nontechnical review of the
eral viable clinical strategies. research findings on the intended
treatment. For depressed patients, for
• Explicitly assess patients’ prognostic example, a clinician could say, “Much
expectations at the beginning of research has shown that people in
treatment. Depending on what is revealed cognitive therapy for their depression
(verbally or through a brief measure), tend to get significantly better than
therapists can verify and validate their people who simply try to deal with
patients’ beliefs and consider behaving their problems on their own.”
in a way that matches patients’ level of • While articulating such outcome
optimism. perspectives, do some foreshadowing
• Tread lightly and empathically in about the process of change. Using the
using strategies to enhance outcome same example, the likelihood of small
expectations. Make a concerted effort setbacks or fluctuations in mood can
to use hope-inspiring statements that be normalized, highlighting that change
neither too quickly threaten a patient’s is often gradual and nonlinear.
belief system or sense of self (Pinel & • Regularly check in on patients’
Constantino, 2003), nor promise an outcome expectations and respond
unrealistic degree or speed of change accordingly. For example, if a depressed
(Kirsch, 1990). Rather, such statements patient has developed unrealistically high
expectations after just a few sessions, the (Richert, 1983). These observations and
therapist should not only provide positive subsequent negotiations might include
feedback to reinforce self-efficacy but also trying to draw out a passive patient and
remind the patient that depression can giving a ballpark figure of the treatment
be recurrent, thus bringing expectations length. The therapist can process decisions
more in line with the nature of the with the patient, thereby providing a
disorder. On the other hand, if a patient rationale, and perhaps setting the
expresses diminished hope, the clinician foundation for a corrective experience.
could help him or her retrieve past
successes to at least partially bolster
future-oriented inspiration (in the References
context of empathy and validation of Abouguendia, M., Joyce, A. S., Piper, W. E., &
Ogrodniczuk, J. S. (2004). Alliance as a media-
the current demoralization). tor of expectancy effects in short-term group
psychotherapy. Group Dynamics: Theory, Research,
and Practice, 8, 3–12.
Treatment Expectations Ægisdóttir, S., & Gerstein, L. (2000). Icelandic and
• Heed the potential value of American students’ expectations about counsel-
preparatory work and socialization to ing. Journal of Counseling & Development, 78,
44–53.
treatment, especially for patients who are
Ægisdóttir, S., Gerstein, L., & Gridley, B. (2000).
inexperienced with therapy. Although The factorial structure of the Expectations
certain treatment expectations could be About Counseling Questionnaire-Brief Form:
diagnostic (e.g., “I expect my therapist to Some serious questions. Measurement and
tell me exactly what to do because I am Evaluation in Counseling and Development, 33,
not able to do for myself ”), others might 3–20.
Apfelbaum, B. (1958). Dimensions of transference
simply reflect therapy naiveté. In the latter in psychotherapy. Berkeley, CA: University of
case, striving to shape or alter treatment California Press.
expectations via socialization could be Arnkoff, D. B., Glass, C. R., & Shapiro, S. J.
clinically indicated. (2002). Expectations and preferences. In
• When patients manifest treatment J. C. Norcross (Ed.), Psychotherapy relationships
role or process expectations that are that work: Therapists contributions and responsive-
ness to patients (pp. 325–46). New York: Oxford
incompatible with the therapist’s, University Press.
investigate the patient’s perspective, Asch, S. E. (1946). Forming impressions of person-
consider how these may reflect the ality. Journal of Abnormal and Social Psychology,
patient’s underlying dynamics, inform 41, 258–90.
the patient of your own perspective, Aubuchon-Endsley, N. L., & Callahan, J. L. (2009).
enter a process negotiation, and let the The hour of departure: Predicting attrition in
the training clinic from role expectancies.
patient choose if the treatment seems Training and Education in Professional Psychology,
appropriate (Van Audenhove & 3, 120–26.
Vertommen, 2000). Austin, J. T., & Vancouver, J. B. (1996). Goal
• When the patient’s expectations constructs in psychology: Structure, process, and
are not manifest, infer the type of role content. Psychological Bulletin, 120, 338–75.
the patient wants by assessing activity Barrios, F. X., & Karoly, P. (1983). Treatment
expectancy and therapeutic change in treat-
level, deference toward the therapist, ment of migraine headache: Are they related?
information presented or omitted, and Psychological Reports, 52, 59–68.
responses to interventions calling for Basoglu, M., Marks, I. M., Swinson, R. P.,
self-understanding and access to feelings Noshirvani, H., O’Sullivan, G., & Kuch, K. (1994).
Pre-treatment predictors of treatment outcome Clarkin, J. F., & Levy, K. N. (2004). The influ-
in panic disorder and agoraphobia treated with ence of client variables on psychotherapy. In
alprazolam and exposure. Journal of Affective M. J. Lambert (Ed.) Bergin and Garfield’s
Disorders, 30, 123–32. handbook of psychotherapy and behavior change
Belaire, C., & Young, J. (2002). Conservative (5th ed., pp. 194–226). New York: John Wiley
Christians’ expectations of non-Christian coun- and Sons.
selors. Counseling and Values, 46, 175–87. Cohen, J. (1988). Statistical power analysis for
Berzins, J. I. (1971). Revision of Psychotherapy the behavioral sciences (2nd ed.). Mahwah, NJ:
Expectancy Inventory. Unpublished manuscript, Erlbaum.
University of Kentucky, Lexington. Collins, J. F., & Hyer, L. (1986). Treatment expec-
Bleyen, K., Vertommen, H., Vander Steene, G., & tancy among psychiatric inpatients. Journal of
Van Audenhove, C. (2001). Psychometric proper- Clinical Psychology, 42, 562–69.
ties of the Psychotherapy Expectancy Inventory— Connolly Gibbons, M. B., Crits-Christoph, P.,
Revised (PEI-R). Psychotherapy Research, 11, 69–83. de la Cruz, C., Barber, J. P., Siqueland, L., &
Bloch, S., Bond, G., Qualls, B., Yalom, I., & Gladis, M. (2003). Pretreatment expectations,
Zimmerman, E. (1976). Patients’ expectations interpersonal functioning, and symptoms in the
of therapeutic improvement and their outcomes. prediction of the therapeutic alliance across sup-
American Journal of Psychiatry, 133, 1457–60. portive-expressive psychotherapy and cognitive
Bootzin, R. R., & Lick, J. R. (1979). Expectancies therapy. Psychotherapy Research, 13, 59–76.
in therapy research: Interpretive artifact or medi- Constantine, M., & Arorash, T. (2001). Universal-
ating mechanism? Journal of Consulting and diverse orientation and general expectations about
Clinical Psychology, 47, 852–55. counseling: Their relation to college students’
Borkovec, T. D., & Costello, E. (1993). Efficacy multicultural counseling expectations. Journal of
of applied relaxation and cognitive-behavioral College Student Development, 42, 535–44.
therapy in the treatment of generalized anxiety Constantino, M. J., Arnow, B. A., Blasey, C., &
disorder. Journal of Consulting and Clinical Agras, W. S. (2005). The association between
Psychology, 61, 611–619. patient characteristics and the therapeutic alli-
Borkovec, T. D., & Nau, S. D. (1972). Credibility ance in cognitive-behavioral and interpersonal
of analogue therapy rationales. Journal of Behavior therapy for bulimia nervosa. Journal of Consulting
Therapy & Experimental Psychiatry, 3, 257–60. and Clinical Psychology, 73, 203–211.
Borkovec, T. B., Newman, M. G., Pincus, A. L., & Constantino, M. J., Klein, R., & Greenberg, R. P.
Lytle, R. (2002). A component analysis of cogni- (2006). Guidelines for enhancing patient expecta-
tive-behavioral therapy for generalized anxiety tions: A companion manual to cognitive therapy for
disorder and the role of interpersonal problems. depression. Unpublished manuscript.
Journal of Consulting and Clinical Psychology, 70, Crits-Christoph, P., Gibbons, M., Losardo, D.,
288–98. Narducci, J., Schamberger, M., & Gallop, R.
Buwalda, F., & Bouman, T. (2008). Predicting (2004). Who benefits from brief psychody-
the effect of psychoeducational group treat- namic therapy for generalized anxiety disorder?
ment for hypochondriasis. Clinical Psychology & Canadian Journal of Psychoanalysis, 12, 301–24.
Psychotherapy, 15, 396–403. Davis, M., & Addis, M. (2002). Treatment expec-
Calsyn, R., Morse, G., Yonker, R., Winter, J., tations, experiences, and mental health func-
Pierce, K., & Taylor, M. (2003). Client choice tioning predict attrition status in behavioural
of treatment and client outcomes. Journal of medicine groups. Irish Journal of Psychology, 23,
Community Psychology, 31, 339–48. 37–51.
Chambless, D. L., Tran, G. Q., & Glass, C. R. Dearing, R., Barrick, C., Dermen, K., &
(1997). Predictors of response to cognitive- Walitzer, K. (2005). Indicators of client engage-
behavioral group therapy for social phobia. ment: Influences on alcohol treatment satis-
Journal of Anxiety Disorders, 11, 221–40. faction and outcomes. Psychology of Addictive
Clark, D. M., Salkovskis, P. M., Hackmann, A., Behaviors, 19, 71–78.
Wells, A., Ludgate, J., & Gelder, M. (1999). Devilly, G., & Borkovec, T. (2000). Psychometric
Brief cognitive therapy for panic disorder: A ran- properties of the credibility/expectancy ques-
domized controlled trial. Journal of Consulting tionnaire. Journal of Behavior Therapy and
and Clinical Psychology, 67, 583–89. Experimental Psychiatry, 31, 73–86.
Dew, S. E., & Bickman, L. (2005). Client expec- analysis study of the importance of patient
tancies about therapy. Mental Health Services involvement in treatment planning. European
Research, 7, 21–33. Archives of Psychiatry & Neurological Sciences,
Duckro, P., Beal, D., & George, C. (1979). Research 236, 269–75.
on the effects of disconfirmed client role expec- Hardy, G. E., Barkham, M., Shapiro, D. A.,
tations in psychotherapy: A critical review. Reynolds, S., Rees, A., & Stiles, W. B.
Psychological Bulletin, 86, 260–75. (1995). Credibility and outcome of cognitive-
Evans, R. L., Smith, K. M., Halar, E. M., & behavioural and psychodynamic-interpersonal
Kiolet, C. L. (1985). Effect of expectation and therapy. British Journal of Clinical Psychology,
level of adjustment on treatment outcome. 34, 555–69.
Psychological Reports, 57(3, Pt. 1), 936–38. Heckhausen, H., & Leppmann, P. (1991).
Frank, J. D. (1961). Persuasion and healing: A com- Motivation and action. New York: Springer-
parative study of psychotherapy. Baltimore: The Verlag Publishing.
Johns Hopkins Press. Heine, R. W., & Trosman, H. (1960). Initial expecta-
Garfield, S. L., & Wolpin, M. (1963). Expectations tions of the doctor-patient interaction as a factor
regarding psychotherapy. Journal of Nervous and in continuance in psychotherapy. Psychiatry, 23,
Mental Disease, 137, 353–62. 275–78.
Gaudiano, B., & Miller, I. (2006). Patients’ expectan- Higginbotham, H. N. (1977). Culture and the
cies, the alliance in pharmacotherapy, and treat- role of client expectancy in psychotherapy.
ment outcomes in bipolar disorder. Journal of In R.W. Brislin (Ed.), Topics in culture learning
Consulting and Clinical Psychology, 74, 671–76. (Vol. 5, pp. 107–24). Honolulu, HI: East-West
Ghosh, A., Marks, I. M., & Carr, A. C. (1988). Center.
Therapist contact and outcome of self-exposure Hochberg, M. G. (1986). Client-therapist role expec-
treatment for phobias: A controlled study. British tations and psychotherapy duration. Dissertation
Journal of Psychiatry, 152, 234–38. Abstracts International, 46(10), 3595–3596B.
Goldfarb, D. (2002). College counseling center Hoehn-Saric, R., Frank, J., Imber, S., Nash, E.,
clients’ expectations about counseling: How they Stone, A., & Battle, C. (1964). Systematic
relate to depression, hopelessness, and actual-ideal preparation of patients for psychotherapy: I.
self-discrepancies. Journal of College Counseling, Effects of therapy behavior and outcome. Journal
5, 142–52. of Psychiatric Research, 2, 267–81.
Goldfried, M. R. (1980). Toward the delinea- Holt, C. S., & Heimberg, R. G. (1990). The
tion of therapeutic change principles. American Reaction to Treatment Questionnaire: Measuring
Psychologist, 35, 991–99. treatment credibility and outcome expectancies.
Goldstein, A. P. (1960). Therapist and client expec- the Behavior Therapist, 13, 213–214, 222.
tation of personality change in psychotherapy. Jenkins, S. J., Fuqua, D. R., & Blum, C. R. (1986).
Journal of Counseling Psychology, 7, 180–84. Factors related to duration of counseling in a
Goldstein, A. P. (1962). Therapist-patient expectan- university counseling center. Psychological Reports,
cies in psychotherapy. New York: Pergamon Press. 58, 467–72.
Goossens, M. E. J. B., Vlaeyen, J. W. S., Hidding, A., Joyce, A. S., McCallum, M., Piper, W. E., &
Kole-Snijders, A., & Evers, S. M. A. A. (2005). Ogrodniczuk, J. S. (2000). Role behavior expec-
Treatment expectancy affects the outcome of tancies and alliance change in short-term indi-
cognitive-behavioral interventions in chronic vidual psychotherapy. Journal of Psychotherapy
pain. Clinical Journal of Pain, 21, 18–26. Practice and Research, 9, 213–25.
Greenberg, R. P., Constantino, M. J., & Bruce, N. Joyce, A. S., Ogrodniczuk, J. S., Piper, W. E., &
(2006). Are expectations still relevant for McCallum, M. (2003). The alliance as media-
psychotherapy process and outcome? Clinical tor of expectancy effects in short-term individ-
Psychology Review, 26, 657–78. ual therapy. Journal of Consulting and Clinical
Greer, F. L. (1980). Prognostic expectations and Psychology, 71, 672–79.
outcome of brief therapy. Psychological Reports, Joyce, A. S., & Piper, W. E. (1998). Expectancy,
46, 973–74. the therapeutic alliance, and treatment out-
Hansson, L., & Berglund, M. (1987). Factors influ- come in short-term individual psychotherapy.
encing treatment outcome and patient satisfac- Journal of Psychotherapy Practice and Research, 7,
tion in a short-term psychiatric ward: A path 236–48.
Kamin, I., & Caughlan, J. (1963). Subjective expe- look at an early study. Research Communica-
riences of outpatient psychotherapy. American tions in Psychology, Psychiatry & Behavior, 1,
Journal of Psychotherapy, 17, 660–68. 367–379.
Karzmark, P., Greenfield, T., & Cross, H. (1983). Mathews, A. M., Johnston, D. W., Lancashire, M.,
The relationship between level of adjustment Munby, M., Shaw, P. M., & Gelder, M. G. (1976).
and expectations for therapy. Journal of Clinical Imaginal flooding and exposure to real phobic
Psychology, 39, 930–32. situations: Treatment outcome with agoraphobic
Katz, E. C., Brown, B. S., Schwartz, R. P., patients. British Journal of Psychiatry, 129, 362–71.
Weintraub, E., Barksdale, W., & Robinson, R. McConaghy, N., Armstrong, M., & Blaszczynski, A. P.
(2004). Role induction: A method for enhanc- (1985). Expectancy, covert sensitization and
ing early retention in outpatient drug-free imaginal desensitization in compulsive sexuality.
treatment. Journal of Consulting and Clinical Acta Psychiatrica Scandinavica, 72, 176–87.
Psychology, 72, 227–34. Meyer, B., Pilkonis, P. A., Krupnick, J. L.,
Kazdin, A. E., & Krouse, R. (1983). The impact Egan, M. K., Simmens, S. J., & Sotsky, S. M.
of variations in treatment rationales on expec- (2002). Treatment expectancies, patient alli-
tancies for therapeutic change. Behavior Therapy, ance, and outcome: Further analyses from the
14, 657–71. National Institute of Mental Health Treatment
Kelley, H. H. (1950). Warm-cold variable in first of Depression Collaborative Research Program.
impressions of persons. Journal of Personality, 18, Journal of Consulting and Clinical Psychology, 70,
431–39. 1051–55.
Kirsch, I. (1985). Response expectancy as a deter- Moene, F., Spinhoven, P., Hoogduin, K., & Dyck, R.
minant of experience and behavior. American (2003). A randomized controlled clinical trial
Psychologist, 40, 1189–1202. of a hypnosis-based treatment for patients with
Kirsch, I. (1990). Changing expectations: A key conversion disorder, motor type. International
to effective psychotherapy. Pacific Grove, CA: Journal of Clinical and Experimental Hypnosis,
Brooks/Cole. 51, 29–50.
Lax, T., Basoglu, M., & Marks, I. M. (1992). Mueller, M., & Pekarik, G. (2000). Treatment dura-
Expectancy and compliance as predictors tion prediction: Client accuracy and its relation-
of outcome in obsessive-compulsive disorder. ship to dropout, outcome, and satisfaction.
Behavioural Psychotherapy, 20, 257–66. Psychotherapy: Theory, Research, Practice, Training,
Lick, J., & Bootzin, R. (1975). Expectancy factors in 37, 117–123.
the treatment of fear: Methodological and theo- Noble, L. M., Douglas, B. C., & Newman, S. P.
retical issues. Psychological Bulletin, 82, 917–31. (2001). What do patients expect of psychologi-
Lipkin, S. (1954). Clients’ feelings and attitudes in cal services? A systematic and critical review of
relation to the outcome of client-centered therapy. empirical studies. Social Science & Medicine, 52,
Psychological Monographs, 68(1, No. 372), 1–30. 985–98.
Lipsey, M. W., & Wilson, D. B. (2001). Practical Oliver, J., Hart, B., Ross, M., & Katz, B. (2001).
meta-analysis. Thousand Oaks, CA: Sage Healthy perfectionism and positive expectations
Publications, Inc. about counseling. North American Journal of
Lorentzen, S., & H glend, P. (2004). Predictors Psychology, 3, 229–42.
of change during long-term analytic group psy- O’Malley, S. S., Foley, S. H., Rounsaville, B. J.,
chotherapy. Psychotherapy and Psychosomatics, 73, Watkins, J. T., Sotsky, S. M., Imber, S. D., et al.
25–35. (1988). Therapist competence and patient out-
MacNair-Semands, R. (2002). Predicting atten- come in interpersonal psychotherapy of depres-
dance and expectations for group therapy. sion. Journal of Consulting and Clinical Psychology,
Group Dynamics: Theory, Research, and Practice, 56, 496–501.
6, 219–228. Patterson, C. L., Uhlin, B., & Anderson, T. (2008).
Martin, P. J., & Sterne, A. L. (1975). Prognostic Clients’ pretreatment counseling expectation as
expectations and treatment outcome. Journal of predictors of the working alliance. Journal of
Consulting and Clinical Psychology, 43, 572–76. Counseling Psychology, 55, 528–34.
Martin, P. J., Sterne, A. L., Claveaux, R., & Pekarik, G., & Wierzbicki, M. (1986). The relation-
Acree, N. J. (1976). Significant factors in the ship between clients’ expected and actual treat-
expectancy-improvement relationship: A second ment duration. Psychotherapy, 23, 532–34.
Persson, G., & Nordlund, C. L. (1983). Expectations of treatment outcome. Psychotherapy Research,
of improvement and attitudes to treatment pro- 18, 481–94.
cesses in relation to outcome with four treatment Secord, P. F. (1958). Facial features and infer-
methods for phobic disorders. Acta Psychiatrica ence processes in interpersonal perception. In
Scandinavica, 68, 484–93. R. Taguiri & L. Petrillo (Eds.), Person percep-
Pinel, E. C., & Constantino, M. J. (2003). Putting tion and interpersonal behavior (pp. 300–315)
self psychology to good use: When social psy- Stanford, CA: Stanford University Press.
chologists and clinical psychologists unite. Shapiro, D. A. (1981). Comparative credibility of
Journal of Psychotherapy Integration, 13, 9–32. treatment rationales: Three tests of expectancy
Price, M., Anderson, P., Henrich, C., & Rothbaum, B. theory. British Journal of Clinical Psychology, 21,
(2008). Greater expectations: Using hierarchi- 111–22.
cal linear modeling to examine expectancy for Shaw, H. (1977). A simple and effective treatment
treatment outcome as a predictor of treatment for flight phobia. British Journal of Psychiatry,
response. Behavior Therapy, 39, 398–405. 130, 229–32.
Richert, A. (1983). Differential prescription for Smeets, R. J. E. M., Beelen, S., Goossens, M. E. J. B.,
psychotherapy on the basis of client role prefer- Schouten, E. G. W., Knottnerus, A., &
ences. Psychotherapy: Theory, Research & Practice, Vlaeyen, J. W. S. (2008) Treatment expectancy
20, 321–329. and credibility are associated with the outcome
Richert, A. J. (1976). Expectations, experiencing of both physical and cognitive-behavioral treat-
and change in psychotherapy. Journal of Clinical ment in chronic low back pain. Clinical Journal
Psychology, 32, 438–44. of Pain, 24, 305–315.
Rickers-Ovsiankina, M. A., Geller, J. D., Berzins, J. I., Sotsky, S. M., Glass, D. E., Shea, M. T., Pilkonis, P. A.,
& Rogers, G. W. (1971). Patient’s role expec- Collins, J. F., Elkin, I., et al. (1991). Patient pre-
tancies in psychotherapy: A theoretical and dictors of response to psychotherapy and phar-
measurement approach. Psychotherapy: Theory, macotherapy: Findings in the NIMH Treatment
Research and Practice, 8, 124–126. of Depression Collaborative Research Program.
Ripley, J., Worthington, E., & Berry, J. (2001). The American Journal of Psychiatry, 148, 997–1008.
effects of religiosity on preferences and expec- Spinhoven, P., & ter Kuile, M. (2000). Treatment
tations for marital therapy among married outcome expectancies and hypnotic susceptibil-
Christians. American Journal of Family Therapy, ity as moderators of pain reduction in patients
29, 39–58. with chronic tension-type headache. International
Rosenthal, D., & Frank, J. D. (1956). Psychotherapy Journal of Clinical and Experimental Hypnosis,
and the placebo effect. Psychological Bulletin, 53, 48, 290–305.
294–302. Stern, R., & Marks, I. (1973). Brief and prolonged
Rosenzweig, S. (1936). Some implicit common exposure: A comparison in agoraphobia patients.
factors in diverse methods of psychotherapy. Archives of General Psychiatry, 28, 270–76.
American Journal of Orthopsychiatry, 6, 412–415. ter Kuile, M. M., Spinhoven, P., & Linssen, C. G.
Safren, S. A., Heimberg, R. G., & Juster, H. R. (1995). Responders and nonresponders to
(1997). Clients’ expectancies and their relation- autogenic training and cognitive self-hypnosis:
ship to pretreatment symptomatology and out- Prediction of short- and long-term success in
come of cognitive-behavioral group treatment tension-type headache patients. Headache: The
for social phobia. Journal of Consulting and Journal of Head and Face Pain, 35, 630–36.
Clinical Psychology, 65, 694–98. Tinsley, H. E. A, Bowman, S. L., & Barich, A.W.
Schneider, W., & Klauer, T. (2001). Symptom level, (1993). Counseling psychologists’ perceptions
treatment motivation, and the effects of inpatient of the occurrence and effects of unrealistic
psychotherapy. Psychotherapy Research, 11, 153–67. expectations about counseling and psychother-
Schoenberger, N. E., Kirsch, I., Gearan, P., apy among their clients. Journal of Counseling
Montgomery, G., & Pastyrnak, S. L. (1997). Psychology, 40, 46–52.
Hypnotic enhancement of a cognitive behavioral Tinsely, H. E. A., Bowman, S. L., & Ray, S. B. (1988).
treatment for public speaking anxiety. Behavior Manipulation of expectancies about counseling
Therapy, 28, 127–40. and psychotherapy: Review and analysis of expec-
Schulte, D. (2008). Patients’ outcome expectancies tancy manipulation strategies and results. Journal
and their impression of suitability as predictors of Counseling Psychology, 35, 99–108.
Tinsley, H. E. A., & Westcot, A. M. (1990). Analysis Vannicelli, M., & Becker, B. (1981). Prediction
of the cognitions stimulated by the items on the outcome in treatment of alcoholism: A study of
Expectations About Counseling—Brief Form: staff and patients. Journal of Studies on Alcohol,
An analysis of construct validity. Journal of 42, 938–50.
Counseling Psychology, 37, 223–26. Walborn, F. S. (1996). Process variables: Four
Tinsley, H. E. A., Workman, K. R., & Kass, R. A. common elements of counseling and psychotherapy.
(1980). Factor analysis of the domain of Pacific Grove, CA: Brooks/Cole.
client expectancies about counseling. Journal of Walitzer, K. S., Dermen, K. H., & Conners, G. J.
Counseling Psychology, 27, 561–70. (1999). Strategies for preparing clients for treat-
Tollinton, H. (1973). Initial expectations and ment: A review. Behavior Modification, 23, 129–51.
outcome. British Journal of Medical Psychology, Weinberger, J., & Eig, A. (1999). Expectancies:
46, 251–57. The ignored common factor in psychotherapy.
Turton, D. (2004). Expectations of counselling: In I. Kirsch (Ed.), How expectancies shape experi-
A comparison between evangelical Christians and ence (pp. 357–82). Washington, DC: American
non-evangelical Christians. Pastoral Psychology, Psychological Association.
52, 507–517. Wenzel, A., Jeglic, E., Levy-Mack, H., Beck, A., &
Van Audenhove, C., & Vertommen, H. (2000). Brown, G. (2008). Treatment attitude and ther-
A negotiation approach to intake and treatment apy outcome in patients with borderline person-
choice. Journal of Psychotherapy Integration, 10, ality disorder. Journal of Cognitive Psychotherapy,
287–99. 22, 250–57.
Van Minnen, A., Arntz, A., & Keijsers, G. P. J. Westra, H. A., Dozois, D. J., & Marcus, M. (2007).
(2002). Prolonged exposure in patients with Expectancy, homework compliance, and initial
chronic PTSD: Predictors of treatment outcome change in cognitive-behavioral therapy for anxi-
and dropout. Behaviour Research and Therapy, ety. Journal of Consulting and Clinical Psychology,
40, 439–57. 75, 363–73.
C HA P TER
19 Attachment Style
Kenneth N. Levy, William D. Ellison, Lori N. Scott, and Samantha L. Bernecker
Attachment style or organization is a to become a reliable and trustworthy com-

