Journal of the Society for Social Work and Research

Volume 2, Issue 2, 104-124

July 2011
ISSN 1948-822X DOI:10.5243/jsswr.2011.6

The Empathy Assessment Index (EAI):

A Confirmatory Factor Analysis of a Multidimensional Model of Empathy
Cynthia A. Lietz
Karen E. Gerdes
Fei Sun
Jennifer Mullins Geiger
M. Alex Wagaman
Elizabeth A. Segal
Arizona State University
Both historically and currently, social work and related fields have expressed considerable interest in the
construct of empathy. However, the ability to define and ultimately measure empathy is limited. This project
validates a revised version of the Empathy Assessment Index (EAI), which is a measure rooted in a social
cognitive neuroscience conceptualization of empathy. To evaluate the instruments psychometric properties, we
administered the 50-item EAI with a five-component model of empathy to a sample of 773 undergraduate
students and community members. We evaluate the EAI psychometrics conducting internal consistency, testretest, and confirmatory factor analyses. Findings indicate that a 17-item five-factor model of the EAI offers the
best fit [CFI = .98; RMSEA = .04 (90% CI (.03; .05); WRMR = .80]. The data do not support using empathic
attitudes as a proxy for actions; however, the findings suggest the EAI functions better when measuring a fourfactor model, offering an important implication for future research.
Keywords: empathy, measurement, psychometrics, confirmatory factor analysis

The concept of empathy has a long history within

social work literature. For years, social work scholars
have suggested practitioners need to develop empathy to
understand and to respond appropriately when working
with diverse populations facing a variety of social
problems. Equally important, increased empathy in
populations served by social workers has a number of
positive outcomes, including increased satisfaction with
relationships, improved parenting, and increased social
well-being (Curtner-Smith et al., 2006; Hoffman, 2000;
Hunter, Figueredo, Becker, & Malamuth, 2007).
Although these assumptions regarding empathy remain
ubiquitous across policy and practice texts, the term
empathy lacks clarity, suggesting a need for further
conceptualization (Duan & Hill, 1996; Gerdes, Segal, &
Lietz, 2010; Pithers, 1999).
Recent advances in social cognitive neuroscience
have indicated empathy is an induction process that can
be described and, therefore, can be measured, offering
important implications regarding a conceptualization of
empathy (Decety & Meyer, 2008; Decety & Moriguchi,
2007). Social workers would benefit from having an
empathy measure that can be applied across social work
settings. For example, social work educators seek to
foster empathic responses from students whereas
practitioners might attempt to cultivate empathy within

Journal of the Society for Social Work and Research

client populations. Having a robust measure of empathy

would enable educators and practitioners working in
these varied contexts to assess levels of empathy
whatever the population of interest. In addition, because
levels of empathy can inform interpersonal interactions
across settings and purposes, a measure of empathy may
offer important applications beyond social work. This
project used confirmatory factor analysis to test revisions
of the Empathy Assessment Index (EAI), which is an
instrument designed to measure a multidimensional
model of empathy grounded in social cognitive
neuroscience (see Gerdes, Lietz, & Segal, 2011).
Social cognitive neuroscience is a relatively new,
interdisciplinary field, which examines topics that have
traditionally been of interest to social science (e.g.,
information processing, emotional regulation) using
approaches that are more typical of neuroscience such as
brain imaging (e.g., functional magnetic resonance
imaging).
Literature Review
The term empathy can be traced from early literature
informing social work (Richmond, 1917; Rogers, 1975;
Towle, 1945) to current discussions describing social
work practice (Boyle, Hull, Mather, Smith, & Farley,
2006; Hepworth, Rooney, Rooney, Strom-Gottfried, &
Larsen, 2006; Saleebey, 2009; Shulman, 2009). The

104

THE EMPATHY ASSESSMENT INDEX

historical development of the empathy concept within

social work suggests that although influential, the
construct lacks clarity and consensus (Gerdes et al.,
2010).
Some theorists have perceived empathy as a capacity
to feel the experiences of another (Kohut, 1959),
suggesting an introspective, cognitive component to
empathy. On the other hand, Rogers (1957) discussed
empathy as a skill that was demonstrated by a therapists
ability to relate with a client without losing his or her
sense of self. Rogers work represented some of the
earliest consideration that differentiation between self
and others may be an important part of empathy. Social
developmental psychology contributed to the
conceptualization of empathy by highlighting the
possibility that empathy encompasses multiple
dimensions (Cliffordson, 2002; Davis, 1996), including
components such as awareness of self and others (Batson
et al., 1997) and emotional regulation (Eisenberg at al.,
1994). Hoffman (2000) added to this discussion and
suggested empathy was not simply multidimensional, but
may also represent a developmental process, progressing
from automatic mimicry (Iacoboni, 2009) to cognitive
processing that involves the ability to imagine the
experiences of another.
More recently, Decety and Jackson (2004)
demonstrated that observable brain activity was linked to
four subjectively experienced components of empathy.
The first component is affective sharing, which
comprises automatic reactions based on a persons
observation of another. The second, self-awareness, is a
persons ability to differentiate the experiences of another
from his or her own. The third component is perspective
taking, which is the cognitive process of imagining the
experiences of another. The fourth component is emotion
regulation that is described as a persons ability to sense
anothers feelings without becoming overwhelmed by the
intensity of this experience. Decety and Moriguchis
(2007) descriptions of these four components increased
the clarity of the overall conceptualization of empathy.
Recent thinking within social work has suggested a
fifth potentially important component to empathy. Gerdes
and Segal (2009) have contended that empathy goes
beyond automatic responses, suggesting the induction
process culminates in a conscious course of action.
Eisenberg (2006, p. 71) referred to this idea as empathyrelated responding, asserting the existence of a
relationship between empathy and prosocial behavior. To
measure this fifth component, the EAI includes a proxy
for empathic action defined as empathic attitudes. The
empathic attitudes component is of particular interest to
social work because the component suggests that

Journal of the Society for Social Work and Research

empathic attitudes may increase the likelihood for taking

empathic action.
Considering the many variations in how empathy is
conceptualized across disciplines, social work often
serves as a bridging discipline. For example, many
psychologists have described empathy as an automatic
emotional response to others behavior or emotions
(Freedberg, 2007). However, social work incorporates an
interdisciplinary approach to understanding empathy as a
biological, emotional, and cognitive response. In
addition, social work is interested in how empathy can be
developed and cultivated among professionals and client
populations. In this way, as with other concepts within
social work, empathy can be conceptualized by drawing
from the strengths of other disciplines to create a holistic
approach.
Measuring Empathy
Efforts to conceptualize and ultimately measure
empathy have been informed by theoretical developments
and advances in neuroscience. However, no instrument
designed to measure empathy has used a four-factor
model grounded in neuroscience, nor have other
measures incorporated the empathic attitudes component
that is relevant to social work. Moreover, previous
measures fell short regarding their psychometric
properties. For example, one of the earliest and most
widely used multidimensional measures of empathy,
Davis (1980) Interpersonal Reactivity Index (IRI)
includes four factors: perspective taking, fantasy,
empathic concern, and personal distress. Critics of this
instrument have argued that the personal distress and
fantasy factors were inadequate to assess levels of
empathy, and that the instrument measured sympathy
rather than empathy. In addition, the IRI was not
validated by further statistical analysis (Cliffordson,
2002). Similarly, in an evaluation of Hogans (1969)
Empathy Scale, Froman and Peloquin (2001) concluded
that the model suffered from questionable test-retest
reliability, low internal consistency, and poor replication
of factor structure.
Measures of empathy that have been developed more
recently have demonstrated improved psychometric
properties; these measures include the Toronto Empathy
Questionnaire (Spreng, McKinnon, Mar, & Levine,
2009); the Basic Empathy Scale (Joliffe & Farrington,
2006); the Ethnocultural Empathy Scale (Wang et al.,
2003); the E-Scale (Leibetseder, Laireiter, & Kller,
2007); and Hojat et al.s (2001) Jefferson Scale of
Physician Empathy. However, some notable limitations
may prevent these measures from fully capturing an
accurate and multidimensional representation of
empathy. For example, the Toronto Empathy
Questionnaire (Spreng et al., 2009) presents empathy

