Stigmatizing Attitudes Toward Obesity in a Representative Population-based Sample
Objective: The aim of this study was to determine stigmatizing attitudes toward obesity in the population, and its related psychological and sociodemographic determinants.
Methods and Procedures: In a representative population-based survey (N = 1,000), computer-assisted telephone interviewing was used to assess stigmatizing attitudes toward obesity, causal attributions of obesity, the labeling of obesity as an illness, perceptions about prevalence, severity, and chronicity of obesity, support of obesity prevention, and sociodemographic characteristics.
Results: Of the 1,000 participants, 23.5% (n = 235) had stigmatizing attitudes toward obesity, 21.5% (n = 215) did not have stigmatizing attitudes toward obesity, and 55.0% (n = 550) had attitudes that were undetermined with respect to stigmatization. Predictors of greater stigmatization were more causal attributions of obesity to individual behavior, less education, and older age, while causal attributions of obesity to heredity and labeling obesity as an illness predicted less stigmatization. Stigmatizing attitudes were significantly associated with stronger overall support of obesity prevention, but less readiness to support prevention financially.
Discussion: Our results indicate that stigmatizing attitudes toward obesity are prevalent in the population. Information about the etiology of obesity and the clinical relevance of this condition could prove useful for destigmatization efforts.
Obese individuals are commonly blamed for their excess weight, are socially disliked, and are the targets of pervasive negative stereotypes such as having a lack of self-discipline (1). As research in this area has been based mostly on samples in selected settings, it remains unclear how widespread stigmatizing attitudes toward obesity are in the population, how they vary across psychological and sociodemographic factors, and whether these attitudes affect one's readiness to support interventive approaches for this condition.
Stigmatizing attitudes toward obese individuals emerge in the context of beliefs about controllability and responsibility for the excess body weight (2,3,4,5,6,7). Such negative attitudes are often grounded in social ideologies, for example, of individualism or political conservatism (2,5). According to attribution theory (3,8), causal attributions play a pivotal role in determining reactions to stigmatized persons; indeed, the more a stigma such as obesity is attributed to internal, controllable causes, the greater are one's negative reactions to it (7,9). In contrast, attributions to external, uncontrollable causal factors are assumed to attenuate negative attitudes. However, inconsistent results have been found from experiments that examined effects of information about the biogenetic risk factors of obesity on stigmatizing attitudes (2,10,11), and evidence for the stigma relevance of attributions to obesogenic environmental factors is lacking. In addition, many studies in this field have been criticized for relying on nonrandom student populations and using fictitious case vignettes or manipulations of causality (1).
Little is known about other potential psychological determinants of the obesity stigma. One study on a number of health problems including obesity found perceptions of perceived severity of health problems to predict greater social rejection (6). Additional psychological determinants of the obesity stigma can be derived from conceptualizations of stigma dimensions and representations of illness (9,12,13). These putative determinants may be associated with less focus on personal controllability and, thereby, less stigmatizing attitudes: awareness about the chronicity/low treatability of this condition; awareness about the high prevalence of obesity; and labeling obesity as an illness. However, it remains unknown how these aspects actually relate to stigmatization.
With regard to sociodemographic determinants, there is some evidence that stigmatizing attitudes toward obesity decrease with older age (14,15) and are unrelated to one's body weight (2,5,15,16,17,18,19), although not consistently (20,21). Whether stigmatizing attitudes decrease with family history of obesity is inconsistent (15,22,23). There is more evidence (2,5,11,16,17,24,25) (4,19,21) that men show greater stigmatization of obesity than women. However, there is lack of evidence on socioeconomic correlates of the obesity stigma. Thus, the psychological and sociodemographic determinants of stigmatizing attitudes toward obesity need further clarification.
Furthermore, it remains largely unclear as to what extent stigmatizing attitudes translate into overt discrimination of obese individuals (26,27). As predicted by attribution theory, attribution of a person's obesity to internal, controllable causes has been shown to provoke negative reactions, of which some, especially less pity, decreased willingness to help the obese person (3,7). Therefore, it may be assumed that stigmatizing attitudes toward obesity, based on causal attributions to behavior, are associated with less support of interventive measures, such as obesity prevention efforts. Less support of obesity prevention with greater stigmatizing attitudes toward obesity could be interpreted as a discriminatory tendency.
