Drug and Alcohol Dependence 143 (2014) 225–231
Contents lists available at ScienceDirect
Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep
Changes in the perception of alcohol-related stigma in Germany over the last two decades Georg Schomerus a,b,∗ , Herbert Matschinger c,d , Michael J. Lucht a,b , Matthias C. Angermeyer e,f a
Department of Psychiatry, Ernst Moritz Arndt University, Greifswald, Germany HELIOS Hanseklinikum Stralsund, Germany c Institute of Social Medicine, Occupational Health and Public Health, University of Leipzig, Germany d Institute of Medical Sociology, Social Medicine and Health Economics, University of Hamburg, Germany e Center for Public Mental Health, Gösing am Wagram, Austria f Department of Public Health and Clinical and Molecular Medicine, University of Cagliari, Italy b
a r t i c l e
i n f o
Article history: Received 13 March 2014 Received in revised form 26 July 2014 Accepted 27 July 2014 Available online 8 August 2014 Keywords: Stigma Substance use disorders Time-trend study Perceived Discrimination and Devaluation Scale
a b s t r a c t Background: Alcohol dependence is a severely stigmatized disorder. Perceived stigma may deter helpseeking and is associated with higher co-morbidity and self-stigma in persons with alcohol dependence. We assess changes in the perception of alcohol-related stigma over 21 years in the general population. Methods: Two representative population surveys using identical methodology were conducted in Germany in 1990 and 2011 (n = 1022 and n = 967), eliciting the perceived discrimination and devaluation of someone with a history of alcohol problems as measured with an adoption of Link’s Perceived Discrimination and Devaluation Scale (aPDDS), and perceived negative stereotypes of an “alcoholic.” Results: Both on item level and using factor scores, attitudes changed significantly between 1990 and 2011. Perceived discrimination and devaluation of someone with a history of alcohol dependence decreased considerably by 0.44 standard deviations (SD). Perceived negative stereotypes related to unpredictability of an “alcoholic” increased slightly by 0.15 SD, while perceived stereotypes related to strangeness decreased (−0.23 SD). Conclusions: Our findings suggest that particularly the image of someone who has received treatment for alcohol dependence has improved in Germany. This parallels increasing acceptance of professional treatment for alcohol dependence among the general population over the last twenty years, and contrasts with overall unchanged negative attitudes toward persons who actually suffer from alcohol problems. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Studies on the development of public attitudes toward people with substance use disorders indicate that they have not changed for the better over the last decades. Attitudes toward persons with alcohol dependence, for example, have remained disturbingly stable: time trend studies from the U.S. and Germany found no meaningful changes in the desire for social distance between 1996 and 2006 (U.S.) or 1990 and 2011 (Germany) (Angermeyer et al., 2013b; Pescosolido et al., 2010). Similarly, personally held stereotypical beliefs about alcohol dependence did hardly change
∗ Corresponding author at: Department of Psychiatry, University Medicine Greifswald, Helios Hanseklinikum Stralsund, Rostocker Chaussee 70 18437 Stralsund, Germany.Tel.: +49 3831 452109; fax: +49 3831 452105. E-mail address: [email protected] (G. Schomerus). http://dx.doi.org/10.1016/j.drugalcdep.2014.07.033 0376-8716/© 2014 Elsevier Ireland Ltd. All rights reserved.
between 1990 and 2011 in Germany (Schomerus et al., in press), or in Great Britain between 1998 and 2003 (Crisp et al., 2005). Some negative changes were observed in the U.S.: here, the proportion of respondents assuming that bad character was a cause for alcohol dependence did increase by 16% between 1996 and 2006. These few time-trend studies of alcohol-related public attitudes have all examined public stigma, i.e., the personal attitudes of the respondents. However, there is another component of stigma that is of high relevance to those suffering from alcohol use disorders: the perception of public stigma. People not only hold personal beliefs about alcohol dependent persons, they also have an idea of what, in their opinion, other people believe. Perceived public stigma and personally held attitudes of the general public are not identical: a study in Australia examining perceived and personal stigma related to depression found very low correlation of the according scales (R = 0.10; Griffiths et al., 2004). So far, there are no
226
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
time-trend studies of the perceived stigma attached to alcohol use disorders. Regarding the perceived stigma of mental disorders, a study from Australia found only small differences in perceived stigma between 2004 and 2011, indicating increasing perceptions of the dangerousness-stereotype related to persons with chronic schizophrenia. In contrast, a recent study from Germany found perceived discrimination and devaluation of “former mental patients” to have diminished considerably since 1990 (Angermeyer et al., 2013a). Here, perceptions of stigma had developed favorably, while personal attitudes particularly toward persons with schizophrenia had deteriorated within the same time period (Angermeyer et al., 2013b). The perceived stigma attached to alcohol use disorders seems to deter help-seeking; in a large, representative sample of adults with alcohol use disorders in the U.S., perception of higher stigma toward persons with alcohol problems was associated with lower lifetime alcohol service use (Keyes et al., 2010). Perceptions of stigma have also been conceptualized as starting point for the formation of personal attitudes, including self-stigma. In his progressive model of self-stigma, Corrigan posits awareness of other persons’ stereotypical beliefs as the beginning of a cascade of stigmatizing cognitions, resulting in agreeing with these stereotypes (personal stigma), and eventually applying these stereotypes to oneself should one develop the stigmatized condition (self-stigma; Corrigan et al., 2011, 2006). This model has also been applied to persons with alcohol dependence, showing that self-stigma is associated with decreased drinking-refusal self-efficacy (Schomerus et al., 2011a). Finally, modified labeling theory highlights that expectations of rejection and discrimination shape the individuals’ experience of a stigmatized disorder (Link et al., 1989). In fact, a large cross-sectional study in the U.S. found high perceived stigma of persons labeled with alcohol use disorders to be associated with negative psychiatric outcomes (Glass et al., 2013b). High perceptions of stigma related to alcohol dependence could thus decrease help-seeking, increase self-stigma and impair outcomes of persons suffering from alcohol use disorders. In this study, we explore how the perception of alcohol-related stigma has developed in Germany over the last two decades. To our knowledge, this is the first time-trend study of perceived alcoholrelated stigma in the general population. We use data from two methodologically identical surveys in 1990 and 2011 to explore how two facets of perceived stigma have developed over time:
(A) perceptions of the salience of negative stereotypes related to alcohol dependence and (B) perceptions of discrimination and devaluation of persons with alcohol dependence. 2. Methods 2.1. Surveys and interview Two population surveys were conducted among German citizens aged 18 years and over, the first in 1990 (n = 3087, response rate 70.0%), the second in 2011 (n = 3642, response rate 64.0%). In both surveys samples were drawn using a random sampling procedure with three stages: (1) sample points, (2) households, and (3) individuals within target households. Target households within sample points were determined according to the random route procedure. Target persons within each household were selected using random digits. Informed consent was considered to have been given when individuals agreed to complete the interview. The fieldwork for the first survey was carried out by GETAS, Hamburg, for the second survey by USUMA, Berlin; both institutes specialize in marketing and social research. The surveys were approved by the Ethics committees of the Universities of Heidelberg (1990) and Greifswald (2011). Before the first survey, the interview had been pretested with 20 persons to ensure maximum understandability of the questions. Results of the survey in 1990 have already been published separately (Angermeyer et al., 1995). In both surveys, the same interview mode (face-to-face, paper–pencil) was used. On both occasions, the interview used identical wording and sequence of questions. In the first part, which is not subject of the present paper, we asked questions related to a case–vignette of a person with different mental disorders. The second part covered issues unrelated to the case–vignette. In both surveys, respondents who had answered questions related to a vignette depicting a person with alcohol dependence received further, vignette-unrelated questions on their perception of the stigma surrounding alcohol dependence, which are the subject of this paper (1990: n = 1022; 2011: n = 1187). Since the survey in 1990 had been conducted before reunification in the old Federal Republic of Germany, we restricted our analyses of the 2011 data to respondents living in the “old” states of Germany (n = 967). Sociodemographic characteristics of the study samples and of the German population of each year are reported in Table 1. 2.2. Measures We assessed two aspects of perceived stigma. First, we asked respondents about the perceived salience of negative stereotypes associated with “alcoholics” among the general public (Angermeyer et al., 1995). Respondents were asked to rate the extent to which they believed the general public would endorse negative attributes of alcohol dependent patients: being short-tempered, unpredictable, aggressive, dependent on others, strange, stupid, scary, untruthful, and dangerous. The rating of each stereotype had to be given on five-point Likert scales with the anchors 1 = ”is certainly true” and 5 = ”is certainly not true”. We reversed scores for our analyses, so that higher scores indicate higher perceptions of stereotypes. Second, we used an adapted version of Link’s Perceived Discrimination and Devaluation Scale (adopted Perceived Discrimination and Devaluation Scale, aPDDS;
Table 1 Socio-demographic characteristics of study samples and sum-scores of perceived stigma measures. 1990
2011
Survey (n = 1022) (%)
Total populationa (%)
Survey (n = 967) (%)
Total populationa (%)
Gender Male Female
47.2 52.8
48.5 51.5
44.7 55.3
48.6 51.4
Age (years) 18–25 26–45 46–60 61+
14.9 36.8 25.9 22.4
12.3 38.0 24.2 25.5
9.0 34.7 24.6 28.7
11.3 31.9 26.9 29.9
Educational attainment Still student No schooling completed 8/9 years of schooling 10 years of schooling 12/13 years of schooling
1.6 2.3 52.1 29.2 14.8
0.4 2.5 55.8 25.8 15.5
0.6 2.0 44.5 37.6 15.3
1.0 4.0 38.5 29.3 27.1
Mean perceived stereotypes sum-scoreb (SD)
30.4
(6.5)
30.2
(6.5)
41.3
(9.4)
37.2
(8.1)
c
Mean aPDDS sum-score (SD) a b c
Data from the Federal Statistical Office of Germany. Range: 9–45. Adopted Perceived Discrimination and Devaluation Scale, range: 12–60; items 5–7, 9, 11–12 reversed.
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231 Link et al., 1989). The original instrument asks about the extent of agreement with statements indicating discrimination and devaluation of former psychiatric patients. It was designed to capture attitudes toward a labeled individual without referring to present symptoms, using descriptions like “a former mental patient,” “a person that has been in a mental hospital,” or “a fully recovered former mental patient” (Link et al., 1989). We adopted the authorized German version (Matschinger et al., 1991) by asking about someone who “has been treated for alcoholism,” “has been in a hospital for alcohol detoxification,” or “a fully recovered former alcoholic.” Items were otherwise similar to the original version, enquiring for example whether most people perceive someone who had been in hospital for alcohol detoxification as failure or as less intelligent than other persons, and to what extent the respondent believes that most people distance themselves from persons treated for alcoholism in various social relationships, like having a patient as a close friend or letting him teach young children in school. The exact wording of the items is given in Table 3. The scale is balanced such that a high level of perceived stigma is indicated by agreement with six of the items and by disagreement with six others. Answers to the items are recorded using 5-point Likert scales with the anchors “totally agree” (1) and “totally disagree” (5). The original scale and its German version have been used in population surveys (e.g., Link et al., 1989; Schomerus et al., 2006) as well as in studies involving patients suffering from mental illness (e.g., Freidl et al., 2012; Lundberg et al., 2007). The first adoption of the PDDS to alcohol dependence was employed in 1990 in Germany in the first of our two surveys (Angermeyer et al., 1995). Meanwhile, several adoptions of the PDDS for alcohol or substance use disorders have been used, and have shown good psychometric properties (Glass et al., 2013a). Because most items of the aPDDS refer to a person that has undergone treatment for alcohol problems, we included items on treatment optimism from the 2011 survey into our analyses. In this survey, we asked respondents to rate their agreement with the following statements regarding the person with alcohol problems described in the vignette: (1) “After undergoing treatment, this person can lead a normal life again,” and (2) “with treatment, this person’s condition is going to improve markedly.” Answers were given on Likert scales with the anchors 1 = certainly true to 5 = certainly not true. We reversed scores, higher scores indicating greater treatment optimism. Additionally we assessed socio-demographic characteristics (gender, age, and educational attainment). 2.3. Statistical analysis We first performed analyses on item level examining to what extent answers to single items had changed between 1990 and 2011. For this reason, we calculated multinomial logit regression models for all items using the STATA modules “mlogit” and “prvalue” (Long and Freese, 2006; Xu and Long, 2005; StataCorp, 2011). To reduce the number of categories, we grouped the five-point Likert scales into three categories as indicated in Tables 2 and 3. To adjust changes between years for demographic differences across samples, we controlled for respondents’ gender, age,
227
and educational attainment. The regression models yield answer probabilities for each category and each year, with control variables held at their means for the combined sample. Using the delta method, we determined the significance of changes by computing ninety-five percent confidence intervals for the differences between both years. To make adjusted predictions comparable to unadjusted percentages, answer probabilities are multiplied by 100 and can thus be read as estimated percentages of respondents endorsing each category. In a second step, we looked for possible latent variables within our scales to determine the effect of time on the expression of these latent variables. With both scales, we first performed an exploratory factor analysis, followed by a number of linear regression analyses using factor scores as dependent variables in order to examine the change of perceived stigma over time. This procedure is inline with the so-called “two-step approach” (Anderson and Gerbing, 1988, 1992; Hayduk, 1996), because we do not estimate the structure and the effect of time and other exogenous variables on the latent variables simultaneously, but rather perform measurement and structural (explanatory) analyses sequentially. For our exploratory factor analysis (EFA), we used the “factor” command and the “pcf” option of STATA, release 12.1 (StataCorp, 2011), which performs a principal component factor analysis.