105

LIETZ et al.
primarily as an emotional process and does not account
for cognitive components required for perspective taking
and selfother awareness. Conversely, although Hogans
(1969) Empathy Scale has been widely used as a measure
for cognitive empathy, this scale fails to incorporate the
critical component of emotion. Other instruments
measure empathy within a specific group, such as the
Jefferson Scale of Physician Empathy (Hojat et al., 2001)
or individuals empathy toward people of different
racial/ethnic backgrounds (Wang et al., 2003). Although
important contributions, the scales reviewed here do not
offer a measure of empathy with wide-reaching
applicability, nor do they reflect the understanding of
empathy based in neuroscience.
The ability to measure empathy in social work
practice settings is relevant to client assessment and the
evaluation of evidence-based practice. For example, low
levels of empathy have been linked to delinquent and
aggressive behavior in adolescents (de Kemp, Overbeek,
De Wied, Engels, & Scholte, 2007) and sex offenders
(Varker & Devilly, 2007; Whittaker, Brown, Bekett, &
Gerhold, 2006). An empathy measure can help guide
treatment planning and evaluation of treatment goals in
settings within social work and within other disciplines
including education, psychology, and medicine.
Development of the EAI
The EAI was developed to overcome the limitations
of current measures of empathy by incorporating five
components of empathy: affective response, selfother
awareness, perspective taking, emotion regulation, and
empathic attitudes. As described by Gerdes and
colleagues (2011), survey design protocols were followed
in the development of items for each of the five
components (DeVellis, 2003). After an exhaustive
literature review on each component and a review of
current items from existing measures of the five
constructs, the researchers composed their own unique
items. The goal was to follow Sartori and Pasinis (2007)
recommendation that item generation should seek to
achieve content validity by creating items that flow
logically and theoretically from the conceptualization of
each component.
Pilot Version of the EAI. The pilot version of the
EAI was a 54-item survey that was administered in
October 2009 to a nonrandom sample of 312 students
(63% response rate; Gerdes et al., 2011). Although this
initial version of the EAI demonstrated some promise, a
confirmatory factor analysis (CFA) could not be
conducted on the five-component pilot data because the
selfother awareness items had unacceptable reliability
analysis ( =.299). Therefore, the researchers performed
an exploratory factor analysis (EFA) using the maximum
likelihood extraction method with oblique rotation.

Journal of the Society for Social Work and Research

The EFA results indicated that in addition to

reconceptualizing the selfother awareness construct,
several items on the emotion regulation component
needed revision or elimination, and the affective response
component needed modification. Before the current data
were collected, the researchers improved the EAI by
eliminating items that did not contribute significant
information to the index and, when needed, developing
new items that, in turn, would be further tested. (See the
Method section for a description of how new items were
generated). This article is based on a second round of
data collected in September 2010, to validate the
psychometric properties of a revised version of the EAI.
Method
Revising the EAI Based on EFA Pilot Study, Focus
Groups, and Expert Reviews
Revisions based on the EFA pilot study. Before
collecting data for the CFA of the EAI, the researchers
returned to the social cognitive neuroscience literature to
reconceptualize the items for the selfother awareness
component, to modify the affective response component,
and to add new items to the emotion regulation and
perspective-taking components. In the EAI pilot test, the
selfother awareness component was broadly interpreted
with an overemphasis on boundaries. For example, the
pilot EAI included five selfother awareness items such
as Getting over-involved in other peoples lives
describes me well, and When a friend has a problem, I
am supportive, but let him/her handle it. In addition, one
selfother awareness item was unintentionally doublebarreled: I am willing to look at my own behavior and
consider how I interact with people. These miscues are
likely responsible for the pilot selfother awareness
components Cronbach alpha of .299.
The researchers reevaluated the selfother awareness
component after carefully considering two important
aspects of the component. First, selfother awareness
allows the empathizer to track the origins of his or her
feelings triggered during the affective arousal phase
(Lamm, Batson, & Decety, 2007). Second, selfother
awareness enables the empathizer to suspend ones own
experience in order to conjure up the thoughts and
feelings of others (Mitchell, 2009, p. 1314); an ability
which is a prerequisite for perspective taking. With this
context in mind, the researchers generated 10 new self
other awareness items, including I can tell the difference
between someone elses feelings and my own, and I am
aware of what other people think of me. We also
eliminated reverse-scored items in the selfother
awareness component because negative items tend to
load on a different factor than positive-worded items
(Brown, 2003), and we decided these reverse scored
items were not necessary for this component.

106

THE EMPATHY ASSESSMENT INDEX

The emotion regulation component also required

modification because the pilot version items were
insufficient. Although we did not choose reverse-scored
items for selfother awareness, we considered the
approach important for the emotion regulation
component. Regulating emotions is about changing the
way people think, which, in turn, changes the way they
feel; overall, a complex cognitive process (Ochsner,
Bunge, Gross & Gabrieli, 2002). Because cognitive
components of empathy require increased thought,
measurement must reflect that complexity and, therefore,
be more cognitively challenging. Developing these items
required creating a cognitive challenge without creating
undue confusion. Brown (2003) recommended that when
reverse-scored items are considered important,
researchers should use equal numbers of positively and
negatively worded items to better control for method
effects. Following this reasoning, the revised emotion
regulation component included five negatively worded
items and six positively worded.
In the next phase of the revision, we sought to
further clarify and refine the number of items in each
component by seeking feedback from focus groups and
content experts.
Revisions based on the focus groups. We
conducted focus groups with social work students and
community professionals to obtain their feedback
regarding the overall experience of taking the EAI as well
as to gather feedback on specific index items. Six focus
groups were conducted with 49 participants; group size
ranged from six to 15 participants. Four groups included
social work students (n = 24) and two groups were
conducted with social work community professionals (n
= 25) who were recruited from the School of Social
Works field instructor list. Before focus group
discussion started, each participant independently
completed the online version of the EAI. The focus group
participants then provided feedback through a structured,
facilitated dialogue that was documented by a note taker.
Participants identified several challenging aspects of
items, including items deemed difficult to answer using
the given the response options; items for which the
meaning of a word or phrase was unclear or could be
interpreted in multiple ways; and items that were too
broad, which hindered participants ability to provide a
concrete response. Participants also provided feedback
regarding the online process of administration. The
sample was chosen based on convenience and because
social workers receive training in empathy, making the
perspectives of the study sample especially beneficial.
However, that same training is also a possible source of
bias, and must be considered a limitation of this
approach.

Journal of the Society for Social Work and Research

Revisions based on the expert reviewers. The

expert reviewers included a sample of three researchers
identified as experts within the area of empathy. The
experts were selected using the following criteria: (a) the
person has published substantial original research in the
area of empathy or self-report instrumentation, and (b)
the person has credentials in developmental psychology,
social cognitive neuroscience, or social work. Each of the
expert reviewers was asked to comment on face and
content validity for the overall measure and identify any
wording and items he or she thought could be
problematic. The feedback from the expert reviewers was
similar to the feedback from focus groups regarding
problems with clarity of specific items.
Two expert reviewers and several focus group
participants recommended the researchers switch from a
5-point Likert scale to a 6-point scale ranging from never
(1) to always (6). The 5-point scale had been chosen for
the EAI based on guidelines for scale development
(DeVellis, 2003), and to provide enough categories to
allow for a meaningful variation in the answers. The
research team debated between 5- and 6-point scales,
opting to go with five response choices in the pilot
administration. The choice for an odd number allowed for
a central or neutral point. However, focus group feedback
indicated the response options did not represent an evenly
divided continuum; specifically, respondents noted that
the responses of frequently and always represented a
greater separation of responses than the other categories
(i.e., these responses were further apart than others).
The experts and participants suggested almost always as
a sixth choice. Therefore, the current version of the EAI
offers a range of six responses. An even number of
choices forces the respondent to make at least a weak
commitment in the direction of one or the other extreme
(DeVellis, 2003, p. 77) and, as such, the variability in
answers may be more meaningful.
Revision of the EAI. Based on the EFA of pilot data
and feedback from the focus groups and expert reviewers,
the revisions of several items and generation of new
items produced a final revision of the EAI that was ready
for testing. To obtain the most valid and useful results
possible, the research team created nine to 11 items
(some potentially redundant) for each component. In a
survival of the fittest process, researchers understood
that the items that induced the most reliable and valid
scores would emerge from the reliability and factor
analyses. This first iteration of the EAI included 50 items
and five components: (a) affective response comprised 10
items; (b) perspective taking comprised nine items; (c)
selfother awareness comprised 10 items; (d) emotion
regulation comprised 11 items; and (e) empathic attitudes
comprised 10 items. (See Appendix [p. 124] for a copy of
the EAI). The survey also gathered data on eight