Further investigation of stigmatizing attitudes toward obesity in the population is indicated in light of the rising prevalence rates of obesity in western industrialized countries (28), and possible detrimental effects of stigmatization on the mental health and well-being of vulnerable obese individuals (29,30,31,32). Examining the psychological and sociodemographic conditions by which stigmatizing attitudes and discriminatory tendencies arise is essential to identify potential starting points for stigma reduction.
Methods and Procedures
Participants and procedures
The survey investigation was conducted in August 2005 by USUMA, an institute specializing in market, opinion, and social research. Methods are detailed in a previous report (33). Sampling was based on random digital dialing using the drawing base of the Association of German Market and Social Research Agencies (ADM) including both registered and nonregistered telephone numbers. Within a randomly selected household, a target person was chosen according to the last birthday method for permanent residents. Using a computer-assisted telephone interviewing procedure, up to six calls were made to establish initial contact with a household as well as further contact to the target person to conduct the interview.
At the outset of the telephone interview, potential participants were informed about the purpose of the call (i.e., to conduct a survey on the topic of obesity) and consent was obtained. There was no compensation for participation. The telephone interview lasted 20 min on an average, included the measures described below, and was pilot tested for feasibility in 10 cases. Interviewers were trained on the entire interviewing procedure (i.e., beginning the interview, obtaining consent, following the instructions, and correct coding of every item).
Following this procedure, 1,836 noninstitutionalized civilian individuals were randomly selected in all states of Germany. Of these, N = 1,000 completed the interview, corresponding to a response rate of 54.5% (583 (31.8%) could not be reached, 157 (8.6%) refused to complete the interview, and 96 (5.2%) were excluded because of incomplete interviews). After weighting, data were representative of the German population aged ≥14 years with regard to age, gender, and state of residence. The study sample consisted of 431 men and 569 women (43.1 and 56.9%, respectively). Participants were 45.9 years old on average (s.d. = 17.9) and had a mean BMI (kg/m2) of 24.5 kg/m2 (s.d. = 4.1), calculated from self-reported height and weight. According to the guidelines of the National Institutes of Health, 29.2% (283/969) of participants were classified as overweight (BMI 25.0–29.9 kg/m2) and 9.1% as obese (88/969; BMI ≥ 30.0 kg/m2) (34). These rates are lower than current prevalence rates of overweight and obesity (40.7 and 18.1%, respectively) (35), which can be attributed to the underestimation of body weight in self-report (36).
Stigmatizing attitudes toward obesity. For assessing stigmatizing attitudes toward obesity, the subscale “Weight Control/Blame” (WCB) from the Antifat Attitudes Test was used (25). This scale contains nine beliefs on whether obese people are responsible for their weight (e.g., “Fat people have no willpower,” “Most fat people are lazy,” “If fat people knew how bad they looked, they would lose weight”). All the items are rated on five-point rating scales ranging from 1 = strongly disagree to 5 = strongly agree (response codes ≥4 indicate definite stigmatizing attitudes). A mean WCB score is then calculated. The WCB scale has adequate internal consistency and convergent validity (23,24,25). The German translation of the WCB scale was controlled through a retranslation procedure by a licensed translator and was piloted in an independent population-based sample of people participating in an Internet-based investigation on the stigmatization of obesity (N = 381). Internal consistency and mean scores of the German version were comparable with the English version (Cronbach α = 0.81; M = 2.66, s.d. = 0.65). The correlation with the Antifat Attitudes Test total score was highly significant (r = 0.81, P < 0.001; A.H., unpublished data).
Psychological and sociodemographic determinants of stigmatization. All potential psychological and sociodemographic determinants of stigmatizing attitudes are listed in Table 1. Psychological variables were specifically designed for this survey and were based on the current literature, as described in a previous report (33).
Table 1. Zero-order associations between stigmatizing attitudes toward obesity and causal attributions, problem identification, sociodemographic variables, and prevention support (N = 1,000)
Causal attributions of obesity were assessed through three scales on perceived risk factors of obesity: behavior, which encompassed eating and activity behavior likely leading to a positive energy balance (e.g., “Binge eating,” “Lack of physical activity;” 5 items); Environment, which referred to the obesogenic food and activity environment (e.g., “Healthy food is too expensive,” “Lack of facilities for outside physical activity;” 5 items); and Heredity (“Obesity is something that is inherited from the parents;” 1 item). All scales were derived from principal components analysis (33). Items were given five-point rating scales (1 = disagree completely to 5 = agree completely).
