Acta Psychiatr Scand 2015: 18

All rights reserved

DOI: 10.1111/acps.12401

2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
ACTA PSYCHIATRICA SCANDINAVICA

Do attitudes towards persons with mental

illness worsen during the course of life? An
age-period-cohort analysis
Schomerus G, Van der Auwera S, Matschinger H, Baumeister SE,
Angermeyer MC. Do attitudes towards persons with mental illness
worsen during the course of life? An age-period-cohort analysis.
Objective: Cross-sectional studies frequently nd higher age associated
with negative attitudes towards persons with mental illness. We explore
whether attitudes worsen over the life span, or follow a cohort pattern.
Method: Using data from three identical population surveys in
Germany from 1990, 2001 and 2011 (combined sample n = 7835), we
performed ageperiodcohort analyses determining the association of
age, time period and birth-cohort with social distance from a person
with either depression (n = 3910) or schizophrenia (n = 3925), using
linear and nonlinear partial least squares regression models.
Results: Social distance increases with age, independent from cohort
and period eects, cumulating to an increase of 2.4 (schizophrenia) and
2.3 (depression) on the 28 point social distance scale over the life span
(youngest to oldest participant). We found a cohort eect in depression,
but not schizophrenia, with decreasing social distance until 1970 and a
slight increase in younger cohorts. Period eects were visible
particularly in schizophrenia, with growing social distance over time.
Conclusion: Considering demographic change and the vulnerability of
older persons to severe outcomes of mental illness such as suicide, the
observed increase of negative attitudes over the life span seems highly
relevant. We discuss the role of conservatism and preferences for
agreeable social contacts in older age.

G. Schomerus1,2, S. Van der

Auwera3, H. Matschinger4,5,
S. E. Baumeister3,*,
M. C. Angermeyer6,7,*
1

Department of Psychiatry, University of Greifswald,

Greifswald, 2HELIOS Hanseklinikum Stralsund,
Stralsund, 3Institute of Community Medicine, University
of Greifswald, Greifswald, 4Institute of Social Medicine,
Occupational Health and Public Health, University of
Leipzig, Leipzig, 5Institute of Health Economics and
Health Services Research, University of Hamburg,
Hamburg, Germany, 6Department of Public Health,
University of Cagliari, Cagliari, Italy and 7Center for
Public Mental Health, Gosing am Wagram, Austria
Key words: ageperiodcohort analysis; stigma; social
distance; socio-emotional selectivity hypothesis
Georg Schomerus, Department of Psychiatry, University
of Greifswald, Rostocker Chaussee 70, 17437 Stralsund,
Germany. E-mail: georg.schomerus@uni-greifswald.de
*These authors contributed equally.

Accepted for publication January 19, 2015

Signicant outcomes

Attitudes towards persons with mental illness become more negative with growing age in all birth

cohorts in Germany.
Against a backdrop of loneliness and higher suicide rates, stigma might have particularly grave consequences in old-age. Older people should thus receive the full attention of antistigma eorts that
have so far largely focused on young persons.

Limitation

Being conducted in Germany, this rst ageperiodcohort analysis of attitudes towards persons with
mental illness is likely valid only for Western industrialized countries.

Introduction

Attitudes towards persons with mental illness

change. Time trend studies among the general population over the last 20 years in Western industri-

alized countries have shown worsening attitudes

towards persons with schizophrenia, stable attitudes towards persons with depression (13), and,
in some contrast to these developments, an overall
decrease of perceived stigma in Germany (4, 5).
1