3. Results Table 1 (bottom) shows the sum-scores of the two perceived stigma measures in both surveys, suggesting that overall, the perception of public negative stereotypes did hardly change, while perceptions of discrimination/devaluation seem to have diminished. Tables 2 and 3 show the predicted answer probabilities and predicted changes between 1990 and 2011 on item level as calculated by multinomial logit models controlling for age, gender and educational attainment. Values changing significantly between years are printed in bold. These answer probabilities can be read as percentages of the population answering in the defined manner. With regard to the perceived salience of negative stereotypes, a majority believed that the general population endorsed the stereotypes “short-tempered,” “unpredictable,” “aggressive,” and “dependent on others” in both years (Table 2). The stereotype found to be least salient was “stupid”, which was rated as salient by one in five respondents. While the perceived salience of the first three stereotypes increased, it decreased for “stupid,” “scary” and “untruthful”
Table 2 The German public’s perception of the salience of negative public stereotypes associated with “alcoholics”. Multinomial logit regression controlling for age, sex and education, predicted percentages. Representative population surveys 1990 and 2011. Stereotype Short-tempered
Unpredictable
Aggressive
Dependent on others
Strange
Stupid
Scary
Untruthful
Dangerous
True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true
1990 (n = 1022) (%)
2011 (n = 967) (%)
Predicted change
95% CI for change
65 20 15 71 18 11 59 26 15 67 22 11 44 26 30 23 37 40 39 29 32 46 30 23 36 35 30
72 21 7 72 21 7 68 25 7 51 31 17 42 34 24 18 24 58 29 34 37 37 32 32 35 40 24
7 1 −8 1 3 −4 10 −2 −8 −16 10 6 −2 8 −6 −5 −13 18 −10 5 5 −10 2 8 0 5 −5
(3, 12) (−3, 4) (−11, −5) (−3, 5) (0, 7) (−7, −2) (5, 14) (−6, 2) (−11, −5) (−20, −11) (6, 14) (3, 9) (−6, 3) (4, 12) (−10, −2) (−9, −1) (−17, −9) (14, 23) (−14, −5) (0, 9) (1, 9) (−14, −5) (−3, 6) (4, 12) (−4, 5) (1, 10) (−9, −1)
Note: CI: Confidence interval. Values that have significantly changed between years are in bold (p < 0.05). Respondents who endorsed the two points on either side of the mid-point of the five-point scales (values 1 + 2 and 4 + 5) were grouped together to the categories “true” and “not true”.
228
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
Table 3 Perceived discrimination and devaluation of persons with alcohol dependence in Germany. Multinomial logit regression controlling for age, sex and education, predicted percentages. Representative population surveys 1990 and 2011. Item 1
Most people would willingly accept someone as a close friend who has been treated for alcoholism
2
Most people believe that a person who has been in a hospital for alcohol detoxification is just as intelligent as the average person Most people believe that someone who has been treated for alcoholism is just as trustworthy as the average citizen
3
4
Most people would accept a fully recovered former alcoholic as a teacher of young children in a public school
5
Most people feel that entering a hospital for alcohol detoxification is a sign of personal failure (R)
6
Most people would not hire someone who has been in treatment for alcoholism to take care of their children, even if he or she had been well for some time (R) Most people think less of a person who has been in a hospital for alcohol detoxification (R)
7
8
Most employers will hire someone who has been in treatment for alcoholism if he or she is qualified for the job
9
Most employers will pass over the application of someone who has been in treatment for alcoholism in favor of another applicant (R) Most people in my community would treat someone who has been in treatment for alcoholism just as they would treat anyone Most young women would be reluctant to date a man who has been hospitalized for alcohol detoxification (R)
10
11
12
Once they know a person was in a hospital for alcohol detoxification, most people will take his opinion less seriously (R)
Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree
1990 (n = 1022) (%)
2011 (n = 967) (%)
31 27 42 38 28 34 28 24 48 25 23 52 62 21 18 58 23 18 63 21 16 20 26 54 72 19 9 24 27 50 56 26 18 37 34 29
48 30 22 59 27 14 46 30 24 31 27 42 48 25 26 52 28 20 51 30 19 26 32 42 63 28 9 36 31 32 59 26 14 26 39 35
95% confidence interval 17 3 −20 20 −1 −20 18 6 −24 6 4 −10 −14 5 9 −6 4 2 −13 9 3 6 6 −12 −9 9 0 12 5 −17 4 0 −4 −11 6 6
(13, 22) (−1, 7) (−24, −16) (16, 25) (−5, 3) (−23, −16) (14, 22) (2, 10) (−28, −20) (2, 10) (0, 8) (−15, −6) (−18, −9) (1, 9) (5, 13) (−11, −2) (1, 8) (−2, 5) (−17, −8) (5, 13) (0, 7) (3, 10) (2, 10) (−17, −8) (−13, −5) (5, 13) (−2, 3) (8, 16) (1, 9) (−22, 13) (−1, 8) (−4, 4) (−7, −1) (−15, −7) (1, 10) (1, 10)
Note: CI: Confidence interval. Values that have changed significantly between years are in bold (p < 0.05). Respondents who endorsed the two points on either side of the mid-point of the five-point scales (values 1 + 2 and 4 + 5) were grouped together to the categories “agree” and “disagree”. R: reversed item.