107

LIETZ et al.
demographic items: sex, age, education level,
race/ethnicity, major (if the respondent was a student) or
occupation (if respondent was not a student), family-oforigins socioeconomic status, and current household
income.
CFA Phase: Participants, Procedures, and Data
Collection
The Arizona State University Institutional Review
Board granted the research team permission to recruit
students and nonstudents to voluntarily participate in the
testing the revised EAI. Because one objective of
developing the EAI was for the instrument to have
applicability across multiple settings, we used a
convenience sampling approach and recruited undergraduate students from a variety of majors as well as recruited
some participants who were not students but who
represented diverse professional backgrounds.
To recruit student participants representing diverse
fields of interest, the researchers approached undergraduate students in several general studies courses in
which students with a variety of majors enroll. These
courses included three course sections of Introduction to
Social Work; two course sections of Statistics for Social
Workers; two sections of The Living World, an
introductory-level biology course; and one section of
Macro Economic Principles. When recruiting the
students and explaining the study, the EAI instrument
was referred to as a human relations survey to avoid
social desirability or reactivity that may have occurred if
participants were given the names of each measure
included in the research. With the exception of one
instructor, all instructors of the above courses agreed to
offer extra credit to students who participated in the
research survey; the economics instructor chose not to
offer extra credit for participation. The offer of extra
credit was a useful recruitment strategy because it served
as an effective incentive to participate. However, this
inducement to participate may represent some bias in the
reporting.
The researchers sent an e-mail invitation to
participate to the 935 undergraduate students enrolled in
the target courses; those interested in participation could
do so by clicking on a hyperlink to a Qualtrics-based
survey. Qualtrics is an online survey software package
that allows participants to access a website with the
revised EAI at their own convenience. The students were
told the survey was voluntary, and that the survey would
remain active for 72 hours from the time the invitation
was issued. Of the 935 students contacted, 688 students
(74% response rate) completed the first administration of
the index. Four days after the first administration of the
EAI, 695 students (all students in the social work classes)
were asked to take the survey again. For this retest, 454

Journal of the Society for Social Work and Research

students (65%) completed the second administration of

the index.
In addition, the research team generated a
convenience snowball sample of 85 nonstudents. The
rationale for including nonstudents held that because the
EAI was intended for use with the general population,
expanding the research to include community
professionals (i.e., nonstudents) had the potential to
increase confidence regarding the robustness of the
index. The nonstudents were also contacted via e-mail,
and invited to participate in the study by clicking on a
link to the Web-based survey. In addition, these potential
participants were invited to forward the invitation and
survey link to colleagues who might be interested in
participating. The sample of nonstudents lived in five
states: Arizona, Illinois, Wisconsin, Colorado, and South
Carolina. Nonstudent participants represented 12
occupations, ranging from construction to finance, with
the majority working in the fields of public health and
social services. The snowball sampling strategy was used
to help increase diversity in the nonstudent sample.
The students and nonstudents were combined
because no important differences were found on the EAI
scores of these subsamples. This finding is explained in
more detail in the Results section. The final analytic
sample included 773 useable surveys for the CFA and
429 surveys for the test-retest analysis. The sample size
(N = 773) met the commonly agreed-on rule that each
item should have 10 or more participants (Nunnally &
Bernstein, 1994). Kim (2005) indicated that the
estimation of sample sizes in structural equation
modeling, given a provided power level, depends on the
number of variables or degrees of freedom, the strength
in relations among variables, and the choices of a
particular fit index (e.g., comparative fit index [CFI], root
mean square error of approximation [RMSEA]). For
example, a five-factor CFA analysis with 15 items needs
417 participants to ensure a power of .80 to detect a CFI
value of .95 or above, whereas the same CFA needs only
185 participants to have the same level of power to
conclude a RMSEA value of .05 or less. The number of
required sample sizes in SEM decreases as the number of
variables and the strength in the relations among these
variables increase (Kim, 2005). For these reasons, the
current sample size (N = 773) appeared to entail sufficient power to detect the differences between the
covariance matrix derived from the data and that from the
hypothesized model.
Additional Measures
Cognitive Emotion Regulation Questionnaire and
the Mindfulness Attention Awareness Scale. The retest
of the EAI (i.e., second round of administration with
n = 429 social work students) included additional items

108

THE EMPATHY ASSESSMENT INDEX

from two other instruments. In addition to the EAI items,

the retest round included nine items from the short-form
of the Cognitive Emotion Regulation Questionnaire
(CERQ-short; Garnefski & Kraaji, 2006) and eight items
from the Mindfulness Attention and Awareness Scale
(MAAS; Brown & Ryan, 2003). The intention was to
keep the number of items on the survey to a minimum
while retaining enough data to demonstrate concurrent
validity for the emotion regulation and selfother
awareness components.
The 18-item CERQ-short focuses on measuring
emotion regulation or coping strategies; this scale does
not include items that focus on a persons ability to
identify his or her emotions or to distinguish emotions.
The CERQ-short has demonstrated Cronbach's alphas
ranging from .62 to .85 and good factorial validity,
discriminative properties, and construct validity
(Garnefski & Kraaji, 2006).
The MAAS is a 15-item scale designed to measure a
persons open or receptive awareness and attention to
what is occurring in the present. Further, the MAAS has
proven to be predictive of self-regulation and selfawareness (Creswell, Way, Eisenberger, & Lieberman,
2007).The MAAS has been validated with student and
general populations, yielding a Cronbach's alpha of .81
and strong psychometric properties, including convergent
and divergent validity with other measures of
psychological well-being (Brown & Ryan, 2003).
Both CERQ-short and MAAS have been validated in
several studies, with diverse samples (Brown & Ryan,
2003; Carlson & Brown, 2005; Garnefski & Kraaij, 2007;
Jermann, Van, d'Acremont, & Zermatten, 2006; Zhu et
al., 2008). Based on theoretical expectation, we
hypothesized that the emotion regulation component of
the revised EAI would correlate with the CERQ-short
items. In addition, based on previous research and
theoretical expectation, we hypothesized that both the
EAI components of emotion regulation and selfother
awareness would correlate with the MAAS items.
Analytic Procedures

analyses. The Qualtrics-based data were first uploaded to

PASW 18 (formerly SPSS) for analysis. Two items were
eliminated (i.e., Q42, emotion regulation, and Q49,
empathic attitudes) because of problems that occurred
when the items were uploaded to Qualtrics. Participants
were not able to view these items properly and, therefore,
did not provide responses for these items. Twenty-one of
the 48 items had one to three missing cases, which would
result in a loss of 27 cases when using the listwise
deletion method. Following recent recommendations
(Allison, 2003; Peugh & Enders, 2004) that listwise
deletion method is prone to biased estimates, we used the
expectation maximization algorithm (Schafer & Graham,
2002) to impute the missing values. A total of 773 cases
with complete information were read in PASW 18 for
descriptive analyses.
Consistent with our intention to refine the EAI, we
conducted psychometric analyses (reliability and validity
tests) to consider the EAI factor structure and determine
the best model fit. We first performed internal
consistency reliability tests to examine the inter-item
relationship and identify items that would increase the
alpha if deleted.
On the basis of the reliability test, CFAs were
performed using Mplus to identify the best measurement
model of the EAI. Mplus was chosen because this
software provides several estimation methods to deal
with ordinal-level data, and we used the default
estimation method WLSMV (weighted least square with
mean- and variance- adjusted chi-square tests) in our
analysis (Muthn & Muthn, 1998-2010).
The WLSMV method generated a weighted matrix
based on the asymptotic variance and covariance of
polychoric correlations of the observed items (Flora &
Curran, 2004). A two-step CFA approach adopted in
previous scale validation studies (e.g., Brown, 2003) was
used. The two-step process is more feasible than a study
replication in that the two-step process enables
researchers to run CFA independently on both samples to
compare and confirm the results (Schumacker & Lomax,
2004).

This study involved three types of analyses: missing

value analysis, descriptive analyses, and psychometric

Journal of the Society for Social Work and Research

109

LIETZ et al.
Table 1
Description of Sample (N = 773)
Variable

f (Valid %)

Students

Community
Professionals

Male

196 (25.6%)

176(25.6%)

20(23.5%)

Female

568 (74.4%)

505(74.4%)

63(76.5%)

African American

47 (6.1%)

41 (6.0%)

6 (7.1%)

American Indian

17 (2.2%)

17 (2.5%)

0 (0%)

Asian American

21 (2.7%)

17 (2.5%)

4 (4.8%)

Caucasian

450 (58.7%)

381 (55.8%)

69 (82.1%)

Latino

130 (16.9%)

128 (18.7%)

2 (2.4%)

Mixed Race

51 (6.6%)

48 (7.0%)

3 (3.6%)

Other

51 (6.6%)

51 (7.5%)

0 (0%)

Sex

Race/Ethnicity

Major
Social Work

151 (22.3%)

Criminal Justice

114 (16.9%)

Psychology

63 (9.3%)

Nursing

26 (3.8%)

Education

31 (4.6%)

Sociology

18 (2.7%)

Undecided

28 (4.1%)

Other

245 (36.2%)

Employment Field
Education, Health, Social Services

41 (55.4%)

Public Administration or Management

13 (17.6%)

Retail

9 (12.2%)

Arts and Entertainment

3 (4.1%)

Finance, Insurance, Real Estate

1 (1.4%)

Other

7 (9.3%)

Participants were randomly divided into two

subsamples of comparable size (n = 389 and n = 384).
With the exception of one item, (i.e., Q13), the second
subsample had a slightly higher mean score than the
first subsample; no significant differences were found
in the mean scores of the other 48 items. Model
Journal of the Society for Social Work and Research

respecifications were made based on CFA results on the

first half sample before the ideal model fit was
identified. The final version of the EAI scale was tested
on the second half of the sample. There seems to be no
established rules regarding the cutoff values for fit
indices (Marsh, Hau, & Wen, 2004). Given the lack of
110