Schomerus et al.
However, attitudes may also change over the
course of life: a review of studies examining social
distance towards persons with mental illness found
age associated with negative attitudes towards persons with mental illness in most countries, older
people expressing more negative attitudes than
younger individuals (6). So far, however, this phenomenon has received little attention. Given the
cross-sectional nature of most studies, it remained
unclear whether the correlation of age and negative
attitudes reects a true age eect, or whether it represents a cohort eect based on improving tolerance towards persons with mental illness in
younger cohorts.
Trend studies using repeated cross-sectional surveys within the same population have so far
focused on time period eects on the prevalence of
stigmatizing attitudes (for example refs. 3, 7). Period eects reect general attitude changes within a
specic time frame, caused for example by economic or societal changes, but potentially also by
population wide antistigma campaigns. These
studies used age as a control variable to account
for demographic changes between surveys, but in
doing so, separating the contribution of age,
cohort and period was not possible. Disentangling
age, cohort and period eects is methodologically
challenging because of the perfect collinearity of
the three time variables (period = cohort + age).
Hence, although the respondents age is controlled
for in most and elicited in virtually all population
studies on attitudes towards mental illness, age or
cohort eects have so far not been examined specically. We do not know whether stigmatizing
attitudes towards persons with mental illness
change regularly over the course of life, nor have
we identied whether they follow a cohort pattern.
A potential role of age in stigma has not been conceptualized.
Knowing whether attitudes towards persons
with mental illness do in fact worsen with growing
age is important given the massive demographic
change towards an ageing society in western industrialized countries. Attitudes of older persons are
becoming more inuential because their share of
the general population is growing (8). If attitudes
towards persons with mental illness deteriorate
over the life span, the average level of tolerance
towards persons with mental illness in a society
will decline simply because the proportion of older
persons increases. Higher stigma in old-age could
aggravate the problem of high suicide rates among
older persons: recent studies show a link between
high stigma levels, high suicide rates and low helpseeking (911), which could be of particular relevance in older persons (12, 13). A true age eect in
2

social distance would demand a new target group

for antistigma activities: older adults.
A cohort eect, in contrast, would imply that
attitudes towards persons with mental illness are
decisively shaped in younger years. People develop
conceptions of mental illness early in life as part of
socialization into their culture (14), and the correlation of age and negative attitudes found in crosssectional studies could accordingly result from
more tolerant socialization experiences of younger
cohorts. A strong cohort eect would encourage
focusing antistigma activities on young people to
create a tolerant climate in young years that shapes
life-long tolerant attitudes and, over time, slowly
permeates society. This would be in line with present antistigma activities: a recent meta-analysis of
RCTs of antistigma interventions found 12 of 33
interventions focusing on young persons such as
tertiary students, but none focusing explicitly on
persons of higher age (15).
Aims of the study

Using data from three repeated cross-sectional surveys in Germany, we want to nd out whether stigmatizing attitudes towards persons with mental
illness increase with age, and to what extent they
follow a cohort pattern.

Material and methods

Surveys

Our study is based on data from three methodologically identical population-based cross-sectional
surveys among German citizens aged 18 years and
over conducted in 1990, 2001 and 2011, results of
which have been published previously (1, 16, 17).
In all surveys, samples were drawn using an identical random sampling procedure with three stages:
(i) sample points (electoral wards), (ii) households,
and (iii) individuals within the target households.
Target households within the sample points were
determined according to the random route procedure, that is, a street was selected randomly as a
starting point from which the interviewer followed
a set route through the area (18). Target individuals within households were selected using random
digits. Informed consent was considered to have
been given when individuals agreed to complete
the interview. Fieldwork was carried out by GETAS (Hamburg) in 1990 and by USUMA (Berlin) in
2001 and 2011. Data in 1990 were gathered in two
waves, wave 1 in the old German states (April
1990) and wave 2 in the new German states
(November 1990) shortly after reunication.

Attitudes over the life span

Interview

All interviews were conducted face-to-face using

pencil and paper. In all surveys, the fully structured interview was identical regarding wording
and the sequence of questions. At the beginning of
the interview, respondents were presented with a
vignette of a diagnostically unlabelled case history
depicting either a person with schizophrenia or
with major depression (see Appendix for the wording of both vignettes). Then, respondents were
asked a series of questions to assess, among other
things, their desire for social distance from the person described. The symptoms described in the
vignette fullled the criteria of DSM-III-R for
schizophrenia or depression and had been validated by ve experts in psychopathology blinded
to the diagnosis. While in 1990 and 2011, the sex of
the individual presented in the vignettes varied at
random, in 2001 only responses to a male vignette
were elicited. For this study, we thus only use
responses elicited with the male vignette.
Sample description