in 2011 compared to 1990. This change in opposite direction is masked by the unchanged mean perceived stereotype sum-score (Table 1), and mirrored in the opposing changes of the two factors derived from the perceived stereotype scale (see Table 4, below). With regard to perceived discrimination and devaluation, all items except one showed less severe perceived stigma in 2011 compared to 1990: only the proportion stating that most young women would be reluctant to date a man who had been hospitalized for detoxification slightly increased from 56% to 59%, which was not significant. All other items showed improvements of perceived public opinion: for example, while in 1990, 62% agreed that most people felt that entering a hospital for detoxification was a sign of personal failure, this proportion decreased to 48% in 2011. Agreement with the statement that most people think less
of someone who has been in a hospital for alcohol detoxification decreased from 63% in 1990 to 51% in 2011. Conversely, agreement that most people would consider someone who has been in a hospital for alcohol detoxification just as intelligent as the average person increased from 38% in 1990 to 59% in 2011 (Table 3). The EFA of the perceived stereotype scale resulted in two factors with an eigenvalue >1 after varimax rotation: factor 1 (eigenvalue 3.71, explained variance 41.2%) comprised three items related to “perceived unpredictability”: short-tempered (factor loading 0.858), unpredictable (0.854), aggressive (0.814). Factor 2 (eigenvalue 1.17, explained variance 13.3%) comprised five items related to “perceived strangeness”: scary (0.761), strange (0.678), stupid (0.661), dangerous (0.574), and untruthful (0.548). Subscales comprising these items had moderate internal consistency (Cronbach’s
Table 4 Perception of stigma attached to persons with alcohol problems. Standardized factor scores regressed on survey year, gender, age, and educational attainment. Representative population surveys 1990 and 2011, n = 1935–1937. “Perceived unpredictability”
Year (2011)a Age (years) Sex (female)b Education Constant Adjusted R2 a
Reference category: 1990. b Reference category: male.
“Perceived strangeness”
“Perceived discrimination/devaluation”
Coef.
p
Coef.
p
Coef.
p
0.15 −0.00 0.09 −0.05 0.08
0.001 0.338 0.048 0.063 0.472
−0.23 −0.00 0.14 −0.04 0.25
0.000 0.104 0.002 0.135 0.031
−0.44 −0.00 −0.00 0.02 0.22
0.000 0.490 0.951 0.598 0.054
0.01
0.02
0.05
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
229
Table 5 Association of treatment optimism regarding a person with alcohol dependence and perceived stigma. Linear regression analyses controlling for age, gender and education, standardized coefficients (Beta). Representative population survey in Germany 2011, n = 936–938.
“After undergoing treatment, this person can lead a normal life again” “With treatment, this person’s condition is going to improve markedly”
alpha, 0.84 and 0.75). The item “dependent on others” did not load meaningfully on any of the two factors (0.336; 0.277). Both factors explained a cumulative variance of 54%. We calculated factor scores for the two factors with a mean of zero and a standard deviation (SD) of 1. Based on the reversed items, higher scores indicate stronger perceived endorsement of the stereotypes. A similar EFA of the aPDDS yielded two factors with an eigenvalue greater than one. After varimax rotation, the first factor represented the six items depicting the opposite of discrimination and devaluation (eigenvalue 5.22, explained variance 43.5%, items 1, 2, 3, 4, 8, and 10, for item wording, see Table 3), while the second factor represented the six inversely worded items (eigenvalue 1.30, explained variance 10.8%, items 5, 6, 7, 9, 11, and 12). This two-dimensional structure caused by oppositely worded items is considered to be an artifact. Looking at the loadings of the EFA the 2nd axis turns out to be the quadratic form of the first one which will result in a two-dimensional structure after rotation (Bekker and De Leeuw, 1988; Heiser, 1985; van Rijckevorsel, 1987). This is also inline with findings from confirmatory factor modeling of the aPDDS in a large population sample in the U.S., which also yielded one dimension of the scale and a method-related factor related to item direction (Glass et al., 2013a). We thus calculated factor scores of the unrotated first factor to capture the single latent dimension of the scale. Item loadings (of the unrotated items) on this factor were (item numbers refer to Table 3): item 1, 0.754; item 2, 0.700; item 3, 0.794; item 4, 0.742; item 5, −0.529; item 6, −0.557; item 7, −0.683; item 8, 0.644; item 9, −0.585; item 10, 0.775; item 11, −0.417; and item 12, −0.630. Higher factor scores indicate greater perceived discrimination and devaluation, reverse items add negatively to this score. Again, the overall mean of this factor score is zero (standard deviation = 1). Conducting the factor analyses in both samples separately revealed a similar factor structure in 1990 and 2011. We regressed the factor score of the aPDDS and the two factor scores of the perceived stereotypes scale separately as dependent variables on year of survey (using the 1990 survey as reference category). To adjust the year effect for demographic changes across samples, the regression analysis controlled for respondents’ gender (male = 0, female = 1), age (years), and educational attainment (unknown = 0; no school completed = 1; 8 + 9 years of schooling = 2; 10 years of schooling = 3; 12 + 13 years of schooling = 4). Table 4 shows the results of the linear regression analyses. Compared to 1990, the perceived salience of unpredictability-related stereotypes had increased in 2011 by 0.15 SD. Perceived strangeness, in turn, had decreased by 0.23 SD. Perceived discrimination and devaluation of persons who had been in treatment for alcohol dependence were lower in 2011 by 0.44 SD compared to 1990. Among the control variables, only gender was related to perceptions of stereotypes: independent from survey year, women perceived more stereotypes both with regard to unpredictability and strangeness than men. Analysis of interaction effects between year of survey on the one side, and gender, age or education of respondents on the other side, did not yield any significant interactions (data not shown). Taken together, both analyses on item level and those using latent variables yielded similar results.