concept that derives from John Bowlby’s panion in the patient’s exploration of his
attachment theory and refers to a person’s or her experiences. According to Bowlby
characteristic ways of relating in intimate (1988), secure attachment behaviors in
caregiving and receiving relationships, par- psychotherapy include the use of the thera-
ticularly with one’s parents, children, and pist as a secure base from which the indi-
romantic partners. From an attachment vidual can freely reflect on his or her
perspective, these individuals are called experience, reflect on the possible contents
attachment figures. The concept of attach- of the minds of significant others, and
ment style involves one’s confidence in the explore the possibility of trying new experi-
availability of the attachment figure so as to ences and engaging in novel behaviors.
use that person as a secure base from which Additionally, Bowlby discussed patients
the individual can freely explore the world turning to the therapist as a safe haven
when not in distress, as well as the use of for comfort and support in times of dis-
this attachment figure as a safe haven from tress. A number of clinical theorists have
which the individual seeks support, protec- elaborated upon Bowlby’s ideas about the
tion, and comfort in times of distress. function of attachment within the thera-
Exploration of the world includes not only peutic relationship (e.g., Farber, Lippert, &
the physical world but also the examina- Nevas, 1995; Farber & Metzger, 2009;
tion of relationships with other people and Obegi, 2008).
the capacity for reflection on one’s internal The association between adult attach-
experience. ment and psychotherapy has been concep-
From its inception, John Bowlby (1982) tualized and examined both with attachment
conceptualized attachment theory as guid- as an outcome variable and attachment as
ing clinical practice. Consistent with this a moderator of treatment outcome. Early
idea, there has been increased interest in findings from this body of research suggest
the application of an attachment theory that patient attachment status may be rele-
perspective to psychotherapy (see Berant vant to the course and outcome of psycho-
& Obegi, 2009; Levy & Kelly, 2009, for therapy and may also change as a result
reviews). Bowlby not only suggested that of psychotherapy. A recent review of this
the psychotherapist can become an attach- literature (Berant & Obegi, 2009) con-
ment figure for the client, but he also cluded that securely attached clients tend
thought it was important for the therapist to benefit more from psychotherapy than
377
insecurely attached clients. However, the smiling, monitoring caregivers, and develop-
findings across these studies have been vari- ing a preference for a few reliable attachment
able, with some studies suggesting that figures) is part of a functional biological
securely attached clients may not necessar- system that increases the likelihood of
ily show more improvement in treatment protection from dangers and predation,
compared with insecurely attached clients comfort during times of stress, and social
(Cyranowski et al., 2002; Fonagy et al., learning. In fact, the fundamental survival
1996). In addition, the strength of the rela- gain of attachment lies not only in eliciting
tion between attachment security and treat- a protective caregiver response, but also in
ment outcome remains unclear. the experience of psychological contain-
This chapter will focus on what is known ment of aversive affect states required for
about the relation between clients’ attach- the development of a coherent and symbol-
ment styles and their success in psychother- izing self (Fonagy, 1999).
apy. First, we will review definitions and The caregiver’s reliable and sensitive pro-
measurement of attachment and provide vision of loving care is believed to result in
clinical examples of attachment patterns what Bowlby called a secure bond between
in psychotherapy. Second, in order to draw the infant and the caregiver. This attach-
an overall conclusion about the relation ment security is conceptualized as deriving
between attachment and treatment out- from repeated transactions with primary
come, we will present an original meta- caregivers, through which the infant is
analysis of the research on the association believed to form internal working models
between clients’ pretreatment attachment (IWMs) of attachment relationships. These
style/organization and psychotherapy out- IWMs include expectations, beliefs, emo-
come. We conclude with limitations of the tional appraisals, and rules for processing
extant research and therapeutic practices or excluding information. They can be
based on the meta-analytic findings. partly conscious and partly unconscious
and need not be completely consistent or
Definitions and Measures coherent. IWMs are continually elaborated;
In developing attachment theory, John with development, they organize personal-
Bowlby turned to a combination of scien- ity and subsequently shape thoughts, feel-
tific disciplines, including psychoanalysis, ings, and behaviors in future relationships.
ethology, cognitive psychology, and devel- Thus, differences in caregiver behavior
opmental psychology, which provided an result in differences in infants’ IWMs,
array of concepts that could explain affec- which in turn are the basis for individual
tive bonding between infants and their differences in the degree to which relation-
caregivers. Bowlby’s theory concerned ships are characterized by security.
both the short-term effects of this relation- Based on Bowlby’s attachment theory,
ship for a sense of felt security and affect Ainsworth and her colleagues (Ainsworth
regulation and the long-term effects of et al., 1978) developed a laboratory method
early attachment experiences on personal- called the Strange Situation in order to eval-
ity development, relationship functioning, uate individual differences in attachment
and psychopathology. He conceptualized security. The Strange Situation involves a
human motivation in terms of behavioral series of short laboratory episodes staged in
systems, a concept borrowed from ethol- a playroom through which the infant, the
ogy, and noted that attachment-related caregiver, and a stranger interact in a com-
behavior in infancy (e.g., clinging, crying, fortable setting and the behaviors of the
infant are observed. Ainsworth and coll- Bost, 1999, for reviews). Temperament
eagues paid special attention to the infant’s may affect the manner in which attachment
behavior upon reunion with the caregiver security is expressed, but temperament does
after a brief separation. Ainsworth (Ainsworth not affect the security of the attachment
et al., 1978) identified three distinct pat- itself (Belsky & Rovine, 1987). For exam-
terns or styles of attachment that have since ple, research has shown that both behavior-
been termed secure (63% of the dyads ally inhibited and temperamentally fearful
tested), anxious-resistant or ambivalent infants are frequently securely attached
(16%), and avoidant (21%). and engage in both secure-base and safe-
In the Strange Situation, secure infants haven behaviors (e.g., Gunnar et al., 1996;
can find the brief separation from the care- Stevenson-Hinde & Marshall, 1999). More
giver and the entrance of the stranger to be importantly, Ainsworth’s original work has
upsetting, but they approach the caregiver been replicated and extended in hundreds
upon his or her return for support, calm of studies with thousands of infants and
quickly upon the caregiver’s return, are toddlers (see review by Fraley, 2002).
easily soothed by the caregiver’s presence, Studies have found strong evidence for the
and go back to exploration without fuss. In influence of attachment patterns on later
contrast, anxious-resistant infants tend to adaptation as well as remarkable continuity
become extremely distressed upon the care- in attachment patterns over time.
giver’s departure, and they ambivalently A growing body of research (e.g.,
approach the caregiver for attention and Grossmann, Grossmann, & Waters, 2005;
comfort upon the caregiver’s return. They Waters et al., 2000) examining attachment
are clingy and dependent, often crying, but continuity suggests that patterns of attach-
they also seem angry and resist their care- ment are both relatively stable over long
giver’s efforts to soothe them. Avoidantly periods of time and subject to change,
attached infants frequently act unfazed or influenced by a variety of factors including
unaware of the caregiver’s departure and ongoing relationships with family mem-
often avoid the caregiver upon reunion. bers, new romantic relationships, traumatic
Sometimes, these infants appear shut down life events, and possibly psychotherapy
and depressed, and at other times, indiffer- (Fraley, 2002; Ricks, 1985; Shaver, Hazan,
ent and overinvested in play (although & Bradshaw, 1988). These findings are
the play has a rote quality rather than a consistent with Bowlby’s (1982) idea that
rich symbolic quality). Despite their out- attachment theory was not limited to
ward appearance of calmness and uncon- infant–parent relationships. He contended
cern, research has shown that avoidant that the attachment system remains active
infants are quite distressed in terms of throughout the life span, from the cradle to
physiological responding, similar to the the grave.
anxious-resistant babies (Sroufe & Waters, Stemming from Bowlby’s contention
1977). that the attachment system remains active
Despite the obvious resemblance of these throughout the life span, various investiga-
patterns to temperament types (Kagan, tors in the mid-1980s began to apply
1998), and consistent with Bowlby’s hypoth- the tenets of attachment theory to the
eses, these attachment behaviors in the study of adult behavior and personality.
Strange Situation experiment are not simply Because these investigators worked inde-
a result of infant temperament (Belsky, Fish, pendently, they often used slightly different
& Isabella, 1991; see Levy, 2005; Vaughn & terms for similar constructs or focused on
l ev y, e l l i s o n , s cot t, b e r n e c k e r 379
different aspects of Bowlby and Ainsworth’s sort in order to identify the three organized
writings. attachment categories. One notable disad-
Mary Main and her colleagues devel- vantage of the Q-set is that there is no
oped the Adult Attachment Interview (AAI; rating for a disorganized attachment dimen-
George, Kaplan, & Main, 1985; Main, sion, nor can it identify the cannot classify
Kaplan, & Cassidy, 1985), a 1-hour attach- category.
ment history interview, noting that features In contrast to Main’s focus on relation-
in interviews with parents reliably predicted ships with parents, Hazan and Shaver
the Strange Situation behavior of their chil- (Hazan & Shaver, 1987, 1990; Shaver,
dren. The interview inquires into “descrip- Hazan, & Bradshaw, 1988), from a social-
tions of early relationships and attachment psychological perspective, extrapolated the
and adult personality” by probing for both childhood attachment paradigm to study
specific corroborative and contradictory attachment in adulthood by conceptualizing
memories of parents and the relationship romantic love as an attachment process.
with parents (Main et al., 1985, p. 98). They translated Ainsworth’s secure, avoidant,
Three major patterns of adult attachment and anxious-ambivalent attachment pat-
were initially identified: secure/autonomous, terns into a paper-and-pencil prototype-
dismissing, and enmeshed/preoccupied. More matching measure of adult attachment styles
recently, two additional categories have (preferring the term anxious-ambivalent
been identified: unresolved and cannot to anxious-resistant). Several other research-
classify. The first three categories parallel ers have altered and extended the original
the attachment classifications originally Hazan and Shaver measure by breaking
identified in childhood of secure, avoidant, out the sentences in the prototypes into
and anxious-resistant (Ainsworth, Blehar, separate items. Factor analyses of these
Waters, & Wall, 1978), and the unresolved multi-item measures found a three-factor
classification parallels a pattern Main later solution (desire for closeness, comfort with
described in infants that she called disorga- dependency, and anxiety about abandon-
nized/disoriented (Main & Solomon, 1986). ment; Collins & Read, 1990) as well as
A number of studies found that AAI clas- a two-factor solution (desire for closeness
sifications based on individuals’ reports of and anxiety about abandonment; Simpson,
interactions with their own parents could 1990).
predict their children’s Strange Situation A number of empirical studies using
classifications (see van IJzendoorn, 1995, Hazan and Shaver’s (1987) measure or
for a review). derivative measures of adult attachment
A 100-item Adult Attachment Q-set was have found that the distribution of adult
derived from the AAI scoring system and attachment styles is similar to those found
has been applied to AAI transcripts (Kobak for infants. Approximately 55% of indi-
et al., 1993). This system identifies secure, viduals are classified as secure, 25% as
preoccupied, and dismissing categories avoidant, and 20% as anxious (see reviews
based on ratings of two dimensions: security by Shaver & Clark, 1994, and Shaver &
vs. anxiety and deactivation vs. hyperactiva- Hazan, 1993).
tion. Hyperactivating emotional strategies In an important development,
are typical of preoccupied individuals, Bartholomew (1990; Bartholomew &
whereas deactivitating strategies are typical Horowitz, 1991) revised Hazan and Shaver’s
of dismissing individuals. Scores are com- three-category classification scheme, propos-
pared to a criterion or “ideal” prototype ing a four-category model that differentiated
between two types of avoidant styles— Brennan, Clark and Shaver (1998) created
fearful and dismissing. Bartholomew’s key the Experiences in Close Relationships
insight was an incongruity between Main’s (ECR) scale, which was derived from a
(Main & Goldwyn, 1998) and Hazan and factor analysis of 60 attachment constructs
Shaver’s conceptions of avoidance. Main’s representing 482 items extracted from a
prototype of the adult avoidant style thorough literature search of measures used
(assessed in the context of parenting) is in and developed for previous attachment
more defensive, denial oriented, and overtly research. The ECR factor structure was
unemotional than Hazan and Shaver’s consistent with Bartholomew and Horowitz’s
avoidant romantic attachment prototype, measure but showed stronger relations with
which seems more vulnerable, conscious of other relevant constructs. Two short forms
emotional pain, and “fearful.” Thus, Main’s of the ECR have also been published
avoidant style is predominantly dismissing, (Fraley, Waller, & Brennan, 2000; Wei
whereas Hazan and Shaver’s avoidant style et al., 2007), with both highly related to
is predominantly fearful. Consistent with the original ECR.
Bowlby’s theory, Bartholomew’s four cate-
gories could be arrayed in a two-dimensional Measures Used in Studies
space, with one dimension being model of in Our Meta-Analysis
self (positive vs. negative) and the other Because research groups have approached
being model of others (positive vs. negative). the conceptualization and assessment of
For secure individuals, models of self and adult attachment patterns with emphasis
others are both generally positive. For pre- on different aspects of Bowlby’s writings,
occupied or anxious-ambivalent individu- researchers have often identified slightly
als, the model of others is positive (i.e., different patterns or used different names
relationships are attractive) but the model for the same dimensions. The measures
of self is not. For dismissing individuals, described below are those used in the stud-
the reverse is true: the somewhat defen- ies included in our meta-analysis.
sively maintained model of self is positive, Adult Attachment Prototype Rating
whereas the model of others is not (i.e., (AAPR; Pilkonis, 1988) is a set of 88 items
intimacy in relationships is regarded with on which an interviewer rates an individu-
caution or avoided). Fearful individuals al’s attachment style. The rating system
have relatively negative models of both self focuses on two dimensions, each with a
and others. Bartholomew also developed number of facets. On the excessive depen-
an interview measure of attachment along dency dimension, which corresponds to
with her self-report measure. The inter- attachment anxiety, responders are com-
view measure, initially referred to as the pared to three prototypes: excessive depen-
Bartholomew Attachment Interview (BAI) dency, borderline features, and compulsive
and later the Family Attachment Interview caregiving. The prototypes on the excessive
(FAI; Bartholomew & Horowitz, 1991), autonomy dimension, which corresponds
covers both relationships with parents (in to attachment avoidance, are defensive sep-
line with the AAI) and relationships with aration, antisocial features, and obsessive-
close friends and romantic partners (in line compulsive features. A secure prototype
with Shaver and Hazan’s work). was later added to the system (Strauss,
In an effort to develop a more definitive Lobo-Drost, & Pilkonis, 1999).
measure of adult attachment and respond Adult Attachment Scale (AAS; Collins &
to the proliferation of attachment measures, Read, 1990) is a self-report instrument
developed by breaking Hazan and Shaver’s attachment prototype. Two underlying
(1987) prototype statements into 21 items. dimensions can be derived either by con-
The number of items in the AAS was later ducting a factor analysis of the items or by
shortened to 18 (Collins, 1996). Individuals using the scores from the four prototype
rate these statements on a 5-point, Likert- items to create linear combinations repre-
type scale (1 = not at all characteristic; 5 = very senting the self- and other-model attach-
characteristic). The subscales include comfort ment dimensions.
with closeness and intimacy (Close), com- Family Attachment Interview (FAI;
fort depending on others (Depend), and Bartholomew & Horowitz, 1991) is a sem-
anxiety about abandonment (Anxiety). istructured interview designed to assess
Responders can be categorized as follows: adult attachment styles based on informa-
those with high Close and Depend scores tion about parents. The probes used in the
and low Anxiety scores are Secure, those interview are remarkably similar to those
with high Anxiety scores and moderate used in the Adult Attachment Interview,
Close and Depend scores are Anxious, and and as such, the FAI scoring system can be
those with low scores on all three subscales used with information generated from the
are Avoidant. There is strong evidence AAI. The FAI scoring is similar to the AAI
throughout the literature for the scale’s reli- in that attachment ratings are based on
ability and validity (Ravitz, Maunder, content of reports as well as reporting style
Hunter, Sthankiya, & Lancee, 2010). (e.g., defensive strategies that emerge during
Relationship Questionnaire (RQ; the interview, coherency of the report).
Bartholomew & Horowitz, 1991) is a self- However, the FAI codes people on four
report questionnaire based on Bartholomew’s attachment styles (secure, fearful, preoccu-
(1990) four-category model of attachment. pied, and dismissing) rather than categoriz-
The RQ consists of four paragraphs describing people into the AAI categories. The
ing each of the attachment prototypes— interviews are coded for each attachment
secure, fearful, preoccupied, and dismissing. pattern on a 9-point scale (1 = no evidence
Participants rate how well each corresponds of characteristics of the prototype; 9 = near
to their romantic relationship pattern, perfect fit with the prototype).
where 1 = not at all like me and 7 = very Attachment Style Questionnaire (ASQ;
much like me. Participants then select the Feeney, Noller, & Hanrahan, 1994) is a
one paragraph that best describes them. 40-item self-report questionnaire rated on a
Relationship Style Questionnaire (RSQ; 6-point, Likert-type scale. It includes sub-
Bartholomew & Horowitz, 1991) contains scales to measure Self-Confidence, Discom-
30 short statements drawn from three other fort with Closeness, Need for Approval,
attachment measures. Participants rate each Preoccupation, and Relationships as Secon-
question on a 5-point Likert scale to indi- dary. The instrument has adequate reliabil-
cate the extent to which each statement ity and has been found to converge with
best describes their characteristic style in other attachment measures and to have pre-
close relationships. Five statements contrib- dictive validity (Ravitz et al., 2010).
ute to the secure and dismissing attachment Reciprocal Attachment Questionnaire (RAQ;
patterns and four statements contribute to West & Sheldon-Keller, 1994) is a 43-item,
the fearful and preoccupied attachment 5-point, Likert-type self-report question-
patterns. Scores for each attachment pat- naire designed to assess nine dimensions of
tern are calculated by taking the mean of adult attachment patterns with significant
the four or five items representing each others. Four pattern subscales—Compulsive
Self-Reliance, Compulsive Care-Giving, Clinical Examples
Compulsive Care-Seeking, and Angry In general, patients with secure attachment
Withdrawal—assess dysfunctional patterns styles have been found to be more collab-
of adult attachment. There are also five orative, receptive, and better able to utilize
attachment dimension subscales: Separa- treatment. In contrast, those with dismis-
tion Protest, Feared Loss, Proximity sive styles have been found to be less
Seeking, and Use and Perceived Availability engaged in treatment. Those with preoccu-
of the attachment figure. The validity and pied states of mind with regard to attach-
reliability of the RAQ have been estab- ment have been found to present as more
lished in both clinical and nonclinical needy in therapy but not necessarily com-
adult populations (West & Sheldon-Keller, pliant with treatment (e.g., Dozier, 1990;
1994). Riggs, Jacobovitz, & Hazen, 2002).
Avoidant Attachment Questionnaire (AAQ;
West & Sheldon-Keller, 1994) is a 22-item, Secure Attachment
5-point, Likert self-report questionnaire Given that secure individuals are more open
developed alongside the RAQ as an alterna- to exploring their surroundings and relation-
tive for individuals who deny having a ships, it is not surprising that evidence sug-
primary attachment figure. The question- gests that persons with autonomous states of
naire assesses four subscales: Maintains mind tend to be open, engaged, collabora-
Distance in Relationships, High Priority on tive, compliant, committed, and proactive in
Self-Sufficiency, Attachment Relationship treatment (Dozier, 1990; Korfmacher, Adam,
is a Threat to Security, and Desire for Close Ogawa, & Egeland., 1997; Riggs et al.,
Affectional Bonds. There is a relative dearth 2002). Although these individuals may
of evidence on its reliability and validity, enter treatment distressed, they tend to be
probably due to the infrequency of its use trusting of therapists. Most importantly,
(Ravitz et al., 2010). they tend to be able to integrate and utilize
Experiences in Close Relationships (ECR; their therapists’ comments. Additionally,
Brennan et al., 1998) is a 36-item, self- anecdotal evidence suggests that they can
report questionnaire that assesses attach- show more gratitude toward the therapist
ment security in close relationships by for providing treatment.
tapping two basic dimensions of attach-
ment organization: anxiety and avoid- Preoccupied Attachment
ance. These two dimensions underlie Because preoccupied individuals can be so
most measures of adult attachment style interpersonally engaged, they often initially
(Brennan et al., 1998) and parallel those appear to be easier to treat. Preoccupied
identified by Ainsworth et al. (1978) as individuals are often so distressed and inter-
underlying patterns of behavior in the personally oriented that they are eager to
Strange Situation. Participants rate the discuss their worries and relationship diffi-
extent to which each item is descriptive culties as well as their own role in these
of their feelings in close relationships on a problems (Dozier, 1990). Because the cha-
7-point scale (1 = not at all to 7 = very otic and contradictory representations of
much). Eighteen items assess attachment self and others of individuals classified as
anxiety and 18 assess attachment avoid- preoccupied are so rich, they may be more
ance. The reliability and validity of the readily and vividly mentalized or repre-
scales have been demonstrated (Brennan sented by the therapist. However, both
et al., 1998). clinical and empirical evidence suggests
that these individuals may be difficult to and highly intelligent, with an Ivy League
treat. In a number of papers, Slade (1999, education, she found herself unable to date
2004) has written about the unique chal- and maintain employment. This was mainly
lenges inherent to working clinically with because, though she was emotionally needy,
preoccupied individuals. She warns that she could not get along with others due to
“Progress is… hard won” (Slade, 1999, frequent angry outbursts. Even at 35, she
p. 588) and that therapists must be pre- was highly dependent on her parents, par-
pared for the “slow creation of structures ticularly for financial support, but also for
for the modulation of affect” (Slade, 1999, emotional support. Her parents were at
p. 586). She contends that change occurs their wits’ end with her and felt she was
over a long period of time from the thera- wasting her life away. Although they were
pist’s long-term emotional availability and traditional and perceived psychotherapy as a
tolerance for chaos. corrupt endeavour practiced by charlatans,
Clients with preoccupied attachment they were willing to pay for psychotherapy.
organization tend to present themselves as The patient’s relationship with her par-
needy but are not more compliant with ents was anchored in two equally uncom-
treatment plans than dismissing individu- fortable extremes that led her to vacillate
als (Dozier, 1990). Those classified as pre- between wanting to live at home and
occupied, as compared with those classified submit to their will, and wanting to break
as dismissive, tend to show less improve- away from their control and become inde-
ment (Fonagy et al., 1996). It is hypothe- pendent and self-reliant. At times, she
sized that the preoccupied patients are more would plead with the therapist in a loud
difficult to treat because their representa- pressured voice, “Dr. X, Dr. X, please,
tional systems are intricately linked with please tell me what to do! Should I try to
emotions that are well-developed and elab- work it out with my parents or should I
orated by entrenched preoccupation with just forget about them?” The patient rap-
difficult events in their lives. This is also idly flipped between desperately wanting to
expressed in terms of their certainty about be close to her parents and feeling as if she
mental states and motivations for others’ could not live without them to wanting to
behaviors. have nothing to do with them. In each of
In our own work, we have found a these stances she would be adamant and
number of difficult aspects related to work- inflexible about her position and then flip
ing with preoccupied individuals that can to the other. She would flip so quickly that
be first identified in the narratives of AAIs. when she was in one mental state she did
These include: (1) unmerited certainty not appear to recall the other mental state.
about mental states; (2) rapid vacillations or However, when she would pose this ques-
oscillations between contradictory mental tion, both mental states were represented
states; (3) current anger and confusion for a brief time.
about time and people; (4) self-blame and In these moments the psychotherapist
derogations. Each of these issues, alone or felt extremely pressured by the patient to
in combination, may leave the therapist provide her with an answer to her quan-
feeling confused and overwhelmed. dary. Any hesitation on the therapist’s part
The following vignette contains aspects was interpreted as withholding valuable
of all four of these issues. The patient was an information from the patient and was met
unmarried 35-year-old woman of Southeast with quick anger. The therapist felt backed
Asian descent. Despite being very attractive into a corner with no good solution.
He did not feel he could simply give the into session one morning and announced,
patient advice. Besides, the solution was to her therapist’s surprise, that she was get-
neither to submit to the parents nor to cut ting married that afternoon. Although he
them off, but rather, to figure out how to had known of her engagement, it had been
have a mutually satisfying relationship with many months since she had brought up
them. He also felt pressured because he any aspect of her upcoming marriage.
realized that these moments where the Additionally, dismissing individuals often
patient had both sides of a conflict repre- become more distressed and confused when
sented were rare and fleeting, and he wanted confronted with emotional issues in ther-
to make use of them, and yet, he was feel- apy (Dozier, Lomax, Tyrell, & Lee, 2001).
ing pressured to answer a question that had Another dismissive patient, when reflecting
no answer and would be both unsatisfying on her experience in therapy, stated:
and infuriating to the patient.
Using his countertransference of being He (the therapist) would start digging into
backed into a corner, the therapist com- things and find out why I was angry, and
mented to the patient that she must feel then I would realize something really made
backed into a corner with no good option me mad, but I didn’t want to be mad.
available. He continued by pointing out With my parents, for example, I didn’t
that if he told her to reconcile with her par- want to be angry at them.
ents, he imagined that she might interpret Finally, therapists working with dismis-
this as if he felt she was wrong, they were sive patients may be pulled into enact-
right, and she should submit to their will ments, where they find themselves in a
and allow herself to be controlled by them. situation analogous to a “chase and dodge”
On the other hand, if he told her to resist sequence with mothers and infants (Beebe
their will, and leave them, she would feel as & Lachmann, 1988), which leaves the
if the therapy was useless, and she would patient feeling intruded upon only to with-
feel terribly abandoned by her parents and draw further. Conversely, those with dis-
more dependent on the therapist. With missing attachment may effectively curtail
both affective states acknowledged, vali- the therapist’s capacity to engage with, visu-
dated in the patient, and tolerated by the alize, or evoke the individual’s representa-
therapist, the patient was able to refrain tional world, or identify with the patient.
from her rapid oscillations long enough to
have a productive discussion and develop a “Unresolved for Trauma or Loss”
more integrated perspective on her situa- Attachment
tion vis-à-vis both her own and her parents’ An individual can be classified as unre-
behaviors. solved on the Adult Attachment Interview
for either loss or trauma experiences. This
Dismissing Attachment classification is unique in that it is given to
Dismissing patients are often resistant to an individual in addition to one of the
treatment, have difficulty asking for help, organized attachment patterns (i.e., secure,
and retreat from help when it is offered preoccupied, or dismissing) and can be
(Dozier, 1990). Indeed, dismissive patients either primary or secondary, depending on
often evoke countertransference feelings of a number of factors. Clinical writers have
being excluded from the patients’ lives suggested that it can be very difficult to
(Diamond et al., 1999, 2003). In our pilot treat those patients who are unresolved for
study, a patient classified as dismissive came trauma or loss on the AAI.
In two studies it was found that between were found first through articles reviewing
32% and 60% of patients with borderline the literature (e.g., Berant & Obegi, 2009)
personality disorder (BPD) were classified and second through a series of PsycINFO
as unresolved (Diamond et al., 2003; Levy searches. These searches, conducted in
et al., 2006). In a randomized clinical trial December 2009, used the intersections of
(Levy et al., 2006), we found a nonsignifi- the terms attachment, interpersonal style,
cant decrease from pretreatment to post- relation∗ style, or the name of an attachment
treatment in the number of patients measure with either therap∗ outcome, psy-
classified as unresolved (32% vs. 22%). chotherap∗ outcome, or outcome. The search
Unpublished data from this trial (Levy, initially returned 10,155 results. After for-
Clarkin, & Kernberg, 2007) suggest that eign-language studies (531), dissertations
those BPD patients who were unresolved (8), and studies that did not include treat-
were more likely to drop out of treatment. ment trials (9,448) were excluded, 168
However, in a small sample of women with articles remained. Many of these were irrel-
childhood sexual and physical abuse-related evant to the topic at hand; only studies that
posttraumatic stress disorder (PTSD), measured attachment and treatment out-
62% of unresolved patients lost their unre- come were included.
solved status following treatment (Stovall- In order to be included in the meta-
McClough & Cloitre, 2003). analyses, studies had to report statistics
showing the relation between patients’ pre-
Meta-Analytic Review treatment attachment security, anxiety, and/
To characterize the relation between adult or avoidance to outcome posttreatment. In
attachment and psychotherapy outcome, order to avoid confounding attachment
we conducted three separate meta-analyses. with therapeutic alliance, reports were not
We hypothesized that attachment anxiety included if the measure of attachment con-
would be negatively related to outcome, cerned client attachment to therapist. For
that attachment avoidance would be nega- many identified studies, statistics describing
tively related to outcome, and that attach- the relation between attachment and out-
ment security would be positively related come were not directly available from the
to outcome. Because research on attach- published report, in which cases the authors
ment is converging on the notion that the of the study were contacted via e-mail and
two dimensions of avoidance and attach- asked to provide these statistics. The corre-
ment underlie adult attachment, we decided sponding authors of 15 primary studies
to focus on them instead of the individ- were contacted, of which 10 responded with
ual attachment categories, which evidence suitable statistics. Our final pool of studies
more variability among assessment meth- analyzed consisted of 14 studies, which
ods. In addition, we examined attachment contained 19 separate therapy samples with
security (which can be conceptualized as a a combined N of 1,467. Table 19.1 lists the
blend of avoidance and anxiety dimen- studies included in the meta-analysis along
sions) because it has often been the focus of with relevant characteristics of their designs
psychotherapy research. and samples.
Inclusion Criteria and Search Strategy Independence of ES Estimates