Our combined study sample from all surveys consisted of 7954 participants. 119 participants had
missing values in one of the analysed variables.
The nal study sample thus included 7835 participants, 3925 receiving the schizophrenia vignette
(1990: N = 873; 2001: N = 2438; 2011: N = 614),
3910 subjects receiving the depression vignette
(1990: N = 802; 2001: N = 2500; 2011: N = 608).
As in both samples from 1990, the population
from the old (West-) German states was underrepresented, we weighted this sample to arrive at a
proportion of 80:20 for old and new German states
for the schizophrenia and depression sample
respectively. This is representative of the overall
German population and identical to the old/newproportions in the samples from 2001 and 2011.
The stigma of mental illness has been conceptualized as a cognitive and emotional process that
results in separation, status loss and discrimination
of aected persons (14). Most studies use the desire
for social distance as a measure of discriminatory
attitudes (6). Social distance scales measure willingness to engage in various everyday activities
with a person with mental illness. In this study, we
use a scale developed by Link et al. (19), encompassing the following social situations: rent a
room, work together, have as neighbour, let take
care of a little child, have marry into family, introduce to friends, recommend for a job. With the
help of ve-point Likert scales, respondents could
indicate to what extent they were willing or unwill-

ing to engage in the proposed relationships. For

our study, we used the sum-score of the scale
(range 028). The scale has excellent reliability and
validity and has been used in numerous studies as
a proxy for behavioural discrimination of persons
with mental illness (6, 20). Cronbachs alpha in our
sample was 0.88.
Statistical analysis

For each of the two disorders, we performed an

ageperiodcohort (APC) analysis to test and
separate associations of age at examination, birthcohort and time period with our dependent variable social distance. Because the three variables are
perfectly collinear (time period - age at examination = birth-cohort), ordinary least-square regression models are not a reasonable option. This
problem is referred to as the identication problem
because only two of the three variables can be put
into a regression model simultaneously. Instead,
we used partial least squares regressions (PLS),
which allows using perfect collinear variables in
one regression model. In PLS, components are
extracted that are weighted combinations of the
original variables. These components are chosen
according to their relationship to the outcome variable. They seek to maximize the covariance
between the outcome and those weighted components. As a result, the rst component has a greater
covariance with the outcome than the second and
so on. The regression coecients for the three variables can directly be obtained from their weights in
each component. For more detailed information
on the identication problem in APC analyses and
the application of PLS see (21, 22). Another way
to deal with linear dependencies in APC models
has been proposed by Yang and Land (23). They
apply a mixed model approach where they separately calculate the eect for each variable of
cohort, period and age. In contrast to this model,
our approach allows to put the covariables in a
joint model to get the separate eects for age,
cohort and period simultaneously. As PLS penalizes variables according to their variance when
extracting the components, covariates were scaled
to have unit variance and zero mean.
First, we estimated linear PLS with age at examination and birth-cohort as continuous covariates
and created dummy variables for time period. To
additionally examine nonlinear relationships for
age and birth-cohort, we used restricted cubic
spline PLS with ve knots for age at examination
(schizophrenia: 21, 34, 46, 60, 76 years; depression:
21, 35, 47, 60, 76 years) and birth-cohort (schizophrenia: 1923, 1940, 1954, 1965, 1981; depression:
3

Schomerus et al.
Table 1. Number of subjects of each age group at each period
Age

1990

Schizophrenia vignette (N = 3925)

<30
205
3039
170
4049
137
5059
141
6069
130
70
90
Depression vignette (N = 3910)
<30
178
3040
147
4049
134
5059
144
6069
123
70
76

2001

2011

402
517
489
340
401
289

76
93
114
104
120
107

439
523
441
356
443
298

79
97
116
112
93
111

Schizophrenia

1924, 1940, 1954, 1965, 1981), which represent the

0.05, 0.275, 0.5, 0.725 and 0.95& respectively. To
assess the model t of linear and nonlinear PLS,
we compared the R2, which is a measure of the variance of the outcome variable that can be
explained by the covariates. Wald tests of coecients of the second, third and fourth spline transformation equal to zero were used to obtain P
values for nonlinearity. A J-test was used to test
whether polynomials for age, birth or period in the
nonlinear model with splines were jointly signicant (24). Signicance was assumed at P < 0.05 in
all analyses. All analyses were performed using the
statistical software R (version 3.0.2, http://www.rproject.org/). Apart from age, cohort and period,
no other independent variables were included in
regression models.
Results

