The role of psychological attribution in responses to weight stigma

Summary Objective Weight discrimination is associated with numerous negative health consequences. Little is known about early‐stage psychological mechanisms that explain variability in responses to weight discrimination among people with obesity. This study tested the hypothesis that attributing negative social evaluation to one's weight would be associated with stigma‐related stress responses (eg, reduced cognitive functioning and self‐esteem, increased negative affect and cortisol), especially among people who had experienced frequent weight discrimination in the past. Methods Adults (N = 109) with obesity were randomly assigned to receive a mildly positive (control) versus negative social evaluation. The extent to which participants attributed the negative evaluation to their physical appearance was assessed, along with negative affect, social and appearance self‐esteem, cognitive functioning and salivary cortisol. Results Participants who had experienced frequent weight discrimination in the past were more likely to attribute the negative evaluation to their appearance. Participants who attributed the negative evaluation to their appearance in turn experienced elevated negative affect, lower appearance self‐esteem and worse cognitive functioning. Conclusions This study is among the first to identify attribution as an early‐stage process underlying responses to weight stigma. Attribution may be a key psychological factor conferring risk for or protection from the negative effects of weight stigma.


| INTRODUCTION
Weight stigmathe social devaluation of people with obesity as expressed through stereotypes, prejudice and discriminationis highly prevalent in the United States. 1 As many as 42% of adults report experiencing weight-based discrimination, with higher rates typically observed among women and people with higher body weight (eg, body mass index [BMI] > 35 kg/m 2 ). [2][3][4] Weight discrimination occurs frequently in interpersonal relationships and is common across employment, education and healthcare settings. 1,5,6 Weight discrimination can range from minor forms of differential treatment such as receiving a disparaging look or an insensitive comment to major acts such as being denied a job promotion because of one's weight. 5,7 Over time, repeated experiences with weight discrimination, coupled with an awareness of negative stereotypes and the socially devalued nature of people with obesity, can lead some individuals to internalize weight bias (ie, engage in self-devaluation and chronically fear being stigmatized for their weight). [8][9][10] Being the target of weight discrimination is associated with a range of negative consequences for psychological and physical health. 9,11,12 Perceived weight discrimination is correlated with higher risk of low self-esteem, poor body image and depression, 11 as well as worse self-reported health, 13 poorer diabetes management, 14 increased risk of dementia 15,16 and increased mortality risk. 17 Moreover, perceived weight discrimination is associated with increased risk for the development and maintenance of obesity. 18,19 For instance, a study by Sutin and colleagues 19 found that adults without obesity at baseline who reported experiencing unfair treatment due to their weight were 2.5 times more likely to meet BMI criteria for obesity 4 years later, even while controlling for baseline BMI. Indeed, in some cases, weight discrimination is a stronger predictor of negative health outcomes than is higher BMI itself. 17,20 Weight discrimination is believed to contribute to poor health and obesity maintenance via cognitive (eg, impaired self-regulation and executive functioning), emotional (eg, psychological distress), physiological (eg, secretion of cortisol as triggered by activation of the hypothalamic-pituitaryadrenal axis) and behavioural (eg, increased eating) pathways that are activated in response to the psychological stress of being stigmatized. 9,13,[21][22][23][24][25][26][27][28] Compared with the long-term health consequences of weight stigma, 9,11,12 less is known about early-stage psychological mechanisms that underlie people's reactions to weight discrimination. The current study investigated whether people's responses to weight stigma are linked with biased patterns of psychological attribution.
Attribution refers to the processes through which people generate inferences about the intentions and motivations (among other factors) that guide people's behaviour. 29,30 Attributions reflect the operation of cognitive schemas that assist people in interpreting and assigning meaning to social interactions. [29][30][31][32] Indeed, attributions play a key role in shaping people's interpretations of the social environment. 29,30 Attributions may be especially important when the motives guiding someone's behaviour are not immediately apparent, as is sometimes the case when people experience weight-based discrimination.
Although weight stigma remains a socially acceptable form of bias 1 that often involves overt references to weight (eg, a celebrity with obesity is explicitly derogated for their weight on social media), many instances of weight stigma are less explicit, and thus, the motives underlying the unfair treatment are unknown. Supporting this notion, qualitative research on people's experiences with obesity documents instances in which participants suspected they were mistreated for their weight although weight was never explicitly mentioned. 7 To illustrate, imagine a situation in which a group of people at a party act disparagingly toward a female partygoer with obesity but stop short of saying anything about her weight. In this example, the woman's interpretation of the situation and, in particular, whether she attributes the group's disparaging behaviour to her weight, will likely affect her responses to their behaviour. In sum, when instances of weight stigma are unclear or ambiguous, attributional processes may play an especially important role in shaping people's interpretations and responses.
Theories of attribution 33 and research on cognitive appraisals and stress 34,35 suggest that people experience heightened stress responses when negative experiences are attributed to internal (vs. external) and controllable (vs. uncontrollable) causes. Obesity is a highly visible, socially devalued trait that is stereotypically viewed as controllable and due to internal causes, and thus, people with obesity are often presumed to be personally responsible for their weight. 5 For these reasons, attributing a negative social encounter to one's body weight should elicit high levels of psychological stress.
Moreover, with repeated exposure to day-to-day weight discrimination, individuals with obesity may become more likely to attribute ambiguous forms of mistreatment or social rejection to their body weight. Frequent experiences with weight-based discrimination are likely to make the possibility of experiencing such discrimination particularly salient in the future. 36 Thus, previous experiences with weight discrimination might increase the likelihood that one would come to anticipate weight stigma and to interpret negative social behaviours as constituting weight-based discrimination. See Figure 1 for the conceptual model depicting this hypothesized relationship and the framework guiding the current research.
The present study used an experimental, lab-based approach to investigate attribution as a psychological mechanism underlying people's responses to ambiguous forms of weight stigma. Adults with obesity were randomly assigned to receive a negative (versus mildly positive) social evaluation, yet the reason for the evaluation was left ambiguous. Participants were asked to rate the extent to which they attributed the evaluation to their physical appearance. Afterwards they completed several measures and tasks designed to assess psychological and physiological stress responses to weight stigma. These indicators are represented in conceptual models describing the proposed pathways through which weight stigma leads to poor health 9,27,28 and have been evaluated in previous experimental studies on weight stigma. 21,22,25,37,38 The study tested the following hypotheses: (1) people who have experienced weight discrimination more (vs. less) frequently in the past would be more likely to attribute the ambiguous negative social evaluation to their appearance; (2) attributing negative social evaluation to one's appearance would be associated with elevated negative affect, lower state self-esteem, decrements in cognitive functioning and higher cortisol reactivity; and (3) appearance attribution would statistically mediate the relationship between experiences with weight discrimination and stress responses to negative evaluation. Sample characteristics are provided in Table 1. The sample was primarily female (81%) and included participants from a variety of ethnic (21% Latino) and racial backgrounds (67% White, 20% Black or African American). Average BMI was 38.43 kg/m 2 (SD = 7.67).

