Introduction

Overweight and obese individuals face prejudice and discrimination in many areas of daily life [1,2]. This stigmatization threatens overweight and obese individuals' psychological and physical health by increasing the risk for depression, anxiety, low self-esteem, and body dissatisfaction [3,4]. The psychological toll of weight stigmatization can be devastating. Overweight youth who are stigmatized about their weight are 2-3 times more likely to engage in suicidal thoughts and behaviors compared with their overweight peers who are not stigmatized [5]. Obese individuals who experience weight stigma are more likely to engage in unhealthy eating behaviors and lower levels of physical activity [6,7] and are less likely to undergo age-appropriate preventive cancer screenings [8,9]. Indeed, negativity toward overweight and obese people is also shown in the health care setting by physicians [10], nurses, medical students, dietitians and psychologists. Similarly, coworkers, employers, teachers, peers, friends, family members and romantic partners show weight bias [1,2].

Weight bias has been documented both explicitly [11,12], through consciously held evaluations which are potentially subject to social desirability and rely on self-report, and implicitly [13–17], through automatic evaluations that occur outside of awareness or conscious control. Some reasons for the existence of weight bias include valuation of a thin ideal for beauty, stereotypes such as laziness associated with obesity, and perceptions of the controllability of weight [18]. However, both explicit and implicit measures have highlighted the pervasiveness of weight bias by showing that, although overweight and obese people possess less weight bias than normal weight and underweight people, they still prefer thin people on average [19]. This suggests that overweight and obese people might find it difficult to identify with and prefer their weight group because of weight stigmatization [20]. Thus, factors other than the individual's identity - which promotes strong tendencies to prefer one’s own groups and social characteristics [21] - might be shaping weight bias.

One of those factors could be the cultural context. Culture powerfully shapes the beliefs and behavior of its members [22]. Existing studies show some variations in weight bias across nations [23]. For example, opposing the overall trend, in some nations popular stereotypes associate obesity with more positive than negative characteristics [24–27]. Overweight and obese people’s persisting negativity toward their own group could be partly influenced by negativity toward overweight people at the national level.

For example, health and beauty are conventionally associated with thinness. Thus, in nations with a high prevalence of overweight and obese people, the starker contrast of models of health and beauty with the general population may amplify the desirability and social status of the thin ideal. This could increase cultural messages, advertising, and focus on the desirability of thinness relative to obesity.

Likewise, the health consequences of obesity are well understood. Obesity increases the likelihood of various diseases, particularly heart disease, type 2 diabetes, obstructive sleep apnea, certain types of cancer, and osteoarthritis [28] and ^,the risk of many physical conditions such as high blood pressure, high blood cholesterol and high triglyceride levels [29]. In nations with a high incidence of overweight people, the public discussion health concerns and consequences might lead to an increase in negativity towards obesity even though more people are obese.

These likely cultural influences suggest the possibility that the relationship between weight bias and obesity across nations could be exactly the opposite of what observed at the individual level. For an overweight or obese individual, the evaluation of oneself, one’s groups, and one’s characteristics predicts greater positivity toward obesity than a normal or underweight individual. For a nation with a high prevalence of overweight and obese people, the negative social and health consequences of obesity may predict greater negativity toward obesity than a nation with low prevalence of overweight people.

To investigate the potential influence of obesity on the weight bias at the national level, we assessed the explicit (i.e., self-reported) and implicit (i.e., automatic or non-conscious) weight bias of citizens from 71 nations and investigated their relationship with the obesity comparing across individuals and nations.

Materials and Methods

Results

Overall, the sample showed a preference for thin people over overweight people both with explicit measures (Explicit mean = 1.00, SD = 1.09; t(311,759) = 514.14, P < 0.0001, Cohen’s d = 0.92) and with the IAT (IAT mean= 0.43, SD = 0.42; t(303,818) = 556.69, P < 0.0001, Cohen’s d = 1.02). Explicit and implicit preferences were positively correlated (r = 0.21, P < 0.0001). Moreover, without exception, each of the 71 nations showed explicit and implicit weight bias favoring thin people over overweight people, with considerable variability across nations (Table S1 in File S1).

At the individual level, across all participants in the entire sample, BMI scores were negatively related with explicit (β = -0.26, t(302,737) = -149.91, P < 0.0001) and implicit (β = -0.16, t(291,354) = -87.52, P < 0.0001) weight bias (Figure 1). Higher BMI scores were associated with weaker explicit and implicit negativity toward overweight people compared to thin people. However, overweight and obese people still showed explicit and implicit preferences for thin people compared to overweight people. BMI scores were negatively – though not always significantly – correlated with both explicit (r = -0.26, P < 0.0001; mean r = -0.15) and implicit (r = -0.16, P < 0.0001; mean r = -0.13) weight bias in every nation except Bolivia and Vietnam implicitly, and Lithuania and Albania explicitly (Table S7 in File S1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Scatter plots of relations of implicit (IAT) and explicit weight bias with BMI at the individual level.