In the combined sample, age at examination ranged from 18 to 93, birth-cohort from 1900 to 1993
and social distance from 0 to 28. Table 1 shows
the number of individuals in each age group for
1990, 2001 and 2011.
In both linear and nonlinear PLS, a two components model showed better model t than a
one-component model for schizophrenia and
depression. We thus report results for the two components models only. The results for the linear PLS
regressions are shown in Table 2. The model with
two PLS components explained 2.46% of the variance (R2) in social distance for schizophrenia and
2.28% for depression (one-component model:
schizophrenia R2 = 1.65%; depression R2 = 2.10%).
Age effects

For both disorders, the linear model showed a signicant age eect (Table 2): Independent from
4

Table 2. Social distance towards a person with schizophrenia or depression: age,

period and cohort effects. Linear partial least squares analysis with two components. Unstandardized coefficients

Coefficient
Age (years)
Cohort (years)
Period (1990 ref.)
2001
2011
R2

95% CI

Depression
Coefficient

95% CI

0.032
0.000

0.025, 0.040
0.006, 0.007

0.031
0.022

0.025, 0.038
0.028, 0.017

1.528
0.995
2.46%

1.140, 1.868
0.580, 1.380

0.435
0.612
2.28%

0.005, 0.834
0.195, 0.971

CI after 1000 nonparametric Bootstrap replicates.

cohort and period eects, we observed a score

increase of 0.032 per year for schizophrenia and
0.031 per year for depression. This cumulates to a
dierence in social distance over the life span
(youngest to oldest participant) of 2.4 for schizophrenia and 2.3 for depression. In the nonlinear
models, Wald tests rejected a nonlinear relation
between social distance and age (schizophrenia,
P = 0.58; depression, P = 0.11), indicating that
there is a linear increase of social distance with
age.
Cohort effects

Linear models showed a cohort eect in depression, but not in schizophrenia (Table 2). Persons
born in later years exhibited lower social distance
towards a person suering from depression
( 0.021 per year), irrespective of age and period
eects. Comparing the earliest and most recent
birth-cohorts (1900 and 1993), this eect cumulated to a dierence of 2.0.
In our nonlinear models, a Wald test conrmed
a nonlinear relationship between birth-cohort and

Fig. 1. The relationship between adjusted social distance and

birth-cohort using PLS with restricted cubic splines. Adjusted
predicted values, keeping age and period eects constant at
their mean.

Attitudes over the life span

social distance in depression (P = 0.002), but not
with schizophrenia (P = 0.38). Figure 1 shows the
predicted social distance from a person with
depression plotted against birth-cohort, keeping
the other two time variables constant at their
mean, calculated by a PLS regression model using
linear terms for age and period and restricted cubic
splines for birth-cohort. The R2 of this nonlinear
model was 2.26%, which was slightly lower than
for the linear model for depression. This nonlinear
cohort eect shows a relatively stable decrease by
1.92 points in social distance from birth-cohort
1925 (16.02, 95% CI: 15.83, 16.20) to birth-cohort
1970 (14.10, 95% CI: 13.94, 14.25), followed by an
increase of 0.57 points to 14.67 in birth-cohort
1990 (95% CI: 14.15, 15.18) (Fig. 1).
Period effects

Because only three points of measurements exist

for time period, we calculated period eects for
these distinct time points. A signicant period
eect could be observed particularly in schizophrenia. People from the more recent surveys had
higher social distance scores than those from an
earlier survey with a score increase of 1.5 between
1990 and 2001 and 1.0 between 1990 and 2011
respectively. A possible decrease in social distance
between 2001 and 2011 was not statistically signicant when 2001 served as reference category. In
depression, only the dierence between 1990 and
2011 was signicant, yielding an increase of social
distance of 0.6.
Discussion