| Procedure
The study took place in a laboratory on campus between 12:00 and 8:00 PM. Sessions lasted approximately 75 minutes and were led by two research assistants: a primary experimenter who assisted the participant and a secondary experimenter who ostensibly assisted the partner, but actually played the partner role during the first interaction. To disguise the purpose of the study, participants were told the study investigated how different modes of communication affect social interactions and health. Further, participants were told: "During today's study you will be interacting with a partner who is in the lab next door. Your first interaction will take place over instant messaging, where you will communicate by text only. Then you will interact by video voice mail, where you will each record and watch a brief video.
And then finally, you will interact with your partner face-to-face. After each interaction you will be asked to provide impressions of your partner and your partner will do the same for you." All participants received relatively positive feedback after the first interaction (ratings of "6" for the two impression items).
Next participants recorded an 3-minute video to further introduce themselves to their partner. A webcam recorded the participant's face and entire upper body (so that their weight was readily apparent) while seated at the computer. Participants were given a list of "getting to know you" questions to facilitate creation of their video (eg, What pets did you have while you were growing up?). Participants then exchanged videos, watched their partner's video F I G U R E 1 Conceptual model: Appearance attribution is proposed to explain the link between experiences with weight discrimination and stigma-related stress responses to ambiguous negative social evaluation (as manipulated in the current experiment)  Note. Percentages may not sum to 100 due to rounding error. T tests and chi-square analyses were used to compare participants across conditions. No significant differences were observed. a Body mass index (BMI in kg/m 2 ) was computed from measured height and weight. Fourteen participants declined to be weighed in the lab and instead provided self-reported weight. For these participants, we used the average of the two self-reported weights (screening survey and lab) when calculating BMI. b Perceived weight was assessed with the following item: "How would you rate your current body size?" (1 = very underweight to 7 = very overweight).
condition received scores of "2" on both impression items (where 1 = very negative) and the following comment: "Her video was fine, but I guess I'm not that excited about talking face-to-face." Notably, no reason was provided for this relatively negative social evaluation.

| Experiences with weight discrimination
Perceived experiences with weight discrimination were assessed with six items adapted from research by Williams and colleagues 42 that have been used in previous weight stigma research. 13 Participants were asked to rate the frequency with which they experience unfair treatment due to their weight on a day-to-day basis (eg, "You are treated with less courtesy or respect than other people because of your weight." And "You receive poorer service than other people at restaurants or stores because of your weight."). Items were accompanied by a 6-point response scale (1 = never, 2 = less than once a year, 3 = a few times a year, 4 = a few times a month, 5 = at least once a week and 6 = almost every day). Items were averaged to create a composite score (α = .84) with higher scores representing greater perceived frequency of experiences with weight discrimination.