Eligible studies were published reports of Effect sizes were considered independent if
psychotherapy outcome in samples of treat- they described results from separate sam-
ment-seeking individuals. These studies ples. In one case, relevant information from
Table 19.1 Summary of Studies Included in Meta-Analysis of Patient Attachment and Outcome
Patients Attachment Therapy Outcome
Study N % Female Age (M) Diagnosis Measure Rater Orientation Duration Measure Rater
(weeks)
Cyranowski et al. 162 100 37.6 MDD RQ C D 14 HRSD NT
(2002)
Johnson & 34 0 42 Marital AQ C D 12 DASsatis C
Talitman (1997)
Lawson & Brossart 49 0 31.73 IPV AAS C I 17 Violence C
(2009)
psyabuse C
Levy et al. (2006) 22 95.5 32.27 BPD ECR C D 52 GAF NT
BDI C
SCL-90-R C
15 93.3 32.53 BPD ECR C CB 52 GAF NT
BDI C
SCL-90-R C
23 95.65 28.48 BPD ECR C D 52 GAF NT
BDI C
SCL-90-R C
(Continued )
387
388

Patients Attachment Therapy Outcome
Study N % Female Age (M) Diagnosis Measure Rater Orientation Duration Measure Rater
(weeks)
Marmarosh et al. 31 71 24.6 Unspecified ECR-S C E 15 SCL-90-R C
(2009)
McBride et al. 27 74.1 40.1 MDD RSQ C D 17 BDI C
(2006)
HAM-D NT
28 72.4 41 MDD RSQ C CB 17 BDI C
HAM-D NT
Meyer et al. (2001) 104 57 34.5 PDNOS AAPR T E 14 GAF NT
HRSD NT
HAMA NT
SCL-90-R C
Muller & 101 64 42.8 PTSD RSQ and RQ C D 8 SCL-90-R C
Rosenkranz (2009) (combined)
TSC-40 C
Reis & Grenyer 58 58.6 45.98 MDD RQ C D 16 HRSD NT
(2004)
Saatsi et al. (2007) 82 72.7 34.92 MDD Vignettes C CB 14 BDI C
Stalker et al. (2005) 114 100 40.6 PTSD RAQ C D 6 SCL-90-R C
MPSS-SR C
18 100 40.6 PTSD AAQ C D 6 SCL-90-R C
MPSS-SR C
Strauss et al. (2006) 476 70 34.4 PD AAPR NT D 10 SCL-90-R C
IIP C
Tasca et al. (2006) 33 100 42.75 BED ASQ C CB 16 EDEbinge NT
33 100 42.75 BED ASQ C D 16 EDEbinge NT
Travis et al. (2001) 59 59 41 Unspecified BARS NT D 21 SCL-90-R C
Note: Raters: C = client, NT = nontreater, T = therapist
Orientations: CB = cognitive-behavioral, D = dynamic, E = eclectic, I = integrative
Diagnoses: BED = binge eating disorder, BPD = borderline personality disorder, IPV = intimate partner violence, MDD = major depressive disorder, PD = personality disorder, PDNOS = personality disorder not
otherwise specified, PTSD = post-traumatic stress disorder
Attachment measures: AAPR = Adult Attachment Prototype Rating, AAI = Adult Attachment Interview, AAS = Adult Attachment Scale, AAQ = Avoidant Attachment Questionnaire, AQ = Attachment Questionnaire,
ASQ = Attachment Style Questionnaire, BARS = Bartholomew Attachment Rating Scale, ECR/ECR-R = Experiences in Close Relationships scale/Experiences in Close Relationships–Revised, RAQ = Reciprocal
Attachment Questionnaire, RSQ = Relationship Scales Questionnaire, RQ = Relationship Questionnaire
Outcome measures: BDI = Beck Depression Inventory, DASsatis= satisfaction subscale of the Dyadic Adjustment Scale, EDEbinge = Eating Disorder Examination assessment of days binged, GAF = Global Assessment
of Functioning, HAMA = Hamilton Rating Scale for Anxiety, HAM-D = Six-Item Hamilton Depression Rating Scale, HRSD = Hamilton Rating Scale for Depression, IIP = Inventory of Interpersonal Problems,
MPSS-SR = Modified PTSD Symptom Scale–Self-Report, psyabuse = psychological abuse subscale of the Conflict Tactics Scale, SCL-90-R = Symptom Checklist–90–Revised, TSC-40 = Trauma Symptom Checklist–40,
violence = subscale of the Conflict Tactics Scale
389
a single sample was available from multiple coded for its degree of approximation to
research reports (Kirchmann et al., 2009; attachment avoidance and attachment anx-
Strauss et al., 2006), so only one statistic iety, and attachment measures were coded
was drawn from these reports. In other for rater (client-rated or observer-rated
cases, separate statistics from multiple attachment). Finally, the following thera-
samples (for example, different treatment pist variables were coded: mean years of
groups) were presented in the same publi- experience, proportion of therapists in the
cation (Levy et al., 2006; McBride, study that was female, and student status.
Atkinson, Quilty, & Bagby, 2006; Stalker,
Gebotys, & Harper, 2005; Tasca et al., Effect Size Estimates
2006). For these studies, multiple effect The effect size statistic used for the current
size estimates were coded and treated as meta-analysis was the Pearson product–
independent. Several studies provided sta- moment correlation coefficient (r) describ-
tistics relating attachment to more than ing the relation between attachment
one outcome measure. These estimates were variables and posttreatment outcome mea-
not considered independent because they sures. In some cases, statistics relating
were derived from the same sample and are attachment to outcome took other forms,
thus likely to display substantial intercor- such as means and standard deviations for
relation. Because we had no a priori reason different attachment groups on outcome
to consider any one of these estimates rep- measures, t-tests of these values, or tables
resentative of the study’s “true” effect size, showing categories of outcome (e.g., how
multiple effect size estimates from the same many individuals had achieved a certain
study were transformed to Z-scores (Hedges symptom score) by attachment group. In
& Olkin, 1985), averaged together, and these cases, statistics were transformed to
then back-transformed and treated as a r-values (using formulas presented in Lipsey
single effect size. & Wilson, 2001). Although it would be
optimal to control for pretreatment correla-
Study Coding tions between attachment and symptom
Coding of the 14 studies was conducted scales, this was not feasible because of incon-
by an advanced graduate student. Several sistent reporting among studies. Thus, all
patient characteristics were coded, includ- correlations used in the current analyses
ing the proportion of the sample that was were zero-order correlations between pre-
female, mean age of the sample, proportion treatment attachment and posttreatment
of the sample that was White or Caucasian, outcome.
and whether the primary diagnosis of the The 14 primary studies differed in a
sample was an Axis I disorder (e.g., major number of ways that could be expected to
depressive disorder) or an Axis II disorder impart a systematic bias onto effect size
(e.g., borderline personality disorder). The estimates. Thus, we made two adjustments
treatment characteristics coded included to the statistics reported in the published
theoretical orientation (cognitive-behavioral studies. Both of these adjustments pertain
or psychodynamic therapies) and length of to the operationalization of attachment and
treatment in weeks. Because the 19 samples outcome. First, each study was adjusted to
included in the current study were offered account for differences in operationaliza-
16 different types of psychotherapy, specific tion of attachment. Measures of attach-
type of treatment was not formally coded. ment vary widely, and the 14 studies
The operationalization of attachment was sampled in the current analysis used 11
separate measures. The current analysis shows the correlations between attach-
focuses on attachment security and the ment measures used in the primary studies
underlying attachment dimensions of with attachment anxiety and avoidance
avoidance and anxiety, and when measures from the ECR.
provide an imperfect assessment of these A second correction was applied to
constructs, the resulting effect size estimate account for artificial dichotomization of
is attenuated (Schmidt, Le, & Oh, 2009). attachment dimensions or dimensional
Therefore, each study was corrected to outcome constructs, which also attenuates
account for how closely its attachment effect size estimates (Schmidt et al., 2009),
measure approximated these dimensions especially if the dichotomy produces an
of attachment. In order to do this, each uneven split between groups (Lipsey &
observed effect size was divided by the cor- Wilson, 2001). For example, if outcome is
relation of the attachment measure used in recovery based on a dimensional symptom
the study with the ECR or ECR-R, which score below a certain cutoff, effect size esti-
probably measures attachment anxiety and mates based on the proportion of individu-
attachment avoidance with the most fidel- als in recovered, and nonrecovered groups
ity. These correlation values were culled are distorted when compared with estimates
from the available literature. Figure 19.1 from dimensionally measured variables.
Preoccupation
0.8 Anxiety
Preoccupation Angry
withdrawal Feared loss of partner
Need for approval
0.6
Compulsive Negative self model
careseeking
Proximity-
seeking Separation
0.4 protest
Preoccupied Fearful
Fearfulness
Compulsive Discomfort
caregiving Relationships with closeness
0.2 as secondary
ECR anxiety
Compulsive RQ
Dismissing
self-reliance
RSQ
0
Close Negative other model RAQ
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
Use partner ASQ
as secure base
−0.2 AAS
Self-confidence
Dismissiveness
Secure
Depend −0.4
Security Availability
of partner
−0.6
−0.8
−1
ECR avoidance
Fig. 19.1 Correlations of ECR Anxiety and Avoidance Scales with other self-report measures of adult attachment.
Note: AAS = Adult Attachment Scale (Collins & Read, 1990),ASQ = Attachment Style Questionnaire (Feeney et al., 1994), ECR = Experiences
in Close Relationships scale (Brennan, Clark, & Shaver, 1998), RAQ = Reciprocal Attachment Questionnaire (West & Sheldon-Keller, 1994),
RSQ = Relationship Scale Questionnaire (Griffin & Bartholomew, 1994)
Hunter and Schmidt’s (1990) correction to were tested as moderators of this effect. These
these values was thus applied. To ensure variables (summarized under “Study Coding”)
that more valid estimates contributed more were designated a priori and related to vari-
to the overall mean than estimates for ance at several different levels, including
which these two artifact corrections were sample variables, treatment descriptors,
large, each effect size estimate was weighted operationalization of attachment, and ther-
not only by sample size but was also assigned apist variables. Moderation analyses were
a weight based on the size of the two arti- conducted via weighted least squares regres-
fact corrections (Hunter & Schmidt, 2004; sion in which each effect size estimate was
Schmidt et al., 2009). assigned a weight based on inverse variance
The influence of outliers is also a con- (Lipsey & Wilson, 2001). Fisher’s Zr trans-
cern because the present study involved a formation (Hedges & Olkin, 1985) was
small but heterogeneous sample of primary used for each effect size estimate before
studies. Outliers were detected by means of regression analyses were conducted because
the sample-adjusted meta-analytic deviancy of the problematic standard error formula-
(SAMD; Huffcutt & Arthur, 1995) statis- tion associated with correlation coefficients
tic, which takes into account the fact that (Lipsey & Wilson, 2001). Effect size esti-
smaller samples are more likely to produce mates used in the regression analyses were
deviant estimates of the population effect the attenuated (uncorrected) values; in
due to simple sampling error. The SAMD order to control for the effects of measure
values associated with each of the primary unreliability and artificial dichotomization,
studies were visually inspected in a scree the multiplier values representing these
plot to determine whether any values were artifacts were used as covariates in each
substantially more deviant than would be regression analysis (Borenstein, Hedges,
expected. Higgins, & Rothstein, 2009). Regression
used random-effects modeling estimated
Analyses via iterative maximum likelihood estima-
The mean effect size was computed as a tion (Wilson, 2005).
weighted average of each independent sam-
ple’s correlation coefficient. The weights were Results
composed of two coefficients: the sample The mean weighted r between attachment
size, so that each study’s contribution to anxiety and psychotherapy outcome was
the overall mean would be inversely propor- −.224 (Cohen’s weighted d = −0.460).
tional to sampling error, and a multiplier Outcomes were coded so that higher num-
based on the artifact corrections made to each bers reflected better outcome. Thus, higher
effect size, so that studies that more nearly attachment anxiety predicted worse out-
approximated the constructs of interest were come after therapy. The 80% credibility
weighted more heavily (Hunter & Schmidt, interval around this estimate ranged from
2004; Schmidt et al., 2009). Random-effects −.158 to −.291 (d = −0.320 to −0.608).
modeling was used for each analysis, given Because a random-effects model was used,
the multiple sources of variability between this range refers not to the distribution of
studies and the resultant implausibility of estimates of a single parameter (r values),
fixed-effects models (for which one fixed but to multiple population parameter (that
population of studies is assumed). is, rho) values. Thus, 80% of the parameter
Several likely predictors of the relation- values describing the relation between
ship between attachment and outcome attachment and anxiety lie in this interval.
The mean weighted r between attach- attachment–outcome effect sizes for anxi-
ment avoidance and treatment outcome ety, avoidance, and security, respectively.
was −.014 (d = −0.028), with an 80% cred- Thus, the effect sizes that were combined
ibility interval of −.165 to .136 (d = −0.335 in each of our meta-analyses could be
to 0.275). This suggests that attachment considered fairly homogeneous after arti-
avoidance had a negligible overall effect on factual sources of variance are accounted
outcomes in psychotherapy. for. Nevertheless, an exploratory analysis of
The mean weighted r between attach- potential moderators was conducted.
ment security and outcome was .182 Unfortunately, for a number of the coded
(d = 0.370), with an 80% credibility inter- variables, the effects of moderator variables
val of .042 to .321 (d = 0.084 to 0.678). could not be estimated because data about
Thus, higher attachment security predicted them were not available from the primary
more favorable outcomes in psychotherapy. studies, or because there was not enough
SAMD values were examined to check variance among the primary studies on the
for the presence of outliers among the effect moderator variable. For two examples, the
size estimates. No outliers could be identi- moderating influence of sample ethnicity
fied among the primary studies’ estimates and therapist level of experience could not
of the relation between outcome and be estimated due to insufficient data or
attachment anxiety, avoidance, or security. variability.
Therefore, all values were retained for fur- No moderators were found to influence
ther analyses. the size of the relation between either
attachment avoidance or attachment anxi-
Moderators and Mediators ety and treatment outcome. However, two
For all three attachment dimensions, sample-level moderators did significantly
homogeneity of effect size estimates was influence the effect of attachment security
tested by means of Hunter and Schmidt’s on outcome. Both the proportion of females
(2004) 75% criterion, which estimates the (Z = 2.78, p < .01) and the mean age
amount of variance in effect sizes that is (Z = 2.02, p < .05) of the patients exerted
due to artifacts (such as imperfect validity an effect, such that the more female and
or reliability of the measures used). If this older the sample, the smaller the observed
value is more than 75% of the total vari- relation between security and outcome. We
ance, the authors suggest that a search for suspect that the effect of gender can be
measureable moderators of the effect size explained by one study (Cyranowski et al.,
may be unproductive because the remain- 2002), which included only women and
ing variance in effect sizes is comparatively found the weakest relation between secu-
small. This method was used because rity and outcome. In fact, running the
homogeneity tests based on a null hypoth- analysis without including this study com-
esis of homogeneity (such as the Q statistic) pletely erased the significant gender effect,
would likely have little power given the with a regression coefficient of nearly zero.
small sample of studies in the current meta- Nonetheless, there are gender differences
analyses. In the current study, a substan- in attachment (i.e., studies suggest that
tial portion of the variance in the corrected more men than women demonstrate inse-
effect size estimates was indeed artifac- cure and dismissing attachment styles;
tual. The artifacts for which we corrected Bartholomew & Horowitz, 1991; Levy,
in the three meta-analyses accounted for Blatt, & Shaver, 1998; Levy & Kelly,
89%, 75%, and 82% of the variance in 2010) that could potentially influence
psychotherapy outcome, and this possibil- (Bakermans-Kranenburg, Juffer, & van
ity might be further explored in future IJzendoorn, 1998; McBride et al., 2006;
research. Tasca et al., 2006). There is preliminary
Additionally, client age emerged as a sig- evidence that dismissive/avoidant clients
nificant moderator, such that the positive may benefit more from treatments that
relation between attachment security and focus on cognitions and behaviors rather
outcome was attenuated in samples that than emotionality and relationships, at least
were older on average. This finding may be in short-term psychotherapy. For instance,
explained by cross-sectional research show- one study examined two versions of a short-
ing that older adults are more likely to be term treatment for promoting maternal sen-
securely attached, and less likely to be fear- sitivity and found that insecure preoccupied
fully attached, than younger adults (Diehl, mothers benefited more from an interven-
Elnick, Bourbeau, & Labouvie-Vief, 1998; tion that included both video feedback and
Mickelson, Kessler, & Shaver, 1997). If this discussion of childhood attachment experi-
is a developmental, rather than cohort- ences, whereas dismissive mothers benefited
based, effect, this difference suggests that more from video feedback without such
some preoccupied individuals become discussions (Bakermans-Kranenburg et al.,
secure (perhaps by finding or creating an 1998). In addition, a study examining
intimate relationship with a trustworthy short-term treatments for depression dem-
other) as they age. Thus, it may be that onstrated that attachment avoidance was
there is a weaker relation between attach- associated with more improvement with
ment and therapy outcome among older short-term cognitive-behavioral therapy
adults because there is less variability in (CBT) and less improvement with short-
their characteristic attachment styles. term interpersonal psychotherapy (IPT;
Theoretical orientation was not a signifi- McBride et al., 2006). Such findings paral-
cant moderator of the effect sizes for anxi- lel the early evidence that interpersonal
ety or avoidance in our meta-analyses. and insight-oriented therapies tend to be
However, our null findings for therapeutic slightly more effective among patients
orientation as a moderator may have been with internalizing coping styles, whereas
due to heterogeneity in the treatments that symptom-focused and skill-building thera-
were grouped together into the same cate- pies tend to be more effective among exter-
gory. For example, in order to have enough nalizing patients (Beutler, Harwood,
studies of the same therapeutic orientation Kimpara, Verdirame, & Blau, this volume,
to combine in a meta-analysis, it was neces- Chapter 17).
sary to combine interpersonal with psycho-
dynamic treatments, individual with group Limitations of the Research
therapies, long-term and short-term treat- There are still relatively few empirical stud-
ments, and inpatient with outpatient treat- ies that have examined how client attach-
ments, although these are really quite ment influences psychotherapy outcome.
different experiences of psychotherapy. In addition, there are few investigations
Nevertheless, the few studies that have regarding matching patients to treatments
examined the interaction between client or therapists based on attachment patterns;
attachment and treatment type in the pre- so few, in fact, that we could not submit
diction of outcome do suggest that clients them to a meta-analysis.
respond differentially to different treat- Furthermore, in order to produce find-
ments based on their attachment style ings that are comparable to one another
and that can be combined to yield mean- However, clients’ attachment security
ingful and clinically relevant conclusions, it also tends to be positively associated with
is important for investigators to use mea- therapeutic alliance, with an average effect
sures of attachment that are well validated size of r = .17 according to a recent meta-
and commonly used in the literature. Some analysis (Diener, Hilsenroth, & Weinberger,
studies have used attachment measures that 2009). Perhaps the capacity to develop a
do not correlate well with other measures positive therapeutic alliance is enhanced
of attachment, and that do not appear to by a client’s level of attachment security.
converge with underlying dimensions of Conversely, the formation of a positive
adult attachment (anxiety and avoidance). therapeutic alliance may serve as one mech-
Another limitation of our meta-analyses anism by which a client’s level of attach-
is that we could not control for the correla- ment security leads to better psychotherapy
tions between attachment and pretreat- outcomes. Finally, an intriguing possibility
ment functioning. The interpretation of is that both attachment security and thera-
posttreatment symptoms as outcome is peutic alliance predict unique aspects of
potentially problematic because it does not psychotherapy outcome.
consider baseline levels or actual change We derive several practice implications
in symptoms as a function of treatment. of the empirical research on attachment
Hence, any association between attach- style and our meta-analysis that can guide
ment and posttreatment functioning may, psychotherapists:
to some degree, reflect the relation between
attachment and psychopathology. Although • Assess the patient’s attachment style.
a number of studies that did control for the Attachment style or organization can
influence of pretreatment functioning on influence the psychotherapy process, the
the association between attachment secu- responses of both patients and therapists,
rity and outcome have reported findings the quality of the therapeutic alliance, and
that are consistent with ours (e.g., Meyer, the ultimate outcome of treatment. Thus,
Pilkonis, Proietti, Heape, & Egan, 2001; therapists should be attuned to indicators
Saatsi, Hardy, & Cahill, 2007; Strauss of a patient’s attachment style. Formal
et al., 2006), the results of the current anal- interviewing or use of reliable self-report
yses should be interpreted with caution in measures can be useful as part of the
that respect. assessment process.
• Understanding a patient’s attachment
Therapeutic Practices organization will provide important clues
The estimated effect sizes for the associa- as to how the patient is likely to respond
tion of both attachment security (r = .18) in treatment and to the therapist. Expect
and attachment anxiety (r = −.22) with longer and more difficult treatment with
treatment outcomes are in the small-to- anxiously attached patients but quicker
moderate range, but just below those found and more positive outcomes with securely
for the association of therapeutic alliance attached patients.
with outcome reported in this volume. • Knowledge of the patient’s attachment
Thus, in these 14 studies, clients’ attach- style can help the therapist anticipate how
ment style appears to contribute almost as the patient may respond to the therapist’s
much variance to psychotherapy outcome interventions and guide the therapist in
as does the alliance, a well-established and calibrating to the patient’s interpersonal
potent predictor of therapeutic change. style. That is, if the patient is dismissing in
his or her attachment, the therapist may Therapists might consider intervening
need to be more engaged. In contrast, if with their patients in an effort to change
the patient is preoccupied in his or her attachment style. Early findings suggest
attachment, the therapist should consider that the focus on the relation between the
a stance designed to help the patient therapist and patient and/or the use of
contain his or her emotional experience. interpretations may be the mechanisms by
This may include explicit articulations of which change in attachment organization
the treatment frame, the provision of more is achieved, at least for severely disturbed
structure to compensate for the patient’s personality-disordered patients (Høglend
tendency to feel muddled, and efforts to et al., 2009; Levy et al., 2006). However,
avoid collusion with the patient who may the early research also demonstrates that a
pull the therapist to engage in more range of treatments may be useful for
emotional/experiential techniques that achieving changes in attachment
only contribute to the patient feeling representations in less disturbed patients
overwhelmed. with neurotic-level or Axis I disorders.
• At the same time, psychotherapists
should not go too far in contrasting References
patients’ attachment styles. Practice and
References marked with an asterisk indicate
research suggest that therapists titrate their
studies included in the meta-analysis.
interpersonal styles so as not to overwhelm Ainsworth, M. S., Blehar, M. C., Waters, E., &
dismissing patients or to appear Wall, S. (1978). Patterns of attachment: A psycho-
disengaged, aloof, or uninterested to logical study of the strange situation. Oxford, UK:
preoccupied patients. Lawrence Erlbaum.
• There is preliminary evidence that Bakermans-Kranenburg, M. J., Juffer, F., & van
Ijzendoorn, M. H. (1998). Interventions with
dismissing individuals do respond to video feedback and attachment discussions: Does
cognitive or interpretive treatments type of maternal insecurity make a difference?
slightly better than interpersonally focused Infant Mental Health Journal, 19(2), 202–219.
treatments, at least in the context of Bartholomew, K. (1990). Avoidance of intimacy:
short-term treatments. With regard to An attachment perspective. Journal of Social and
patients who score high on both the Personal Relationships, 7(2), 147–78.
Bartholomew, K., & Horowitz, L. M. (1991).
attachment anxiety and avoidance Attachment styles among young adults: A test of
dimensions (fearful avoidant clients), it is a four-category model. Journal of Personality and
especially important to attend to the Social Psychology, 61(2), 226–44.
structure of their internal working models, Beebe, B., & Lachmann, F. M. (1988). The contri-
as findings suggest that there is much bution of mother-infant mutual influence to the
variation in this group’s functioning in origins of self- and object representations. Psycho-
analytic Psychology, 5(4), 305–37.
therapy and outcome. Belsky, J., Fish, M., & Isabella, R. A. (1991).
• Attachment style can be modified Continuity and discontinuity in infant negative
during treatment, even in brief treatments and positive emotionality: Family antecedents
and for patients with severe attachment and attachment consequences. Developmental
difficulties, such as those suffering from Psychology, 27(3), 421–31.
borderline personality disorder. Therefore, Belsky, J., & Rovine, M. (1987). Temperament and
attachment security in the strange situation: An
change in attachment can be empirical rapprochement. Child Development,
conceptualized as a proximal outcome, not 58(3), 787–95.
just a predictive patient characteristic, and Berant, E., & Obegi, J. H. (2009). Attachment-
could be considered a goal of treatment. informed psychotherapy research with adults.
In J. H. Obegi & E. Berant (Eds.), Attachment Patient-therapist attachment: Impact on the thera-
theory and research in clinical work with adults peutic process and outcome. In M. Cortina & M.
(pp. 461–69). New York: Guilford Press. Marrone (Eds.), Attachment theory and the psycho-
Bernier, A., & Dozier, M. (2002). The client- analytic process (pp. 127–78). Philadelphia: Whurr
counselor match and the corrective emotional Publishers.
experience: Evidence from interpersonal and Diehl, M., Elnick, A. B., Bourbeau, L. S., &
attachment research. Psychotherapy: Theory, Labouvie-Vief, G. (1998). Adult attachment
Research, Practice, Training, 39(1), 32–43. styles: Their relations to family context and per-
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & sonality. Journal of Personality and Social
Rothstein, H. R. (2009). Introduction to meta- Psychology, 74(6), 1656–69.
analysis. Chichester, UK: Wiley. Diener, M. J., Hilsenroth, M. J., & Weinberger, J.
Bowlby, J. (1978). Attachment theory and its thera- (2009). A primer on meta-analysis of correlation
peutic implications. Adolescent Psychiatry, 6, 5–33. coefficients: The relationship between patient-
Bowlby, J. (1982). Attachment and loss: Retrospect reported therapeutic alliance and adult attach-
and prospect. American Journal of Orthopsychiatry, ment style as an illustration. Psychotherapy
52(4), 664–78. Research, 19, 519–26.
Bowlby, J. (1988). A secure base: Parent-child attach- Dolan, R. T., Arnkoff, D. B., & Glass, C. R.
ment and healthy human development. New York: (1993). Client attachment style and the psycho-
Basic Books. therapist’s interpersonal stance. Psychotherapy: Theory,
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Research, Practice, Training, 30(3), 408–412.
Self-report measurement of adult attachment: Dozier, M. (1990). Attachment organization and
An integrative overview. In J. A. Simpson and treatment use for adults with serious psycho-
W. S. Rholes (Eds.), Attachment theory and close pathological disorders. Development and Psycho-
relationships (pp. 46–76). New York: Guilford pathology, 2(1), 47–60.
Press. Dozier, M., Lomax, L., Tyrrell, C. L., & Lee, S. W.
Clarkin, J. F., & Levy, K. N. (2003). The influ- (2001). The challenge of treatment for clients
ence of client variables on psychotherapy. In with dismissing states of mind. Attachment &
M. Lambert (Ed.), Handbook of psychotherapy Human Development, 3(1), 62–76.
and behavior change (5th ed., pp. 194–226). Farber, B. A., Lippert, R. A., & Nevas, D. B.
New York: Wiley & Sons. (1995). The therapist as attachment figure.
Collins, N. L, & Read, S. J. (1990). Adult attach- Psychotherapy: Theory, Research, Practice, Training,
ment, working models, and relationship quality 32(2), 204–212.
in dating couples. Journal of Personality and Farber, B. A., & Metzger, J. A. (2009). The therapist
Social Psychology, 58(4), 644–63. as secure base. In J. H. Obegi & E. Berant (Eds.),
Conradi, H. J., Gerisma, C., van Duijn, M., & Attachment theory and research in clinical work
de Jonge, P. (2006). Internal and external valid- with adults (pp.46–70). New York: Guilford
ity of the Experiences in Close Relationships Press.
questionnaire in an American and two Dutch Feeney, J. A., Noller, P., & Hanrahan, M. (1994).
samples. European Journal of Psychiatry, 20(4), Assessing adult attachment. In M. B. Sperling &
258–69. W. H. Berman (Eds.), Attachment in adults: Clinical
∗
Cyranowski, J. M., Bookwala, J., Feske, U., Houck, P., and developmental perspectives (pp. 128–52).
Pilkonis, P., Kostelnik, B., et al. (2002). Adult New York, NY: Guilford Press.
attachment profiles, interpersonal difficulties, Fonagy, P. (1999). Attachment, the development
and response to interpersonal psychotherapy of the self, and its pathology in personality disor-
in women with recurrent major depression. ders. In J. Derksen, C. Maffei, & H. Groen
Journal of Social and Clinical Psychology, 21(2), (Eds.), Treatment of personality disorders (pp. 53–68).
191–217. Dordrecht, Netherlands: Kulwer Academic
Diamond, D., Clarkin, J., Levine, H., Levy, K. N., Publishers.
Foelsch, P. & Yeomans, F. (1999). Borderline Fonagy, P., Leigh, T., Steele, M., Steele, H., Kennedy, R.,
conditions and attachment: A preliminary Mattoon, G., et al. (1996). The relation of
report. Psychoanalytic Inquiry, 19(5), 831–84. attachment status, psychiatric classification, and
Diamond, D., Clarkin, J. F., Stovall-McClough, K. C., response to psychotherapy. Journal of Consulting
Levy, K. N., Foelsch, P. A., Levine, H., et al. (2003). and Clinical Psychology, 64(1), 22–31.
Fraley, R. C. (2002). Attachment stability from Hunter, J. E., & Schmidt, F. L. (1990). Dichotomi-
infancy to adulthood: Meta-analysis and dynamic zation of continuous variables: The implications
modeling of developmental mechanisms. for meta-analysis. Journal of Applied Psychology,
Personality and Social Psychology, 6(2), 123–51. 75, 334–49.
Fraley, R. C., Waller, N. G., & Brennan, K. A. Hunter, J. E., & Schmidt, F. L. (2000). Fixed effects
(2000). An item response theory analysis of self- vs. random effects meta-analysis models:
report measures of adult attachment. Journal of Implications for cumulative research knowledge.
Personality and Social Psychology, 78(2), 350–65. International Journal of Selection and Assessment,
∗
Fuertes, J. N., Mislowack, A., Brown, S., 8(4), 275–92.
Gur-Arie, S., Wilkinson, S., & Gelso, C. J. Hunter, J. E., & Schmidt, F. L. (2004). Methods of
(2007). Correlates of the real relationship in meta-analysis: Correcting error and bias in research
psychotherapy: A study of dyads. Psychotherapy findings (2nd ed.). Thousand Oaks, CA: Sage.
∗
Research, 17(4), 423–30. Johnson, S. M, & Talitman, E. (1997). Predictors
George, C., Kaplan, M., & Main, M. (1985). Adult of success in emotionally focused marital ther-
Attachment Interview. Unpublished manuscript, apy. Journal of Marital & Family Therapy, 23(2),
Department of Psychology, University of 135–52.
California, Berkeley. Kagan, J. (1998). Three seductive ideas. Cambridge,
Griffin, D. W., & Bartholomew, K. (1994). Models MA: Harvard University Press.
of the self and other: Fundamental dimen- Kilmann, P. R., Laughlin, J. E., Carranza, L. V.,
sions underlying measures of adult attachment. Downer, J. T., Major, S., & Parnell, M. M.
Journal of Personality and Social Psychology, 67(3), (1999). Effects of an attachment-focused group
430–445. preventive intervention on insecure women.
Grossmann, K. E., Grossmann, K., & Waters, E. Group Dynamics: Theory, Research, and Practice,
(Eds.). (2005). Attachment from infancy to adult- 3(2), 138–47.
hood: The major longitudinal studies. New York: Kirchmann, H., Mestel, R., Schreiber-Willnow, K.,
Guilford Publications. Mattke, D., Seidler, K. P., Daudert, E., et al.
Gunnar, M. R., Broderson, L., Nachmias, M., (2009). Associations among attachment charac-
Buss, K., & Rigatuso, J. (1996). Stress reactivity teristics, patients’ assessment of therapeutic fac-
and attachment security. Developmental Psycho- tors, and treatment outcome following inpatient
biology, 29, 191–204. psychodynamic group psychotherapy. Psycho-
Hardy, G. E., Stiles, W. B., Barkham, M., & therapy Research, 19(2), 234–48.
Startup, M. (1998). Therapist responsiveness to Kivlighan, D. M., Patton, M. J., & Foote, D.
client interpersonal styles during time-limited (1998). Moderating effects of client attachment
treatments for depression. Journal of Consulting on the counselor experience–working alliance
and Clinical Psychology, 66(2), 304–312. relationship. Journal of Counseling Psychology,
Hazan, C., & Shaver, P. R. (1987). Romantic 45(3), 274–78.
love conceptualized as an attachment process. Kobak, R. R., Cole, H. E., Ferenz-Gillies, R.,
Journal of Personality and Social Psychology, 52(3), Fleming, W. S., & Gamble, W. (1993).
511–24. Attachment and emotion regulation during
Hazan, C., & Shaver, P. R. (1990). Love and work: mother-teen problem solving: A control theory
An attachment-theoretical perspective. Journal of analysis. Child Development, 64(1), 231–45.
Personality and Social Psychology, 59(2), 270–80. Korfmacher, J., Adam, E., Ogawa, J., & Egeland, B.
Hedges, L. V., & Olkin, I. (1985). Statistical meth- (1997). Adult attachment: Implications for the
ods for meta-analysis. Orlando, FL: Academic therapeutic process in a home visitation inter-
Press. vention. Applied Developmental Science, 1(1),
Horowitz, L. M., Rosenberg, S. E., & Bartholomew, K. 43–52.
∗