Three scales were designed for the assessment of perceived prevalence, severity, and low treatability of obesity: prevalence, which asked for prevalence estimates of obesity in adults and youth (e.g., “What proportion (%) of the adult population in Germany is obese?”); significance, which addressed the perceived severity of the obesity-related sequelae (e.g., “Obesity increases the risk for diseases such as diabetes or cancer”); and modifiability, which assessed the perceived treatability or chronicity of obesity (e.g., “On average, how much of their body weight (%) can participants in behavioral weight-loss programs reduce?”). All scales were confirmed by principal components analysis and contained two items each (five-point rating scales (1 = disagree completely to 5 = agree completely) or % estimates) (33).
Labeling obesity as an illness and the belief of individual or societal responsibility for this condition were operationalized through one item each (“Obesity is an illness” (1 = disagree completely to 5 = agree completely) and “Is obesity a problem that requires individual or societal solutions?” (1 = completely individual to 5 = completely societal)).
Sociodemographic variables consisted of age, BMI (both continuous), gender (women, men), family history of obesity (no obese first-degree relative, ≥1 obese first-degree relative), highest educational degree (<13 years of education, ≥13 years of education), net household income per month (<€2,000, ≥€2,000), and residence (eastern part, western part of Germany; all dichotomous with the second category being assigned the higher score).
Support of obesity prevention. Support of obesity prevention was operationalized by agreement with a range of preventive measures (33) including information-based measures (e.g., “Campaigns for healthy eating and physical activity;” 4 items), regulatory measures (e.g., “Restricting advertisement for unhealthy food such as sweets or chips;” 3 items), and childhood-focused measures (e.g., “More voluntary sports programs in schools;” 4 items). Level of agreement with the 11 prevention-support items was assessed with five-point rating scales (1 = disagree completely to 5 = agree completely). Further, readiness to support obesity prevention financially was assessed through the item, “For financing preventive measures, how much money would you be willing to spend more per year, e.g., in form of insurance fees or taxes?” (for response categories, see Table 1).
In addition to the response categories described above, a “no response” code was used if the interviewee felt unable to answer a question.
Potential correlates of stigmatizing attitudes toward obesity were first analyzed for zero-order association with the mean WCB score, using Pearson's r or Spearman's ρ correlation coefficients for continuous or ordinal data, respectively. First, the effect size of the correlations was evaluated according to Cohen (small: r ≥ 0.10, medium: r ≥ 0.30, large: r ≥ 0.50) (37). Second, the variables showing significant associations with the mean WCB score were retained for regression analysis. Stepwise multiple regression analysis was used to predict stigmatizing attitudes toward obesity with the continuous WCB score as the outcome variable (effect size evaluation: small: R2 ≥ 0.02, medium: R2 ≥ 0.15, large: R2 ≥ 0.35) (37). Third, for the analysis of zero-order associations between stigmatization (i.e., the mean WCB score) and financial prevention support, Spearman's ρ correlation coefficients were calculated. For descriptive purposes, associations between prevention support and other psychological and sociodemographic variables were computed (r or ρ, respectively). In all analyses, “no response” codes were treated as missing values. A two-tailed α of 0.01 was applied for all statistical tests.
Stigmatizing attitudes toward obesity in the population
The average WCB score was 3.01, indicating neither agreement nor disagreement with stigmatizing attitudes overall (s.d. = 0.68, median = 3; scale range: 1 = strongly disagree to 5 = strongly agree). This score was slightly higher than that in previous student samples (25) and that in an independent population-based sample recruited for an Internet-based investigation on the obesity stigma (M = 2.66, s.d. = 0.65; see Measures). To describe the pattern of response, participants were categorized as follows: those agreeing with stigmatizing attitudes (WCB scores ≥3.50, i.e., response codes ≥4, “definite stigmatizing attitudes”) (25), those disagreeing with stigmatizing attitudes (WCB scores ≤2.49, i.e., response codes ≤2), and those neither agreeing nor disagreeing with them (2.50 ≤WCB scores ≤3.49, i.e., response codes of 3). Accordingly, 23.5% of participants showed definite stigmatizing attitudes (235/1,000), 21.5% did not show stigmatizing attitudes (215/1,000), and 55.0% neither agreed nor disagreed with responses indicative of stigmatizing attitudes (550/1,000).