Summarizing our ndings, our study shows a consistent age eect in depression and schizophrenia.
Over the life span, there is a considerable increase
of social distance towards persons with mental illness, irrespective of time period or birth-cohort.
Cohort eects, in contrast, dier between disorders. While attitudes towards a person with schizophrenia were similar across birth-cohorts, attitudes
towards a person with depression became more
tolerant in younger cohorts until the 1970 cohort,
but then deteriorated again. Our ndings also conrm the previously observed period eect of rising
social distance particularly towards persons with
schizophrenia, this increase occurring predominantly between 1990 and 2001 (1).
Before discussing the implications of our ndings, the limitations of our study have to be considered. First, focusing on time variables and social
distance, our study did not account for other predictors of social distance such as certain illness

beliefs or contact (16, 2527), which is reected in

the overall low explained variance in our models.
Introducing previous contact to a person with
mental illness as additional independent variable
was not possible, because this was not consistently
elicited in the 1990 surveys. Generally, we
refrained from adding sociodemographic control
variables to our analyses, because the necessary
introduction of interaction eects between these
variables and the three time variables would have
made our analyses very complex. Absent of any
specic hypothesis on the interaction of sociodemographic variables with age, period and cohort
eects this seemed not justied.
Second, to arrive at three samples of sucient
size, we used data from respondents from both
East and West Germany. Previous studies have
shown both dierences and similarities in illness
beliefs in both regions (28), and time trends have
thus been examined separately for both parts of
the country (4, 7). An explorative analysis introducing interaction terms between region (East vs.
West) and the three time variables, however, did
not reveal signicant interaction eects. Therefore,
we pooled data from East and West, assuming that
our model is equally valid for East and West Germany. However, we do not know whether attitudes
in other countries follow similar age, cohort and
period patterns, our results thus apply only to Germany. Although a recent meta-analysis of time
trends in Western industrialized countries found
that overall, attitudes towards persons with mental
illness seem to evolve in a similar manner; comparative data on age and cohort eects are missing
(2). Finally, as we could only use data on male
vignettes across all three surveys, we do not know
how the gender of the person described did inuence our results. Future research is needed to
determine whether the observed increase in social
distance in elderly people is driven by stereotypes
such as perceived dangerousness, which could be
particularly relevant when a male person is
described.
We saw a cohort eect in attitudes towards
depression that is in line with observations of generally more liberal attitudes among persons born
1950 and later (29). Nevertheless, the slight
increase of social distance in younger cohorts born
after 1970 is troubling, indicating that recent educational eorts aiming to create more tolerant
mental health-related attitudes particularly among
young persons have not yet succeeded at the population level. Notably, attitudes towards a person
with schizophrenia did not follow a cohort pattern.
Regarding schizophrenia, socialization experiences
in young years seem to be less relevant for the
5

Schomerus et al.
formation of attitudes than time and age eects. A
reason for this dierence could be that symptoms
of depression are closer to everyday experiences,
making their perception more amenable to dierences in cultural framing. Symptoms of schizophrenia, in contrast, are outside most peoples
everyday experiences and are perceived equally
strange and frightening regardless of subtle cultural changes across dierent birth-cohorts. Our
ndings demonstrate that current eorts to
improve attitudes of young persons need to intensify and improve. A more tolerant cultural climate
with regard to mental illness among younger
cohorts does not seem easily accomplished in both
depression and schizophrenia.
Why do attitudes towards persons with mental
illness worsen over the life span? Two general models describe changes of attitudes during the course
of life, the impressionable years model and the
life-long openness model (30). The impressionable years model postulates that attitudes change
particularly in young years and then remain relatively stable (31). This model would explain cohort
eects in attitudes towards mental illness and is
consistent at least with our ndings regarding
depression. However, the impressionable years
model clearly does not explain the linear age eect
found in both disorders. The second model, the
lifelong openness model (30) postulates that people are open to attitude change lifelong, depending
on their personal experiences (32). Generally, this
is more consistent with our data, but it still
remains unclear why attitudes change to the worse.
Previous research has shown that experience of
mental illness and contact to people with mental
illness are generally associated with more tolerant
attitudes and not with a stronger desire for social
distance (16); hence, accumulating personal experience with mental illness over the life span should
not increase social distance. However, as mentioned among the limitations of this study, we were
unable to control our models for personal contact
with people with mental illness.
Two dierent perspectives on attitudes and
social behaviour over the life span may explain our
results. First, studies of political attitudes have
consistently found growing conservatism with
growing age. A cross-sectional study of 30 000 persons in the U.S. found a sharp increase in conservative attitudes particularly between age 40 and 60
(33). Growing social distance towards a person
with symptoms of mental illness who is behaving
in a non-conforming manner might thus reect
more conservative attitudes.
The second observation explaining our ndings
is more closely linked to the actual content of the
6