| Attributions
Attributions for partner feedback were modelled after items from Major and colleagues. 43 Participants indicated the extent to which they thought their partner's feedback was due to "your appearance," in addition to other plausible attributions ("your personality, your partner's personality, your gender, and your partner's gender"), and two distractor items included to support the cover story (eg, "your speaking style"). The phrase "your appearance" (rather than "your weight") was used deliberately to mask the study's focus on weight stigma.
Attributions were rated on a 5-point scale (1 = not at all, 2 = a little bit, 3 = somewhat, 4 = very much and 5 = extremely) with higher scores indicating greater attribution to that domain.

| Negative affect
An adapted version of the Discrete Emotions Questionnaire (DEQ) 44 was used to assess emotions after the experimental manipulation.
Five emotions (anger, sadness, happiness, fear and anxiety) were assessed along with and an additional emotion relevant to weight stigma (ie, shame). 26,45 Each emotion was assessed with three to five adjectives. Participants indicated the extent to which they felt each emotion at that moment (1 = not at all, 2 = slightly, 3 = somewhat, 4 = moderately, 5 = quite a bit, 6 = very much and 7 = extremely). A composite score for each emotional state was computed by taking the average of the relevant adjectives. A total negative affect composite score was computed by averaging the anger, sadness, anxiety, fear and shame adjectives (α = .87). Higher scores indicate more emotion/negative affect.

| State self-esteem
The state self-esteem scale 46

| Cognitive functioning
The Stroop task was used to assess cognitive functioning. 47 The task is used widely for assessing people's ability to inhibit an automatic response (reading) in favour of performing a more controlled task (colour naming). Participants were presented with a series of cards. Each card (trial) contained 30 words (the names of various colours) arranged in rows. Reading from left to right, participants were instructed to name the ink colour of each word as quickly as possible while avoiding mistakes. On filler trials, the ink colour and word name were matched (eg, the word "red" was printed in red ink), so naming the colour was simple. On incongruent trials, however, the ink colour and word name were mismatched (eg, the word "red" was printed in blue ink). This required participants to suppress the automatic response of reading the word. The task thus measures cognitive control with longer reaction times indicating that a participant was less able to suppress the automatic response. Participants began with two abbreviated practice trials followed by 10 trials, three of which were filler (congruent) trials and seven of which were incongruent trials.
The experimenter recorded response time in seconds for each trial.
Cognitive functioning scores were calculated by computing the average time taken to complete incongruent trials. Higher response latencies reflected lower cognitive functioning. Scores were not calculated for two participants who failed to follow instructions. Two additional participants had very high response times (>3 SDs above the mean) and were excluded from analyses. Participants were instructed to refrain from eating, drinking any beverages besides water, smoking or exercising within the 2-hour period before their session. Five participants reported eating, drinking and/or smoking prior to the session; however, findings remained the same whether these participants were included/excluded so they are included for all analyses. One participant with a very high follow-up cortisol value (>3 SDs above the mean) was excluded from analysis.

| Salivary cortisol
Baseline and follow-up cortisol values were not skewed (skewness values were below 1.0); thus, variables were not subjected to a log transformation before analysis.

| Statistical analysis
Analyses were performed using IBM SPSS Statistics 26 and the PRO-CESS macro. 48 To assess whether randomization was successful, t tests and chi-square analyses were used to compare baseline sample characteristics across the two conditions. Ordinary least squares regression analysis was used to examine whether experimental condition interacted with experiences with weight discrimination to affect appearance attribution. To evaluate the form of the interaction, the effect of weight discrimination on appearance attribution was examined for participants in the control vs. negative feedback condition. All predictor variables were centred prior to analysis. Bootstrap confidence intervals were estimated using 5,000 samples.