(1993). Interpersonal problems, attachment Lawson, D. M., & Brossart, D. F. (2009).
styles, and outcome in brief dynamic psycho- Attachment, interpersonal problems, and treat-
therapy. Journal of Consulting and Clinical ment outcome in group therapy for intimate
Psychology, 61(4), 549–60. partner violence. Psychology of Men & Masculinity,
Huffcutt, A. I., & Arthur, W. (1995). Development 10(4), 288–301.
of a new outlier statistic for meta-analytic data. Levy, K. N. (2005). The implications of attach-
Journal of Applied Psychology, 80, 327–34. ment theory and research for understanding
borderline personality disorder. Development and Society for Research in Child Development,
Psychopathology, 17, 959–86. 50(1–2), 66–104.
Levy, K. N., Beeney, J. E., Wasserman, R. H., & Main, M., & Solomon, J. (1986). Discovery of
Clarkin, J. F. (2010). Alliance at the borderline: an insecure-disorganized/disoriented attachment
The relationship between adult attachment and pattern. In T. B. Brazelton & M. W. Yogman
therapeutic alliance in borderline personality disor- (Eds.), Affective development in infancy (pp. 95–124).
der. Unpublished manuscript, Department of Westport, CT: Ablex Publishing.
Psychology, Pennsylvania State University, Mallinckrodt, B., Porter, M. J., & Kivlighan, D. M.
University Park. (2005). Client attachment to therapist, depth
Levy, K. N., Blatt, S. J., & Shaver, P. R. (1998). of in-session exploration, and object relations
Attachment styles and parental representations. in brief psychotherapy. Psychotherapy: Theory,
Journal of Personality and Social Psychology, 74(2), Research, Practice, Training, 42(1), 85–100.
∗
407–419. Marmarosh, C. L., Gelso, C. J., Markin, R. D.,
Levy, K. N., Clarkin, J. F., & Kernberg, O. F. Majors, R., Mallery, C., & Choi, J. (2009). The
(2007). The relationship between unresolved real relationship in psychotherapy: Relationships
attachment and therapy dropout in borderline per- to adult attachments, working alliance, transfer-
sonality disorder. Unpublished manuscript, ence, and therapy outcome. Journal of Counseling
Department of Psychology, Pennsylvania State Psychology, 56(3), 337–50.
∗
University, University Park. McBride, C., Atkinson, L., Quilty, L. C.,
Levy, K. N., & Kelly, K. M. (2009). Using inter- Bagby, R. M. (2006). Attachment as moderator
views to assess adult attachment. In J. H. Obegi of treatment outcome in major depression:
& E. Berant (Eds.), Attachment theory and A randomized control trial of interpersonal psy-
research in clinical work with adults (pp. 121–52). chotherapy versus cognitive behavior therapy.
New York: Guilford Press. Journal of Consulting and Clinical Psychology,
Levy, K. N., & Kelly, K. M. (2010). Sex differences 74(6), 1041–54.
∗
in jealousy: A contribution from attachment Meyer, B., Pilkonis, P. A., Proietti, J. M., Heape, C. L.,
theory. Psychological Science. 21, 168–73. & Egan, M. (2001). Attachment styles and per-
Levy, K. N., Meehan, K. B., Kelly, K. M., sonality disorders as predictors of symptom
Reynoso, J. S., Clarkin, J. F., & Kernberg, O. F. course. Journal of Personality Disorders, 15(5),
(2006). Change in attachment patterns and 371–89.
reflective function in a randomized control trial Mickelson, K. D., Kessler, R. C., & Shaver, P. R.
of transference-focused psychotherapy for bor- (1997). Adult attachment in a nationally repre-
derline personality disorder. Journal of Consulting sentative sample. Journal of Personality and Social
and Clinical Psychology, 74, 1027–40. Psychology, 73(5), 1092–1106.
∗
Levy, M. B., & Davis, K. E. (1988). Lovestyles and Muller, R. T., & Rosenkranz, S. E. (2009).
attachment styles compared: Their relations to Attachment and treatment response among
each other and to various relationship character- adults in inpatient treatment for posttraumatic
istics. Journal of Social and Personal Relationships, stress disorder. Psychotherapy: Theory, Research,
5(4), 439–71. Practice, Training, 46(1), 82–96.
Lipsey, M. W., & Wilson, D. B. (2001). Practical Obegi, J. H. (2008). The development of the client-
meta-analysis. Thousand Oaks, CA: Sage. therapist bond through the lens of attachment
Main, M. (1995). Recent studies in attachment: theory. Psychotherapy: Theory, Research, Practice,
Overview, with selected implications for clinical Training, 45(4), 431–46.
work. In S. Goldberg, R. Muir, & J. Kerr (Eds.), Pilkonis, P. A. (1988). Personality prototypes among
Attachment theory: Social, developmental, and depressives: Themes of dependency and autonomy.
clinical perspectives (pp. 407–74). Hillsdale, NJ: Journal of Personality Disorders, 2(2), 144–52.
Analytic Press. Ravitz, P., Maunder, R. G., Hunter, J. A.,
Main, M., & Goldwyn, R. (1998). Adult attach- Sthyankiya, B., & Lancee, W. J. (2010). Adult
ment classification system. Unpublished manu- attachment measures: A 25-year review. Journal
script. University of California, Berkeley. of Psychosomatic Research. Advance online publi-
Main, M., Kaplan, N., & Cassidy, J. (1985). Security cation. doi: 10.1016/j.jpsychores.2009.08.006
∗
in infancy, childhood, and adulthood: A move to Reis, S., & Grenyer, B. F. S. (2004). Fearful attach-
the level of representation. Monographs of the ment, working alliance and treatment response
for individuals with major depression. Clinical psychopathology and intervention (pp. 181–206).
Psychology & Psychotherapy, 11(6), 414–24. Mawah, NJ: Lawrence Erlbaum Associates.
Ricks, M. H. (1985). The social transmission of Sroufe, L. A., & Waters, E. (1977). Heart rate as a
parental behavior: Attachment across genera- convergent measure in clinical and developmen-
tions. Monographs of the Society for Research in tal research. Merrill-Palmer Quarterly, 23(1),
Child Development, 50(1–2), 211–27. 3–27.
∗
Riggs, S. A., Jacobovitz, D., & Hazen, N. (2002). Stalker, C. A., Gebotys, R., & Harper, K. (2005).
Adult attachment and history of psychotherapy Insecure attachment as a predictor of outcome
in a normative sample. Psychotherapy: Theory, following inpatient trauma treatment for women
Research, Practice, Training, 39(4), 344–53. survivors of childhood abuse. Bulletin of the
Romano, V., Fitzpatrick, M., & Janzen, J. (2008). Menninger Clinic, 69(2), 137–56.
The secure-base hypothesis: Global attachment, Stevenson-Hinde, J., & Marshall, P. J. (1999).
attachment to counselor, and session exploration Behavioral inhibition, heart period, and respira-
in psychotherapy. Journal of Counseling Psychology, tory sinus arrhythmia: An attachment perspec-
55(4), 495–504. tive. Child Development, 70, 805–816.
∗
Saatsi, S., Hardy, G. E., & Cahill, J. (2007). Stovall-McClough, K. C., & Cloitre, M. (2003).
Predictors of outcome and completion status in Reorganization of unresolved childhood trau-
cognitive therapy for depression. Psychotherapy matic memories following exposure therapy.
Research, 17(2), 185–95. Annals of the New York Academy of Sciences, 1008,
Schmidt, F. L., Le, H., & Oh, I. S. (2009). 297–99.
∗
Correcting for the distorting effects of study Strauss, B. M., Kirchmann, H., Eckert, J., Lobo-
artifacts in meta-analysis. In H. Cooper, Drost, A. J., Marquet, A., Papenhausen, R., et al.
L. V. Hedges, & J. C. Valentine (Eds.), The hand- (2006). Attachment characteristics and treat-
book of research synthesis and meta-analysis (2nd ment outcome following inpatient psychother-
ed.). New York: Sage. apy: Results of a multisite study. Psychotherapy
Shaver, P. R., & Clark, C. L. (1994). The psycho- Research, 16(5), 573–86.
dynamics of adult romantic attachment. In Strauss, B. M., Lobo-Drost, A. J., & Pilkonis, P. A.
J. M. Masling & R. F. Bornstein (Eds.), Empirical (1999). Einschätzung von Bindungsstilen bei
perspectives on object relations theory (pp. 105–56). Erwachsenen: Erste Erfahrungen mit der
Washington, DC: American Psychological deutschen Version einer Prototypenbeurteilung.
Association. [Evaluation of attachment styles in adults: First
Shaver, P. R., & Hazan, C. (1993). Adult romantic results with the German version of a prototype
attachment: Theory and evidence. In D. Perlman evaluation.] Zeitschrift für Klinische Psychologie,
& W. Jones (Eds.), Advances in personal relation- Psychiatrie und Psychotherapie, 47, 347–364.
∗
ships (Vol. 4, pp. 29–70). London, UK: Tasca, G. A., Ritchie, K., Conrad, G., Balfour, L.,
Kingsley. Gayton, J., Lybanon, V., et al. (2006). Attach-
Shaver, P. R., Hazan, C., & Bradshaw, D. (1988). ment scales predict outcome in a randomized
Love as attachment. In R. J. Sternberg & controlled trial of two group therapies for binge
M. L. Barnes (Eds.), The psychology of love eating disorder: An aptitude by treatment inter-
(pp. 68–99). New Haven, CT: Yale University action. Psychotherapy Research, 16(1), 106–21.
∗
Press. Travis, L. A., Bliwise, N. G., Binder, J. L., &
Simpson, J. A. (1990). Influence of attachment style Horne-Moyer, H. L. (2001). Changes in clients’
on romantic relationships. Journal of Personality attachment styles over the course of time-limited
and Social Psychology, 59(5), 971–80. dynamic psychotherapy. Psychotherapy: Theory,
Slade, A. (1999). Attachment theory and research: Research, Practice, Training, 38(2), 149–59.
Implications for the theory and practice of indi- Ulberg, R., Marble, A., & Høglend, P. (2009). Do
vidual psychotherapy with adults. In J. Cassidy gender and level of relational functioning influ-
& P. R. Shaver (Eds.), Handbook of attachment: ence the long-term treatment response in
Theory, research, and clinical applications dynamic psychotherapy? Nordic Journal of
(pp. 575–94). New York: Guilford Press. Psychiatry, 63(5), 412–419.
Slade, A. (2004). Two therapies: Attachment orga- van Ijzendoorn, M. H. (1995). Adult attach-
nization and the clinical process. In L. Atkinson ment representations, parental responsiveness,
& S. Goldberg (Eds.), Attachment issues in and infant attachment: A meta-analysis on
the predictive validity of the Adult Attach- longitudinal study. Child Development, 71(3),
ment Interview. Psychological Bulletin, 117(3), 684–89.
387–403. Wei, M., Russell, D. W., Mallinckrodt, B., &
Vaughn, B. E., & Bost, K. K. (1999). Attachment Vogel, D. L. (2007). The Experiences in Close
and temperament: Redundant, independent, or Relationship Scale (ECR)-short form: Reliability,
interacting influences on interpersonal adapta- validity, and factor structure. Journal of Personality
tion and personality development? In J. Cassidy Assessment, 88(2), 187–204.
& P. R. Shaver (Eds.), Handbook of attach- West, M. L., & Sheldon-Keller, A. E. (1994).
ment: Theory, research, and clinical applications Patterns of relating: An adult attachment perspec-
(pp. 198–225). New York: Guilford Press. tive. New York: Guilford Press.
Waters, E., Merrick, S., Treboux, D., Crowell, J., Wilson, D. B. (2005). Meta-analysis macros for SAS,
& Albersheim, L. (2000). Attachment security SPSS, and Stata. Retrieved December 11, 2009,
in infancy and early adulthood: A twenty-year from http://mason.gmu.edu/∼dwilsonb/ma.html
C HA P TER
20 Religion and Spirituality
Everett L. Worthington, Jr., Joshua N. Hook, Don E. Davis, and Michael A. McDaniel
One relationship factor that can potentially needed (Norcross, 2002). The increase in
affect the outcome of psychotherapy is the number, variety, and rigor of outcome
match or mismatch between a client’s studies evaluating R/S psychotherapies
religious or spiritual (R/S) beliefs and the allows for a far more rigorous evaluation of
type of psychotherapy. Some R/S clients the effectiveness of tailoring psychotherapy
desire R/S-tailored or accommodated treat- to a patient’s R/S convictions.
ment. Others can comfortably accept a In this chapter, we first define R/S and
secular treatment. Even for those who do discuss how these constructs are generally
not request R/S treatment, some might measured. Second, we offer clinical examples
benefit from the contextualization of treat- that illustrate how psychotherapy might be
ment in their R/S framework. accommodated for one’s R/S beliefs. Third,
There has been an increase in outcome we present data from a meta-analysis exam-
studies examining psychotherapies that ining the effectiveness of R/S psychother-
incorporate R/S beliefs (Hook et al., 2010; apy. Fourth, we discuss patient contributions
Pargament & Saunders, 2007; Post & to the effectiveness of R/S psychotherapy.
Wade, 2009; Smith, Bartz, & Richards, Fifth, we note several limitations of the
2007; Worthington & Aten, 2009). At present body of research. Finally, we give
the time of the first edition of Psychothe- recommendations for therapists based on
rapy Relationships That Work (Norcross, the present research evidence.
2002), there were only 11 outcome studies
examining an R/S psychotherapy, making Definitions and Measures
conclusions based on this set of studies Although the terms religion and spiritu-
necessarily tenuous (Worthington & ality have historically been closely linked
Sandage, 2001). Furthermore, these studies (Sheldrake, 1992), current conceptualiza-
were limited to mainly Christian or Muslim- tions make important distinctions between
accommodative cognitive-behavioral inter- religion and spirituality. Religion can be
ventions. Thus, it was difficult to generalize defined as adherence to a belief system and
to other types of R/S psychotherapies. As practices associated with a tradition and
such, tailoring psychotherapy to the R/S community in which there is agreement
beliefs of clients was judged to have prom- about what is believed and practiced (Hill
ising empirical support, but it was sug- et al., 2000). Spirituality, in contrast, can
gested that more research on this topic was be defined as a more general feeling of
402
closeness and connectedness to the sacred. many methods and goals as secular psycho-
What one views as sacred is often a socially therapy but also incorporates methods or
influenced perception of either (a) a divine goals that are R/S in nature. For example,
being or object or (b) a sense of ultimate in addition to using cognitive or behavioral
reality or truth (Hill et al.). Many people techniques to alleviate depression, a clini-
experience their spirituality in the context cian practicing R/S psychotherapy might
of religion, but not all do. conceptualize using an R/S framework and,
Four types of spirituality have been iden- within that framework, use methods such
tified on the basis of the type of sacred as prayer or religious imagery. Besides
object (Davis, Hook, & Worthington, pursuing goals that are psychological, a
2008; Worthington, 2009; Worthington & client in R/S psychotherapy might also
Aten, 2009). First, religious spirituality work toward spiritual goals, such as becom-
involves a sense of closeness and connection ing more like Jesus Christ, or adhering
to the sacred as described by a specific reli- more closely to the teachings of Buddha.
gion (e.g., Christianity, Islam, Buddhism). R/S outcome variables, such as spiritual
This type of spirituality fosters a sense well-being, might be important in psycho-
of closeness to a particular god or higher therapy when clients’ reasons for attending
power. Second, humanistic spirituality therapy and criteria for evaluating therapy
involves a sense of closeness and connection include spiritual goals. Accordingly, the
to humankind. This type of spirituality outcome measures used in the subsequent
develops a sense of connection to a general review and meta-analysis fall into two
group of people, often involving feelings of categories. First, almost all studies use a
love, altruism, or reflection. Third, nature psychological outcome variable. A study
spirituality involves a sense of closeness and examining R/S psychotherapy for depres-
connection to the environment or to nature. sion, for example, might use the Beck
For example, one might experience wonder Depression Inventory (BDI; Beck, Ward,
by witnessing a sunset or experiencing a Mendelson, Mock, & Erbaugh, 1961).
natural wonder such as the Grand Canyon. Second, many studies also use a measure of
Fourth, cosmos spirituality involves a sense spirituality. For example, a study examin-
of closeness and connection with the whole ing R/S psychotherapy for unforgiveness
of creation. This type of spirituality might might use not only a primary psychological
be experienced by meditating on the mag- measure of forgiveness but also a secondary
nificence of creation, or by looking into the measure of spiritual well-being (Ellison,
night sky and contemplating the vastness 1983).
of the universe. The majority of studies in the present
Psychotherapy has been defined as the review measured R/S beliefs simply by
“informed and intentional application of identification (i.e., the participant self-
clinical methods and interpersonal stances identified as Christian). Some studies used
derived from established psychological a measure of R/S beliefs or commitments
principles for the purpose of assisting (e.g., Religious Orientations Scale, Allport
people to modify their behaviors, cogni- & Ross, 1967; Religious Commitment
tions, emotions, and/or other personal Inventory-10, Worthington et al., 2003)
characteristics in directions which the and employed a minimum cutoff score as a
participants deem desirable” (Norcross, criterion for inclusion in the study. This
1990, p. 218). R/S psychotherapy shares ensured that the participants in the study
wo rt h i n g to n , h o o k , d av i s , mc d a n i e l 403
were at least moderately engaged with their she wanted to incorporate R/S issues in
R/S beliefs. A few studies (e.g., Razali, her psychotherapy. As Dana and her thera-
Aminah, & Kahn, 2002) used a measure pist explored and modified her negative
of R/S beliefs or commitments and also core beliefs, they discussed how Dana
measured the extent to which R/S treat- thought God viewed her. Several passages
ments had different effects for participants of the Bible comforted Dana and helped
who were more (or less) committed. her realize that, even though she viewed
herself negatively, God and other people
Clinical Examples loved and accepted her as she was.
We now provide several case examples
of R/S psychotherapy from different theo- Case Example 2:
retical and R/S perspectives. Spiritual Self-Schema
Therapy for Addiction
Case Example 1: Spiritual self-schema therapy integrates
Christian-Accommodative cognitive-behavioral techniques with
Cognitive Therapy for Depression Buddhist psychological principles (Avants
The cognitive model of depression empha- & Margolin, 2004). The goal of this psy-
sizes the role of maladaptive cognition in chotherapy is to modify a person’s self-
both the causes and treatment of depres- schema. When a self-schema is activated,
sion (Beck, 1972). Christian-accommodative beliefs about the self energize specific
cognitive therapy for depression retains the behaviors. This psychotherapy attempts to
main features of the secular theory yet facilitate a shift from an “addict” self-
places the psychotherapy in a religious con- schema to a “spiritual” self-schema that fos-
text. For example, the rationale for psycho- ters mindfulness, compassion, and doing
therapy, the homework assignments, and no harm to self or others (Margolin et al.,
the challenging of negative automatic 2007). Psychotherapy sessions focus on
thoughts and core beliefs are integrated aspects of the Buddhist Noble Eightfold
with and based on biblical teachings regard- Path, which include training in mindful-
ing the self, world, and future (Pecheur & ness, morality, and wisdom.
Edwards, 1984). Dave (age 47) did not ascribe to a reli-
Dana (age 31) was a Christian female gion. He considered himself to be spiri-
who presented to psychotherapy with sev- tual. After he lost his job because he failed
eral symptoms of depression, including a drug test due to cocaine use, he checked
feelings of sadness, sleeping more than into a rehabilitation facility. He had been
usual, low energy, weight gain, and loss of dependent on drugs and alcohol on and off
interest in everyday activities. As psycho- for 30 years. During psychotherapy, Dave
therapy progressed, Dana explored negative was taught about the wandering nature of
beliefs about herself. Her most problem- the mind, and how this contributed to his
atic core belief was that she was worth- addict self-schema. If Dave did not work
less and no one would ever love and accept to control his mind, he usually thought
her as she was. These beliefs seemed related of using drugs. Dave practiced a medita-
to a difficult childhood. She had been tion technique called anapanasati, which
physically abused by her mother, who even- involves sitting silently with eyes closed
tually abandoned her. Dana was a commit- and focusing on the sensations experienced
ted Christian. At intake she stated that while breathing naturally. Dave improved
his concentration and mindfulness with gratitude to God for forgiving her helped
practice. Over time, he developed disci- her forgive her father.
pline over his maladaptive thoughts.
Case Example 4:
Case Example 3: Muslim-Accommodative
Christian-Accommodative Cognitive Therapy for Anxiety
Forgiveness Therapy Similar to Christian-accommodative cog-
REACH is a model of promoting forgive- nitive therapy for depression, Muslim-
ness that involves five steps: recall the hurt, accommodative cognitive therapy for
develop empathy toward the offender, give anxiety retains Beck’s cognitive model
an altruistic gift of forgiveness, commit to (Beck, Rush, Shaw, & Emery, 1979), aug-
forgive, and hold on to the forgiveness menting it with spiritual strategies and
(Worthington, 1998). Christian versions of interventions. For example, psychothera-
REACH actively encourage clients to access pists work with clients to identify and chal-
their religious beliefs while moving toward lenge negative thoughts and beliefs using
forgiveness (Lampton et al., 2005; Rye the Koran and Hadith (sayings and cus-
et al., 2005). Clients are encouraged to toms of the Prophet) as guidance (Razali,
view forgiveness as a collaborative process Aminah, & Khan, 2002). Clients are
with God and to consider prayer or use of encouraged to cultivate feelings of close-
Scripture in forgiving. ness to Allah, pray regularly, and read the
Lisa (age 20) was a Christian female who Koran.
struggled to forgive her father. Her father Hasan (age 35) was a highly committed
had several extramarital affairs when Lisa Muslim male, diagnosed with generalized
was younger, which precipitated her par- anxiety disorder. He became worried every
ents’ divorce when Lisa was 7. Lisa’s father day, and his anxiety interfered with his
was unreliable when Lisa was growing up. marriage and job. In psychotherapy, Hasan
He regularly broke promises, such as failing acknowledged that he did not believe the
to attend birthday parties or soccer games. world was a safe place, and he felt as if he
Lisa harbored resentment and anger toward had to worry or else something terrible
her father. During her junior year of col- might happen. The psychotherapist helped
lege, she concluded that her unforgiveness Hasan examine the evidence for and against
was a problem. Even though her father was his thoughts. Hasan and his psychothera-
not a part of her life, most days Lisa woke pist worked together to develop religious
up actively angry, stressed, and upset toward coping strategies and discover religious
her father. She attended a group psychoed- truths to counteract his anxious thoughts.
ucational workshop for people struggling For example, it helped Hasan to remember
with forgiveness. During the workshop, the that he believed that Allah was always in
group leader led Lisa and seven other people control, and that he could trust in Allah to
through the steps to promote forgiveness. be with him and comfort him.
Group members shared with each other
how they had been hurt and worked toward Meta-Analytic Review
developing empathy for their offender. The Past research assessing the efficacy and
group also discussed God’s role in forgive- specificity of R/S psychotherapies has been
ness, which helped Lisa realize the extent mixed. McCullough (1999) evaluated
that God and others had forgiven her. Lisa’s the efficacy of Christian-accommodative
psychotherapies for depression and con- treatment are equivalent in regard to
cluded that the R/S psychotherapies worked theoretical orientation and duration of
as well, but not better than established treatment but differ in whether they are
secular therapies. Hook and colleagues accommodated to R/S clients.
(2010) reached a similar conclusion in their Comparison conditions may differ in
review of empirically supported R/S psycho- strength, so these studies most rigorously
therapies. They found some evidence for test whether it is helpful to tailor
the efficacy of R/S psychotherapies. Thus, psychotherapy to a client’s R/S faith.
R/S psychotherapies performed better than
control groups and equal to established Method
secular psychotherapies. However, review- Inclusion Criteria. Studies included in the
ers found little evidence for the specificity present meta-analysis met a definition of
of R/S psychotherapies—that R/S psycho- psychotherapy (Norcross, 1990), and all
therapies consistently outperformed estab- studies explicitly integrated R/S consider-
lished secular psychotherapies. However, in ations into psychotherapy. All studies
a recent meta-analysis, Smith and associ- included in the present review used random
ates (2007) found evidence for the positive assignment and compared an R/S treat-
effects of R/S psychotherapies even when ment with either (a) a no-treatment control
compared with alternate treatments. condition or (b) an alternate treatment. We
In the present meta-analytic study, we excluded studies of (a) 12-step groups such
sought to determine the extent to which as Alcoholics Anonymous, (b) meditation
tailoring the psychotherapy relationship or mindfulness interventions that were not
to the client’s R/S faith is efficacious. We explicitly R/S, (c) R/S interventions such as
address this at three levels. intercessory prayer that were not contextu-
alized in a psychotherapy, and (d) one-
• First we compare outcomes of session “workshop-type” interventions.
clients in R/S psychotherapy versus Literature Search. We conducted our
clients in no-treatment control groups. literature search by (a) using two or more
Studies using comparative designs control computer databases (listed in the next
for possible confounding variables present paragraph), (b) manually searching the
in less rigorous designs. The use of control references of previous meta-analyses and
groups provides for credible inference reviews, and (c) contacting relevant research-
concerning the efficacy of R/S ers for file-drawer studies. We included
psychotherapies. both published and unpublished studies.
• Second, we compare outcomes of Effect sizes from published studies tend to
clients in R/S psychotherapy versus clients be larger than effect sizes from unpublished
in alternate psychotherapies. These types studies, so limiting the review to published
of studies not only control for possible studies may exacerbate publication bias
confounding variables but also provide (Lipsey & Wilson, 2001).
some evidence for the specificity of R/S First, we identified studies by searching
psychotherapies. the PsychINFO, Social Sciences Citation
• Third, we compare outcomes of Index, and Dissertation Abstracts International
clients in R/S psychotherapy versus clients databases up until December 1, 2009. The
in alternate psychotherapies that use a search used the key terms [counseling
dismantling design. In these studies, the OR therapy] AND [religio∗ OR spiritu∗]
R/S psychotherapy and the comparison AND [outcome]. Second, we used previous
reviews of the literature (Harris, Thoresen, which the R/S condition and the compari-
McCullough, & Larson; Hodge, 2006; son condition were identical in theoretical
Hook et al., 2010; McCullough, 1999; orientation and duration of treatment.
Smith et al., 2007; Worthington, Kurusu, Coding. The coding of studies included
McCullough, & Sandage, 1996; Worthington sample size, as well as information neces-
& Sandage, 2001) to identify relevant sary to calculate the d and standard error of
studies. Third, we contacted the correspond- the d (e.g., means, standard deviations).
ing author from each study identified to Also coded were potential moderators
inquire about studies we may have missed, including study design characteristics,
including unpublished file-drawer studies. treatment characteristics, and measurement
Effect Size. The effect size used in this characteristics. Study design characteristics
study was the standardized mean differ- coded involved source of data (published
ence (d ). The standardized mean difference or unpublished). An effect for source of
is a standard deviation metric with zero data would suggest that publication bias
indicating no mean group difference. The could be present, which might limit the
value of d summarizes the posttest differ- conclusions that could be drawn from the
ence between the R/S condition and the meta-analysis. Treatment characteristics
comparison condition. A positive d indi- included treatment format (e.g., group,
cates that the R/S condition performed individual), problem rated (e.g., depres-
better, on average, than the comparison; sion, anxiety), theoretical orientation (e.g.,
a negative d indicates that the comparison cognitive, behavioral), and type of R/S
condition performed better. faith commitment (e.g., Christian, Muslim,
Missing Data. Some studies did not general spirituality). Measurement charac-
contain sufficient data for the calculation teristics involved type of measure (e.g.,
of effect sizes. For each study with insuffi- psychological, spiritual).
cient data to calculate the effect size, we Data Analysis. Data analysis was
requested missing data from the corre- conducted using Comprehensive Meta-
sponding author. If the necessary data Analysis Version 2.2 (Borenstein, Hedges,
could not be obtained, we excluded the Higgins, & Rothstein, 2005). Random-
study from the analysis. effects models were used because we had no
Outcome of Search. Overall, a total of reason to believe that the population effect
51 samples from 46 separate studies evalu- sizes were invariant. Consistent with random-
ated R/S psychotherapy. Eleven samples effects models, studies were weighted by
employed both a control condition and an the sum of the inverse sampling variance
alternate treatment, resulting in 62 total plus tau-squared (Borenstein, Hedges,
comparisons. Of these comparisons, 5 did Higgins, & Rothstein, 2009). Separate
not have sufficient information to calculate analyses were conducted for psychological
the effect size, and 6 did not come from a and spiritual outcomes. For studies that
study that employed random assignment to reported more than one effect size, we used
condition, leaving 51 valid comparisons the measure that best assessed the goal of
for analysis. Of these comparisons, 22 the specific psychotherapy. For example,
compared R/S psychotherapy to a control if a study purported to examine R/S
condition, and 29 compared R/S psycho- cognitive-behavioral therapy for depres-
therapy to an alternate treatment. Of these sion, a measure such as the Beck Depression
29 comparisons, 11 comparisons were Inventory was chosen and other measures,
identified that used a dismantling design in such as anxiety or general distress, were
ignored. In addition, measures that had magnitude at follow-up, although these
been subjected to peer review were chosen results should be treated with caution
over non-peer-reviewed measures. because of the low number of studies
reporting such data.
Results Our second analysis examined whether
The total number of participants from patients in R/S psychotherapies showed
the 51 samples was 3,290 (1,524 from R/S greater improvement than those in alter-
psychotherapies, 921 from alternate psycho- nate psychotherapies on both psychological
therapies, and 845 from no-treatment and spiritual outcomes. This was largely
control conditions). Descriptive information the case (psychological d = 0.26; spiritual
for all studies is summarized in Table 20.1. d = 0.41). Participants in R/S psychothera-
R/S psychotherapies addressed problems pies outperformed alternate treatments on
in a variety of areas. A wide range of R/S psychological and spiritual outcomes. These
perspectives were represented, although the differences in outcomes were largely main-
most common perspectives were Christi- tained at follow-up, although these results
anity, Islam, and general spirituality. Many should be treated with caution because
theoretical orientations were represented, of the small number of studies reporting
although the most common theories were such data.
cognitive, cognitive-behavioral, and mind- Our third analysis was limited to studies
body-spirit. that used a dismantling design in which the
The meta-analytic results for psycho- R/S and alternate treatment had the same
logical and spiritual outcomes are summa- theoretical orientation and duration of
rized in Table 20.2. The first column lists treatment. For psychological outcomes,
the level of comparison. Columns 2 through there was little difference between condi-
6 list the posttest results. The second and tions (d = 0.13). For spiritual outcomes,
third columns list the number of partici- participants in R/S psychotherapies outper-
pants (N) and studies (k). The fourth and formed participants in alternate psycho-
fifth columns list the mean d and 95% therapies at posttest (d = 0.33). This
confidence interval for the observed d. difference in outcome was maintained at
The sixth column lists I 2, the ratio of true follow-up, although this result should be
heterogeneity to total variation in observed treated with caution because of the low
effect sizes. Columns seven through eleven number of studies reporting such data.
list the follow-up results using the same In summary, the meta-analytic results
format. present clear findings about the effective-
Our first analysis examined whether ness of religious and spiritual tailoring.
patients in R/S psychotherapies showed Consistent with Smith et al. (2007), there
greater improvement than would patients was some evidence that R/S psychothera-
in no-treatment control conditions on both pies outperformed alternate psychothera-
psychological and spiritual outcomes. This pies on both psychological and spiritual
was largely the case (psychological d = 0.45; outcomes. However, this finding is difficult
spiritual d = 0.51). Participants in R/S to interpret because comparison treatments
psychotherapies outperformed no-treatment varied in quality. When the analysis was
control conditions on psychological and limited to studies that used a dismantling
spiritual outcomes. These differences in design—studies in which the R/S condition
outcomes were maintained at a smaller and alternate condition utilized the same
Table 20.1 Descriptive Information for All Studies
Study Published Design Random N RS N Alt N Ctl Belief R/S Problem Theory d (vs. Alt) d (vs. Ctl)
Azhar & Varma (1995a) Y C Y 15 15 NA Muslim R Depression Cognitive-behavioral .75 NA
Azhar & Varma (1995b) Y C Y 32 32 NA Muslim R Depression Cognitive-behavioral .27 NA
Azhar et al. (1994) Y C Y 31 31 NA Muslim R Anxiety Cognitive-behavioral .28 NA
Baker (2000) Y C Y 47 NA 47 General S Depression Pastoral care NA NC
Barron (2007) N D Y 20 19 NA General R Depression Cognitive-behavioral .73 NA
Bay et al. (2008) Y C Y 85 NA 85 General S Heart disease Pastoral care NA .21
Bowland (2008) N C Y 21 NA 22 General S Trauma Spiritual NA .56
Byers et al. (in press) Y C N 20 NA 19 Christian R Lack of hope Installation of hope NA .10
Chan, Ho et al. (2006) Y C Y 27 16 17 General S Breast cancer Body-mind-spirit .69 −.06
Chan, Ng et al. (2006) Y C Y 69 NA 115 General S Anxiety Body-mind-spirit NA NC
Cole (2005) Y C Y 9 NA 7 General S Cancer Spiritual NA −.52
Combs et al. (2000) Y C Y 30 NA 32 Christian R Marital Cognitive-behavioral NA .89
Gibbel (2010) N D Y 24 19 22 General S Depression Cognitive .56 .61
Hart & Shapiro (2002) N C Y 28 26 NA General S Unforgiveness 12-step .78 NA
Hawkins et al. (1999) Y D N 18 11 NA Christian R Depression Cognitive-behavioral .48 NA
Ho et al. (2009) Y C Y 26 33 NA General S Breast cancer Body-mind-spirit .09 NA
Hsiao et al. (2007) Y C Y 14 12 NA General S Depression Body-mind-spirit NC NA
(Continued)
409
410