Perceived discrimination/ devaluation
Perceived unpredictability
Perceived strangeness
Beta
p
Beta
p
Beta
p
−0.30 −0.23
Contents lists available at ScienceDirect
Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep
Changes in the perception of alcohol-related stigma in Germany over the last two decades Georg Schomerus a,b,∗ , Herbert Matschinger c,d , Michael J. Lucht a,b , Matthias C. Angermeyer e,f a
Department of Psychiatry, Ernst Moritz Arndt University, Greifswald, Germany HELIOS Hanseklinikum Stralsund, Germany c Institute of Social Medicine, Occupational Health and Public Health, University of Leipzig, Germany d Institute of Medical Sociology, Social Medicine and Health Economics, University of Hamburg, Germany e Center for Public Mental Health, Gösing am Wagram, Austria f Department of Public Health and Clinical and Molecular Medicine, University of Cagliari, Italy b
a r t i c l e
i n f o
Article history: Received 13 March 2014 Received in revised form 26 July 2014 Accepted 27 July 2014 Available online 8 August 2014 Keywords: Stigma Substance use disorders Time-trend study Perceived Discrimination and Devaluation Scale
a b s t r a c t Background: Alcohol dependence is a severely stigmatized disorder. Perceived stigma may deter helpseeking and is associated with higher co-morbidity and self-stigma in persons with alcohol dependence. We assess changes in the perception of alcohol-related stigma over 21 years in the general population. Methods: Two representative population surveys using identical methodology were conducted in Germany in 1990 and 2011 (n = 1022 and n = 967), eliciting the perceived discrimination and devaluation of someone with a history of alcohol problems as measured with an adoption of Link’s Perceived Discrimination and Devaluation Scale (aPDDS), and perceived negative stereotypes of an “alcoholic.” Results: Both on item level and using factor scores, attitudes changed significantly between 1990 and 2011. Perceived discrimination and devaluation of someone with a history of alcohol dependence decreased considerably by 0.44 standard deviations (SD). Perceived negative stereotypes related to unpredictability of an “alcoholic” increased slightly by 0.15 SD, while perceived stereotypes related to strangeness decreased (−0.23 SD). Conclusions: Our findings suggest that particularly the image of someone who has received treatment for alcohol dependence has improved in Germany. This parallels increasing acceptance of professional treatment for alcohol dependence among the general population over the last twenty years, and contrasts with overall unchanged negative attitudes toward persons who actually suffer from alcohol problems. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Studies on the development of public attitudes toward people with substance use disorders indicate that they have not changed for the better over the last decades. Attitudes toward persons with alcohol dependence, for example, have remained disturbingly stable: time trend studies from the U.S. and Germany found no meaningful changes in the desire for social distance between 1996 and 2006 (U.S.) or 1990 and 2011 (Germany) (Angermeyer et al., 2013b; Pescosolido et al., 2010). Similarly, personally held stereotypical beliefs about alcohol dependence did hardly change
∗ Corresponding author at: Department of Psychiatry, University Medicine Greifswald, Helios Hanseklinikum Stralsund, Rostocker Chaussee 70 18437 Stralsund, Germany.Tel.: +49 3831 452109; fax: +49 3831 452105. E-mail address: [email protected] (G. Schomerus). http://dx.doi.org/10.1016/j.drugalcdep.2014.07.033 0376-8716/© 2014 Elsevier Ireland Ltd. All rights reserved.
between 1990 and 2011 in Germany (Schomerus et al., in press), or in Great Britain between 1998 and 2003 (Crisp et al., 2005). Some negative changes were observed in the U.S.: here, the proportion of respondents assuming that bad character was a cause for alcohol dependence did increase by 16% between 1996 and 2006. These few time-trend studies of alcohol-related public attitudes have all examined public stigma, i.e., the personal attitudes of the respondents. However, there is another component of stigma that is of high relevance to those suffering from alcohol use disorders: the perception of public stigma. People not only hold personal beliefs about alcohol dependent persons, they also have an idea of what, in their opinion, other people believe. Perceived public stigma and personally held attitudes of the general public are not identical: a study in Australia examining perceived and personal stigma related to depression found very low correlation of the according scales (R = 0.10; Griffiths et al., 2004). So far, there are no
226
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
time-trend studies of the perceived stigma attached to alcohol use disorders. Regarding the perceived stigma of mental disorders, a study from Australia found only small differences in perceived stigma between 2004 and 2011, indicating increasing perceptions of the dangerousness-stereotype related to persons with chronic schizophrenia. In contrast, a recent study from Germany found perceived discrimination and devaluation of “former mental patients” to have diminished considerably since 1990 (Angermeyer et al., 2013a). Here, perceptions of stigma had developed favorably, while personal attitudes particularly toward persons with schizophrenia had deteriorated within the same time period (Angermeyer et al., 2013b). The perceived stigma attached to alcohol use disorders seems to deter help-seeking; in a large, representative sample of adults with alcohol use disorders in the U.S., perception of higher stigma toward persons with alcohol problems was associated with lower lifetime alcohol service use (Keyes et al., 2010). Perceptions of stigma have also been conceptualized as starting point for the formation of personal attitudes, including self-stigma. In his progressive model of self-stigma, Corrigan posits awareness of other persons’ stereotypical beliefs as the beginning of a cascade of stigmatizing cognitions, resulting in agreeing with these stereotypes (personal stigma), and eventually applying these stereotypes to oneself should one develop the stigmatized condition (self-stigma; Corrigan et al., 2011, 2006). This model has also been applied to persons with alcohol dependence, showing that self-stigma is associated with decreased drinking-refusal self-efficacy (Schomerus et al., 2011a). Finally, modified labeling theory highlights that expectations of rejection and discrimination shape the individuals’ experience of a stigmatized disorder (Link et al., 1989). In fact, a large cross-sectional study in the U.S. found high perceived stigma of persons labeled with alcohol use disorders to be associated with negative psychiatric outcomes (Glass et al., 2013b). High perceptions of stigma related to alcohol dependence could thus decrease help-seeking, increase self-stigma and impair outcomes of persons suffering from alcohol use disorders. In this study, we explore how the perception of alcohol-related stigma has developed in Germany over the last two decades. To our knowledge, this is the first time-trend study of perceived alcoholrelated stigma in the general population. We use data from two methodologically identical surveys in 1990 and 2011 to explore how two facets of perceived stigma have developed over time:
(A) perceptions of the salience of negative stereotypes related to alcohol dependence and (B) perceptions of discrimination and devaluation of persons with alcohol dependence. 2. Methods 2.1. Surveys and interview Two population surveys were conducted among German citizens aged 18 years and over, the first in 1990 (n = 3087, response rate 70.0%), the second in 2011 (n = 3642, response rate 64.0%). In both surveys samples were drawn using a random sampling procedure with three stages: (1) sample points, (2) households, and (3) individuals within target households. Target households within sample points were determined according to the random route procedure. Target persons within each household were selected using random digits. Informed consent was considered to have been given when individuals agreed to complete the interview. The fieldwork for the first survey was carried out by GETAS, Hamburg, for the second survey by USUMA, Berlin; both institutes specialize in marketing and social research. The surveys were approved by the Ethics committees of the Universities of Heidelberg (1990) and Greifswald (2011). Before the first survey, the interview had been pretested with 20 persons to ensure maximum understandability of the questions. Results of the survey in 1990 have already been published separately (Angermeyer et al., 1995). In both surveys, the same interview mode (face-to-face, paper–pencil) was used. On both occasions, the interview used identical wording and sequence of questions. In the first part, which is not subject of the present paper, we asked questions related to a case–vignette of a person with different mental disorders. The second part covered issues unrelated to the case–vignette. In both surveys, respondents who had answered questions related to a vignette depicting a person with alcohol dependence received further, vignette-unrelated questions on their perception of the stigma surrounding alcohol dependence, which are the subject of this paper (1990: n = 1022; 2011: n = 1187). Since the survey in 1990 had been conducted before reunification in the old Federal Republic of Germany, we restricted our analyses of the 2011 data to respondents living in the “old” states of Germany (n = 967). Sociodemographic characteristics of the study samples and of the German population of each year are reported in Table 1. 2.2. Measures We assessed two aspects of perceived stigma. First, we asked respondents about the perceived salience of negative stereotypes associated with “alcoholics” among the general public (Angermeyer et al., 1995). Respondents were asked to rate the extent to which they believed the general public would endorse negative attributes of alcohol dependent patients: being short-tempered, unpredictable, aggressive, dependent on others, strange, stupid, scary, untruthful, and dangerous. The rating of each stereotype had to be given on five-point Likert scales with the anchors 1 = ”is certainly true” and 5 = ”is certainly not true”. We reversed scores for our analyses, so that higher scores indicate higher perceptions of stereotypes. Second, we used an adapted version of Link’s Perceived Discrimination and Devaluation Scale (adopted Perceived Discrimination and Devaluation Scale, aPDDS;
Table 1 Socio-demographic characteristics of study samples and sum-scores of perceived stigma measures. 1990
2011
Survey (n = 1022) (%)
Total populationa (%)
Survey (n = 967) (%)
Total populationa (%)
Gender Male Female
47.2 52.8
48.5 51.5
44.7 55.3
48.6 51.4
Age (years) 18–25 26–45 46–60 61+
14.9 36.8 25.9 22.4
12.3 38.0 24.2 25.5
9.0 34.7 24.6 28.7
11.3 31.9 26.9 29.9
Educational attainment Still student No schooling completed 8/9 years of schooling 10 years of schooling 12/13 years of schooling
1.6 2.3 52.1 29.2 14.8
0.4 2.5 55.8 25.8 15.5
0.6 2.0 44.5 37.6 15.3
1.0 4.0 38.5 29.3 27.1
Mean perceived stereotypes sum-scoreb (SD)
30.4
(6.5)
30.2
(6.5)
41.3
(9.4)
37.2
(8.1)
c
Mean aPDDS sum-score (SD) a b c
Data from the Federal Statistical Office of Germany. Range: 9–45. Adopted Perceived Discrimination and Devaluation Scale, range: 12–60; items 5–7, 9, 11–12 reversed.
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231 Link et al., 1989). The original instrument asks about the extent of agreement with statements indicating discrimination and devaluation of former psychiatric patients. It was designed to capture attitudes toward a labeled individual without referring to present symptoms, using descriptions like “a former mental patient,” “a person that has been in a mental hospital,” or “a fully recovered former mental patient” (Link et al., 1989). We adopted the authorized German version (Matschinger et al., 1991) by asking about someone who “has been treated for alcoholism,” “has been in a hospital for alcohol detoxification,” or “a fully recovered former alcoholic.” Items were otherwise similar to the original version, enquiring for example whether most people perceive someone who had been in hospital for alcohol detoxification as failure or as less intelligent than other persons, and to what extent the respondent believes that most people distance themselves from persons treated for alcoholism in various social relationships, like having a patient as a close friend or letting him teach young children in school. The exact wording of the items is given in Table 3. The scale is balanced such that a high level of perceived stigma is indicated by agreement with six of the items and by disagreement with six others. Answers to the items are recorded using 5-point Likert scales with the anchors “totally agree” (1) and “totally disagree” (5). The original scale and its German version have been used in population surveys (e.g., Link et al., 1989; Schomerus et al., 2006) as well as in studies involving patients suffering from mental illness (e.g., Freidl et al., 2012; Lundberg et al., 2007). The first adoption of the PDDS to alcohol dependence was employed in 1990 in Germany in the first of our two surveys (Angermeyer et al., 1995). Meanwhile, several adoptions of the PDDS for alcohol or substance use disorders have been used, and have shown good psychometric properties (Glass et al., 2013a). Because most items of the aPDDS refer to a person that has undergone treatment for alcohol problems, we included items on treatment optimism from the 2011 survey into our analyses. In this survey, we asked respondents to rate their agreement with the following statements regarding the person with alcohol problems described in the vignette: (1) “After undergoing treatment, this person can lead a normal life again,” and (2) “with treatment, this person’s condition is going to improve markedly.” Answers were given on Likert scales with the anchors 1 = certainly true to 5 = certainly not true. We reversed scores, higher scores indicating greater treatment optimism. Additionally we assessed socio-demographic characteristics (gender, age, and educational attainment). 2.3. Statistical analysis We first performed analyses on item level examining to what extent answers to single items had changed between 1990 and 2011. For this reason, we calculated multinomial logit regression models for all items using the STATA modules “mlogit” and “prvalue” (Long and Freese, 2006; Xu and Long, 2005; StataCorp, 2011). To reduce the number of categories, we grouped the five-point Likert scales into three categories as indicated in Tables 2 and 3. To adjust changes between years for demographic differences across samples, we controlled for respondents’ gender, age,
227
and educational attainment. The regression models yield answer probabilities for each category and each year, with control variables held at their means for the combined sample. Using the delta method, we determined the significance of changes by computing ninety-five percent confidence intervals for the differences between both years. To make adjusted predictions comparable to unadjusted percentages, answer probabilities are multiplied by 100 and can thus be read as estimated percentages of respondents endorsing each category. In a second step, we looked for possible latent variables within our scales to determine the effect of time on the expression of these latent variables. With both scales, we first performed an exploratory factor analysis, followed by a number of linear regression analyses using factor scores as dependent variables in order to examine the change of perceived stigma over time. This procedure is inline with the so-called “two-step approach” (Anderson and Gerbing, 1988, 1992; Hayduk, 1996), because we do not estimate the structure and the effect of time and other exogenous variables on the latent variables simultaneously, but rather perform measurement and structural (explanatory) analyses sequentially. For our exploratory factor analysis (EFA), we used the “factor” command and the “pcf” option of STATA, release 12.1 (StataCorp, 2011), which performs a principal component factor analysis.