THE EMPATHY ASSESSMENT INDEX

rules, we used the following standards as guidelines for

a good model fit: normed chi-square (NC; i.e., the ratio
of chi-square vs. df 2.00); CFI .95; the weighted
root mean residual (WRMR < .95); and RMSEA < .06
with a confidence interval; Hu & Bentler, 1999;
Schreiber, Stage, King, Nora, & Barlow, 2006; Yu,
2002). Finally, the test-retest reliability and convergent
validity analyses were conducted using the whole
sample data (N = 773) to provide additional evidence of
the psychometric strength of the 17-item EAI index
Results
Sample Description
Table 1 describes the gender and race/ethnicity of
the sample (N = 773) as well as the major field of
student participants and occupation of community
professionals. The average age of the sample was 21.37
years. Although students were heavily recruited from
social work courses, the student participants
represented a variety of majors, with only 22% of the
students identifying social work as their major.
Considering the EAI was being developed for use
beyond social work, the research team was pleased that
most participants were involved in occupations or
majors beyond social work. However, a limitation is
that the sample was primarily female (74.4%),
Caucasian (58.7%), and over represents an
undergraduate perspective.
Mean Differences Across Demographic Variables
and the Five EAI Components
The use of t-tests indicated no statistically
significant differences existed between the student and
professionals groups across the means of the five EAI
components. However, t-test findings indicated two
statistically significant findings based on sex
differences. Females scored higher on selfother
awareness (p = .046) and perspective taking (p = .025).
These results are not unusual in that females tend to
score higher than males on self-report empathy
measures (Jolliffe & Farrington, 2006).
One-way ANOVA results identified no statistically
significant differences across racial/ ethnic groups on
four of the five components (i.e., affective response,
selfother awareness, perspective taking, and emotion
regulation). However, African Americans (41.19) and
Latinos (41.06) scored higher than Caucasians (37.74)
on the empathetic attitude component (p = .001).
Students identifying as social work majors also scored
higher on empathic attitudes (F = 7.667, p = .001) than

Journal of the Society for Social Work and Research

students who identified other majors. However, no

other statistically significant differences were found
across majors. Finally, participants who grew up in
poor or working class families scored higher on the
perspective-taking component (F = 6.885, p = .001) and
empathic attitudes (F = 7.667, p = .001) component
than participants who grew up in middle class, upper
class, or wealthy families. The ANOVA results indicate
that among the study sample, people of color, people
who grew up in lower income families, and students
majoring in social work majors had higher levels of
empathic attitudes than their counterparts.
Internal consistency reliability analysis. An internal consistency analysis was performed on the 50item, five-component EAI using Cronbachs alpha. The
results of the analysis for each component were as
follows: affective response ( = .84); perspective taking
( = .82); emotion regulation ( = .72); selfother
awareness ( = .70); and empathic attitudes ( = .81).
These results indicate excellent internal consistency for
the affective response, perspective taking, and empathic
attitudes components (Streiner, 2003). The emotion
regulation and self-other awareness components had
acceptable internal consistency. Parenthetically, the
CERQ-short and MAAS items had Cronbachs alpha
scores of .66 and .75, respectively. Emotion regulation
and selfother awareness scales have traditionally had
more difficulty achieving excellent internal consistency
(see Corcoran, 1982). Therefore, the research team
concluded the levels of internal consistency on the
emotion regulation and self-other awareness components were reasonable.
The items within components that, if deleted,
would result in higher alphas were flagged and
considered for elimination. In all, eight items were
eliminated based on the reliability analysis, particularly
interitem correlation coefficients and alpha if item
deleted results. The eliminated items were Q4, Q10,
Q11, Q13, Q15, Q25, Q28, and Q48.
CFA
After these items were eliminated, the CFA was
conducted on the remaining 40 items using Mplus
software. Consistent with Byrnes (1989) suggestion
that model specification should be based on theory as
well as the modification indices, we compared
several alternative models on the first half of the
sample (n = 389). The CFA models and results are
shown in Table 2.

111

LIETZ et al.
Table 2
Model Fit Indices of the CFA on the First Half of the Sample (n = 389)
Model Details

2 /Df

2 /Df
Ratio

CFI

WRMR

40 item five-factor model

2943.00/730**

4.03

.77

2.19

.088(.085;.092)

24-item five-factor model

694.11/242**

2.87

.92

1.33

.069(.063;.075)

24 item five-factor model with eight

error covariance added

400.14/234**

1.71

.96

.94

.043(.035;.050)

17 item five-factor model

213.70/109**

1.96

.97

.88

.05(.040; .060)

17 item five-factor model with

correlated error

185.16/107**

1.73

.98

.80

.04 (.033; .054)

RMSEA
(90% CI)

** p < .01

The 40-item five-factor model was not a good fit

(CFI = .77, WRMR = 2.19, RMSEA = .088). An additional 16 items were eliminated based on the findings for
the 40-item model, including small factor loadings (i.e.,
under .40), cross-loadings (items that could load on more
than one factor), highly correlated items, and items that
could cause the highest expected decrease in chi-square
values according to the modification index (Kline, 2010).
The eliminated items were Q14, Q18, Q32, and Q40 from
the affective response component; items Q6 and Q23
from the self-other awareness component; items Q7, Q31,
Q39, and Q45 from the emotion regulation component;
items Q1, Q34, and Q37 from the perspective-taking
component; and items Q5, Q19, and Q26 from the
empathic attitudes component. The researchers then
tested a 24-item measure using the five-factor model.
However, based on the good-fit criterion, the 24-item
five-factor model was still not an ideal fit (CFI = .92;
WRMR = 1.33; and RMSEA = .069). Therefore, in
keeping with the model modification index, several
measurement error covariances were added in the fivefactor model. The 24-item five-factor model with eight
error covariances added improved the model fit (CFI
=.96, WRMR =.94; RMSEA =.043).
The multiple error covariances indicated that the
items could be further refined. Therefore, seven more
items were removed (Q9, Q38, Q41, Q22, Q8, Q12, and
Q33). These items were removed based on the modifica-

Journal of the Society for Social Work and Research

tion index and the possibility of cross-loadings. For

example, the measurement error of Q8 is correlated with
the measurement error of Q2 and Q3; indicating Q8
might share some commonality with Q2 and Q3, each of
which represents a different latent construct. To avoid
item redundancy, Q8 was eliminated. As shown in Figure
2, the 17-item five-factor model was a good fit (CFI .97,
WRMR =.88; RMSEA =.05). The 17-item five-factor
model comprised three affective response items, three
self-other awareness items, four perspective-taking items,
four emotion regulation items, and three empathic
attitude items. As expected, all the factor loadings on the
five latent constructs were statistically significant (see
Table 3). Further review of the model fit modification
indices suggested an addition of two error correlations
(i.e., Q24 and Q20, Q43 and Q36). Both item Q20
(Watching a happy movie makes me feel happy) and
item Q24 (When I am with a happy person, I feel happy
myself) assessed affect sharing and loaded on affective
response, suggesting the influence of external forces on
personal affect. Although item Q36 (Friends view me as
a moody person) and item Q43 (I can imagine what a
character is feeling in a well-written book) loaded on
different factors (i.e., emotion regulation and perspective
taking, respectively), both items tap sensitivity to
emotions. For these reasons, the research team considered
the two error correlations to be conceptually sound.
Adding these two error covariances improved the model
fit (CFI =.98; WRMR =.80; and RMSEA=.04).

112

THE EMPATHY ASSESSMENT INDEX

Table 3
Confirmatory Factor Analysis of the 17-Item Model on the First Half of the Sample
CFA Model
Latent and Observed Variables

R2

Affective Response (AR)

Q20

.69

.48

Q24

.80

.64

Q44

.76

.58

Q17

.69

.48

Q21

.69

.48

Q29

.66

.44

Q27

.63

.40

Q35

.62

.38

Q36

.46

.21

Q47

.61

.37

Q2

.41

.17

Q30

.68

.46

Q43

.66

.44

Q50

.67

.45

Q3

.38

.14

Q16

.89

.79

Q46

.77

.59

SelfOther Awareness (SOA)

Emotion Regulation (ER)

Perspective Taking (PT)

Empathic Attitudes (EA)

Note. is the standardized factor loading of the observed variable on the latent construct. The
CFA also yielded results regarding the relationships among the five components (see Table 4).
As expected, the five factors are indeed inter-correlated with each other (p < .01).