social distance scale, the readiness to engage in

everyday situations of social contact. Research on
choices of social partners in older age shows that
older people develop a preference for familiar
social partners that do not trigger negative aect
(34). This phenomenon has been conceptualized in
the socio-emotional selectivity hypothesis of
social contacts in older age (35). According to this
theory, reduced social contacts in older people are
not the result of general disengagement, but of
active selection of those social partners that trigger
positive emotions (34). Quite conceivably the
growing desire for social distance towards persons
with severe mental illness reects this active selection process of positive social contacts. For those
who experience mental illness, however, a tendency
to avoid contact with people who confer negative
feelings is likely aggravating the social isolation
that frequently accompanies mental illness (3638).
Two points emerge as desiderata of future
research in the stigma of mental illness among
elderly people. First, the social distance scale used
in this study encompasses situations that may be of
dierent importance in dierent stages of life:
marrying into ones family might imply marriages
of ones own children in older age and become thus
more relevant, while letting someone look after
ones children may become less relevant. Although
the scale showed strong internal consistency (0.88)
in the entire sample, future research should look
more closely on age-specic indicators of social
distance, which might also be of greater relevance
for stigma experiences of older persons with mental illness.
Second, although not mentioning a specic age,
the stimulus used in our studies implicitly referred
to a person of working age, because among other,
age-neutral symptoms, problems at work were
mentioned in both the schizophrenia and the
depression vignette (see Appendix for the full
wording of the two vignettes). Hence, older persons might have felt social distance towards this
person simply because of an age dierence. Inconsistent with this hypothesis, however, we observed
a linear trend of increasing social distance across
the entire life span and did not nd a pronounced
increase specically around retirement age. The
use of age-specic vignettes in future research
would advance our understanding of stigma in
older age.
Stigma in old-age is thus likely amplifying the
burden of mental illness in a population group that
is particularly vulnerable to social isolation and
may contribute to severe outcomes of mental illness such as suicides (39). Our study indicates that
the relation between age and stigma means more

Attitudes over the life span

than just a confounding variable in data analysis.
More research is needed to further understand the
reasons for the increase of social distance towards
persons with mental illness over the life span and
to identify promising, age-specic strategies to
improve attitudes not only in young persons.
Given the growing proportion of older people in
western societies, targeting the stigma of mental illness in persons of higher age seems urgent.
Acknowledgement
This study was funded by the Fritz-Thyssen-Stiftung
(Az. 10.11.2.175).

Declaration of interest
GS reports having received speakers honoraria from Lundbeck. MCA reports Lecturer fees from AstraZeneca, JanssenCilag, Eli Lilly, and Pzer. Research grants from GlaxoSmithKline and Lundbeck. All other authors declare no potential conict of interest.

References
1. Angermeyer MC, Matschinger H, Schomerus G. Attitudes
towards psychiatric treatment and people with mental illness: changes over two decades. Br J Psychiatry
2013;203:146151.
2. Schomerus G, Schwahn C, Holzinger A et al. Evolution of
public attitudes about mental illness: a systematic review
and meta-analysis. Acta Psychiatr Scand 2012;125:440
452.
3. Pescosolido BA, Martin JK, Long JS, Medina TR, Phelan
JC, Link BG. A disease like any other? A decade of
change in public reactions to schizophrenia, depression,
and alcohol dependence. Am J Psychiatry 2010;167:1321
1330.
4. Angermeyer MC, Matschinger H, Carta MG, Schomerus
G. Changes in the perception of mental illness stigma in
Germany over the last two decades. Eur Psychiatry
2014;29:390395.
5. Schomerus G, Matschinger H, Lucht MJ, Angermeyer MC.
Changes in the perception of alcohol-related stigma in
Germany over the last two decades. Drug Alcohol Depend
2014;143:225231.
6. Jorm AF, Oh E. Desire for social distance from people with
mental disorders. Aust N Z J Psychiatry 2009;43:183200.
7. Angermeyer MC, Holzinger A, Matschinger H. Mental
health literacy and attitude towards people with mental illness: a trend analysis based on population surveys in the
eastern part of Germany. Eur Psychiatry 2009;24:225232.
8. Tinker A. The social implications of an ageing population.
Mech Ageing Dev 2002;123:729735.
9. Reynders A, Kerkhof AJ, Molenberghs G, van Audenhove
C. Attitudes and stigma in relation to help-seeking intentions for psychological problems in low and high suicide
rate regions. Soc Psychiatry Psychiatr Epidemiol
2014;49:231239.
10. Schomerus G, Evans-Lacko S, Rusch N, Mojtabai R, Angermeyer M, Thornicroft G. Collective levels of stigma and
national suicide rates in 25 European countries. Epidemiol
Psychiatr Sci 2014. DOI: http://dx.doi.org/10.1017/
S2045796014000109. [Epub ahead of print].