3. Results Table 1 (bottom) shows the sum-scores of the two perceived stigma measures in both surveys, suggesting that overall, the perception of public negative stereotypes did hardly change, while perceptions of discrimination/devaluation seem to have diminished. Tables 2 and 3 show the predicted answer probabilities and predicted changes between 1990 and 2011 on item level as calculated by multinomial logit models controlling for age, gender and educational attainment. Values changing significantly between years are printed in bold. These answer probabilities can be read as percentages of the population answering in the defined manner. With regard to the perceived salience of negative stereotypes, a majority believed that the general population endorsed the stereotypes “short-tempered,” “unpredictable,” “aggressive,” and “dependent on others” in both years (Table 2). The stereotype found to be least salient was “stupid”, which was rated as salient by one in five respondents. While the perceived salience of the first three stereotypes increased, it decreased for “stupid,” “scary” and “untruthful”
Table 2 The German public’s perception of the salience of negative public stereotypes associated with “alcoholics”. Multinomial logit regression controlling for age, sex and education, predicted percentages. Representative population surveys 1990 and 2011. Stereotype Short-tempered
Unpredictable
Aggressive
Dependent on others
Strange
Stupid
Scary
Untruthful
Dangerous
True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true True Undecided Not true
1990 (n = 1022) (%)
2011 (n = 967) (%)
Predicted change
95% CI for change
65 20 15 71 18 11 59 26 15 67 22 11 44 26 30 23 37 40 39 29 32 46 30 23 36 35 30
72 21 7 72 21 7 68 25 7 51 31 17 42 34 24 18 24 58 29 34 37 37 32 32 35 40 24
7 1 −8 1 3 −4 10 −2 −8 −16 10 6 −2 8 −6 −5 −13 18 −10 5 5 −10 2 8 0 5 −5
(3, 12) (−3, 4) (−11, −5) (−3, 5) (0, 7) (−7, −2) (5, 14) (−6, 2) (−11, −5) (−20, −11) (6, 14) (3, 9) (−6, 3) (4, 12) (−10, −2) (−9, −1) (−17, −9) (14, 23) (−14, −5) (0, 9) (1, 9) (−14, −5) (−3, 6) (4, 12) (−4, 5) (1, 10) (−9, −1)
Note: CI: Confidence interval. Values that have significantly changed between years are in bold (p < 0.05). Respondents who endorsed the two points on either side of the mid-point of the five-point scales (values 1 + 2 and 4 + 5) were grouped together to the categories “true” and “not true”.
228
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
Table 3 Perceived discrimination and devaluation of persons with alcohol dependence in Germany. Multinomial logit regression controlling for age, sex and education, predicted percentages. Representative population surveys 1990 and 2011. Item 1
Most people would willingly accept someone as a close friend who has been treated for alcoholism
2
Most people believe that a person who has been in a hospital for alcohol detoxification is just as intelligent as the average person Most people believe that someone who has been treated for alcoholism is just as trustworthy as the average citizen
3
4
Most people would accept a fully recovered former alcoholic as a teacher of young children in a public school
5
Most people feel that entering a hospital for alcohol detoxification is a sign of personal failure (R)
6
Most people would not hire someone who has been in treatment for alcoholism to take care of their children, even if he or she had been well for some time (R) Most people think less of a person who has been in a hospital for alcohol detoxification (R)
7
8
Most employers will hire someone who has been in treatment for alcoholism if he or she is qualified for the job
9
Most employers will pass over the application of someone who has been in treatment for alcoholism in favor of another applicant (R) Most people in my community would treat someone who has been in treatment for alcoholism just as they would treat anyone Most young women would be reluctant to date a man who has been hospitalized for alcohol detoxification (R)
10
11
12
Once they know a person was in a hospital for alcohol detoxification, most people will take his opinion less seriously (R)
Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree Agree Undecided Disagree
1990 (n = 1022) (%)
2011 (n = 967) (%)
31 27 42 38 28 34 28 24 48 25 23 52 62 21 18 58 23 18 63 21 16 20 26 54 72 19 9 24 27 50 56 26 18 37 34 29
48 30 22 59 27 14 46 30 24 31 27 42 48 25 26 52 28 20 51 30 19 26 32 42 63 28 9 36 31 32 59 26 14 26 39 35
95% confidence interval 17 3 −20 20 −1 −20 18 6 −24 6 4 −10 −14 5 9 −6 4 2 −13 9 3 6 6 −12 −9 9 0 12 5 −17 4 0 −4 −11 6 6
(13, 22) (−1, 7) (−24, −16) (16, 25) (−5, 3) (−23, −16) (14, 22) (2, 10) (−28, −20) (2, 10) (0, 8) (−15, −6) (−18, −9) (1, 9) (5, 13) (−11, −2) (1, 8) (−2, 5) (−17, −8) (5, 13) (0, 7) (3, 10) (2, 10) (−17, −8) (−13, −5) (5, 13) (−2, 3) (8, 16) (1, 9) (−22, 13) (−1, 8) (−4, 4) (−7, −1) (−15, −7) (1, 10) (1, 10)
Note: CI: Confidence interval. Values that have changed significantly between years are in bold (p < 0.05). Respondents who endorsed the two points on either side of the mid-point of the five-point scales (values 1 + 2 and 4 + 5) were grouped together to the categories “agree” and “disagree”. R: reversed item.