Psychological and sociodemographic correlates of stigmatization
As presented in Table 1, causal attributions of obesity to behavior were positively associated with the mean WCB score (r = 0.31, indicating medium effect size; P < 0.001). On the other hand, causal attributions of obesity to heredity and labeling obesity as an illness were significantly associated with less stigmatization (small effect sizes; P < 0.001). Additional significant correlates of greater stigmatizing attitudes were greater perceived prevalence and significance of obesity as a health problem, stronger belief of individual responsibility, and the sociodemographic factors older age and lower education (small effect sizes; P < 0.001). Overall, stigmatization was slightly, but significantly associated with greater causal attributions of obesity to the environment (r = 0.09; P < 0.01). A post hoc analysis for explanation of the latter result showed that causal attributions of obesity to the environment were significantly associated with causal attributions to behavior (r = 0.16; P < 0.0001), but not with causal attributions to heredity (r = 0.06; P > 0.01). Causal attributions to behavior and causal attributions to heredity were unrelated (r = −0.03; P > 0.01).
Prediction of stigmatizing attitudes toward obesity
The nine variables showing significant zero-order associations with the mean WCB score (P < 0.01) were entered into a regression equation to predict stigmatizing attitudes toward obesity. The final model retained five variables that accounted for a total of 18.0% of the variance, indicating a medium-size prediction effect (F (5, 954) = 43.62, P < 0.0001; see Table 2). Causal attributions of obesity to behavior showed the greatest positive contribution to the explanation of variance (10.0%, indicative of a small effect size); further significant predictors of greater stigmatization were less education, not labeling obesity as an illness, older age, and fewer causal attribution of obesity to heredity (≤3.0%). The single predictors were not significantly interrelated (except the associations of labeling obesity as an illness with younger age (r = −0.14) and greater attributions of obesity to heredity (r = 0.11); P < 0.01).
Table 2. Multiple linear regression analysis: predicting stigmatizing attitudes toward obesity (N = 1,000)a
Stigmatization and support of obesity prevention
As presented in Table 1, greater WCB scores were significantly associated with greater support of obesity prevention (small effect size; P < 0.001), but less readiness to support prevention financially (P < 0.01). When determining associations between readiness to support prevention financially, and psychological and sociodemographic variables, lower financial support was significantly associated with older age, greater belief of individual responsibility, and lower net household income (r = −0.17, 0.13, 0.19, all P < 0.01). A post hoc analysis of associations between significant correlates of financial prevention support showed that older age was significantly associated with lower net household income and greater belief of individual responsibility (r = −0.10 or −0.18, respectively, both P < 0.01). Variables determining prevention support are described in ref. (33).
This study sought to investigate stigmatizing attitudes toward obesity and psychological and sociodemographic determinants related to these attitudes in a representative population-based sample. Stigmatization was found to be higher than in previous student samples (25). Clear stigmatizing attitudes, that is, agreement to negative statements such as “Fat people have no willpower,” were detected in 23.5% of participants, while 21.5% of participants did not agree with these stigmatizing attitudes. The high rate of those who neither agreed nor disagreed with stigmatizing attitudes (55.0%) is noteworthy. It is unlikely that this rate was attributable to the interview-based assessment of stigmatization because a separate Internet-based investigation also yielded a high rate of undetermined responses (49.1%, 187/381; A.H. unpublished data; see Measures). Neither agreeing nor disagreeing with stigmatizing statements could be seen as a covert form of stigmatization that may surface when viewed as justified, for example, when it seems appropriate to express stigmatizing attitudes, and drawbacks are unlikely (27,38).