11. R
usch N, Zlati A, Black G, Thornicroft G. Does the
stigma of mental illness contribute to suicidality? Br J Psychiatry 2014;205:257259.
12. Shahnawaz Z, Reppermund S, Brodaty H et al. Prevalence
and characteristics of depression in mild cognitive impairment: the Sydney Memory and Ageing Study. Acta Psychiatr Scand 2013;127:394402.
13. Nivoli AMA, Murru A, Pacchiarotti I et al. Bipolar disorder in the elderly: a cohort study comparing older
and younger patients. Acta Psychiatr Scand 2014;130:
364373.
14. Link BG, Phelan JC. Conceptualizing stigma. Annu Rev
Sociol 2001;27:363385.
15. Griffiths KM, Carron-Arthur B, Parsons A, Reid R. Eectiveness of programs for reducing the stigma associated
with mental disorders. A meta-analysis of randomized
controlled trials. World Psychiatry 2014;13:161175.
16. Angermeyer MC, Matschinger H. Social distance towards
the mentally ill: results of representative surveys in the
Federal Republic of Germany. Psychol Med 1997;27:131
141.
17. Angermeyer MC, Beck M, Matschinger H. Determinants of the publics preference for social distance from
people with schizophrenia. Can J Psychiatry 2003;48:663
668.
18. Gabler S, Hoffmeyer-Zlotnik JHP. Stichproben in der umfragepraxis. [Samples in survey practice.]. Opladen: Westdeutscher Verlag, 1997.
19. Link BG, Cullen FT, Frank J, Wozniak JF. The social
rejection of former mental patients understanding why
labels matter. Am J Sociol 1987;92:14611500.
20. Corrigan PW, Edwards AB, Green A, Diwan SL, Penn DL.
Prejudice, social distance, and familiarity with mental illness. Schizophr Bull 2001;27:219225.
21. Jiang T, Gilthorpe MS, Shiely F et al. Age-period-cohort
analysis for trends in body mass index in Ireland. BMC
Public Health 2013;13:889.
22. Tu Y-K, Kramer N, Lee W-C. Addressing the identication problem in age-period-cohort analysis: a tutorial on
the use of partial least squares and principal components
analysis. Epidemiology 2012;23:583593.
23. Yang Y, Land KC. A mixed models approach to the ageperiod-cohort analysis of repeated cross-section surveys,
with an application to data on trends in verbal test scores.
Sociol Methodol 2006;36:7597.
24. Mackinnon JG, White H, Davidson R. Tests for model specication in the presence of alternative hypotheses: Some
further results. Journal of Econometrics. 1983;21:5370.
25. Schomerus G, Matschinger H, Angermeyer MC. Continuum beliefs and stigmatizing attitudes towards persons
with schizophrenia, depression and alcohol dependence.
Psychiatry Res 2013;209:665669.
26. Schomerus G, Matschinger H, Angermeyer MC. Causal
beliefs of the public and social acceptance of persons with
mental illness: a comparative analysis of schizophrenia,
depression and alcohol dependence. Psychol Med
2014;44:303314.
27. Kvaale EP, Gottdiener WH, Haslam N. Biogenetic explanations and stigma: a meta-analytic review of associations
among laypeople. Soc Sci Med 2013;96:95103.
28. Angermeyer MC, Matschinger H. Lay beliefs about mental
disorders: a comparison between the western and the eastern parts of Germany. Soc Psychiatry Psychiatr Epidemiol
1999;34:275281.
29. Harding DJ, Jencks C. Changing attitudes toward premarital sex: cohort, period, and aging eects. Public Opin Q
2003;67:211226.