in 2011 compared to 1990. This change in opposite direction is masked by the unchanged mean perceived stereotype sum-score (Table 1), and mirrored in the opposing changes of the two factors derived from the perceived stereotype scale (see Table 4, below). With regard to perceived discrimination and devaluation, all items except one showed less severe perceived stigma in 2011 compared to 1990: only the proportion stating that most young women would be reluctant to date a man who had been hospitalized for detoxification slightly increased from 56% to 59%, which was not significant. All other items showed improvements of perceived public opinion: for example, while in 1990, 62% agreed that most people felt that entering a hospital for detoxification was a sign of personal failure, this proportion decreased to 48% in 2011. Agreement with the statement that most people think less
of someone who has been in a hospital for alcohol detoxification decreased from 63% in 1990 to 51% in 2011. Conversely, agreement that most people would consider someone who has been in a hospital for alcohol detoxification just as intelligent as the average person increased from 38% in 1990 to 59% in 2011 (Table 3). The EFA of the perceived stereotype scale resulted in two factors with an eigenvalue >1 after varimax rotation: factor 1 (eigenvalue 3.71, explained variance 41.2%) comprised three items related to “perceived unpredictability”: short-tempered (factor loading 0.858), unpredictable (0.854), aggressive (0.814). Factor 2 (eigenvalue 1.17, explained variance 13.3%) comprised five items related to “perceived strangeness”: scary (0.761), strange (0.678), stupid (0.661), dangerous (0.574), and untruthful (0.548). Subscales comprising these items had moderate internal consistency (Cronbach’s
Table 4 Perception of stigma attached to persons with alcohol problems. Standardized factor scores regressed on survey year, gender, age, and educational attainment. Representative population surveys 1990 and 2011, n = 1935–1937. “Perceived unpredictability”
Year (2011)a Age (years) Sex (female)b Education Constant Adjusted R2 a
Reference category: 1990. b Reference category: male.
“Perceived strangeness”
“Perceived discrimination/devaluation”
Coef.
p
Coef.
p
Coef.
p
0.15 −0.00 0.09 −0.05 0.08
0.001 0.338 0.048 0.063 0.472
−0.23 −0.00 0.14 −0.04 0.25
0.000 0.104 0.002 0.135 0.031
−0.44 −0.00 −0.00 0.02 0.22
0.000 0.490 0.951 0.598 0.054
0.01
0.02
0.05
G. Schomerus et al. / Drug and Alcohol Dependence 143 (2014) 225–231
229
Table 5 Association of treatment optimism regarding a person with alcohol dependence and perceived stigma. Linear regression analyses controlling for age, gender and education, standardized coefficients (Beta). Representative population survey in Germany 2011, n = 936–938.
“After undergoing treatment, this person can lead a normal life again” “With treatment, this person’s condition is going to improve markedly”
alpha, 0.84 and 0.75). The item “dependent on others” did not load meaningfully on any of the two factors (0.336; 0.277). Both factors explained a cumulative variance of 54%. We calculated factor scores for the two factors with a mean of zero and a standard deviation (SD) of 1. Based on the reversed items, higher scores indicate stronger perceived endorsement of the stereotypes. A similar EFA of the aPDDS yielded two factors with an eigenvalue greater than one. After varimax rotation, the first factor represented the six items depicting the opposite of discrimination and devaluation (eigenvalue 5.22, explained variance 43.5%, items 1, 2, 3, 4, 8, and 10, for item wording, see Table 3), while the second factor represented the six inversely worded items (eigenvalue 1.30, explained variance 10.8%, items 5, 6, 7, 9, 11, and 12). This two-dimensional structure caused by oppositely worded items is considered to be an artifact. Looking at the loadings of the EFA the 2nd axis turns out to be the quadratic form of the first one which will result in a two-dimensional structure after rotation (Bekker and De Leeuw, 1988; Heiser, 1985; van Rijckevorsel, 1987). This is also inline with findings from confirmatory factor modeling of the aPDDS in a large population sample in the U.S., which also yielded one dimension of the scale and a method-related factor related to item direction (Glass et al., 2013a). We thus calculated factor scores of the unrotated first factor to capture the single latent dimension of the scale. Item loadings (of the unrotated items) on this factor were (item numbers refer to Table 3): item 1, 0.754; item 2, 0.700; item 3, 0.794; item 4, 0.742; item 5, −0.529; item 6, −0.557; item 7, −0.683; item 8, 0.644; item 9, −0.585; item 10, 0.775; item 11, −0.417; and item 12, −0.630. Higher factor scores indicate greater perceived discrimination and devaluation, reverse items add negatively to this score. Again, the overall mean of this factor score is zero (standard deviation = 1). Conducting the factor analyses in both samples separately revealed a similar factor structure in 1990 and 2011. We regressed the factor score of the aPDDS and the two factor scores of the perceived stereotypes scale separately as dependent variables on year of survey (using the 1990 survey as reference category). To adjust the year effect for demographic changes across samples, the regression analysis controlled for respondents’ gender (male = 0, female = 1), age (years), and educational attainment (unknown = 0; no school completed = 1; 8 + 9 years of schooling = 2; 10 years of schooling = 3; 12 + 13 years of schooling = 4). Table 4 shows the results of the linear regression analyses. Compared to 1990, the perceived salience of unpredictability-related stereotypes had increased in 2011 by 0.15 SD. Perceived strangeness, in turn, had decreased by 0.23 SD. Perceived discrimination and devaluation of persons who had been in treatment for alcohol dependence were lower in 2011 by 0.44 SD compared to 1990. Among the control variables, only gender was related to perceptions of stereotypes: independent from survey year, women perceived more stereotypes both with regard to unpredictability and strangeness than men. Analysis of interaction effects between year of survey on the one side, and gender, age or education of respondents on the other side, did not yield any significant interactions (data not shown). Taken together, both analyses on item level and those using latent variables yielded similar results.
Perceived discrimination/ devaluation
Perceived unpredictability
Perceived strangeness
Beta
p
Beta
p
Beta
p
−0.30 −0.23