Journal of the Society for Social Work and Research

113

LIETZ et al.
Table 4
Standardized Correlation Coefficients Among Latent Factors on the First Sample

Affective
Response

Affective
Response

SelfOther
Awareness

Emotion
Regulation

Perspective
Taking

Empathic
Attitudes

1.00

.58**

.32**

.54**

.18**

1.00

.58**

.86**

.23**

1.00

.62**

.19**

1.00

.20**

SelfOther
Awareness
Emotion
Regulation
Perspective
Taking
Empathic
Attitudes
**p < .01

1.00

Finally, the 24-item and 17-item five-factor models

were applied to the second half of the sample (n = 384).
As shown in Table 5, the model fit was satisfactory for
the 17-item five-factor model with two error covariances

added (CFI = .95, WRMR = .97, RMSEA = .06). Thus,

the 17-item EAI yielded the best model fit for the data in
this study.

Table 5
Model Fit Indices of the CFA on the Second Subsample (n = 384)
2 /Df

2 /Df

24 item five-factor model

694.11/242**

Ratio
2.87

.87

1.55

(90% CI)
.08(.077; .089)

17 item five-factor model

301.68/109**

2.77

.93

1.07

.07(.059; .077)

17 item five-factor model with

error covariance

185.16/107**

1.73

.95

.97

.06(.050; .069)

Model Details

CFI

WRMR

RMSEA

**p<.01
Comparisons Across Demographic Variables and
Components of the 17-item EAI
The use of t-tests indicated there were no statistically
significant differences between students and nonstudents.
The selfother awareness component approached
statistical significance with females scoring higher
(difference = .493, p = .061). One-way ANOVA results
indicated that the only statistically significant difference
among race or ethnic groups was on the empathic attitude
component (F = 14.01, p = .001) on which African
Americans (m = 12.68) and Latinos (m = 12.55) scored
higher than Caucasians (m = 11.45). Likewise, the only
statistically significant difference between college majors
was found for the empathic attitude component (F =
8.508, p = .001) with social workers having the highest
Journal of the Society for Social Work and Research

mean (m = 12.87). Psychology majors had a mean of

12.15, criminal justice 11.59, sociology 11.55, education
11.45 and nursing had a mean of 11.37. Empathic attitude
was the only component for which a statistically
significant difference was found for family-of-origin
socioeconomic status (F = 7.426, p = .001). Participants
who identified as poor or working class (m = 12.22)
scored higher on the empathic component than
participants who identified as middle class (m = 11.69),
upper-middle class (m = 11.29), or wealthy (m =
10.47).
The mean difference in the total 17-item EAI score
by sex was statistically significant (p = .05), with females
having higher scores. However, when the empathic
attitude component was removed, the statistically
114

THE EMPATHY ASSESSMENT INDEX

significant difference between the sexes disappeared. No

statistically significant differences in total mean scores
were found between students and nonstudents. In
addition, the mean differences of total scores across
race/ethnicity were not statistically significant with or
without the empathic attitude component. This finding is
important because it provides evidence that the EAI may
be useful with diverse populations. Total mean
differences across majors was statistically significant
(p =. 003) primarily due to social work majors high
mean (m = 75.15). However, when the empathic attitude
component was removed, the statistical significance
disappeared (p = .20).
Finally, the total 17-item EAI mean score difference
across family-of-origins socioeconomic status was
statistically significant with (p = .01) and without (p =
.024) the empathic attitude component. Highest means
were found for participants who identified as poor or
working class, and the mean scores dropped successively
as socioeconomic status increased. This finding is
consistent with recent research that has suggested socioeconomic status influences the ways in which people
experience emotion and shapes empathic accuracy
(Kraus, Cote, & Keltner, 2010; Kraus & Keltner, 2009;
Kraus, Piff, & Keltner, 2009).
Reliability and Test-Retest Analysis
Cronbachs alpha was used to perform an internal
consistency analysis on the 17- item EAI ( = .823). In
addition, the alphas for each component are as follows:
the affective response three-item component ( = .751);
the cognitive-based component (11 items; = .785); and
the three-item empathic attitude component ( = .671).
These results indicate acceptable internal consistency.
Finally, a test-retest reliability study was carried out
using the data from the students who finished both
administrations of the survey within one week of each
other. All of the component or component scores for the
17-item EAI for both administrations were significantly
correlated (n = 429; p = .001). The correlation coefficients were as follows: affective response (r = .743,
Spearmans rho = .739); selfother awareness (r = .686,
Spearmans rho = .670); emotion regulation (r = .759,
Spearmans rho = .748); perspective taking (r = .771,
Spearmans rho = .776); empathic attitude (r = .792

Journal of the Society for Social Work and Research

Spearmans rho = .759). These are considered to be

strong correlations indicating healthy test-retest reliability
(Cohen, 1988).
Construct Validity Tests
The pilot administration of the EAI (Gerdes et al.,
2011) included items from the empathic concern and
perspective-taking components of the Interpersonal
Reactivity Index (Davis, 1980; 1983). The items were
used to demonstrate concurrent validity for the EAIs
affective response, perspective-taking and empathic
attitudes components. In each case, the results indicated
statistically significant correlations between the scales
with Pearson rs ranging between .48 and .75. For the
current research, the team used correlation coefficients to
analyze the concurrent validity of two components from
the 17-item EAI: ER and selfother awareness.
The first hypothesis predicted a positive relationship
between the emotion regulation component and the
CERQ-short. The emotion regulation component and the
CERQ-short had a statistically significant correlation (n =
429, r = .507, Spearmans rho = .493, p = .001). These
findings indicate a moderately strong positive correlation
between the two scales (Cohen, 1988). Keep in mind that
the CERQ-short is designed to measure emotion
regulation coping strategies whereas the emotion
regulation component is designed to measure a persons
emotion regulation in the context of empathy. It was also
hypothesized that the four-item emotion regulation and
the three-item selfother awareness components would
be negatively correlated with the MAAS score (a lower
MAAS score indicates more mindfulness). The emotion
regulation and MAAS were modestly correlated (n = 424,
r = -.267, Spearmans rho = -.270, p = .001). This result
is acceptable given that the emotion regulation and
MAAS are measuring independent constructs that are
theoretically related but not identical (Creswell et. al,
2007). The selfother awareness and MAAS had a
moderate negative correlation (n = 424, r = -.396,
Spearmans rho = -.435, p = .001). Presumably, the
correlation would have been stronger if the researchers
had been measuring self-awareness only rather than self
other awareness (Decety & Sommerville, 2003).
However, this expected result provides initial evidence
for construct validity.

115

LIETZ et al.
Figure 1. The 24-item CFA model

Q9
Q20

.46

.75

Affective
Response

.80

Q24

.98

Q44

.51

Q17
Q21
Q29

.68
.66

Self/Other
Awareness

.62

.31

.47

Q38

.49
.72

.66

Q41
Q22
Q27
Q35

.42
.65

Emotion
Regulation

.65

.90

.47

Q36
.58
.61

Q47
Q2
Q8

.42
.63
.65

Q30
Q43

.24

Perspective
Taking

.21

.66

.69

.24

.18

Q50
Q3
.36

Q12
Q16

Q33

.55
Empathic
Attitudes

.88
.53

.77

Q46

Journal of the Society for Social Work and Research

116

THE EMPATHY ASSESSMENT INDEX

Discussion
After examining the internal consistency of the 48item EAI, researchers used CFA techniques to test a 40item five-factor model of empathy. The 40-item EAI had
insufficient model compatibility; therefore, items were
sequentially eliminated until criteria for a good model fit
were met. Elimination was based on several criteria: (a)
non-significant factor loadings ( < .40), (b) items that
loaded significantly on more than one factor, (c) highly
correlated items (r >. 80), and (d) items that would
cause the highest expected decrease in chi-square values.
The resulting model was a 17-item five-factor model that
achieved good model fit with half the sample and
reasonable model fit with the other half of the sample.
The 17-item EAI version has excellent internal
consistency and strong test-retest reliability. Convergent
validity correlation coefficients for the CERQ-short and
the MAAS were statistically significant.
The CFA highlighted weak to moderate (.32 to .58)
intercorrelations between the affective response
component and the three cognitive components (i.e.,
emotion regulation, selfother awareness, perspective
taking). The intercorrelations between the three cognitive
components ranged from strong to very strong (.58 to
.86). Although the intercorrelations between the empathic
attitude component and the four other components were
statistically significant, the intercorrelations were very
modest (.18 to .23).
It is not surprising that the perspective taking and
selfother awareness components have the highest
intercorrelation (r = .86). The social cognitive
neuroscience literature makes it clear that accurate
perspective taking requires selfother awareness or that
an essential part of empathy is to recognize the other
person as like the self while maintaining a clear
separation between self and other (Ruby & Decety,
2004, p. 988). It is more difficult to establish
discriminant validity between two constructs that are so
closely intertwined and function simultaneously, though
separately, in the brain. However, the literature supports
the assumption that perspective taking and selfotherawareness are isolable and observable neural networks
that operate in conjunction with each other (Decety &
Grezes, 2006; Ruby & Decety, 2003, 2004).
SelfOther awareness has a strong correlation with
both perspective taking and affective response. The same
personal boundary must be maintained to feel what the
other person is feeling (affective response) while
maintaining a clear separation between self and other