Iler (2001) Y C Y 25 NA 24 General S COPD Pastoral care NA .61
Jackson (1999) N C Y 14 NA 13 Christian R Unforgiveness Promote empathy NA .91
Johnson et al. (1994) Y D Y 13 16 NA Christian R Depression Rational-emotive −.53 NA
Johnson & Ridley (1992) Y D Y 5 5 NA Christian R Depression Rational-emotive .32 NA
Lampton et al. (2005) Y C N 42 NA 23 Christian R Unforgiveness REACH NA .95
Lee et al. (2009) Y C Y 69 NA 79 General S Colon cancer Body-mind-spirit NA 1.23
Liu et al. (2008) Y C Y 12 NA 16 General S Breast cancer Body-mind-spirit NA .66
Margolin et al. (2006) Y C Y 30 30 NA Buddhist S Drug use Spiritual self-schema .64 NA
Margolin et al. (2007) Y C Y 14 11 NA Buddhist S Drug use Spiritual self-schema .27 NA
McCain et al. (2008) Y C Y 68 65 57 General S Stress, HIV Spiritual growth .24 −1.56
Miller et al. (2008)1 Y C Y 27 27 NA General S Substance use Spiritual guidance -.41 NA
Miller et al. (2008)2 Y C N 31 34 NA General S Substance use Spiritual guidance .17 NA
Nohr (2001) N D Y 35 23 14 General S Stress Cognitive-behavioral .02 .30
Pecheur & Edwards (1984) Y D Y 7 7 7 Christian R Depression Cognitive .57 2.06
Propst (1980) Y D Y 7 10 11 Christian R Depression Cognitive NC .95
Propst et al. (1992)1 Y D Y 10 9 11 Christian R Depression Cognitive-behavioral −.30 .93
Propst et al. (1992)2 Y D Y 9 10 11 Christian R Depression Cognitive-behavioral 1.44 1.47
Razali et al. (2002)1 Y C Y 45 40 NA Muslim R Anxiety Cognitive −.35 NA
(Continued)
Razali et al. (2002)2 Y C Y 42 38 NA Muslim R Anxiety Cognitive .13 NA
Razali et al. (1998)1 Y C Y 54 49 NA Muslim R Anxiety Cognitive .31 NA
Razali et al. (1998)2 Y C Y 52 48 NA Muslim R Depression Cognitive .32 NA
Richards et al. (2006) Y C Y 43 35 NA General S Eating disorders Spiritual .58 NA
Rosmarin et al. (2010) N C Y 36 42 47 Jewish R Anxiety Cognitive-behavioral .23 .45
Rye & Pargament (2002) Y D Y 19 20 19 Christian R Unforgiveness REACH .35 1.50
Rye et al. (2005) Y D Y 50 49 50 Christian R Unforgiveness REACH −.03 .28
Scott (2001) Y D N 15 3 NA Christian R Breast cancer Cognitive-behavioral .21 NA
Stratton et al. (2008) Y C N 22 NA 29 Christian R Unforgiveness REACH NA .09
Targ & Levine (2002) Y C Y 72 60 NA General S Breast cancer Body-mind-spirit .14 NA
Toh & Tan (1997) Y C Y 22 NA 24 Christian R Various Lay counseling NA .71
Tonkin (2005) Y D Y 9 9 NA Christian R Eating disorders Cognitive-behavioral −2.00 NA
Trathen (1995)1 N C Y 23 NA 22 Christian R Premarital PREP NA .05
Trathen (1995)2 N C Y 23 NA 22 Christian R Premarital PREP NA .10
Yang et al. (2009) Y C Y 17 19 NA General S Depression Body-mind-spirit NC NA
Zhang et al. (2002) Y C Y 46 48 NA Taoist S Anxiety Cognitive .85 NA
Note: RS = religious or spiritual psychotherapy; Alt = alternate psychotherapy, Ctl = control condition; Y = Yes; N = No; C = comparative design; D = dismantling design; NA = not applicable; R = religious; S = spiritual;
NC = not able to calculate effect size.

Table 20.2 Overall Results for Psychological and Spiritual Outcomes
Posttest Follow-up
Comparison N k d 95% CI I2 N k d 95% CI I2
Psychological Outcomes
Control 1,280 22 .45 0.15 to 0.75 83.84 602 8 .21 −0.43 to 0.86 92.62
Alternate 1,718 29 .26 0.10 to 0.41 57.47 610 13 .25 0.05 to 0.45 28.74
Dismantling 387 11 .13 − 0.26 to 0.52 67.87 277 8 .22 −.09 to 0.52 30.34
Spiritual Outcomes
Control 600 8 .51 0.19 to 0.84 71.18 317 4 .25 −.03 to 0.52 25.87
Alternate 707 14 .41 0.18 to 0.65 53.95 222 6 .32 −0.10 to 0.74 56.62
Dismantling 235 7 .33 0.07 to 0.59 0 126 4 .38 −0.16 to 0.91 51.96
Note: The symbol N is the sample size summed across studies. The k is the number of effect sizes summarized. The d is the weighted mean d
across samples. The 95% CI is the confidence interval for the mean d. The I2 is the percentage of the observed variance that reflects real
differences in effect sizes.
theoretical orientation and duration of tendency for studies available to the reviewer
psychotherapy—patients in R/S psycho- to be systematically different from studies
therapies outperformed patients in alter- that were unavailable such that conclusions
nate psychotherapies on spiritual outcomes may be biased. In our study, published stud-
but not on psychological outcomes. That is, ies had slightly higher effect sizes than unpub-
participants in R/S psychotherapies showed lished studies (see Table 20.3), although in
similar reductions in psychological symp- no case was this difference significant. Addi-
toms as did participants in similar alternate tionally, we used the trim and fill procedure
psychotherapies (e.g., similar reductions in (Duval & Tweedie, 2000) to estimate the
depression) but showed better results on effects of publication bias. The trim and fill
spiritual variables (e.g., greater increases in procedure estimates the number of missing
spiritual well-being). studies due to publication bias and statisti-
cally imputes these studies, recalculating the
Publication Bias overall effect size. The effect sizes were some-
We conducted a series of analyses to deter- what reduced using this procedure, but
mine whether our results were affected by the overall conclusions did not change (see
publication bias. Publication bias refers to the Table 20.4). In summary, the results of the
Table 20.3 Comparison of Published and Unpublished Studies

Level of specificity k published d published 95% CI k d 95% CI
published unpublished unpublished unpublished
Comparison with control 15 .49 .06 to 0.92 7 .41 .20 to 0.62