The most significant predictor of greater stigmatizing attitudes was causal attributions of obesity to individual behavior. This finding confirms previous research that has documented perceived onset-controllability as a main determinant of stigmatizing attitudes toward obesity (2,3,4,5,6,7), and is consistent with the prediction of attribution theory (3,8). Also consistent with this theory, causal attributions of heredity were found to be predictive of less stigmatization, with a very small effect size. Such a small association may explain the inconsistent results from experimental studies that have examined the stigma-reducing potential of information about the biogenetic causes of obesity (2,10,11). In contrast, causal attributions to the environment were slightly associated with greater stigmatization. A post hoc analysis showed that causal attributions to the environment were significantly associated with behavioral attributions, but not with genetic attributions. This suggests that environmental factors were assumed to be within, rather than outside, one's control. Thus, the association between greater stigmatization and environmental attributions is not contradictory to attribution theory. Indeed, participants may have focused on perceived unhealthy behavioral choices of obese persons given the obesogenic environment. Further, labeling obesity as an illness was identified as a predictor of less stigmatization. Thus, acknowledging the clinical relevance of obesity, including its genetic basis, is associated with increased acceptance rather than devaluation, an aspect that potentially deserves consideration in the enduring classification debate on obesity (39,40).
Among the sociodemographic determinants, lower education and older age were found to be predictive of greater stigmatization of obesity. The latter result is consistent with some previous findings (14,15). Older persons with less education have a higher risk for obesity (35) and its related medical sequelae (28). It is therefore possible that they view obesity as a greater threat than younger, more educated persons, and thus present more stigmatizing attitudes toward other people and their own obesity (41). Future research is desirable to further elucidate the social-cognitive processes of stigmatizing attitudes toward obesity.
Unexpectedly, those with greater stigmatizing attitudes showed more prevention support. Thus, the more people view obesity as an undesirable condition, the more they agree that measures ought to be taken to prevent it. However, consistent with the stigmatization correlate of greater individual responsibility for the solution of the obesity problem (2,5), stigmatizing attitudes toward obesity were associated with less readiness to financially contribute to prevention of obesity. Financial constraints and, relatedly, higher age, were also associated with lowered readiness to support prevention financially. Thus, it was not clearly found whether stigma-related discriminatory tendencies included a potential exclusion of obese individuals from preventive care.
The results bear several clear implications for stigma reduction. It appears to be particularly important to target the view that obesity is mainly behaviorally caused. Information about risk factors that are not directly within personal control, especially genetic risk factors, may prove fruitful, but may only produce small effects. Further, characterizing obesity as an illness, for example, in the case of extreme obesity, could be emphasized in destigmatization approaches. On the other hand, our findings did not support the use of information on environmental risk factors, severity, prevalence, or chronicity/low treatability of obesity as ways to destigmatize obesity. Finally, those with lower education and older age could represent potential target groups of destigmatization interventions. As they are also important target groups of obesity prevention (42), destigmatization could be included as a goal in obesity prevention efforts in these groups (43).
The results need to be interpreted while considering the strengths and limitations of this study. Strengths are the population-wide assessment of stigmatizing attitudes toward obesity and the identification of psychological and sociodemographic determinants in a sample representative with regard to age, gender, and state of residence. A validated stigmatization scale was used that focused on lack of willpower and blame, aspects that are central to the obesity stigma (44). Although this scale was found to be highly associated with multiple aspects of stigmatization (see Measures), its focus on willpower and blame may have led associations between this scale and causal attributions to be higher than associations between other aspects of stigmatization, such as antipathy, and causal attributions. Further, many methodological precautions were undertaken to avoid sampling biases, for example, the use of random digit dialing methodology with coverage of nonregistered telephone numbers (33). However, a comparison of the study data to the 2005 population data from the Federal Statistical Office on net household income and education showed that lower socioeconomic groups may have been under-represented. Finally, ethnicity, which has been found to moderate stigmatizing attitudes toward obesity in previous studies (45,46), was not assessed in this study. Ethnic variation of the obesity stigma could, for example, be present in Turkish migrants (47), one of the largest migrant groups in Germany.
Overall, our results indicate that stigmatizing attitudes toward obesity are prevalent in the population. Tackling causal beliefs on the controllability and highlighting the clinical relevance of this condition could be useful starting points of destigmatization efforts.
This project was funded from internal resources within the Department of Medical Psychology and Medical Sociology, University of Leipzig, Germany. A.H. was supported by educational grant 01GP0491 from the German Federal Ministry of Education and Research. External funding body had no input into protocol development, data collection, analyses, or interpretation.
The authors declared no conflict of interest.