Journal of the Society for Social Work and Research

(Decety & Jackson, 2004; Decety & Meyer, 2008). Given

this reality, it might be assumed that the correlation
between the two components would be even higher. The
moderately strong intercorrelation may be the result of
the lack of content validity in the affective response
component. After elimination procedures, the three-item
affective response component was limited to one aspect
of mirroring (i.e., happiness). In the next round of testing
of the EAI, the researchers will revise the affective
response component by adding items to increase the
content validity. Suggestions for these additional items
include I understand other peoples emotional signals
and I am good at judging other peoples emotional
states, which could be added to Hearing laughter makes
me smile to bolster the dimensions of measuring
affective response.
It is not unexpected that the empathic attitude
component has the lowest intercorrelations with the other
components, with values ranging from .18 to .23. The
empathic attitude component is the only factor in the
model not grounded in the social cognitive neuroscience
literature. The empathic attitude items are a set of social
justice attitudes that are intended to be reflective of, and a
proxy for, the commitment to change and action that is
integral to the social work profession. However, based on
these results, it can be argued that attitudes are simply not
a good proxy for action. Further research is needed that
can measure action that is implemented (i.e., actually
taken) rather than estimating action through a set of
attitudes.
There are limitations to this study. The sampling
strategy for the focus groups may have overemphasized
the social work perspective, especially considering the
intended purpose was to develop the EAI as an
instrument for application across multiple disciplines and
settings. The use of convenience sampling for the
administration phase was also a limitation in that this
technique over represented the perspectives of
undergraduate students. In addition, the sample was
predominantly female and Caucasian; therefore, the
sample is not representative of the area in which the
study was conducted. Although self-report instruments
are an efficient way to collect data, this mechanism is
prone to social desirability and bias. The use of extra
credit as a participation incentive may have also affected
participant reporting. Finally, the sample is relatively
high functioning considering that all participants were
either students or currently employed. The index should
be tested on clinical samples to see whether the EAI is
applicable for high-risk groups.

117

LIETZ et al.
Figure 2. The 17-item CFA model

Q20
Q24

Affective
Response

Q44
Q17
Q21

SelfOther/
Awareness

Q29

Q27
Q35

Emotion
Regulation

Q36
Q47

Q2
Q30

Perspective
Taking

Q43
Q50
Q3
Q16

Empathic
Attitudes

Q46

Note. Factor loadings and correlations between factors are reported in Table 3 and Table 4, respectively

Journal of the Society for Social Work and Research

118

THE EMPATHY ASSESSMENT INDEX

Implications
The EAI has many useful applications in practice,
policy, and educational settings in social work and other
disciplines. Students in social work are encouraged to use
empathy in client assessment, interaction, and
interventions. However, the essential components of
empathy are not consistently incorporated into social
work curriculum (Gerdes, Segal, Jackson, & Mullins,
2011). For example, social workers are rarely taught that
the simple act of mirroring clients triggers a biological
response that can enhance the ability to develop rapport
and establish increased level of awareness of the clients
feeling state. Understanding the full range of empathic
abilities, including affective response, selfother
awareness, and the ability to perspective take and
regulate emotions may be helpful for practitioners. With
the use of the EAI, students and professionals can
develop better understandings of boundaries and use
these ideas to guide their practice. In addition, by having
a clear understanding of these emotional and cognitive
mechanisms, social workers may be better able to
manage stress, compassion fatigue, and burnout.
In addition, recent social cognitive neuroscience
research has described the potential for increasing brain
elasticity and the brains ability to be retrained, which has
introduced important implications for clinical
intervention with clients dealing with trauma, brain
injuries, and autism. In these cases, the EAI could be
used to identify specific components of treatment and
assess progress towards goals. Similarly, for client
populations thought to have lower levels of empathy,
such as sex and interpersonal violence offenders, the EAI
offers a tool to treatment programs that are considering
how best to incorporate the new science of empathy in
ways that can be evaluated.
Conclusion
In conclusion, the findings indicate the 17-item fivefactor self-report EAI is capable of generating reliable
and sufficiently valid scores. However, the evidence also
indicates that a purely social cognitive neuroscience fourfactor model of empathy (i.e., affective response, self
other awareness, perspective taking and emotion
regulation) may yield more useful results. The fifth
component, empathic attitudes, was used as a proxy for
action. The use of empathic attitudes, which was a social
justice extension of the social cognitive neuroscience
model of empathy, was not supported by the findings.
Therefore, before the next round of data collection, the
researchers intend to revise the EAI to a 20-item fourfactor model by eliminating the empathic attitude
component. The affective response component will be
modified to include five items that demonstrate improved
content validity over the current three-item component.

Journal of the Society for Social Work and Research

The emotion regulation component appears to be

sufficient, whereas the perspective taking and selfother
awareness components also require some revision of
terminology and possible inclusion of a new item to
strengthen the construct.
Author Note
This research was supported by a grant from the
Samuel and Lois Silberman Foundation.
Cynthia A. Lietz is an assistant professor in the
School of Social Work at Arizona State University.
Karen E. Gerdes is an associate professor in the
School of Social Work at Arizona State University;
kegerdes@asu.edu
Fei Sun is an assistant professor in the School of
Social Work at Arizona State University;
fei.sun.1@asu.edu
Jennifer Mullins Geiger is a doctoral student in the
School of Social Work at Arizona State University;
jennifer.mullins@asu.edu
M. Alex Wagaman is a second-year doctoral
student in the School of Social Work at Arizona State
University; mary.wagaman@asu.edu
Elizabeth A. Segal is a professor in the School of
Social Work at Arizona State University;
esegal@asu.edu
Correspondence regarding this article should be
sent to Cynthia Lietz at clietz@asu.edu
References
Allison, P. D. (2003). Missing data techniques for
structural equation modeling. Journal of Abnormal
Psychology, 112, 545-557. doi:10.1037/0021843X.112.4.545
Batson, C. D., Polycarpou, M., Harmon-Jones, E.,
Imhoff, H., Mitchener, E., Bednar, L.,
Highberger, L. (1997). Empathy and attitudes: Can
feeling for a member of a stigmatized group
improve feelings toward the group? Journal of
Personality and Social Psychology, 72, 105-118.
doi:10.1037/0022-3514.72.1.105
Boyle, S. W., Hull Jr., G. H., Mather, J. H., Smith, L.
L., & Farley, O. W. (2006). Direct practice in
social work. Boston, MA: Pearson.
Brown, T. A. (2003). Confirmatory factor analysis of
the Penn State Worry Questionnaire: Multiple
factors or method effects? Behavioral Research

119

LIETZ et al.
and Therapy, 41, 1411-1426. doi:10.1016/S00057967(03)00059-7
Brown, K.W., & Ryan, R.M. (2003). The benefits of
being present: Mindfulness and its role in
psychological well-being. Journal of Personality
and Social Psychology, 84, 822-848.
doi:10.1037/0022-3514.84.4.822
Byrne, B. M. (1989). A primer of Lisrel: Basic
applications and programming for confirmatory
factor analytic models. New York, NY: Springer
Verlag.
Carlson, L. E., & Brown, K. W. (2005). Validation of
the mindful attention awareness scale in a cancer
population. Journal of Psychosomatic Research,
58, 29-33. doi:10.1016/j.jpsychores.2004.04.366
Cliffordson, C. (2002). The hierarchical structure of
empathy: Dimensional organization and relations
to social functioning. Scandinavian Journal of
Psychology, 43, 49-59. doi:10.1111/14679450.00268

Davis , M. H. (1996). Empathy: A social psychological

approach. Boulder, CO: Westview.
http://www.questia.com/PM.qst?a=o&d=98667200
de Kemp, R.A.T., Overbeek, G., de Wied, M., Engels,
R.C.M.E., & Scholte, R.H.J. (2007). Early
adolescent empathy, parental support, and
antisocial behavior. Journal of Genetic
Psychology, 168(1), 5-18.
doi:10.3200/GNTP.168.1.5-18
Decety, J. (2007). A social cognitive neuroscience
model of human empathy. In E.Harmon-Jones &
P. Winkielman (Eds.), Social neuroscience:
Integrating biological and psychological
explanations of social behavior (pp. 246-270).
New York: Guilford Press.
Decety, J., & Grezes, J. (2006). The power of
simulation: Imagining ones own and others
behavior. Brain Research, 1079, 4-14.
doi:10.1016/j.brainres.2005.12.115

Cohen, J. (1988). Statistical power for the behavioral

sciences (2nd ed.). New York, NY: Routledge.