Note. Each point in the plots represents the average preference of participants as a function of their BMI. The weight bias scores ranges from +2 to -2 for the IAT and from +3 to -3 for the explicit, with 0 indicating no relative preference between thin people over overweight people. More positive scores indicate a preference for thin people over overweight people, while more negative scores indicate a preference for overweight people over thin people. Vertical bars signify standard error. Data for participants with BMI greater than 60 were not included in the plot (0.15%). The regression line was computed on the original and not on the average data.

https://doi.org/10.1371/journal.pone.0083543.g001

An opposite relationship between obesity and implicit weight bias was found at the national level, for all three of the operationalizations of obesity (BMI, national percentage of people underweight, national percentage of people overweight). Results of the regressions showed that greater obesity in a nation was associated with stronger implicit negativity toward overweight people compared to thin people [national BMI: β = 0.54, t(66) = 5.15, P < 0.0001; national percentage of underweight people: β = -0.60, t(48) = -5.08, P < 0.0001; national percentage of overweight people: β = 0.60, t(58) = 5.71, P < 0.0001; Figure 2]. Demonstrating robustness, this relationship persisted after removing outliers, [BMI: β= 0.43, t(64) = 3.81, P < 0.0001; national percentage of underweight people: β = -0.49, t(46) = -3.76, P < 0.0001; national percentage of overweight people: β = 0.51, t(56) = 4.42, P < 0.0001] and after including 4 nation-level covariates to the models [BMI: β = 0.36, t(64) = 3.04, P < 0.01; national percentage of underweight people: β = -0.45, t(46) = -3.13, P < 0.01; national percentage of overweight people: β = 0.51, t(56) = 4.31, P < 0.0001]. Moreover, the relationship persisted after adding covariates (Table 1). Across three operationalizations of obesity, and three regressions for each examining the robustness of the relationship to outliers and covariates, the relationship between obesity and implicit weight bias across nations was strong and persistent. Further, supporting information shows additional robustness checks indicating that the obesity-weight bias relationship persists with alternative weighting strategies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Scatter plots of relations of implicit (IAT) and explicit weight bias as a function of three weight indicators (BMI, percentage of overweight and underweight people) at the national level.

Note. The weight bias scores ranges from +2 to -2 for the IAT and from +3 to -3 for the explicit, with 0 indicating no relative preference between thin people over overweight people. More positive scores indicate a preference for thin people over overweight people, while more negative scores indicate a preference for overweight people over thin people.

https://doi.org/10.1371/journal.pone.0083543.g002

Across models, none of the covariates were significant predictors. However, after removing the health expenditure per capita from the regression models, which was highly correlated with GDP (r = 0.87, P < 0.0001; Table S8 in File S1), we found that the GDP was a significant predictor of implicit weight preferences in each three regression models with national weight indicators. When, instead, we removed GDP from regression then health expenditure per capita was a significant predictor of implicit weight bias, but only in the regression model with percentage of underweight people (Table S9 in File S1). This result indicated that richer nations showed stronger weight bias but importantly it did not account for the relationship between the weight indicators and implicit preferences. Notably, no statistically significant effect emerged when implicit measures were replaced by explicit measures in all conducted regression analyses (Table 1; Table S9 in File S1).

Discussion

We found that (a) each of the 71 nations investigated showed implicit and explicit preferences for thin people compared to overweight people, (b) at the individual level, higher obesity (i.e., increased individual BMI scores) was associated with weaker implicit and explicit weight bias but on average overweight people still showed persistent preferences for thin people (c) at the national level, higher obesity (i.e., increased national BMI scores and percentages of overweight people) was associated with stronger implicit weight bias whereas no relation was observed with explicit measures, and (d) independent of weight indicators, national wealth was a significant predictor, with richer nations showing stronger implicit preferences for thin people compared to overweight people.

As the effects we found in our study were observed implicitly, but not explicitly, they suggest that the influence of prevalence of obesity on weight bias at the national level may be consciously denied or rejected. Explicit attitudes indicate evaluations that people are able and willing to report while implicit attitudes are inferred with behavioral measures that assess associations that exist in memory [30] and may thus be less influenced by conscious intentions. Indeed, implicit weight attitudes – in addition to being modestly positively related with explicit weight attitudes [17] – can differ in important ways from self-reported evaluations [14–16]. Furthermore, a recent study showed that self-report measures for investigating cross-national differences is weakened by factors compromising the validity of self-report [42]. In particular, the reference-group effect is the tendency for people to respond to subjective self-report items by comparing themselves with standards from their culture [43,44]. This compromises the sensitivity of detecting differences across nations. This is less likely to affect the implicit measures because participants’ evaluations are estimated from behavioral performance, not based on introspective processes that involve such decision-making biases.