Comparison with alternate 23 .26 .10 to 0.41 6 .19 −.34 to 0.71
Comparison with alternate 7 .18 −.24 to 0.60 4 −.06 −.91 to 0.80
(dismantling)
Note: The symbol k refers to the number of effect sizes summarized. The statistic d is the weighted mean standardized mean difference across
samples. The 95% CI is the confidence interval of the weighted mean standardized difference.
Table 20.4 Results for Trim and Fill Analyses
Posttest
Comparison K+ d adj 95% CI
Psychological Outcomes
Control 7 .15 −.13 to 0.44
Alternate 4 .17 .01 to 0.33
Dismantling 1 .03 −.37 to 0.43
Spiritual Outcomes
Control 0 .51 .19 to 0.84
Alternate 3 .25 .03 to 0.51
Dismantling 1 .26 −.01 to 0.53
Note: The K + is the number of the studies imputed by the trim and fill procedures. The symbol d adj is the weighted mean d of the
distribution of d that contains both the observed and the imputed effects.
publication bias analyses indicate that it may One patient characteristic that might
be more difficult for studies on R/S psycho- be especially pertinent is the client’s R/S
therapies with small magnitude or negative commitment. In the vast majority of studies,
results to be published. These results should the participants have identified with a par-
be taken with caution, as these analyses were ticular religion or spirituality under investi-
conducted with a low number of studies. gation; for instance, a study on Christian
accommodative psychotherapy for depres-
Moderators sion would recruit only Christian partici-
We tested three moderators of interest— pants. However, people differ in their level
treatment format (individual vs. group), of R/S commitment. For some, R/S beliefs
target problem (psychological, forgiveness, may be little more than a tradition or
or health), and type of R/S faith commit- demographic characteristic, whereas for
ment (religious vs. spiritual). All moderator others R/S beliefs may be the driving force
analyses were conducted on psychological behind their core values, life goals, and
outcomes at posttest. None of the modera- everyday behaviors. Thus, religious commit-
tors were statistically significant. That is, ment is likely more important than beliefs
none of these variables accounted for appre- or a religious demographic identification
ciable variance in the effect size estimates in (Worthington, 1988). We suggest that
the reviewed studies. including R/S beliefs into psychotherapy
may be more important for clients that are
Patient Contributions highly R/S committed than for clients who
The research reviewed in the present meta- are less R/S committed. There is modest
analysis focused on the psychotherapist’s support for this hypothesis in a recent effec-
contribution to the relationship. That is, tiveness—not randomized clinical trial—
analysis has addressed the question of study (Wade, Worthington, & Vogel, 2007).
whether it is helpful to tailor the psycho- Unfortunately, this hypothesis has not
therapy to the client’s religious and spiritual been addressed frequently enough to be
proclivities. However, characteristics of indi- tested in the present review. The vast major-
vidual clients probably also affect tailoring. ity of studies have simply required that
participants identify with the particular research, but it is sometimes difficult to
religion that is integrated with the psycho- accomplish in studies of R/S psychotherapy.
therapy or indicate that they are open to Religion is an emotionally charged topic for
participate in a psychotherapy that includes many people, and thus, highly religious
spirituality. Two studies (Nohr, 2001; people may be less willing to be randomized
Razali, Aminah, & Khan, 2002) assessed to a secular treatment, and adamantly nonre-
the efficacy of R/S psychotherapies using ligious people may not be willing to be
clients with different levels of religious randomized to a religious treatment.
commitment. But their findings were Another limitation of this meta-analysis
mixed. Thus, there is not sufficient research was publication bias. Our analyses indi-
on this patient factor to make viable conclu- cated that some studies indicating negative
sions or clinical recommendations. or null findings for R/S psychotherapies
may have been unpublished, literally sitting
Limitations of the Research in a file-drawer somewhere. There are several
There are limitations of the research on R/S possible reasons for publication bias in this
psychotherapies. First, although the quality literature. First, much of the research on
of studies has improved in the past several R/S psychotherapy is conducted by research-
years, some studies still suffered from less ers who have religious orientations. Author
rigorous study designs and low power. In decisions may be a cause of the apparent
particular, there were relatively few compar- publication bias. When the results of a
isons (n = 11 with psychological effect sizes; study do not support the efficacy of R/S
n = 7 with spiritual effect sizes) that met psychotherapy or yield an estimate of effi-
the criteria for a dismantling design, mean- cacy that is small, it may be that the authors
ing they compared R/S psychotherapy with tend not to submit the paper for publica-
an alternate psychotherapy that was the tion. Second, when the research is pub-
same in theoretical orientation and dura- lished, some of the it has been published in
tion. These types of studies are especially religiously oriented journals. Editors and
important because they best answer the reviewers for journals with a religious theme
empirical question of whether it improves may accept papers that are supportive of
efficacy to incorporate R/S beliefs in an R/S psychotherapy more frequently than
existing psychotherapy for R/S clients. those that are not. Third, editors may be
Studies that compare R/S psychotherapy reluctant to publish comparative studies
with a completely different type of psycho- that report null findings because it is diffi-
therapy can be rigorous as well. However, cult to determine whether these results
if participants in the R/S psychotherapy reflect (a) no true difference between condi-
outperform participants in the alternate tions or (b) problems in the study design
psychotherapy, it is difficult to discern and implementation (e.g., low power).
whether this occurred because (a) the
specific R/S elements caused the differen-
tial outcomes or (b) something else that
To conclude, we offer several concrete appli-
was different between the two psychothera-
cations for clinical practice based on the
pies caused the differential outcomes.
findings from our meta-analytic review.
Many studies with comparative designs
used random assignment to conditions, but • R/S psychotherapy works. The
some did not. Random assignment to condi- research evidence is consistent that R/S
tions is the gold standard of psychotherapy psychotherapies are efficacious at
improving both psychological and accommodating patient preferences
spiritual outcomes, and there is some modestly enhances treatment outcomes
evidence that these gains are maintained and decreases premature termination by
at follow-up. Thus, R/S psychotherapies a third (Swift, Callahan, & Vollmer,
should be viewed as a valid alternative Chapter 15, this volume).
treatment option for R/S clients. • We hypothesize that incorporating
• The addition of R/S beliefs R/S beliefs or practices into psychotherapy
or practices to an established secular might be more efficacious with clients
psychotherapy does not reliably improve who are highly religiously or spiritually
psychological outcomes for R/S clients committed. Few studies have addressed
over and above the effects of the this hypothesis, but there is no research
established secular psychotherapy alone. or clinical evidence to suggest that R/S
Although there was some evidence that psychotherapies produce worse outcomes
R/S psychotherapies outperformed than secular therapies for these patients.
alternate psychotherapies, that difference Thus, we recommend that practitioners
was reduced when the analysis was limited consider offering R/S treatment to highly
to studies that used a dismantling design. religious or spiritual patients.
Thus, at this time there is no empirical
basis to recommend R/S psychotherapies References
over established secular psychotherapies An asterisk (∗) indicates studies included in
when the primary or exclusive treatment the meta-analysis.
outcome is symptom remission. Allport, G. W., & Ross, J. M. (1967). Personal
• R/S psychotherapies offer religious orientation and prejudice. Journal of
spiritual benefits to clients that are not Personality and Social Psychology, 5, 432–33.
Avants, S. K., & Margolin, A. (2004). Development
present in secular psychotherapies. The of Spiritual Self-Schema (3-S) therapy for the
meta-analytic results indicate that patients treatment of addictive and HIV risk behavior:
in R/S psychotherapies showed more A convergence of cognitive and Buddhist psycho-
improvement on spiritual outcomes than logy. Journal of Psychotherapy Integration, 14(3),
did patients in alternate psychotherapies, 253–289.
∗
even when this analysis was limited to Azhar, M. Z., & Varma, S. L. (1995a). Religious
psychotherapy as management of bereavement.
studies that used a dismantling design. Acta Psychiatrica Scandinavica, 91, 233–35.
Thus, for those patients and contexts in ∗
Azhar, M. Z., & Varma, S. L. (1995b). Religious psy-
which spiritual outcomes are highly chotherapy in depressive patients. Psychotherapy
valued, R/S psychotherapy can be and Psychosomatics, 63, 165–73.
∗
considered a treatment of choice. Azhar, M. Z., Varma, S. L., & Dharap, A. S. (1994).
• The incorporation of R/S beliefs or Religious psychotherapy in anxiety disorder
patients. Acta Psychiatrica Scandinavica, 90,
practices into psychotherapy should follow 1–3.
the desires and needs of the particular ∗
Baker, D. C. (2000). The investigation of pastoral
client. Psychotherapists are encouraged care interventions as a treatment for depression
to ask about R/S beliefs and commitment among continuing care retirement community
as part of the intake process and to residents. Journal of Religious Gerontology, 12,
incorporate them into psychotherapy 63–85.
∗
Barron, L. W. (2007). Effect of religious coping skills
(a) as they feel comfortable and (b) in training with group cognitive-behavioral therapy
line with the preferences of the particular for treatment of depression. Unpublished doctoral
client. Research summarized elsewhere dissertation, Northcentral University, Prescott
in this volume demonstrates that Valley, Arizona.
∗ roles of attachment to God, religious coping, and
Bay, P. S., Beckman, D., Trippi, J.,
Gunderman, R., & Terry, C. (2008). The effect viewing the transgression as a desecration. Journal
of pastoral care services on anxiety, depression, of Psychology and Christianity, 27, 293–301.
hope, religious coping, and religious problem Duval, S. J., & Tweedie, R. L. (2000). A non-
solving styles: A randomized controlled study. parametric “trim and fill” method of accounting
Journal of Religious Health, 47, 57–69. for publication bias in meta-analysis. Journal of
Beck, A. T. (1972). Depression: Causes and treatment. the American Statistical Association, 95, 89–98.
Philadelphia: University of Pennsylvania Press. Ellison, C. W. (1983). Spiritual well-being:
Beck, A. T., Rush, A. J., Shaw, B. F., & Emery, G. Conceptualization and measurement. Journal of
(1979). Cognitive therapy of depression. New York: Psychology and Theology, 11, 330–40.
∗
Guilford Press. Gibbel, M. R. (2010). Evaluating a spiritually
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., integrated intervention for depressed college students.
& Erbaugh, J. (1961). An inventory for measur- Unpublished doctoral dissertation, Bowling
ing depression. Archives of General Psychiatry, Green State University, OH.
4, 561–71. Harris, A. H. S., Thoresen, C. E., McCullough, M. E.,
Borenstein, M., Hedges, L., Higgins, J., & & Larson, D. B. (1999). Spiritually and religiously
Rothstein, H. R. (2005). Comprehensive meta- oriented health interventions. Journal of Health
analysis, Version 2. Englewood, NJ: Biostat, Inc. Psychology, 4, 413–33.
∗
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Hart, K. E., & Shapiro, D. A. (2002, August).
Rothstein, H. R. (2009). Introduction to meta- Secular and spiritual forgiveness interventions
analysis. Chichester, UK: Wiley. for recovering alcoholics harboring grudges.
∗
Bowland, S. E. (2008). Evaluation of a psycho- Paper presented at the Annual Convention of the
social-spiritual intervention with older women American Psychological Association, Chicago.
∗
survivors of interpersonal trauma. Unpublished Hawkins, R. S., Tan, S., & Turk, A. A. (1999).
doctoral dissertation, Washington University in Secular versus Christian inpatient cognitive-
St. Louis, MO. behavioral therapy programs: Impact on depres-
∗ sion and spiritual well-being. Journal of Psychology
Byers, A. B., Underwood, L. A., & Hardy, V. L. (in
press). Spirituality group with female offenders: and Theology, 27, 309–318.
Impacting hope. Journal of Correctional Health Hill, P. C., Pargament, K. I., Hood, R. W., Jr.,
Care, in press. McCullough, M. E., Swyeres, J. P., Larson, D. B.,
∗ et al. (2000). Conceptualizing religion and
Chan, C. L. W., Ho, R. T. H., Lee, P. W. H., Cheng,
J. Y. Y., Leung, P. P. Y., Foo, W., et al. (2006). A spirituality: Points of commonality, points of
randomized controlled trial of psychosocial inter- departure. Journal for the Theory of Social
ventions using the psychophysiological framework Behavior, 30, 51–77.
∗
for Chinese breast cancer patients. Journal of Ho, T. H., Lo, P. H. Y., & Chan, C. L. W. (2009).
Psychosocial Oncology, 24, 3–26. The efficacy of the body-mind-spirit interven-
∗ tion and social support groups on chinese breast
Chan, C. H. Y., Ng, E. H. Y., Chan, C. L. W.,
Ho, & Chan, T. H. Y. (2006). Effectiveness of cancer patients. In Y. M. Lee, C. L. W. Chan,
psychosocial group intervention for reducing S. M. Ng, & P. P. Y. Leung.(Eds.), Integrative
anxiety in women undergoing in vitro fertiliza- Body-Mind-Spirit Social Work: An Empirically
tion: A randomized controlled study. Fertility Based Approach to Assessment and Treatment
and Sterility, 85, 339–46. (pp. 217–34). New York: The Oxford University
∗ Press.
Cole, B. S. (2005). Spiritually-focused psycho-
therapy for people diagnosed with cancer: Hodge, D. R. (2006). Spiritually modified cognitive
A pilot outcome study. Mental Health, Religion, therapy: A review of the literature. Social Work,
and Culture, 8, 217–226. 51, 157–66.
∗ Hook, J. N., Worthington, E. L., Jr., Davis, D. E.,
Combs, C. W., Bufford, R. K., Campbell, C. D., &
Halter, L. L. (2000). Effects of cognitive-behav- Jennings, D. J., II., Gartner, A. L., & Hook, J. P.
ioral marriage enrichment: A controlled study. (2010). Empirically supported religious and
Marriage and Family: A Christian Journal, 3, spiritual therapies. Journal of Clinical Psychology,
99–111. 66, 46–72.
∗
Davis, D. E., Hook, J. N., & Worthington, E. L., Jr. Hsiao, F. H., Yang, T. T., Chen, C. C., Tsai, S. Y.,
(2008). Relational spirituality and forgiveness: The Wang, K. C., Lai, Y. M., et al. (2007).
The comparison of effectiveness of two modalities ∗
Margolin, A., Beitel, M., Schuman-Olivier, Z., &
of mental health nurse follow-up programmes for Avants, S. K. (2006). A controlled study of a
female outpatients with depression in Taipei, spirituality-focused intervention for increasing
Taiwan. Journal of Clinical Nursing, 16, 1141–50. motivation for HIV prevention among drug
∗
Iler, W. L. (2001). The impact of daily visits from users. AIDS Education and Prevention, 18,
chaplains on patients with chronic obstructive 311–22.
pulmonary disease (COPD): A pilot study. ∗
Margolin, A., Schuman-Olivier, Z., Beitel, M.,
Chaplaincy Today, 17, 5–11. Arnold, R. M., Fulwiler, C. E., & Avants, S. K.
∗
Jackson, R. E. (1999). Reducing shame through (2007). Journal of Clinical Psychology, 63, 979–99.
forgiveness and empathy: A group therapy approach ∗
McCain, N. L., Gray, D. P., Elswick, R. K., Jr.,
to promoting prosocial behavior. Unpublished Robins, J. W., Tuck, I., Walter, J. M., et al.
doctoral dissertation, Fuller Theological (2008). A randomized clinical trial of alternative
Seminary, Pasadena, CA. stress management interventions in persons with
∗
Johnson, W. B., DeVries, R., Ridley, C. R., HIV infection. Journal of Consulting and Clinical
Pettorini, D., & Peterson, D. R. (1994). The Psychology, 76, 431–41.
comparative efficacy of Christian and secular McCullough, M. E. (1999). Research on religion-
rational-emotive therapy with Christian clients. accommodative counseling: Review and meta-
Journal of Psychology and Theology, 22, 130–40. analysis. Journal of Counseling Psychology, 46,
∗
Johnson, W. B., & Ridley, C. R. (1992). Brief 92–98.
Christian and non-Christian rational-emotive ∗
Miller, W. R., Forcehimes, A., O’Leary, M. J., &
therapy with depressed Christian clients: An LaNoue, M. D. (2008). Spiritual direction in
exploratory study. Counseling and Values, 36, addiction treatment: Two clinical trials. Journal
220–29. of Substance Abuse Treatment, 35, 434–42.
∗ ∗
Lampton, C., Oliver, G. J., Worthington, E. L., Jr., Nohr, R. W. (2001). Outcome effects of receiving
& Berry, J. W. (2005). Helping Christian college a spiritually informed vs. a standard cognitive-
students become more forgiving: An interven- behavioral stress management workshop.
tion study to promote forgiveness as part of a Unpublished doctoral dissertation, Marquette
program to shape Christian character. Journal of University, Milwaukee, WI.
Psychology and Theology, 33, 278–90. Norcross, J. C. (1990). An eclectic definition of
∗
Lee, A. M., Chan, C. L. W., Ho, A. H. Y., psychotherapy. In J. K. Zeig & W. M. Munion
Wang, C. N., Tang, V. Y. H., Lau S. S. M., et al. (Eds.), What is psychotherapy? Contemporary
(2009). The efficacy of the body-mind-spirit perspectives (pp. 218–20). San Francisco:
intervention model on improving the quality of Jossey-Bass.
life and psychological well-being of chinese Norcross, J. C. (Ed.). (2002). Psychotherapy
patients with colorectal cancer: A preliminary relationships that work: Therapist contributions
report. In Y. M. Lee, C. L. W. Chan, S. M., Ng, and responsiveness to patients. New York: Oxford
& P. P. Y. Leung (Eds.), Integrative body-mind- University Press.
spirit social work: An empirically based approach to Pargament, K. I. (2007). Spiritually integrated
assessment and treatment (pp. 236–43). New York: psychotherapy: Understanding and addressing the
The Oxford University Press. sacred. New York: Guilford Press.
Lipsey, M. W., & Wilson, D. B. (2001). Practical Pargament, K. I., & Saunders, S. M. (2007).
meta-analysis. Thousand Oaks, CA: Sage. Introduction to the special issue on spirituality
∗
Liu, C. J., Hsiung, P. C., Chang, K. J., Liu, Y. F., and psychotherapy. Journal of Clinical Psychology,
Want, K. C., Hsiao, F. H., et al. (2008). A study 63, 903–907.
on the efficacy of body-mind-spirit group ther- ∗
Pecheur, D. R., & Edwards, K. J. (1984).
apy for patients with breast cancer. Journal of A comparison of secular and religious versions of
Clinical Nursing, 17, 2539–49. cognitive therapy with depressed Christian
Luborsky, L., Diguer, L., Seligman, D. A., college students. Journal of Psychology and
Rosenthal, R., Krause, E. D., Johnson, S., et al. Theology, 12, 45–54.
(1999). The researcher’s own therapy allegiances: Post, B. C., & Wade, N. G. (2009). Religion and
A “wild card” in comparisons of treatment spirituality in psychotherapy: A practice-friendly
efficacy. Clinical Psychology: Science and Practice, review of research. Journal of Clinical Psychology:
6, 95–106. In Session, 65, 131–46.
∗ Forgiveness interventions as spiritual develop-
Propst, L. R. (1980). The comparative efficacy of
religious and nonreligious imagery for the treatment strategies: Comparing forgiveness workshop
ment of mild depression in religious individuals. training, expressive writing about forgiveness,
Cognitive Therapy and Research, 4, 167–78. and retested controls. Journal of Psychology and
∗ Christianity, 27, 347–57.
Propst, L. R., Ostrom, R., Watkins, P., Dean, T., &
∗
Mashburn, D. (1992). Comparative efficacy of Targ, E. F., & Levine, E. G. (2002). The efficacy
religious and nonreligious cognitive-behavioral of a mind-body-spirit group for women with
therapy for the treatment of clinical depression breast cancer: A randomized clinical trial. General
in religious individuals. Journal of Consulting and Hospital Psychiatry, 24, 238–48.
∗
Clinical Psychology, 60, 94–103. Toh, Y., & Tan, S. (1997). The effectiveness
∗ of church-based lay counselors: A controlled
Razali, S. M., Aminah, K., & Khan, U. A. (2002).
Religious-cultural psychotherapy in the manage- outcome study. Journal of Psychology and
ment of anxiety patients. Transcultural Psychiatry, Christianity, 16, 260–67.
∗
39, 130–36. Tonkin, K. M. (2005). Obesity, bulimia, and binge-
∗ eating disorder: The use of a cognitive behavioral and
Razali, S. M., Hasanah, C. I., Aminah, K., &
Subramaniam, M. (1998). Religious-sociocultural spiritual intervention. Unpublished doctoral dis-
psychotherapy in patients with anxiety and sertation, Bowling Green State University, OH.
∗
depression. Australian and New Zealand Journal Trathen, D. W. (1995). A comparison of the
of Psychiatry, 32, 867–72. effectiveness of two Christian premarital counseling
∗ programs (skills and information-based) utilized by
Richards, P. S., Berrett, M. E., Hardman, R. K., &
Eggett, D. L. (2006). Comparative efficacy of evangelical Protestant churches. Unpublished
spirituality, cognitive, and emotional support doctoral dissertation, University of Denver, CO.
groups for treating eating disorder inpatients. Wade, N. G., Worthington, E. L., Jr., & Vogul, D. L.
Eating Disorders, 14, 401–415. (2007). Effectiveness of religiously tailored inter-
∗ ventions in Christian therapy. Psychotherapy
Rosmarin, D. H., Pargament, K. I., Pirutinsky, S.,
& Mahoney, A. (2010). A randomized controlled Research, 17, 91–105.
evaluation of a spiritually-integrated cognitive- Worthington, E. L., Jr. (1988). Understanding
behavioral intervention for subclinical anxiety the values of religious clients: A model and its
among religious Jews. Unpublished manuscript, application to counseling. Journal of Counseling
Bowling Green State University, OH. Psychology, 35, 166–74.
∗ Worthington, E. L., Jr. (1998). The pyramid model
Rye, M. S., & Pargament, K. I. (2002). Forgiveness
and romantic relationships in college: Can it heal of forgiveness: Some interdisciplinary specula-
the wounded heart? Journal of Clinical Psychology, tions about unforgiveness and the promotion of
58, 419–41. forgiveness. In E. Worthington (Ed.), Dimensions
∗ of forgiveness: Psychological research and theological
Rye, M. S., Pargament, K. I., Pan, W.,
Yingling, D. W., Shogren, K. A., & Ito, M. perspectives (pp. 107–38). Philadelphia: Templeton
(2005). Can group interventions facilitate Foundation Press.
forgiveness of an ex-spouse? A randomized Worthington, E. L., Jr. (2009). A just forgiveness:
clinical trial. Journal of Consulting and Clinical Responsible healing without excusing injustice.
Psychology, 73, 880–92. Downers Grove, IL: InterVarsity Press.
∗ Worthington, E. L., Jr., & Aten, J. D. (2009).
Scott, S. (2001). Faith supportive group therapy
and symptom reduction in Christian breast cancer Psychotherapy with religious and spiritual clients:
patients. Unpublished doctoral dissertation, An introduction. Journal of Clinical Psychology:
Regent University, Virginia Beach, VA. In Session, 65, 123–30.
Sheldrake, P. (1992). Spirituality and history: Worthington, E. L., Jr., Kurusu, T. A.,
Questions of interpretation and method. New York: McCullough, M. E., & Sandage, S. J. (1996).
Crossroads. Empirical research on religion and psychothera-
Smith, T. B., Bartz, J., & Richards, P. S. (2007). peutic processes and outcomes: A 10-year review
Outcomes of religious and spiritual adaptations and research prospectus. Psychological Bulletin,
to psychotherapy: A meta-analytic review. 119, 448–87.
Psychotherapy Research, 17, 643–55. Worthington, E. L., Jr., & Sandage, S. J. (2001).
∗ Religion and spirituality. Psychotherapy: Theory,
Stratton, S. P., Dean, J. B., Nonneman, A. J.,
Bode, R. A., & Worthington, E. L., Jr. (2008). Research, Practice, Training, 38, 473–478.
Worthington, E. L., Jr., Wade, N. G., Hight, T. L., The effect of psychotherapy added to pharmaco-
Ripley, J. S., McCullough, M. E., Berry, J. W., therapy on cortisol responses. Journal of Nervous
Schmitt, M. M., Berry, J. T., et al. (2003). and Mental Disease, 197, 401–406.
∗
The religious commitment inventory—10: Zhang, Y., Young, D., Lee, S., Li, L., Zhang, H.,
Development, refinement, and validation of a Xiao, Z., et al. (2002). Chinese Taoist cognitive
brief scale for research and counseling. Journal of psychotherapy in the treatment of generalized
Counseling Psychology, 50, 84–96. anxiety disorder in contemporary China.
∗
Yang, T. T., Hsiao, F. H., Wang, K. C., Ng, S. M., Transcultural Psychiatry, 39, 115–29.
Ho, R. T. H., Chan, C. L. W., et al. (2009).
PART
4
Conclusions and
Guidelines
C HA P TER
Evidence-Based Therapy Relationships:
21 Research Conclusions and Clinical Practices
John C. Norcross and Bruce E. Wampold
We shall not cease from exploration • Practice and treatment guidelines

And the end of all our exploring should explicitly address therapist
Will be to arrive where we started behaviors and qualities that promote a
And know the place for the first time. facilitative therapy relationship.
—T. S. Eliot (“Little Gidding” in Four Quartets) • Efforts to promulgate best practices or
evidence-based practices (EBPs) without
Having traversed more than two dozen including the relationship are seriously
meta-analyses and arrived at the end of this incomplete and potentially misleading.
book, we have the opportunity to present • Adapting or tailoring the therapy
the interdivisional Task Force conclusions relationship to specific patient
and to reflect on its work. Like the tireless characteristics (in addition to diagnosis)
traveler in Eliot’s poem, we have rediscov- enhances the effectiveness of treatment.
ered the therapy relationship and know it, • The therapy relationship acts in
again, for the first time. concert with treatment methods, patient
This closing chapter presents the conclu- characteristics, and practitioner qualities
sions and recommendations of the second in determining effectiveness; a
Task Force on Evidence-Based Therapy comprehensive understanding of effective
Relationships. These statements reaffirm (and ineffective) psychotherapy will
and, in several instances, update those of consider all of these determinants and
the earlier Task Force (Norcross, 2001, their optimal combinations.
2002). We then offer some final thoughts • The following table summarizes the
on what works, what doesn’t work, and Task Force conclusions regarding the
clinical practice. evidentiary strength of (a) elements of the
therapy relationship primarily provided by
Conclusions of the Task Force the psychotherapist and (b) methods of
• The therapy relationship makes adapting psychotherapy to particular
substantial and consistent contributions to patient characteristics.
psychotherapy outcome independent of • These conclusions do not by
the specific type of treatment. themselves constitute a set of practice
• The therapy relationship accounts for standards but represent current scientific
why clients improve (or fail to improve) at knowledge to be understood and applied
least as much as the particular treatment in the context of all the clinical evidence
method. available in each case.
423
Elements of the relationship Methods of adapting
Demonstrably effective Alliance in individual psychotherapy Reactance/Resistance level
Alliance in youth psychotherapy Preferences
Alliance in family therapy Culture
Cohesion in group therapy Religion and spirituality
Empathy
Collecting client feedback
Probably effective Goal consensus Stages of change
Collaboration Coping style
Positive regard
Promising but insufficient Congruence/Genuineness Expectations
research to judge Repairing alliance ruptures Attachment style
Managing countertransference
Recommendations of the Task Force 5. Practitioners are encouraged to adapt