Schomerus et al.
30. Tyler TR, Schuller RA. Aging and attitude change. J Pers
Soc Psychol 1991;61:689697.
31. Sears DO. Life-stage eects on attitude change, especially
among the elderly. In: Kiesler SB, Morgan JN, Oppenheimer
VK, eds. Aging: social change. New York: Academic
Press, 1981:183204.
32. Visser PS, Krosnick JA. Development of attitude strength
over the life cycle: surge and decline. J Pers Soc Psychol
1998;75:13891410.
33. Truett KR. Age dierences in conservatism. Pers Individ
Dif 1993;14:405411.
34. Fredrickson BL, Carstensen LL. Choosing social partners:
how old age and anticipated endings make people more
selective. Psychol Aging 1990;5:335.
35. Carstensen LL, Isaacowitz DM, Charles ST. Taking time
seriously: a theory of socioemotional selectivity. Am Psychol 1999;54:165181.
36. Link B, Struening E, Neese-Todd S, Asmussen S, Phelan J.
On describing and seeking to change the experience of
stigma. Am J Psychiatr Rehabil 2002;6:201231.
37. Busch M, Neuner B, Aichberger C et al. Depressive Symptomatik und Inanspruchnahme von Gesundheits- und
Pfelgeleistungen bei Personen im Alter ab 50 Jahren in
Deutschland. Psychiatr Prax 2013;40:214219.
38. Link BG, Struening EL, Rahav M, Phelan JC, Nuttbrock
L. On stigma and its consequences: evidence from a longitudinal study of men with dual diagnoses of mental illness
and substance abuse. J Health Soc Behav 1997;38:177
190.
39. Riedel-Heller SG, Luppa M. Depression im Alter was
tragt die aktuelle epidemiologische Forschung bei? Psychiatr Prax 2013;40:173175.

Appendix
Vignette schizophrenia

Imagine that you hear the following about an

acquaintance with whom you occasionally spend
your leisure time:
Within the past months, your acquaintance
appears to have changed. More and more, he
retreated from his friends and colleagues, up to the
point of avoiding them. If someone managed to
involve him in a conversation, he would address
only one single topic: the question whether some
people had the natural gift of reading other peoples thoughts. This question became his sole concern. In contrast with his previous habits, he
stopped taking care of his appearance and looked
increasingly untidy. At work, he seemed absentminded and frequently made mistakes. As a conse-

quence, he has already been summoned to his

boss.
Finally, your acquaintance stayed away from
work for an entire week without an excuse. Upon
his return, he seemed anxious and harassed. He
reports that he is now absolutely certain that people cannot only read other peoples thoughts, but
that they also directly inuence them. He was,
however, unsure who would steer his thoughts. He
also said that, when thinking, he was continually
interrupted. Frequently, he would even hear those
people talk to him, and they would give him
instructions. Sometimes, they would also talk to
each other and make fun of whatever he was doing
at the time. The situation was particularly bad at
his apartment, he claimed. At home, he would
really feel threatened and would be terribly scared.
Hence, he had not spent the night at his place for
the past week, but rather he had hidden in hotel
rooms and hardly dared to go out.
Vignette major depressive disorder

Imagine that you hear the following about an

acquaintance with whom you occasionally spend
your leisure time:
Within the past 2 months, your acquaintance
has changed in his nature. As opposed to previously, he is down and sad without being able to
make out a tangible reason for his feeling low. He
appears serious and worried. There is nothing that
will make him laugh anymore. He hardly ever
talks, and if he is saying something, he is speaking
in a low voice about the worries he has with regard
to his future. Your acquaintance feels useless and
has the impression to do everything wrong. All
attempts to cheer him up have failed. He lost all
interest in things and is not motivated to do anything. He complains of often waking up in the middle of the night and not being able to get back to
sleep. He feels exhausted and without energy in the
morning already. He says that he encounters diculty in concentrating on his job. In contrast with
previous times, everything takes him very long. He
hardly manages his workload. As a consequence,
he has already been summoned to his boss.