Our results suggest that the degree of negativity toward overweight and obese people at the national level is in sharp contrast to individuals’ social identity factors that promote preferences for oneself and one’s group identities and characteristics. Whereas the relationship between obesity and weight bias was negative across individuals, it was positive across nations. This is an instance of Simpson’s paradox [45], in which a relation present at one level of analysis (across individuals) is reversed when examined at another level of analysis (across national aggregates of individuals). Unlike some social groups that are evaluated more positively when they are present in greater numbers, the increased prevalence of obesity in a nation does not seem to attenuate negative implicit social attitudes toward overweight people but exaggerates them. As a consequence, for example, an obese person living in the United States (where a relatively high proportion of the population is overweight) is, on average, likely to have an implicit preference for thin people over overweight people that is as strong as a thin person living in India (where a relatively small proportion of the population is overweight).

The mechanism for this effect is not indicated directly by these data and no existing theoretical frameworks anticipate this particular pattern of effects. However, there are some existing theories and areas of research that offer relevant insights. Here, we offer relatively speculative possibilities that will require additional research to clarify the mediating mechanisms of the present results. For example, as noted in the introduction, the greater prevalence of obesity in a nation may increase the negativity toward overweight and obese people by eliciting more frequent national discussion about the health risks and medical costs of obesity, ultimately exacerbating the stigmatization of obesity. Likewise, the high prevalence of obesity in a nation may reinforce the desirability of the thin ideal or of being normal weight by increase prevalence and emphasis on advertisements for diet plans, healthy foods, and gym memberships against the demographic trend of increasing obesity. Further, the association of thinness with social status – health and beauty – may produce even stronger negativity against obesity when this signal of status is rarer.

An additional speculative explanation of our findings may be provided by “the just world hypothesis” which asserts that people generally get what they deserve as the result of a universal force that restores moral balance [46,47]. Based on this proposal, the bias toward overweight and obese people might represent a “fair” consequence for their negative (i.e. unhealthy) actions. The different weight bias across nations might indicate differences in a just world belief at the national level. That is, countries with stronger beliefs in a just world might show a greater derogation of overweight and obese people.

Potentially supporting this interpretation is the fact that just world beliefs are a function not only of personal experience, but also of societal functionalism (i.e. a country’s structural and societal factors) [48]. Differences, indeed, have been found at the national and cultural level [49,50]. Furnham [50], for instance, found that nations who have more property, wealth and power had stronger just world beliefs, whereas nations who have little or no power and wealth would have unjust world beliefs. However, we did not measure just world beliefs in order to evaluate this possibility in the present data.

Differences in weight bias across countries may be also explained by attribution theory, which declares that people use information to arrive at casual explanations of events [51]. Specifically, according to Heider [52] people can explain behaviors by internal attributions, such as personal or individual features, or by external attributions, such as situational and environmental variables. Several studies showed that overweight and obese people are stigmatized because their weight is perceived to be caused by factors within personal control (e.g., overeating and lack of exercise) than external factors [53]. Societal attributions about the causes of obesity thus seem to contribute significantly to expressions of weight stigma. A speculative application of these findings to our results is that countries with greater negative weight attitudes might present also greater internal attributions of obesity. Indeed, it has been demonstrated that culture affects how people make attributions [54]. Follow-up research could investigate variation in attributions for obesity across nations.

We also observed a separate relationship between national wealth and implicit weight bias. A possible explanation is that the wealth in a nation may alter the meaning of obesity as an indicator of status. In wealthy nations, obesity may be negatively related with wealth perhaps because high-fat foods are cheaper than alternatives [55]. In this context, being overweight may be perceived as being of a lower economic status. In less wealthy nations, obesity may be positively associated with wealth because access to food resources indicate comparatively high economic status. Thus, the different contextualized meaning of obesity may account for the observed relationship between wealth and implicit weight preferences across nations. These explanations suggest that the national context, such as the prevalence of obesity and wealth in a nation, may influence the attitudes toward obesity.

In addition to several novel results the present study has some limitations. Indeed, it is important to point out that while we have offered a directional interpretation – that the prevalence of obesity and wealth influence implicit weight bias – the data themselves are correlational. The data cannot unambiguously support a causal conclusion, and there may be as yet unidentified third variables that are responsible for the observed relationships. A second limitation is that we did not examine the universe of possible covariates to test the robustness of present relationships, and potential explanatory factors. Thus, in addition to the four covariates that we considered it is possible that other variables could still be identified that account for part or most of the observed relationship.

A third limitation of our study is that while the sample is very large, it is not a random selection of the world, or any national population. Selection biases might be present in the sample of people that learned about the web site, volunteered to participate, and selected the weight task. While we cannot conclusively exclude that selection bias is present in our sample, we cannot identify any plausible reason why variation in selection biases across nations would covary with the variables of interest. Nonetheless, replication with other sampling contexts will be particularly useful to increase confidence in the observed relationships.

Taken together, our results point to the subtle influence of national factors in shaping social evaluations. Formation and change of implicit attitudes is not just a matter of influencing intentions; it also requires consideration of the social realities that shape minds without intention.

Supporting Information