General Recommendations or tailor psychotherapy to those specific
patient characteristics in ways found to be
1. We recommend that the results and
demonstrably and probably effective.
conclusions of this second Task Force be
6. Practitioners are encouraged to
widely disseminated in order to enhance
routinely monitor patients’ responses to
awareness and use of what “works” in the
the therapy relationship and ongoing
therapy relationship.
treatment. Such monitoring leads to
2. Readers are encouraged to interpret
increased opportunities to reestablish
these findings in the context of the
collaboration, improve the relationship,
acknowledged limitations of the Task
modify technical strategies, and avoid
Force’s work.
premature termination.
3. We recommend that future Task
7. Concurrent use of evidence-based
Forces be established periodically to
therapy relationships and evidence-based
review these findings, include new
treatments adapted to the patient is likely
elements of the relationship, incorporate
to generate the best outcomes.
the results of non-English language
publications (where practical), and
update these conclusions.
Training Recommendations
8. Training and continuing education
Practice Recommendations programs are encouraged to provide
4. Practitioners are encouraged to make competency-based training in the
the creation and cultivation of a therapy demonstrably and probably effective
relationship, characterized by the elements elements of the therapy relationship.
found to be demonstrably and probably 9. Training and continuing education
effective, a primary aim in the treatment programs are encouraged to provide
of patients. competency-based training in adapting
424 co n c lu s i o n s a n d g u i d e l i n e s
psychotherapy to the individual patient in 15. Mental health organizations as a
ways that demonstrably and probably whole are encouraged to educate their
enhance treatment success. members about the improved outcomes
10. Accreditation and certification associated with using evidence-based
bodies for mental health training therapy relationships, as they frequently
programs should develop criteria for now do about evidence-based treatments.
assessing the adequacy of training in 16. We recommend that the American
evidence-based therapy relationships. Psychological Association and other
mental health organizations advocate for
the research-substantiated benefits of a
Research Recommendations
nurturing and responsive human
11. Researchers are encouraged to relationship in psychotherapy.
progress beyond correlational designs 17. Finally, administrators of mental
that associate the frequency of relationship health services are encouraged to attend to
behaviors with patient outcomes to the relational features of those services.
methodologies capable of examining the Attempts to improve the quality of care
complex associations among patient should account for treatment relationships
qualities, clinician behaviors, and and adaptations.
treatment outcome. Of particular
importance is disentangling the patient
contributions and the therapist What Works
contributions to relationship elements The process by which the preceding con-
and ultimately outcome. clusions on which relationship elements and
12. Researchers are encouraged to adaptation methods are effective requires
examine the specific mediators and some elaboration as these conclusions tend
moderators of the links between the to be the most cited and controversial find-
relationship elements and treatment ings of the Task Force. These conclusions
outcome. represent the consensus of expert panels
13. Researchers are encouraged to composed of five judges who indepen-
address the observational perspective dently reviewed and rated the empirical
(i.e., therapist, patient, or external rater) evidence. They evaluated, for each relation-
in future studies and reviews of “what ship element or adaptation method, the
works” in the therapy relationship. previous research summary and the new
Agreement among observational meta-analysis according to the following
perspectives provides a solid sense of criteria: number of empirical studies, con-
established fact; divergence among sistency of empirical results, independence
perspectives holds important of supportive studies, magnitude of associ-
implications for practice. ation between the relationship element and
outcome, evidence for causal link between
relationship element and outcome, and the
Policy Recommendations ecological or external validity of research.
14. APA’s Division of Psychotherapy, Their respective ratings of demonstrably
Division of Clinical Psychology, and other effective, probably effective, or promis-
practice divisions are encouraged to ing but insufficient research to judge were
educate its members on the benefits of then combined to render a consensus. In
evidence-based therapy relationships. this way, we added a modicum of rigor
n o rc ro s s , wa m p o l d 425
and consensus to the process, which was must be accompanied by implementation
admittedly less so in the first edition of materials, training, and support resources
the book. that are ready to use by the public. By these
The consensus deemed six of the rela- standards, practically all of the relationship
tionship elements as demonstrably effec- elements and adaptation methods in this
tive, three as probably effective, and three volume would be considered demonstrably
as promising but insufficient research to effective, if not for the requirement of a
judge. The consensus of another panel randomized clinical trial, which is neither
deemed four adaptation methods as demon- clinically nor ethically feasible for the vast
strably effective, two as probably effective, majority of the relationship elements.
and two as promising but insufficient In important ways, the criteria for rela-
research to judge. We were impressed by the tionship elements are more rigorous.
skepticism and precision of the panelists (as Whereas the criteria for designating treat-
scientists ought to be). At the same time, ments as evidence based relies on only one
we were impressed by the disparate and or two studies, the evidence for relation-
perhaps elevated standards against which ship elements and adaptation methods dis-
these relationship elements were evaluated. cussed here are based on comprehensive
Consider the evidentiary strength meta-analyses of many studies (in excess of
required for psychological treatments to 50 in several cases), spanning various treat-
be considered demonstrably efficacious in ments and research groups. The studies
two influential compilations of evidence- used to establish evidence-based treatments
based treatments. The Division of Clinical are, however, clinical trials, which are often
Psychology’s Subcommittee on Research- designated as the “gold standard” for estab-
Supported Treatments (www.div12.org/ lishing evidence. Nevertheless, these stud-
PsychologicalTreatments/index.html) requires ies are often plagued by confounds such
two between-group design experiments as researcher allegiance, cannot be blinded,
demonstrating that a psychological treat- and often contain bogus comparisons
ment is either (a) statistically superior to a (Luborsky et al., 1999; Mohr et al., 2009;
pill or psychological placebo or to another Wampold, 2001; Wampold et al., 2010).
treatment or (b) equivalent to an already The point here is not to denigrate the crite-
established treatment in experiments with ria used to establish evidence-based treat-
adequate sample sizes. The studies must ments, but to underscore the robust
have been conducted with treatment man- scientific standards by which these relation-
uals and conducted by at least two different ship elements and adaptation methods have
investigators. The typical effect size of those been evaluated.
studies was often smaller than the effects A further research complication, but a
for the relationship elements reported in clinical strength, concerns responsiveness.
this book. For listing in SAMHSA’s National Research on the effectiveness of the psycho-
Registry of Evidence-based Programs and therapy relationship is constrained by ther-
Practices (www.nrepp.samhsa.gov), only evi- apist responsiveness—the ebb and flow of
dence of statistically significant behavioral clinical interaction. Responsiveness refers
outcomes demonstrated in at least one study, to therapist behavior that is affected by
using an experimental or quasi-experimental emerging context, and occurs on many
design, that has been published in a peer- levels, including choice of a treatment
reviewed journal or comprehensive evalua- method, case formulation, strategic use of
tion report is needed. The intervention the self, and then adjusting those to meet
the emerging, evolving needs of the client what does not. In the following section, we
in any given moment (Stiles, Honos-Webb, highlight those therapist relational behav-
& Surko, 1998). Effective psychotherapists iors that are ineffective, perhaps even hurt-
are responsive to the different needs of their ful, in psychotherapy.
clients, providing varying levels of relation- One means of identifying ineffective
ship elements in different cases and, within qualities of the therapeutic relationship is
the same case, at different moments. to simply reverse the effective behaviors.
Successful responsiveness can confound Thus, what do not work are poor alliances
attempts to find naturalistically observed in individual psychotherapy, lack of cohe-
linear relations of outcome with therapist sion in group therapy, and discordance in
behaviors (e.g., cohesion, positive regard). couple and family therapy. Paucity of empa-
Because of such problems, the statistical thy, collaboration, consensus, and positive
relations between the relationship and out- regard predict treatment dropout and fail-
come cannot always be trusted. By being ure. The ineffective practitioner will neither
clinically attuned and flexible, psychother- seek nor respond to client feedback, will
apists make it more difficult in research ignore alliance ruptures, and will not be
studies to discern what works. aware of his/her countertransference. And
In this volume, the relationship elements less effective psychotherapists will rarely
and adaptation methods are presented as tailor or customize treatment to patient
separate, stand-alone practices. But as every characteristics beyond diagnosis.
seasoned psychotherapist knows, this is Another means of identifying ineffec-
certainly never the case in clinical work. tive qualities of the relationship is to scour
The alliance in individual therapy and the research literature and conduct polls of
cohesion in group therapy never act in iso- experts. Here are several behaviors to avoid
lation from other relationship behaviors, according to that research (Duncan, Miller,
such as empathy or support. Nor does it Wampold, & Hubble, 2010) and a Delphi
seem humanly possible to cultivate a strong poll (Norcross, Koocher, & Garofalo,
relationship with a patient without ascer- 2006):
taining his/her feedback on the therapeutic
process and understanding the therapist’s • Confrontations. Controlled research
countertransference. Likewise, adapting trials, particularly in the addictions field,
treatment to a patient characteristic rarely consistently find a confrontational style to
occurs in isolation from other elements, be ineffective. In one review (Miller,
such as forming a collaborative relationship Wilbourne, & Hettema, 2003),
with the patient. Stage of change, reactance confrontation was ineffective in all 12
level, culture, preferences, and the like identified trials. By contrast, expressing
all interconnect as we try to tailor therapy empathy, rolling with resistance,
to the unique, complex individual. In developing discrepancy, and supporting
short, while the relationship elements self-efficacy, characteristic of Motivational
and adaptation methods featured in this Interviewing, have demonstrated large
book “work,” they work together and effects with a small number of sessions
interdependently. (Lundahl & Burke, 2009).
• Negative Processes. Client reports and
What Doesn’t Work research studies converge in warning
Translational research is both prescriptive therapists to avoid comments or behaviors
and proscriptive; it tells us what works and that are hostile, pejorative, critical,
rejecting, or blaming (Binder & Strupp, needs of the client, not by imposing a
1997; Lambert & Barley, 2002). Procrustean bed onto unwitting
Therapists who attack a client’s consumers of psychological services.
dysfunctional thoughts or relational We should all avoid the crimes of
patterns need, repeatedly, to distinguish Procrustes, the legendary Greek
between attacking the person versus her innkeeper who would cut the long
behavior. limbs of clients or stretch short limbs to
• Assumptions. Psychotherapists who fit his one-size bed.
assume or intuit their client’s perceptions • Singularity. In the quest to adapt
of relationship satisfaction and treatment psychotherapy, some psychotherapists
success frequently misjudge these aspects. become enamored of a single matching
By contrast, therapists who specifically protocol and apply that match to virtually
and respectfully inquire about their every patient who crosses their path. They
client’s perceptions frequently enhance are convinced that a single adaptation, be
the alliance and prevent premature it the patient’s reactance, diagnosis,
termination (Lambert & Shimokawa, culture, or stage of change, is the exclusive
this volume, Chapter 10). means of tailoring treatment to a
• Therapist Centricity. A recurrent lesson successful outcome. However, the research
from process-outcome research is that the appraised in this book convincingly
client’s observational perspective on the demonstrates that many adaptations
therapy relationship best predicts outcome succeed. We must also guard against
(Orlinsky, Ronnestad, & Willutzki, 2004). imposing the Procrustean bed when we
Psychotherapy practice that relies on the adapt psychotherapy; one size, even in
therapist’s observational perspective, while adaptation or tailoring, never works
valuable, does not predict outcome as well. for all clients.
Therefore, privileging the client’s • Flexibility without Fidelity. The desire
experiences is central. to be flexible and responsive with patients
• Rigidity. By inflexibly and excessively frequently gives rise to a clinical dilemma
structuring treatment, the therapist risks (Norcross, Hogan, & Koocher, 2008).
empathic failures and inattentiveness to Flexibility to the patient’s preferences or
clients’ experiences. Such a therapist is culture offers the promise that it “fits” but
likely to overlook a breach in the not necessarily of research support for that
relationship and mistakenly assume she preferred treatment. Fidelity to a research-
has not contributed to that breach. supported treatment offers the promise
Dogmatic reliance on particular relational that it “works” but not necessarily with
or therapy methods, incompatible with that particular patient or population.
the client, imperils treatment (Ackerman Errors in either direction can portend
& Hilsenroth, 2001). clinical failure, but after half a book
• Procrustean Bed. As the field of dedicated to the benefits of treatment
psychotherapy has matured, using an adaptation, we should note the downside
identical therapy relationship (and of ignoring brand-name therapies that
treatment method) for all clients is now possess considerable empirical evidence.
recognized as inappropriate and, in select Practitioners can become overly flexible
cases, even unethical. The efficacy and when employing a treatment without any
applicability of psychotherapy will be research evidence or when adapting a
enhanced by tailoring it to the unique treatment in ways that markedly deviate
from its established effectiveness. While bring themselves—their origins, cultures,
the research supports adaptation in many personalities, psychopathology, expecta-
cases, the research also recommends tions, biases, defenses, and strengths—to
fidelity to treatments as found effective in the human relationship. Some will judge
controlled research. We need to balance that relationship to be a precondition of
flexibility with fidelity. change and others a process of change, but
We can optimize therapy relationships all agree that it is a relational enterprise.
by simultaneously using what works and Fourth and final, how we create and cul-
studiously avoiding what does not work. tivate that powerful human relationship
can be guided by the fruits of research. As
Carl Rogers (1980) compellingly demon-
Concluding Thoughts strated, there is no inherent tension between
In the culture wars of psychotherapy that a relational approach and a scientific one.
pit the therapy relationship against the Science can and should inform us about
treatment method (Norcross & Lambert, what works in psychotherapy—be it a treat-
this volume, Chapter 1), it is easy to choose ment method, an assessment measure, a
sides, ignore disconfirming research, and patient behavior, an adaptation method, or
lose sight of our superordinate commit- yes, a therapy relationship.
ment to patient benefit. Instead, let us con-
clude, like T. S. Eliot, by “arriving where we References
started” and underscoring four incontro- Ackerman, S. J., & Hilsenroth, M. J. (2001). A
vertible but oft-neglected truths about psy- review of therapist characteristics and techniques
chotherapy relationships. negatively impacting the therapeutic alliance.
First, the interdivisional Task Force was Psychotherapy, 38, 171–185.
Binder, J. L., & Strupp, H. H. (1997). “Negative
commissioned in order to augment patient process”: A recurrently discovered and underesti-
benefit. We continue to explore what works mated facet of therapeutic process and out-
in the therapy relationship and what works come in the individual psychotherapy of
when we adapt that relationship to (transdi- adults. Clinical Psychology: Science and Practice,
agnostic) patient characteristics. That remains 4, 121–139.
our collective aim: improving patient suc- Duncan, B. L., Miller, S. D., Wampold, B. E., &
Hubble, M. A. (Eds.) (2010). Heart & soul of
cess, however measured and manifested in change in psychotherapy (2nd ed.). Washington,
a given case. DC: American Psychological Association.
Second, psychotherapists have always Lambert, M. J., & Barley, D. E. (2002). Research
aimed to integrate the idiographic and the summary on the therapeutic relationship and
nomothetic, the particular and the general, psychotherapy outcome. In J. C. Norcross (Ed.),
in their craft. One means of doing so is to Psychotherapy relationships that work (pp. 17–32).
New York: Oxford.
adapt psychotherapy to the particulars of Luborsky, L., Diguer, L., Seligman, D. A.,
the individual patient according to gener- Rosenthal, R., Krause, E. D., Johnson, S., et al.
alities identified by research. We can offer (1999). The researcher’s own therapy allegiances:
research-supported methods of individual- A “wild card” in comparisons of treatment effi-
izing psychotherapy to the entire person cacy. Clinical Psychology: Science and Practice, 6,
and his/her singular situation. 95–106.
Lundahl, B., & Burke, B. L. (2009). The effective-
Third, psychotherapy is at root a human ness and applicability of motivational interviewing:
relationship. Even when “delivered” via dis- A practice-friendly review of four meta-analyses.
tance or on a computer, psychotherapy is an Journal of Clinical Psychology: In Session, 11,
irreducibly human encounter. Both parties 1232–1245.
Miller, W. R., Wilbourne, P. L., & Hettema, J. E. Norcross, J. C., Koocher, G. P., & Garofalo, A.
(2003). What works? A summary of alcohol (2006). Discredited psychological treatments
treatment outcome research. In R. K. Hester & and tests: A Delphi poll. Professional Psychology:
W. R. Miller (Eds.), Handbook of alcoholism Research & Practice, 37, 515–522.
treatment approaches: Effective alternatives (3rd Orlinsky, D. E., Ronnestad, M. H., & Willutzki, U.
ed., pp. 13–63). Boston: Allyn & Bacon. (2004). Fifty years of psychotherapy process-
Mohr, D. C., Spring, B., Freedland, K. E., outcome research: Continuity and change. In
Beckner, V., Arean, P., Hollon, S. D., et al. M. J. Lambert (Ed.), Handbook of psychotherapy
(2009). The selection and design of control con- and behavior change, (5th ed.). New York: Wiley.
ditions for randomized controlled trials of psy- Rogers, C. R. (1980). A way of being. Boston:
chological interventions. Psychotherapy and Houghton Mifflin.
Psychosomatics, 78, 275−284. Stiles, W. B., Honos-Webb, L., & Surko, M.
Norcross, J. C. (Ed.). (2001). Empirically supported (1998). Responsiveness in psychotherapy. Clinical
therapy relationships: Summary Report of the Psychology: Science and Practice, 5, 439–458.
Division 29 Task Force. Psychotherapy, 38(4). Wampold, B. E. (2001). The great psychotherapy
Norcross, J. C. (Ed.). (2002). Psychotherapy relationships debate: Models, methods, and findings. Mahwah,
that work. New York: Oxford University Press. NJ: Lawrence Erlbaum.
Norcross, J. C., Hogan, T. P., & Koocher, G. P. Wampold, B. E., Imel, Z. E., Laska, K. M.,
(2008). Clinician’s guide to evidence-based prac- Benish, S., Miller, S. D., Flückiger, C., et al.
tices: Mental health and the addictions. New York: (2010). Determining what works in the treatment
Oxford University Press. of PTSD. Clinical Psychology Review, 8, 923–933.
INDEX
AAI. See Adult Attachment Interview DAS, 362t–365t American Psychiatric Association
AAS. See Adult Attachment Scale; Alberta externalization compared to, 349–50 DSM disorders, 7
Alliance Scale Project MATCH study, 7, 8, 345 Practice Guidelines for the Treatment of
adolescent. UKATT, 345 Psychiatric Disorders, 9–10
alliance formation difficulty, 83 Alcohol Use Questionnaire (AUQ), 291 American Psychological Association
CBT strategies, 85 alliance. See also alliance ruptures; specific (APA), 3–6, 14–15
clinical examples, 74–75 alliance types or groups cultural context, 316–17
definitions and measures, 71–74 adolescence period difficulty, 83 Division of Clinical Psychology’s
deterioration after treatment, 203–4 Bordin’s working alliance concept, 27, Subcommittee on Research-
emotional bond, 70–74 74, 94, 207 Supported Treatments, 426
ES calculation, 77, 81f concept popularity, 7 Guidelines for Providers of
meta-analysis, 76–77 developmental issues, 72–73 Psychological Services to Ethnic,
meta-analytic findings, 77, 80, 81–83 difficulty in adolescent Linguistic, and Culturally Diverse
moderators and mediators, 80–81 psychotherapy, 83 Populations, 316
pretreatment predictors, 83–85 -outcome link, 8 Guidelines on Multicultural
prior reviews, 75–76 participant’s perceptions of, 56 Education, Training, Research,
pushing as undermining to, 88 pretreatment predictors for Practice, and Organizational
research limitations, 87–88 adolescents, 83–85 Change for Psychologists, 316
research reports, 78t–79t reconceptualization as Task Force on evidence-based practice,
TASC and WAI, 73–74 collaboration, 26–27 3–6, 8–11, 12t, 14–18, 316–17,
Task Force conclusions about, 424f therapeutic relationship compared 423–25, 424f
therapeutic practices, 88 to, 56 Template for Developing
therapist strategies, 85–87 therapy early development of, 56 Guidelines, 8–9
Adult Attachment Interview (AAI), 380, tracking fluctuations in, 225 anapanasati (meditation technique),
387t–389t two views of therapeutic 404–5
Adult Attachment Prototype Rating relationship, 71 Anker, M. G., 208, 213
(AAPR), 381, 387t–389t Alliance Observation Coding System anxiety
Adult Attachment Scale (AAS), 381–82, (AOCS), 41 Beck Anxiety Inventory, 362t–365t
387t–389t, 391f alliance ruptures GAD, 362t–365t, 405
affirmation, 6, 168 client self-report of, within HRSA, 42, 362t–365t, 387t–389t
client contribution, 180, 182 session, 224–25 Muslim and Christian therapy for,
clinical examples, 172–76 clinical examples, 227–29 402, 404, 405
definitions and measures, 169–72 definitions and measures, 224 STAI, 42, 209, 241, 362t–365t
ES coding, 178–79, 179t meta-analysis, 229–33, 231t, 232t Taylor Manifest Anxiety Scale, 209
literature search and study selection, moderators and mediators, 231, 233 anxious-ambivalent adult attachment
177–78 observer-based methods of tracking, styles, 380, 391f
meta-analytic findings, 179–80 225–26 APA. See American Psychological
moderators, 178, 180, 181t outcome compared to, 231t Association
previous reviews, 176–77 prevalence of, 226, 227t Apfelbaum, B., 357
research limitations, 182–83 research limitations, 235 aptitude by treatment interaction (ATI),
therapeutic practice, 183–84 research studies, 233–35 262–63
Agnew, R. M., 41, 234 resolution training and AQ. See Alliance Questions; Attachment
Agnew Relationship Measure (ARM), supervision, 232t Questionnaire
41, 234 Rupture Resolution Rating System, Aspland, H., 234–35
Agoraphobic Cognitions Questionnaire 226, 227t ASQ. See Attachment Style Questionnaire
(ACQ), 42 Task Force conclusions about, 424f Assessment for Signal Cases (ASC), 210
Ainsworth, M. S., 378–79 therapeutic practices, 235–36 assumptions, 428
alarm status, 209–10, 211f, 214–19, tracking fluctuations in alliance ATI. See aptitude by treatment interaction
216f, 219f scores, 225 attachment, 377
Alberta Alliance Scale (AAS), 41 under-reported, 225 definitions and measures, 378–81
alcohol dependency, 36t, 280–81, Alliance Weakenings and ECR factor, 381
289t, 305, 341–42 Repairs (AWR), 41 ES estimates, 390–92
431
attachment (Cont’d) Bowlby, John, 377–78 definitions and measures, 378–81
measures, 381–83 BPRS. See Brief Psychiatric Rating Scale ES estimates, 390–92
meta-analysis, 386–90, 387t–389t Brehm, J. W., 263 measures, 381–83
meta-analytic findings, 392–93 Brief Outpatient Psychopathology Scale meta-analysis, 386–90, 387t–389t
moderators and mediators, 393–94 (BOPS), 42 meta-analytic findings, 392–93
research limitations, 394–95 Brief Psychiatric Rating Scale (BPRS), 42 moderators and mediators, 393–94
research studies, 387t–389t Buddhist, 404 preference compared to, 309–10
Task Force conclusions about, 424f research limitations, 394–95
therapeutic practices, 395–96 California Psychotherapy Alliance Scale research studies, 387t–389t
Attachment Questionnaire (AQ), (CALPAS), 28, 41, 49–50, 225 Task Force conclusions about, 424f
387t–389t California Therapeutic Alliance Rating therapeutic practices, 395–96
Attachment Style Questionnaire (ASQ), Scale (CALTARS), 41 client contributions. See also client
382, 387t–389t, 391f Carkhuff, R. R., 190 feedback; patient expectations
attendance, family and couples therapy, Carter, J. A., 4 affirmation, 180
101–2 CBT. See cognitive behavioral therapy congruence, 198
autonomous/secure pattern, of adult Center of Epidemiologic Studies coping styles, 348–49
attachment styles, 380 Depression Scale (CES-D), 42 countertransference, 254
autonomy, excessive, 381–82 CFT. See couple and family therapy couple and family therapy, 103
Avoidant-Attachment Questionnaire child culture compared to, 326–27
(AAQ), 383, 387t–389t, 391f alliance formation difficulty, 83 empathy, 143–44
avoidant style, of adult attachment, CBT strategies, 85 goal consensus, 163
380–81 child deterioration after treatment, patient expectations and, 367–68
203–4 positive regard, 180, 182
BAI. See Bartholomew Attachment clinical examples, 74–75 preference, 311–12
Interview; Beck Anxiety Inventory definitions and measures, 71–74 reactance/resistance, 274
Barber, J. P., 344 emotional bond, 70–74 religious and spiritual beliefs, 413–14
Barkham, M., 207, 234 ES calculation, 77, 81f client feedback, 212f
Barrett-Lennard, G. T., 189 meta-analysis, 76–77 computer-assisted feedback-driven
Barrett-Lennard Relationship Inventory meta-analytic findings, 77, 80, 81–83 system, 207
(BLRI), 41, 189, 242 moderators and mediators, 80–81 definitions and feedback systems,
Bartholomew, K., 380–81 pretreatment predictors, 83–85 206–10
Bartholomew Attachment Interview prior reviews, 75–76 ES computation and dependent
(BAI), 381 “pushing” as undermining to, 88 measures, 213–14
Battery of Interpersonal Capabilities research limitations, 87–88 flagged clients, 205
(BIC), 42 research reports, 78t–79t group therapy cohesion, 126t
BDI. See Beck Depression Inventory TASC and WAI, 73–74 meta-analysis, 212–13
Bech-Rafaelson Mania Scale (BRMS), Task Force conclusions about, 424f meta-analytic findings, 214–19, 216f,
362t–365t therapeutic practices, 88 219f
Beck Anxiety Inventory (BAI), 362t–365t therapist strategies, 85–87 OQ-analyst screen shot, 211f
Beck Depression Inventory (BDI), 42, Christian, 413–14 previous reviews of, 204–5
51–52, 209, 362t–365t, CBT for anxiety, 402, 405 RCTs to explain, 3, 14–15
387t–389t, 403 evangelical compared to research limitations, 219
Bedi, R. P., 45 nonevangelical, 368 Task Force conclusions about, 424f
Bennett, D., 234 Chu, B., 85 therapeutic practices, 219–20
Bergin, A. E., 8 Clark, C. L., 381 video feedback, 394
Berlin Alliance Scale (BAS), 41 Clarkin, J. F., 344 Client Involvement Scale (CIS), 41
Bern Post Session Report (BPSR), 41 client. See also client characteristics; client client preference. See preference
Beutler, L. E., 343, 344 contributions; client feedback; client’s theory of change, 207
BIC. See Battery of Interpersonal client variables; patient client variables, 301, 326. See also group
Capabilities expectations; preference therapy; outcome variance
Bickman, L., 76, 205 alarm status, 209–10, 211f, 214–19, client characteristics, 309–10
Big Five personality factors, 339 216f, 219f client contributions, 311–12
binge eating disorder (BED), 387t–389t approval seeking by, 357 clinical examples, 303–6
BLRI. See Barrett-Lennard Relationship Big Five personality factors, 339 definitions and measures, 302–3
Inventory Client Involvement Scale, 41 dropout rates, 51–52, 307, 308t
Body Sensation Questionnaire (BSQ), 42 Client Support Tool, 214–19, 216f, ES matched compared to nonmatched
Bongar, B., 344 219f groups, 309f
borderline personality disorder (BPD), cultural characteristics, 325 meta-analytic findings, 307–8
386, 387t–389t motivation for change compared to moderators, 308–9
Bordin, E. S., 27, 74 therapy outcome, 8 research limitations, 312–13
couple and family psychotherapy, 94 self-report of ruptures, 224–25 search strategy, 306
SRS based on therapeutic alliance various inclient therapeutic alliance study coding, 306–7
concept, 207 scales, 41 Task Force conclusions about, 424f
Borkovec, T., 355–56 client characteristics, 12–14, 13f, 325, 377 therapeutic practices, 313
432 index
treatment preference interview, leader variables, 122 contractual features, of alliance, 71–72
303t, 304 measures, 110–15, 111t–113t, 114t Coordination Scale (CS), 41
clinical examples member variables, 122 coping style
affirmation, 172–76 meta-analysis, 117 clinical examples, 341–43
attachment styles, 383–86 meta-analytic findings, 118, 129 definitions, 338
child and adolescent psychotherapy, moderators and mediators, 120–24 ES calculation, 345
74–75 raters of outcome data, 121–22 internalizing compared to
congruence, 190–94 research limitations, 124 externalizing, 336–38
coping style, 341–43 search strategy, 117 literature search, 343–45
countertransference, 242–43 study characteristics, 119t–120t, measuring of, 339–40
couple and family therapy, 95–97 121–22 meta-analysis, 343–48
culture, 319–20 Task Force conclusions about, 424f meta-analytic findings, 348
empathy, 136–38 therapeutic practices, 124–25 patient contributions, 348–49
goal consensus, 158–59 collaboration. See also specific measures of research limitations, 349–50
group therapy cohesion, 115–17 collaboration and goal consensus Task Force conclusions about, 424f
individual psychotherapy alliance, 28, alliance reconceptualization as, 26–27 therapeutic practices, 350
42–45 child and adolescent psychotherapy therapy focus and, 346t–347t
patient expectations, 357–58 alliance, 88 CORE. See Clinical Outcomes in Routine
positive regard, 172–76 client contribution, 163 Evaluation
preference, 303–6 clinical examples, 158–59 Countertransference Factors Inventory
reactance/resistance, 266–68 concept of, 154t–157t, 157–58 (CFI), 242
religious and spiritual beliefs, 404–6 definitions and measures, 153–58, countertransference (CT) management
rupture repairs, 227–29 154t–156t adult attachment styles, 385–86
stage of change, 283–85 ES estimation, 161 client contribution, 254
Clinical Outcomes in Routine Evaluation file drawer analyses, 162–63 clinical examples, 242–43
(CORE), 207 goal consensus and, 154t–157t, CT manifestation, 251–52
Clinical Psychology (Division 12) 157–58 definitions, 240–41
Subcommittee on Research- inclusion criteria and study selection, measuring of, 241–42
Supported Treatments, 16, 159–60 meta-analysis, 243
18, 426 meta-analysis, 161 meta-analytic findings, 249–54
Clinical Support Tool (CST), 210, 212f, meta-analytic findings, 161–62 outcome for, 245t, 248t, 249–51
214–19, 216f, 219f moderators and mediators, 163 research limitations, 254–55
coding, 41, 45, 139, 160, 178, 194 reality-based, 26 research studies, 246t–247t
attachment styles, 390 research limitations, 163–64 search strategy, 243–44
client culture, 321 Task Force conclusions about, 424f statistical methods, 244, 249
coping styles, 345 therapeutic practices, 164–65 study coding, 244
countertransference, 244 youth model feature, 72 Task Force conclusions
group therapy cohesion, 117–18 Collaborative Interaction Scale (CIS), 226 about, 424f
patient preference, 306 Colli, A., 226 therapeutic practices, 255–56
religious and spiritual beliefs, 407 comparative effectiveness research couple and family therapy (CFT)
ruptures and resolutions, 226 (CER), 8–9 attendance and retention,
Therapy Process Observation Coding COMPASS Treatment Outcome 101–2
System-Alliance Scale, 74 Systems, 207 case examples, 95–97
cognitive behavioral therapy (CBT), 52, Comprehensive Meta-Analysis, Version client contributions, 103
74, 346t–347t, 362t–365t, 394 2.0, 230–31 clinical examples, 95–97
Christian-accommodative Comprehensive Psychopathological Rating couples therapy, 103–4
intervention, 402, 405 Scale (CPRS), 42 definitions and measures, 92–95
depression treatment, 86 compulsive caregiving, 391f meta-analysis, 97, 99–100
group, 82–83 compulsive careseeking, 391f meta-analytic findings, 102–3
Muslim-accommodative Conflict Tactics Scale, 387t–389t moderators and mediators,
intervention, 402 confrontations, 427 100–101
therapist flexibility with congruence. See also therapist research limitations, 104–5
adolescents, 85 client contribution, 198 research reports, 98–99
Cohen, J., 12t clinical examples, 190–94 Task Force conclusions about, 424f
weighted d, 270, 286, 307, 308, 322, definitions, 187–89 therapeutic practices, 105–6
325t, 345, 348, 363, 392 measures, 189–90, 196–98 couple therapy. See couple and family
cohesion meta-analysis, 194–96 therapy
clinical examples, 115–17 meta-analytic findings, 196 Couple Therapy Alliance Scale
coding and analysis, 117–18 moderators, 196–98 (CTAS), 94
correlations and range by outcome rating scales for, 193 Creed, T., 85
measure and cohesion, 122f research limitations, 198–99 Crits-Christoph, P., 230, 232t
ES for cohesion-outcome research reports, 195t Cronbach’s alpha, 208
relationship, 121f Task Force conclusions about, 424f CT. See countertransference management
group variables, 122–23 therapeutic practices, 199–200 CTAS. See Couple Therapy Alliance Scale
index 433
culture Dyadic Adjustment Scale (DAS), alcohol dependency and, 349–50
centrality of, 316–17 387t–389t internalizing compared to, 336–38
client contributions, 326–27 Dysfunctional Attitudes Scale Eysenck, H. J., 339
clinical examples, 319–20 (DAS), 42
culturally adapted treatments, 317–19 Facilitative Alliance Inventory (FAI), 41
definitions and measures, 317–18 EBP. See evidence-based practice Family Attachment Interview
limitations of research, 327–28 EBPP. See evidence-based practice in (FAI), 381, 382
meta-analysis, 320–22 psychology family therapy
meta-analytic findings, 322 effect size (ES), 49, 345 attendance and retention, 101–2
moderators and mediators, 322, 325 evidence-based practice, 10–11, 12t case examples, 95–97
previous meta-analysis, 320 group cohesion, 121f client contributions, 103
religion compared to patient Eliot, T. S., 423 couples therapy, 103–4
expectations, 368 emotional bond, 70–74 couple therapy meta-analytic
research reports, 323t–325t empathy findings, 102
Task Force conclusions about, 424f BLRI Empathy Scale, 242 definitions and measures, 92–95
therapeutic practices, 328–29 client contribution, 143–44 meta-analysis, 97, 99–100
culture wars, evidence-based practice clinical examples, 136–38 meta-analytic findings, 102–3
compared to, 3–5, 18, 429 definitions, 132–34 moderators and mediators, 100–101
ES estimation, 139 research limitations, 104–5
DAS. See Dyadic Adjustment Scale; inclusion criteria, 138 research reports, 98–99
Dysfunctional Attitudes Scale measures, 134–36 Task Force conclusions about, 424f
DeNisi, A., 204 meta-analytic findings, 140 therapeutic practices, 105–6
Department of Health, Great Britain, 27 moderators and mediators, 140–43, Family Therapy Alliance Scale (FTAS), 94
dependency, excessive, 381–82, 391f 141t, 142t FAQs. See frequently asked questions
depression research limitations, 144–45 fearful style, of adult attachment, 380–81,
adolescent, 86 research reports, 138–39, 139t–140t 391f, 394
BDI, 42, 51–52, 209, 362t–365t, Task Force conclusions about, 424f Fear of Flying Inventory (FFI), 362t–365t
387t–389t, 403 therapeutic practices, 145–47 Fear of Negative Evaluation (FNE),
HRSD, 42, 387t–389t English language, 317 362t–365t
NIMH, 15 enmeshed/preoccupied pattern, of adult feedback. See client feedback
treatments or therapy relationship attachment styles, 380 Fields, S., 76
for, 15 ES. See effect size file drawer bias, 48, 162–63
Zung, 42, 209 ethnic minority groups, 84 Fisher, R. A., 47
Depuy General Well-Being Scale culture impact on therapist-client Fisher’s Z, 47, 48f, 77, 140, 140t,
(DGWBS), 362t–365t relationship, 316–17 141t, 302
Devilly, G., 356 therapist availability flexibility, 17, 85
Diener, M., 99–100 demographics, 317 without fidelity, 428–29
directiveness, 183, 262–63, 268–70, European Brain Injury Questionnaire Flight Fear Questionnaire (FFQ),
272–75, 343 (EBIQ), 42 362t–365t
measuring resistance and, evidence-based practice (EBP) Frank, A., 354, 369
264–66, 271t APA Task Force on, 3–6, 8–11, 12t, frequently asked questions (FAQs), 15–18
patient reactance matched to, 276 14–18, 316–17, 423–25, 424f Freud, A., 70, 72
dismissing pattern, of adult attachment codification of, 6–9 Freud, S.
styles, 380–81, 385, 391f culturally adapted treatments, 317–19 countertransference concept, 239
disorganized/disoriented pattern, of adult culture wars ended by, 3–5, 18, 429 reality-based collaboration, 26
attachment styles, 380 effect size interpretation, 10–11, 12t treatment adaptation, 9
Division 12 (Clinical Psychology), FAQs about, 15–18 funnel plot, 48–49, 48f
Subcommittee on Research- future of, 18–19
Supported Treatments, 16, multiple determinants in, 3, 10 GAD. See generalized anxiety disorder
18, 426 relationship elements, 8–9 GAS. See Global Assessment Scale
Dodo bird verdict, 7, 25, 70 treatment adaptation in, 9–10 Gaston, L., 207
Drinking Abstinence Scale (DAS), UC therapy, 82 Gelso, C. J., 4
362t–365t evidence-based practice in psychology gender, role of, 84,
Drinking per Day (DpD), 42 (EBPP), 3 102, 104
dropout rates, 51–52, 307, 308t evidence-based therapy relationships, General Health Questionnaire (GHQ), 42
drug abuse Task Force conclusions, generalized anxiety disorder (GAD),
DTES, 42 423–25, 424f 362t–365t, 405
National Institute on Drug Abuse expectation of helpfulness (EH), 41 genuineness. See also therapist
Collaborative Cocaine Treatment expectations. See patient expectations client contribution, 198
Study, 15 Experiences in Close Relationships (ECR) clinical examples, 190–94
SAMHSA, 426 scale, 381, 383–91, 391f definitions, 187–89
Drug Taking Evaluation Scale experiential/humanistic therapy, measures, 189–90, 196–98
(DTES), 42 362t–365t meta-analysis, 194–96
Duncan, B. L., 207–9, 208f externalizing meta-analytic findings, 196
434 index
moderators, 196–98 Halkides, G., 189 inpatient therapeutic alliance scales
rating scales for, 193 halo effect, 52–53, 52t (ITAS), 41
research limitations, 198–99 Hamilton Rating Scale for Anxiety institutional review board (IRB), 15
research reports, 195t (HRSA), 42, 362t–365t, INT. See interpersonal/relational
Task Force conclusions 387t–389t therapy
about, 424f Hamilton Rating Scale for Depression internalizing, 336–39
therapeutic practices, 199–200 (HRSD), 42, 387t–389t internalizing ratio (IR), 339
Global Assessment Scale (GAS), 42, Handelsman, J., 76 internal working models
362t–365t Hannover, W., 206 (IWMs), 378
goal consensus, 41 Hardy, G. E., 234–35 interpersonal/relational therapy (INT),
client contribution, 163 Harper, H., 234r 362t–365t
clinical examples, 158–59 Harwood, M., 343 attachment avoidance, 394
coding of study characteristics, Hawley, K. M., 73 Interpersonal Variables Rating Scale
160–61 Hazan, C., 380–82 (IVRS), 41
collaboration concept, 154t–157t, Hedges, L., 195, 213 intimate partner violence (IPV),
157–58 Heimberg, R. G., 369 387t–389t
definitions and measures, 153–58, Helping Alliance Counting Signs (HAcs), Inventory of Interpersonal Problems (IIP),
154t–156t 41 209, 362t–365t, 387t–389t
ES estimation, 161 Helping Alliance Questionnaire-II Inventory of Interpersonal Problems-
file drawer analyses, 162–63 (HAQ-II), 209 Circumplex (IIP-C), 362t–365t
inclusion criteria and study selection, Helping Alliance Questionnaire-Self- IR. See internalizing ratio
159–60 Rated (HAq), 28, 41, 49–50 IRB. See institutional review board
meta-analysis, 161 Helping Alliance Scale-Rated (HAr), 41
meta-analytic findings, 161–62 Helping Relationship Questionnaire Jacobson, N. S., 213
moderators and mediators, 163 (HRQ), 41 Jolkovski, M. P., 252
research limitations, 163–64 Henry, W. P., 7 Journal of Clinical Psychology, 344
Task Force conclusions about, 424f Hilsenroth, M., 99–100 Journal of Consulting and Clinical
therapeutic practices, 164–65 Holman, J., 343 Psychology, 344
Google Scholar, 77 Holt, C. S., 369 Journal of Counseling Psychology, 344
Greenson, R. R., 26–27 Hook, J. N., 406 Jungbluth, N., 86
Groningen Illness Attitudes Scale (GIAS), Horan, F. P., 207
362t–365t Horvath, A. C., 45 Kagan, J., 339, 349
Group Psychotherapy Interventions Howard, K., 207 Karver, M., 70, 75–76, 80–81, 86
Rating Scale (GPIRS), 125, Hunter, J. E., 99, 393 Kaufman, N., 83
125t–126t Hunter’s & Schmidt’s aggregation Kendall, P., 85
group therapy procedures, 47 Kiesler, D., 190
clinical examples, 115–17 Klauer, T., 366
coding and analysis, 117–18 IIP. See Inventory of Interpersonal Klein, D. N., 15
correlations and range by outcome Problems Kluger, A. N., 204
measure and cohesion, 122f Impact of Event Scale (IES), 42 Kordy, H., 206
ES for cohesion-outcome individual psychotherapy Kraus, D. R., 207
relationship, 121f analysis methods, 46–47
group variables, 122–23 clinical examples, 28, 42–45 Lambert, M. J., 8, 213
leader variables, 122 concept attractiveness, 25 leader, group therapy cohesion, 122, 126t
measures, 110–15, 111t–113t, 114t data sources, 45–46 Lietaer, G., 188
member variables, 122 definitions and reconceptualizations, limitations, of research. See research
meta-analysis, 117 26–27 limitations
meta-analytic findings, 118, 129 ES variability, 49 Lingiardi, V., 226
moderators and mediators, 120–24 measures, 27, 51–52, 51f, 52t Llewelyn, S., 234–35
raters of outcome data, 121–22 meta-analytic findings, 47–49, 48f Luborsky, L., 26–27
research limitations, 124 patterns over time, 54–55 Lunnen, K., 214
search strategy, 117 precursors of, 7–8
study characteristics, 119t–120t, research limitations, 55 MAc meta-analysis package, 47
121–22 research reports, 29t–42t Main, Mary, 380
Task Force conclusions about, 424f sources-of-alliance assessment, major depressive disorder (MDD),
therapeutic practices, 124–25 50–51 387t–389t
Guidelines for Providers of Psychological Task Force conclusions about, 424f McCullough, M. E., 405
Services to Ethnic, Linguistic, and therapeutic practices, 56–57 mediators
Culturally Diverse Populations, time-of-alliance assessment, 50 affirmation, 178, 180, 181t
316 types of treatments, 52 attachment styles, 393–94
Guidelines on Multicultural Education, ineffective relational behaviors, 427–29 child and adolescent psychotherapy,
Training, Research, Practice, and infant-parent relationship, 378–79 80–81
Organizational Change for Inpatient Task and Goal Agreement client culture, 322, 325
Psychologists, 316 (ITGA), 41 client preference, 308–9
index 435
mediators (Cont’d) 265, 269, 271t client characteristics compared to
congruence, 196–98 coping styles, 339, 341, 342, 344–45 outcome variance, 12–13, 13f
couple and family therapy, MMPI-2. See Minnesota Multiphasic client motivation for change
100–101 Personality Inventory compared to, 8
empathy, 140–43, 141t, 142t moderators countertransference and, 245t,
goal consensus, 163 affirmation, 178, 180, 181t 248t, 249–51
group therapy cohesion, 120–24 attachment styles, 393–94 Outcome Questionnaire-45 (OQ-45),
individual psychotherapy measures, child and adolescent psychotherapy, 207, 209, 210–11, 213–14
49–50, 50f 80–81 screen shot, 211f
patient expectations, 366–67 client culture, 322, 325 TAU compared to, 216f
positive regard, 178, 180, 181t client preference, 308–9 Outcome Rating Scale (ORS), 207–9,
religious and spiritual beliefs, 413 congruence, 196–98 208f
ruptures, 231, 233 couple and family therapy, 100–101 outcome variance
stage of change, 290–91 empathy, 140–43, 141t, 142t client characteristics compared to,
mental health services, demographic goal consensus, 163 12–14, 13f
availability of, 316–17 group therapy cohesion, 120–24 effect size variability, 49
mental health vital sign, 209–10, 211f individual psychotherapy measures, Interpersonal Variables Rating
MET. See motivational enhancement 49–50, 50f Scale, 41
therapy patient expectations, 366–67
meta-analysis positive regard, 178, 180, 181t parent, 77
attachment styles, 386–90, 387t–389t religious and spiritual beliefs, 413 in child and adolescent treatment
child and adolescent psychotherapy, ruptures, 231, 233 outcome, 88
76–77 stage of change, 290–91 in child or adolescent psychotherapy,
client culture, 320–22 motivational enhancement therapy 71, 73
client feedback, 214–19, 216f, 219f (MET), 345 -infant relationship, 378–79
congruence, 194–96 Muenz, L. R., 344 Partners for Change Outcome
countertransference, 243 Multicenter Collaborative Study for the Management System (PCOMS),
couple and family therapy, 97, 99–100 Treatment of Panic Disorder, 7–8 207–9, 208f, 212–13
group therapy cohesion, 117 Multnomah Community Ability Scale client feedback results, 214–19,
reactance/resistance, 268–70 (MCAS), 42 216f, 219f
religious and spiritual beliefs, 406–8, Muran, J. C., 234 patient expectations, 41. See also outcome
412–13, 412t, 413t Muslim, 402, 404, 405 expectations
ruptures, 229–33, 231t, 232t client contribution and, 367–68
stage of change, 285–88, 287t National Institute for Mental Health clinical examples, 357–58
summary of findings, 81–83 (NIMH), 15 definitions, 354–56
meta-analytic findings National Institute on Drug Abuse EAC, 356
affirmation, 179–80 Collaborative Cocaine Treatment Expectations About Counseling, 356
attachment styles, 392–93 Study, 15 group therapy cohesion, 125t
child and adolescent psychotherapy, Nau, S. D., 355–56 limitations of research, 368–69
77, 80, 81–83 negative processes, 427–28 measures, 355–56
client culture, 322 Neuropsychology Alliance Scale mediators, 366–67
client feedback, 214–19, 216f, 219f (NAS), 41 religions compared to, 368
collaboration, 161–62 Nielsen, S. L., 8 research reports, 362t–365t
congruence/genuineness, 196 NIMH. See National Institute for Mental research review, 358–66
coping styles, 345, 348 Health Task Force conclusions about, 424f
countertransference management, NNT. See number needed to treat therapeutic practices, 370–71
243, 249–54 Non Standard Instrument (NSI), 41 treatment expectations, 356–58, 368
couple and family therapy, 102–3 number needed to treat (NNT), 12t Patient Prognostic Expectancy Inventory
empathy, 140 (PPEI), 356, 362t–365t
goal consensus, 161–62 Observer Alliance Scale (OAS), 41 Patients’ Therapy Expectation and
group therapy cohesion, 118, 129 observer-based methods, 225–26 Evaluation (PATHEV), 356
individual psychotherapy, 47–49, 48f odd ratio (OR), 214 Pattern of Individual Change Scores
positive regard, 179–80 Ogles, B. M., 8, 214 (PICS), 42
preference, 307–8 Okiishi, J., 8 Paul, Gordon, 9
reactance/resistance level, 272–73 Olkin, I., 195 PCOMS. See Partners for Change
religion and spirituality, 408, 412–13, OQ-45. See Outcome Questionnaire-45 Outcome Management System
412t, 413t OQ Psychotherapy Quality Management Penn State Worry Questionnaire (PSWQ),
ruptures repair, 231, 231t System (OQ System), 214–19, 362t–365t
stage of change, 291–93 216f, 219f Personal Adjustment and Role Skills Scale
for trim and fill analysis, 413t Orlinsky, David, 6 (PARS), 362t–365t
Michelson, A., 343 ORS. See Outcome Rating Scale Personal Report of Confidence as a
Miller, S. D., 207–9, 208f, 214 Osler, William, 9 Speaker (PRCS), 362t–365t
Minnesota Multiphasic Personality outcome expectations, 354–55. See also person of therapist, 6–9, 10. See also
Inventory (MMPI-2), 195t, 209, patient expectations countertransference; therapist
436 index
Persuasion and Healing (Frank), 354 characteristics compared to outcome Task Force conclusions about, 424f
pharmacotherapy, therapy relationship variance, 12–13, 13f therapeutic practices, 275–76
compared to, 15 child and adolescent psychotherapy, Reaction to Treatment Questionnaire
Positive and Negative Syndrome Scale 77, 80, 81–83 (RTQ), 356, 369
(PNSS), 42 client culture, 322 reactive compared to proactive behaviors,
positive regard, 6, 168 client feedback, 214–19, 216f, 219f 339
client contribution, 180, 182 collaboration, 161–62 Real Relationship Inventory, 190
clinical examples, 172–76 congruence/genuineness, 196 Reciprocal Attachment Questionnaire
definitions and measures, 169–72 coping styles, 345, 348 (RAQ), 382–83, 391f
ES coding, 178–79, 179t countertransference management, Reese, R. J., 214
literature search and study selection, 243, 249–54 Reich, A., 252
177–78 couple and family therapy, 102–3 Relationship Questionnaire (RQ), 382
meta-analytic findings, 179–80 empathy, 140 Relationship Scale Questionnaire (RSQ),
moderators, 178, 180, 181t goal consensus, 161–62 382, 391f
previous reviews, 176–77 group therapy cohesion, 118, 129 Reliable Change Index (RCI), 213
research limitations, 182–83 individual psychotherapy, 47–49, 48f religion and spirituality (R/S)
Task Force conclusions about, 424f motivation for change compared to client contributions, 413–14
therapeutic practice, 183–84 therapy outcome, 8 clinical examples, 404–6
postsession questionnaire (PSQ), Outcome Questionnaire-45, 207, 209, coding, 407
225–26 210–11, 211f, 213–14, 216f data analysis, 407–8
Post Therapy Questionnaire (PTQ), 42 positive regard, 179–80 definitions and measures, 402–4
posttraumatic stress disorder (PTSD), preference, 307–8 ES, 407
386, 387t–389t reactance/resistance level, 272–73 inclusion criteria, 406–7
Practice Guidelines for the Treatment of religion and spirituality, 408, 412–13, meta-analysis method, 406–8
Psychiatric Disorders (American 412t, 413t meta-analytic findings, 408, 412–13,
Psychiatric Association), 9–10 ruptures repair, 231, 231t 412t, 413t
preference, 301 stage of change, 291–93 moderators, 413
client characteristics, 309–10 for trim and fill analysis, 413t publication bias, 412–13, 412t
client contributions, 311–12 psychotherapy relationships. See therapist; research limitations, 414
clinical examples, 303–6 specific group research reports, 409t–410t
definitions and measures, 302–3 PsycINFO database, 45–46, 77, 229, 344, study selection, 409t–410t
dropout rates, 51–52, 307, 308t 386 Task Force conclusions about, 424f
ES matched compared to nonmatched PSYNDEX German language therapeutic practices, 414–15
groups, 309f database, 46 trim and fill analysis results, 413t
meta-analytic findings, 307–8 PTSD. See posttraumatic stress disorder research limitations
moderators, 308–9 publication bias, 412–13, 412t affirmation, 182–83
research limitations, 312–13 attachment styles, 394–95
search strategy, 306 Questionnaire on Attitudes Toward Flying child and adolescent psychotherapy,
study coding, 306–7 (QATF), 362t–365t 87–88
Task Force conclusions about, 424f client feedback, 219
therapeutic practices, 313 race, 84, 316–17 congruence, 198–99
treatment preference interview, randomized controlled/clinical trial coping styles, 349–50
303t, 304 (RCT), 3, 14–15 countertransference, 254–55
pretreatment predictors, 83–85 reactance and psychotherapy couple and family therapy, 104–5
Prigatano Alliance Scale (PAS), 41 directiveness, 271t culture, 327–28
proactive compared to reactive raters, of outcome data empathy, 144–45
behaviors, 339 developing measures for, 189 expectations, 368–69
process expectations, 357 group therapy cohesion, 121–22 goal consensus, 163–64
Procrustean bed, 428 individual psychotherapy, 52 group therapy cohesion, 124
Project MATCH Research RCI. See Reliable Change Index individual psychotherapy, 55
Group, 7, 8, 345 RCT. See randomized positive regard, 182–83
psychodynamic therapy, 362t–365t controlled/clinical trial preference, 312–13
psychotherapy directiveness, reactance reactance, 261–62 reactance/resistance, 274–75
compared to, 271t client contribution, 274 religious and spiritual beliefs, 414
Psychotherapy Evaluation Questionnaire clinical examples, 266–68 ruptures, 235
(PEQ), 362t–365t definitions, 263–64 stage of change, 293
Psychotherapy Expectancy Inventory ES calculation, 270, 272 Task Force on Evidence-Based
(PEI), 357 measures of, 264–66 Therapy Relationships, 14–15
Psychotherapy Expectancy Inventory- meta-analysis, 268–70 research studies/reports, for
Revised (PEI-R), 357 meta-analytic findings, 272–73 meta-analysis
psychotherapy outcomes previous reviews, 268 adolescent psychotherapy, 78t–79t
affirmation, 179–80 psychotherapy directiveness and, 271t affirmation, 177–78
alliance-outcome link, 8 research limitations, 274–75 attachment styles, 381–83,
attachment styles, 392–93 research studies, 273–74 387t–389t
index 437
research (Cont’d) Rupture Resolution Scale, 226 stage-matched treatments, 291–93
child and adolescent psychotherapy, Russell, R. L., 86 stage of change
78t–79t clinical examples, 283–85
client culture, 323t–325t safe environment, for CFT, 93–94 definitions and measures,
congruence/genuineness, 195t Safran, J. D., 234 279–81, 281t
countertransference management, SAMD. See sample-adjusted meta-analytic ES, 288, 289t–290t, 290
243, 246t–247t deviancy meta-analysis, 285–88, 287t
couple and family therapy, 98–99 SAMHSA. See Substance Abuse and meta-analytic findings, 291–93
couple therapy, 98–99 Mental Health Services moderators, 290–91
empathy, 138–39, 139t–140t Administration previous meta-analysis, 285
individual psychotherapy, sample-adjusted meta-analytic deviancy processes of change, 281–83
29t–42t (SAMD), 302, 392–93 research limitations, 293
patient expectations, 362t–365t Sapyta, J., 204–5 research studies, 287t
positive regard, 177–78 SASB. See Structural Analysis of Social stage-matched treatments, 291–93
reactance/resistance level, 273–74 Behavior tailoring treatments to, 283
religious and spiritual beliefs, Satisfaction with Life Scale (SWLS), 42 Task Force conclusions about, 424f
409t–410t Scale for Suicidal Ideation (SSI), therapeutic practices, 293–96
rupture repair, 233–35 362t–365t State-Trait Anxiety Inventory (STAI), 42,
stage of change, 287t Schmidt, F. L., 99, 393 209, 241, 362t–365t
residual gain score (RGS), 41 Schneider, W., 366 Sterba, R. F., 26
resistance, 261–62 SCL/BSI. See Symptom Checklist 90, Stiles, W. B., 234–35
client contribution, 274 Brief Symptom Inventory Stone, L., 26
clinical examples, 266–68 Second Sheffield Psychotherapy Project, Strange Situation, 378–79
definitions, 263–64 234–35 Structural Analysis of Social Behavior
ES calculation, 270, 272 secure/autonomous pattern, of adult (SASB), 41
measures of, 264–66 attachment styles, 380 Structural Equation Modeling
meta-analysis, 268–70 self-confidence, 381–91, 391f (SEM), 345
meta-analytic findings, 272–73 Personal Report of Confidence as a Strupp, Hans, 5
previous reviews, 268 Speaker, 362t–365t study characteristics/design. See also
psychotherapy directiveness Rosenberg Self-Esteem Index, 42 coding
for, 271t self-disclosure, 110, 111t, 121, 126t, 199. affirmation, 177–78
research limitations, 274–75 See also countertransference collaboration, 160–61
research studies, 273–74 difficult, 183 culture and, 325–26
Task Force conclusions SEM. See Structural Equation Modeling goal consensus, 160–61
about, 424f separation protest, 391f. See also group therapy cohesion, 119t–120t,
therapeutic practices, 275–76 attachment 121–22
resolution training/supervision, 232t SEPI. See Society for the Exploration of preference, 306–7
retention, family and couples therapy, Psychotherapy Integration Subcommittee on Research-Supported
101–2 Session Evaluation Questionnaire (SEQ), Treatments, 16, 18, 426
Richard, P. S., 206 42, 226 Substance Abuse and Mental Health
Riemer, M., 205 Session Rating Scale (SRS), 207 Services Administration
rigidity, 428 SFT. See solution-focused therapy (SAMHSA), 426
RII. See Role Induction Interview Shapiro, D. A., 234 suicidal ideation, 174, 250, 254, 365
Robbins, M. S., 252 Shaver, P. R., 380–82 adolescent, 86
Rogerian practice, 71 Shimokawa, K., 218 Symptom Checklist 90, Brief Symptom
congruence concept central to, Shirk, S., 70, 75–76, 80–81, 86 Inventory (SCL/BSI), 42,
187–88 singularity, 9, 428 51–52, 209, 362t–365t,
evidence-based therapy relationships Slade, A., 384 387t–389t
compared to, 16 Smith, T. B., 406, 408 Systematic Treatment Selection-Clinician
genuineness, 188 Social Adjustment Scale (SAS), 209, Rating Form (STS-CRF), 342
person-centered treatment, 25 362t–365t Systematic Treatment Selection
Rogers, Carl. See also Rogerian practice Society for the Exploration of Therapy Process Rating Scale
therapeutic presence of, 188–89 Psychotherapy Integration (STS TPRS), 41
role expectations, 356–57. See also (SEPI), 25 System for Observing Family Therapy
patient expectations SOFTA. See System for Observing Family Alliances (SOFTA), 94
Role-Functioning Questionnaire (RFQ), Therapy Alliances
362t–365t solution-focused therapy (SFT), tailoring treatment, 10. See also
Role Induction Interview (RII), 369 346t–347t attachment; culture; preference;
role preparation, 126t Song, Xiaoxia, 343 reactance; stage of change; Task
Rosenberg Self-Esteem Index sources-of-alliance assessment, 50–51 Force on Evidence-Based Therapy
(RSEI), 42 Sparks, J. A., 207 Relationships
R/S. See religion and spirituality spirituality. See religion and spirituality Paul’s iconic question, 9
Rupture Resolution Rating System (3Rs), split alliances, 94 Procrustean bed compared to, 428
226, 227t SRS. See Session Rating Scale R/S beliefs, 402
438 index
stage of change, 283 Therapist Appraisal Questionnaire, 241 clinical examples, 341–43
therapist flexibility, 17, 85, 428–29 therapist support. See also clinical definitions, 338
treatment preference interview, examples; countertransference; ES calculation, 345
303t, 304 positive regard; therapeutic internalizing compared to
Target Complaints (TC), 42 practices; specific group externalizing, 336–38
Task Force on Evidence-Based Therapy availability demographics, 316–17 literature search and inclusion data,
Relationships centricity of therapist, 428 343–45
APA Template for Developing culture impact on, 316–17 measuring of, 339–40
Guidelines, 8–9 directiveness, 183, 262–66, 268–70, meta-analysis, 343–48
client’s cultural context and, 316–17 271t, 272–76, 343 meta-analytic findings, 348
conclusions of, 423–25, 424f efficacy empirically validated, 6–8 patient contributions, 348–49
ES statistics interpretation, 10–11, 12t emotional bond with child, 70–74 research limitations, 349–50
FAQs about objectives and ineffective relational behaviors, Task Force conclusions about, 424f
results, 15–18 427–29 therapeutic practices, 350
group therapy cohesion, 424f overly optimistic, 206 therapy focus compared to coping
Guidelines for Providers of person of therapist in, 6–9, 10 style, 346t–347t
Psychological Services to Ethnic, play therapy, 71 treatment outcome
Linguistic, and Culturally Diverse Project MATCH study of, 7, 8, 345 affirmation, 179–80
Populations, 316 real relationship, 188 attachment styles, 392–93
Guidelines on Multicultural tailoring treatment to client, 9–10, 17, child and adolescent psychotherapy,
Education, Training, Research, 85, 283, 402, 428–29 77, 80, 81–83
Practice, and Organizational Therapist Appraisal Questionnaire, client culture, 322
Change for Psychologists, 316 241 client feedback, 214–19, 216f, 219f
limitations of, 14–15 validation of efficacy, 6–8 collaboration, 161–62
origin and challenges of, 3–6 Therapy Experience Questionnaire congruence/genuineness, 196
policy recommendations for APA, 425 (TEQ), 234 coping styles, 345, 348
Template for Developing Therapy Process Observation Coding countertransference management,
Guidelines, 8–9 System-Alliance Scale, 74 243, 249–54
treatment adaptation, 9–10 therapy relationship. See also couple and family therapy, 102–3
TAU. See treatment as usual countertransference; therapeutic empathy, 140
Taylor Manifest Anxiety Scale, 209 practices; therapist; specific group goal consensus, 161–62
Thematic Apperception Test (TAT), APA Task Force on, 3–6, 8–11, group therapy cohesion, 118, 129
362t–365t 12t, 14–18, 316–17, individual psychotherapy,
therapeutic alliance. See therapeutic 423–25, 424f 47–49, 48f
practices; therapist; specific group codification of evidence-based positive regard, 179–80
Therapeutic Alliance Rating Scale practices, 6–9 preference, 307–8
(TARS), 41 diamond analogy, 6 reactance/resistance level, 272–73
Therapeutic Alliance Scale for Children Gelso and Carter’s definition, 4 religion and spirituality, 408, 412–13,
(TASC), 73–74 RCT treatment methods 412t, 413t
Therapeutic Bond Scale (TBS), 41 compared to, 3 ruptures repair, 231, 231t
therapeutic practices Timed Behavior Checklist (TBC), stage of change, 291–93
affirmation, 183–84 362t–365t for trim and fill analysis, 413t
attachment styles, 395–96 time-of-alliance assessment, 50 treatment preference interview,
child and adolescent Tinsley, H. E. A., 369–70 303t, 304
psychotherapy, 88 transtheoretical model, 279, 282 trim and fill analysis, 413t
client feedback, 219–20 Trauma Symptom Checklist-40 (TSC-40), Truax Relationship Questionnaire (TRQ),
congruence, 199–200 387t–389t 189, 213
coping styles, 350 treatment adaptation, 10. See also Truax Self-Congruence Scale, 189–90,
countertransference, 255–56 attachment; culture; preference; 193t–194t
couple and family therapy, 105–6 reactance; stage of change; Task unbalanced alliances, 94, 106
empathy, 145–47 Force on Evidence-Based Therapy under-reported ruptures, 225
group therapy cohesion, 124–25 Relationships United Kingdom Alcohol Treatment Trial
individual psychotherapy, 56–57 Paul’s iconic question, 9 (UKATT), 345
patient expectations, 370–71 Procrustean bed compared to, 428 unresolved pattern, of adult attachment
positive regard, 183–84 R/S beliefs, 402 styles, 380, 385–86
reactance/resistance, 275–76 stage of change, 283 usual care (UC) therapy, 82
religious and spiritual beliefs, therapist flexibility, 17, 85, 428–29 validation, of therapist efficacy, 6–8.
414–15 treatment preference interview, See also evidence-based practice
ruptures, 235–36 303t, 304 Vanderbilt II study, 230
stage of change, 293–96 treatment as usual (TAU), 213–15 Vanderbilt Psychotherapy Process Scale
therapeutic relationship, alliance OQ-45 compared to, 216f (VPPS), 28, 41, 49–50
compared to, 56 treatment expectations, 356–58, 368. Vanderbilt Therapeutic Alliance Rating
Therapeutic Relationship Scale See also patient expectations Scale (VTAS), 41, 95
(TRS), 41 treatment fit variability, of effect size, 49
index 439
Veterans Adjustment Scale Global Weinberger, J., 99–100 Working Alliance Inventory-short version
Adjustment Score (VETS Global), Weisz, J. R., 73 (WAI-S), 41
362t–365t Wisconsin Behavior Inventory Working Alliance Scale (WASc), 41
Video Rating Scale (VRS), 362t–365t (WBI), 42 Working Alliance Survey (WASu), 41
videotaped sessions, 394 Wisconsin Personality Disorder Inventory Working Behavior Inventory (WBI), 42
couple and family therapy, 106 (WPDI), 42 Yale-Brown Obsessive-Compulsive Scale
vital sign, mental health, 209–10, 211f Wisconsin Schizophrenia Project, (Y-BOCS), 42
WAI. See Working Alliance Inventory 189–90 Zetzel, E. R., 26
weighted d, 270, 286, 307, 308, 322, Working Alliance Inventory (WAI), 28, Zung’s Self-Rating of Depression (Zung),
325t, 345, 348, 363, 392 41, 49–50, 73, 225 42, 209
440 index