Decety, J., & Jackson, P. L. (2004). The functional

architecture of human empathy. Behavioral and
Cognitive Neuroscience Reviews, 3, 71-100.
doi:10.1177/1534582304267187

Corcoran, K. J. (1982). An exploratory investigation

into selfother differentiation: Empirical evidence
for a monistic perspective of empathy.
Psychotherapy: Theory, Research & Practice, 19,
63-68. doi:10.1037/h0088418

Decety, J., & Meyer, M. (2008). From emotion

resonance to empathic understanding: A social
developmental neuroscience account. Development
and Psychopathology, 20, 1053-1080.
doi:10.1017/S0954579408000503

Creswell, J. D., Way, B. M., Eisenberger, N. I., &

Lieberman, M. D. (2007). Neural correlates of
dispositional mindfulness during affect labeling.
Psychosomatic Medicine, 69, 560-565.
doi:10.1097/PSY.0b013e3180f6171f

Decety, J., & Moriguchi, Y. (2007). The empathic

brain and its dysfunction in psychiatric
populations: Implications for intervention across
different clinical conditions. BioPsychoSocial
Medicine, 1(22), 1-21. doi:10.1186/1751-0759-122

Curtner-Smith, M. E., Culp, A. M., Culp, R., Scheib,

C., Owen, B. S., Tilley, A., Coleman, P. W.
(2006). Mothers parenting and young
economically disadvantaged childrens relational
and overt bullying. Journal of Child and Family
Studies, 15(2), 177-189. doi:10.1007/s10826-0059016-7
Davis, M. H. (1980). A multidimensional approach to
individual differences in empathy. JSAS Catalog
of Selected Documents in Psychology, 10, 85.
http://www.uv.es/~friasnav/Davis_1980.pdf
Davis, M. H. (1983). Measuring individual differences
in empathy: Evidence for a multidimensional
approach. Journal of Personality and Social
Psychology, 44(1), 113-126. doi:10.1037/00223514.44.1.113

Journal of the Society for Social Work and Research

Decety, J., & Sommerville, J. A. (2003). Shared

representations between self and other: A social
cognitive neuroscience view. Trends in Cognitive
Sciences, 7(12), 527-533.
doi:10.1016/j.tics.2003.10.004
DeVellis, R. F. (2003). Scale development: Theory and
applications. Thousand Oaks, CA: Sage.
Duan, C., & Hill, C. E. (1996). The current state of
empathy research. Journal of Counseling
Psychology, 43(3), 261-274. doi:10.1037/00220167.43.3.261
Eisenberg, N. (2006). Empathy-related responding and
prosocial behaviour. In G. Bock & J. Goode (Eds.),
Empathy and fairness: Novartis Foundation
Symposium 278 (pp. 71-88). Chichester, UK: John
120

THE EMPATHY ASSESSMENT INDEX

Wiley & Sons. doi:10.1002/9780470030585.ch6

http://onlinelibrary.wiley.com/book/10.1002/97804
70030585
Eisenberg, N., Fabes, R. A., Murphy, B., Karbon, M.,
Maszk, P., Smith, M., Suh, K. (1994). The
relations of emotionality and regulation to
dispositional and situation empathy-related
responding. Journal of Personality and Social
Psychology, 66, 776-797. doi:10.1037/00223514.66.4.776
Flora, D. B., & Curran, P. J. (2004). An empirical
evaluation of alternative methods of estimation for
confirmatory factor analysis with ordinal data.
Psychological Methods, 9, 466-491.
doi:10.1037/1082-989X.9.4.466
Freedberg , S. (2007). Re-examining empathy: A
relational feminist point of view. Social Work,
52(3), 251-259.
http://findarticles.com/p/articles/mi_hb6467/is_3_5
2/ai_n29371634/
Froman, R. D., & Peloquin, S. M. (2001). Rethinking
the use of the Hogan Empathy Scale: A critical
psychometric analysis. American Journal of
Occupational Therapy, 55, 566-572.
doi:10.5014/ajot.55.5.566
Garnefski, N., & Kraaji, V. (2006). Cognitive Emotion
Regulation Questionnaire Development of a
short 18-item version (CERQ-short). Personality
and Individual Differences, 41, 1045-1053.
doi:10.1016/j.paid.2006.04.010
Garnefski, N., & Kraaij, V. (2007). The Cognitive
Emotion Regulation Questionnaire: Psychometric
features and prospective relationships with
depression and anxiety in adults. .European
Journal of Psychological Assessment, 23, 141-149.
doi:10.1027/1015-5759.23.3.141
Gerdes, K. E., Lietz, C. A., & Segal, E. A. (2011).
Measuring empathy in the 21st century: The
development of an empathy index rooted in social
cognitive neuroscience and social justice. Social
Work Research, 35(2), 83-93.
http://findarticles.com/p/articles/mi_hb6481/is_2_3
5/ai_n57667649/pg_2/
Gerdes, K. E., & Segal, E. A. (2009). A social work
model of empathy. Advances in Social Work,
10(2), 114-127.
http://journals.iupui.edu/index.php/advancesinsocia
lwork/article/viewArticle/235

Journal of the Society for Social Work and Research

Gerdes, K. E., Segal, E. A., Jackson, K. F., & Mullins,

J. L. (2011). Teaching empathy: A framework
rooted in social cognitive neuroscience and social
justice. Journal of Social Work Education, 47, 109131. doi:10.5175/JSWE.2011.200900085
Gerdes, K. E., Segal, E. A., & Lietz, C. A. (2010).
Conceptualising and measuring empathy. British
Journal of Social Work, 40(7), 2326-2343.
doi:10.1093/bjsw/bcq048
Hepworth, D., Rooney, R., Rooney, G., StromGottfried, K., & Larsen, J. (2006). Direct social
work practice: Theory and skills. Belmont, CA:
Thomson/Brooks & Cole.
Hoffman, M. L. (2000). Empathy and moral
development: Implications for caring and justice.
New York, NY: Cambridge University Press.
http://catdir.loc.gov/catdir/samples/cam032/990296
69.pdf
Hogan, R. (1969). Development of an empathy scale.
Journal of Consulting and Clinical Psychology, 33,
307-316. doi:10.1037/h0027580
Hojat, M., Magione, S., Nasca, T. J., Cohen, M. J.,
Gonnella, J. S., Erdmann, J. B., Magee, M.
(2001). The Jefferson Scale of Physician Empathy:
Development and preliminary psychometric data.
Educational and Psychological Measurement, 61,
349-365. doi:10.1177/00131640121971158
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit
indexes in covariance structure analysis:
Conventional criteria versus new alternatives.
Structural Equation Modeling, 6(1), 1-55.
doi:10.1080/10705519909540118
Hunter, J. A., Figueredo, A. J., Becker, J. V., &
Malamuth, N. (2007). Non-sexual delinquency in
juvenile sex offenders: The mediating and
moderating influences of emotional empathy.
Journal of Family Violence, 22, 43-54.
doi:10.1007/s10896-006-9056-9
Iacoboni, M. (2009). Imitation, empathy, and mirror
neurons. Annual Review of Psychology, 60, 653670. doi:10.1146/annurev.psych.60.110707.163604
Jermann, F., Van, d. L., d'Acremont, M., & Zermatten,
A. (2006). Cognitive Emotion Regulation
Questionnaire (CERQ): Confirmatory factor
analysis and psychometric properties of the French
translation. European Journal of Psychological
Assessment, 22(2), 126-131. doi:10.1027/10155759.22.2.126

121

LIETZ et al.
Jolliffe, D., & Farrington, D. P. (2006). Development
and validation of the Basic Empathy Scale. Journal
of Adolescence, 29, 589-611.
doi:10.1016/j.adolescence.2005.08.010

Richmond, M. (1917). Social diagnosis . New York,

NY: Russell Sage Foundation.
http://www.archive.org/details/socialdiagnosis00ric
hiala

Kim, K. H. (2005). The relation among fit indexes,

power, and sample size in structural equation
modeling. Structural Equation Modeling, 12(3),
368-390. doi:10.1207/s15328007sem1203_2

Rogers, C. (1957). The necessary and sufficient

conditions for therapeutic personality change.
Journal of Consulting Psychology, 21, 95-103.
doi:10.1037/h0045357

Kline, R. B. (2010). Principles and practice of

structural equation modeling (3rd ed.). New York,
NY: Guilford Press.

Rogers, C. (1975). Empathic: An unappreciated way of

being. Counseling Psychologist, 5(2), 2-10.
doi:10.1177/001100007500500202

Kohut, H. (1959). Introspection, empathy, and

psychoanalysis: An examination of the relationship
between the mode of observation and therapy.
Journal of the American Psychoanalytic
Association, 7, 459-483.
doi:10.1177/000306515900700304

Ruby, R., & Decety, J. (2003). What you believe

versus what you think they believe: a
neuroimaging study of conceptual perspectivetaking. European Journal of Neuroscience, 17,
2475-2480. doi:10.1046/j.1460-9568.2003.02673.x

Kraus, M. W., Cote, S., & Keltner, D. (2010). Social

class, contextualism, and empathic accuracy.
Psychological Science, 21(11), 1716-1723.
doi:10.1177/0956797610387613
Kraus, M. W., & Keltner, D. (2009). Signs of
socioeconomic status: A thin-slicing approach.
Psychological Science, 20, 99-106.
doi:10.1111/j.1467-9280.2008.02251.x
Kraus, M. W., Piff, P. K., & Keltner, D. (2009). Social
class, sense of control, and social explanation.
Journal of Personality and Social Psychology, 97,
992-1004. doi:10.1037/a0016357
Leibetseder, M., Laireiter, A.-R., & Kller, T. (2007).
Structural analysis of the E-scale. Personality and
Individual Differences, 42, 547-561.
doi:10.1016/j.paid.2006.08.002
Muthn, L. K., & Muthn, B. O. (1998-2010). Mplus
uerss guide (6th ed.). Los Angeles, CA: Muthn &
Muthn.
http://www.statmodel.com/download/usersguide/M
plus%20Users%20Guide%20v6.pdf
Nunnally, J. C., & Bernstein, I. H. (1994).
Psychometric theory (3rd ed.). New York, NY:
McGraw-Hill.
Peugh, J. L., & Enders, C. K. (2004). Missing data in
educational research: A review of reporting
practices and suggestions for improvement. Review
of Educational Research, 74(4), 525-556.
doi:10.3102/00346543074004525
Pithers, W. (1999). Empathy: Definition, enhancement,
and relevance to the treatment of sexual abusers.
Journal of Interpersonal Violence, 14(3), 257-284.
doi:10.1177/088626099014003004
Journal of the Society for Social Work and Research

Ruby, R., & Decety, J. (2004). How would you feel

versus how do you think she would feel? A
neuroimaging study of perspective-taking with
social emotions. Journal of Cognitive
Neuroscience, 16(6), 988-999.
doi:10.1162/0898929041502661
Saleebey, D. (2009). The strengths perspective in
social work practice (5th ed.). Boston, MA:
Pearson/Allyn & Bacon.
Sartori, R., & Pasini, M. (2007). Quality and quantity
in test validity: How can we be sure that
psychological tests measure what they have to?
Quality & Quantity, 41, 359-374.
doi:10.1007/s11135-006-9006-x
Schafer, J. L., & Graham, J. W. (2002). Missing data:
Our view of the state of the art. Psychological
Methods, 7, 147-177. doi:10.1037/1082989X.7.2.147
Schreiber, J. B., Stage, F. K., King, J., Nora, A., &
Barlow, E. A. (2006). Reporting structural equation
modeling and confirmatory factor analysis results:
A review. Journal of Educational Research, 99(6),
323-338. doi:10.3200/JOER.99.6.323-338
Schumacker, R. E., & Lomax, R. G. (2004). A
beginners guide to structural equation modeling.
Mahwah, NJ: Lawrence Erlbaum Associates.
http://www.4shared.com/document/tBYndtkU/A_
Beginners_Guide_to_Structura.html
Shulman, L. (2009). The skills of helping individuals,
families, groups, and communities. Belmont, CA:
Brooks/Cole/Cengage Learning.
Spreng, R. N., McKinnon, M. C., Mar, R. A., &
Levine, B. (2009). The Toronto Empathy
Questionnaire: Scale development and initial
122

THE EMPATHY ASSESSMENT INDEX

validation of a factor-analytic solution to multiple

empathy measures. Journal of Personality
Assessment, 91(1), 62-71.
doi:10.1080/00223890802484381
Streiner, D. L. (2003). Starting at the beginning: An
introduction to coefficient alpha and internal
consistency. Journal of Personality Assessment,
80(1), 99-103.
doi:10.1207/S15327752JPA8001_18
Towle, C. (1945). Common human needs. Washington,
DC: Federal Bureau of Assistance.
http://hdl.handle.net/2027/mdp.39015072125506
Varker, T., & Devilly, G. J. (2007). Types of empathy
and adolescent sexual offenders. Journal of Sexual
Aggression, 13(2), 139-149.
doi:10.1080/13552600701661573
Wang, Y., Davidson, M. M., Yakushko, O. F., Savoy,
H. B., Tan, J.A., & Bleier, J. K. (2003). the scale of
ethnocultural empathy: Development, validation,

Journal of the Society for Social Work and Research

and reliability. Journal of Counseling Psychology,

50, 221-234. doi:10.1037/0022-0167.50.2.221
Whittaker, M. K., Brown, J., Beckett, R., & Gerhold,
C. (2006). Sexual knowledge and empathy: A
comparison of adolescent child molesters and nonoffending adolescents. Journal of Sexual
Aggression, 12, 143-154.
doi:10.1080/13552600600823621
Yu, C.Y. (2002). Evaluating cutoff criteria of model fit
indices for latent variable models with binary and
continuous outcomes. (Doctoral dissertation,
University of California, Los Angeles).
http://www.statmodel.com/download/Yudissertatio
n.pdf
Zhu, X., Auerbach, R. P., Yao, S., Abela, J.R.Z., Xiao,
J., & Tong, X. (2008). Psychometric properties of
the cognitive emotion regulation questionnaire:
Chinese version. Cognition & Emotion, 22, 288307. doi:10.1080/02699930701369035

123

LIETZ et al.
Appendix
The Empathy Assessment Index (EAI) = 50 items (final 17 items are in bold)
Five components: Affective Response (AR), Emotion Regulation (ER), Perspective Taking (PT), SelfOther
Awareness (SOA), and Empathic Attitudes (EA)
____________________________________________________________________________________
Q1
Q2
Q3
Q4
Q5
Q6
Q7
Q8
Q9
Q10
Q11
Q12
Q13
Q14
Q15
Q16
Q17
Q18
Q19
Q20
Q21
Q22
Q23
Q24
Q25
Q26
Q27
Q28
Q29
Q30
Q31
Q32
Q33
Q34
Q35
Q36
Q37
Q38
Q39
Q40
Q41
Q42
Q43
Q44
Q45
Q46
Q47
Q48
Q49
Q50

I am open to listening to the points of view of others. (PT)

I can imagine what its like to be in someone elses shoes. (PT)
If a person is poor, I believe it is the result of bad personal choices. (EA)
When I see a stranger crying, I feel like crying. (AR)
I believe unemployment is brought on by individuals failures. (EA)
I can tell how I am feeling emotionally by noticing how my body feels. (SOA)
When something exciting happens, I get so excited I feel out of control. (ER)
I consider other peoples point of view in discussions. (PT)
Seeing someone dance makes me want to move my feet. (AR)
When someone insults me or verbally attacks me, I dont let it bother me. (ER)
I am not aware of how I feel about a situation until after the situation is over. (SOA)
I believe poverty is brought on by individuals failures.
(EA)
When a friend is sad and it affects me deeply, it does not interfere with my own quality of life. (ER)
When I see a friend crying, I feel like crying. (AR)
I feel what another person is feeling, even when I do not know the person. (AR)
I believe adults who are poor deserve social assistance. (EA)
I am aware of my thoughts. (SOA)
When I am with a sad person, I feel sad myself. (AR)
I believe government should support our well-being. (EA)
Watching a happy movie makes me feel happy. (AR)
I can tell the difference between someone elses feelings and my own. (SOA)
I have angry outbursts. (ER)
I have a physical reaction (such as shaking, crying or going numb) when I am upset. (SOA)
When I am with a happy person, I feel happy myself. (AR)
When I care deeply for people, it feels like their emotions are my own. (SOA)
I think society should help out children in need. (EA)
When I am upset or unhappy, I get over it quickly. (ER)
I can imagine what it is like to be poor. (PT)
I can explain to others how I am feeling. (SOA)
I can agree to disagree with other people. (PT)
I get overwhelmed by other peoples anxiety.
(ER)
When a friend is happy, I become happy. (AR)
I believe government should be expected to help individuals. (EA)
I like to view both sides of an issue. (PT)
Emotional evenness describes me well. (ER)
Friends view me as a moody person. (ER)
It is easy for me to see other peoples point of view. (PT)
I am aware of how other people think of me. (SOA)
When I get upset, I need a lot of time to get over it. (ER)
When a friend is sad, I become sad. (AR)
I can distinguish my friends feelings from my own. (SOA)
I have large emotional swings. (ER)
I can imagine what the character is feeling in a well written book. (PT)
Hearing laughter makes me smile. (AR)
I rush into things without thinking. (ER)
I think society should help out adults in need. (EA)
I watch other peoples feelings without being overwhelmed by them. (ER)
I am comfortable helping a person of a different race or ethnicity than my own. (EA)
I believe the United States economic system allows for anyone to get ahead. (EA)
I can simultaneously consider my point of view and another persons point of view. (PT)

Journal of the Society for Social Work and Research

124