Psychotherapy Relationships That Work - Evidence-Based Responsiveness (2011) - Narcross PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Psychotherapy Relationships That Work - Evidence-Based Responsiveness (2011) - Narcross PDF

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

© Oxford University Press, Inc. 2011

Psychotherapy relationships that work : evidence-based

Emma and Owen

Arnold A. Lazarus, Ph.D.

A cordial welcome to the second edition of a single professional association (Division

Jennifer Alonso, B.S. AC Del Re, M.A.

Part One • Introduction

Part Two • Eﬀective Elements of the Therapy Relationship:

Part Four • Conclusions and Guidelines

1 Evidence-Based Therapy Relationships

John C. Norcross and Michael J. Lambert

an important role is expectancy—the pla- How to improve psychotherapy outcome?

2 Alliance in Individual Psychotherapy

Adam O. Horvath, AC Del Re, Christoph Flückiger, and Dianne Symonds

−1.5 −1.0 −0.5 0.0 0.5 1.0 1.5

Fig. 2.1 Funnel plot of the ESs in the meta analysis.

Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

perspectives of the alliance provided better Outcome Measures

Dropout 15 0.164 0.062 0.262 .001

BDI as an outcome measure. In consider- the alliance–outcome relation among them

Stephen R. Shirk and Marc S. Karver

Early (k = 113) Mid (k = 33) Late (k = 36) Average (k = 52)

4 Alliance in Couple and Family Therapy

Myrna L. Friedlander, Valentín Escudero, Laurie Heatherington, and Gary M. Diamond

5 Cohesion in Group Therapy

Gary M. Burlingame, Debra Theobald McClendon, and Jennifer Alonso

Cohesion is the most popular of several by providing a tabular summary of

Effect Lower Upper Z- p- Correlation and 95% CI

Favors negative relationship Favors positive relationship

Fig. 5.1 Weighted eﬀect size for cohesion-outcome relationship

Table 5.4 Group Psychotherapy Intervention Rating Scale (GPIRS)

Robert Elliott, Arthur C. Bohart, Jeanne C. Watson, and Leslie S. Greenberg

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 133

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 135

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 137

Table 6.1 Study Characteristics

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 139

Results Moderators and Mediators

Table 6.2 Empathy–Outcome Correlations: Overall Summary Statistics

Weighted mean r 0.22∗ 0.33 0.30∗ 0.13

Table 6.3 Mean Eﬀects across Theoretical Orientation

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 141

(individualized to satisfaction ratings), and possibility that empathy is slightly more

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 143

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 145

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 147

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 149

e l l i ot t, b o h a rt, wats o n , g re e n b e rg 151

7 Goal Consensus and Collaboration

Georgiana Shick Tryon and Greta Winograd

This chapter focuses on one element of the Deﬁnitions and Measures

Eﬀect Size (95% CI)

8 Positive Regard and Aﬃrmation

Barry A. Farber and Erin M. Doolin

Author Note. We gratefully acknowledge time—and still an enormously inﬂuen-

Table 8.1 Eﬀect Sizes between Positive Regard and Outcome

Quintana & Meara (1990) 48 .02 [−.26, 0.31]

Rabavilas, Boulougouris, & Perissaki (1979) 36 .09 [−.25, 0.41]

Gregory G. Kolden, Marjorie H. Klein, Chia-Chiang Wang, and Sara B. Austin

Congruence or genuineness is a relational accurately represented by his awareness of

Negatively valenced items