| RESULTS
Sample characteristics are provided in Table 1. Participants were randomly assigned to receive control (n = 54) or negative feedback (n = 55). Sample characteristics did not differ by experimental condition, indicating successful randomization. Descriptive statistics for primary outcomes by experimental condition are provided in Table 2.
Results from the analysis predicting appearance attribution from experimental condition, experiences with weight discrimination and their interaction are presented in Table 3. Experiences with weight discrimination and experimental condition interacted to predict appearance attribution. When participants received negative feedback, there was a positive relationship between weight discrimination and appearance attribution such that participants who experienced more frequent weight discrimination were more likely to attribute the (negative) feedback to their appearance. In contrast, when participants received control feedback, there was a negative relationship between weight discrimination and appearance attribution, such that participants who experienced more frequent weight discrimination were less likely to attribute the (positive) feedback to their appearance.
Results from analyses predicting primary outcome variables from experimental condition, appearance attribution and their interaction are presented in Table 4. As predicted, an interaction between experimental condition and appearance attribution was observed for negative affect, appearance self-esteem and cognitive functioning.
Examining the form of interaction indicated that appearance attribution was not associated with negative affect, appearance self-esteem or cognitive functioning among participants who received control feedback. However, among participants who received negative feedback, those with higher appearance attribution scores reported more  Table 4). Similarly, a repeated measures analysis of variance indicated that cortisol values did not change from baseline to follow-up and neither the main effects of condition or appearance attribution nor the interaction between those variables were statistically significant (results not shown).
Next, the complete moderated mediation model (Figure 1  Note. Negative affect was assessed as the extent to which participants were experiencing five negative emotions (anger, sadness, fear, anxiety and shame) after receiving the experimental manipulation. Each emotion was assessed with multiple adjectives rated on a 7-point scale (1 = not at all to 7 = extremely). Adjectives were combined to create a single composite with higher scores representing more negative affect. Social and appearance self-esteem were each assessed on a 5-point scale (1 = not at all to 5 = extremely) with higher scores representing higher state self-esteem. Cognitive functioning scores were calculated by computing the average time in seconds taken to complete the incongruent trials.
Higher scores (ie, longer response times) indicate worse performance.
T A B L E 3 Results from the regression analysis predicting appearance attribution from experimental condition, experiences with weight discrimination and their interaction Finally, ancillary analyses were conducted to assess whether the observed effects were specific to appearance attribution. Means and standard deviations for all attribution measures are provided in Table 5. Relative to participants in the negative feedback condition, participants in the control condition were more likely to attribute the partner feedback to their personality, their gender and their partner's gender. The only attribution for which a significant difference in the opposite direction was observed was appearance attribution. Participants in the negative feedback condition were more likely to attribute the partner feedback to their appearance than were participants in the control feedback condition.

| DISCUSSION
The present study illustrates the important role of attribution in people's responses to weight stigma. In a carefully controlled laboratory experiment, participants with obesity who reported experiencing  Findings from the present study fit with theories that distinguish between people's experiences with unfair treatment (also referred to as "enacted stigma") and the tendency to internalize the negative stereotypes and attitudes that society holds toward a socially devalued trait (also referred to as "felt" or "self-directed" stigma). 5 was never mentioned in the current study: participants in the experimental condition received a negative evaluation from their partner, yet the reason for the evaluation was not specified. Whether participants attributed the negative evaluation to their weight depended on their previous experiences with weight discrimination. Participants who perceived themselves as having been exposed more frequently to weight discrimination in the past were more likely to attribute the negative evaluation to their appearance. Such perceptions likely reflect a blend of people's previous experiences with weight discrimination as well as their level of internalized weight stigma. 5,9,10 The current research also extends the literature by focusing directly on individual differences in responses to weight stigma among people with obesity. Previous experimental studies on weight stigma have tended to recruit participants with a range of BMIs (eg, from "average weight" to "obese") and focus on comparing people who perceive themselves to be overweight (or have a BMI indicating "overweight") to people who perceive themselves to be of normal weight (or have a BMI indicating "normal weight"). 21 Unlike previous studies, 21,25 no effects on cortisol reactivity were observed. Null findings may have been a function of study design features such as the wide range of study times during which cortisol was assessed (between 12-8 pm), the absence of a recovery saliva sample and variability in participant age. In addition, it is possible that the negative feedback manipulation was not powerful enough to elicit change in stress hormones. Further research is needed to better understand individual variability in physiological reactions to ambiguous instances of weight discrimination using additional biomarkers of stress (eg, alpha-amylase and pro-inflammatory cytokines) beyond cortisol reactivity. 54 The current study has several limitations. Participants were asked to rate the extent to which they attributed the feedback to their appearance rather than their weight. That appearance attributions were pronounced among participants who reported more frequent past experiences with weight discrimination suggests that participants were, in fact, thinking about their weight. Nevertheless, it is possible that participants may also have been considering other aspects of their physical appearance (eg, attractiveness and skin colour). Another limitation is that only one form of negative social evaluation (ie, a negative first impression among unacquainted individuals in a laboratory setting) was used. Future work should evaluate the extent to which the current findings generalize to other social contexts. Additionally, although weight discrimination tends to be more prevalent among women, 2,4,37 the sample included a relatively small proportion of male participants. As research has documented the negative effects of weight discrimination for men's health, 55 it is important that men are well represented in future research. Finally, although findings may have implications for the downstream effects of weight stigma, such long-term outcomes were not examined in the present study. Future research would benefit from investigating the role of weight-related attribution in contributing to poor mental and physical health outcomes associated with weight stigma.
Despite these limitations, this study is one of the first to identify an early-stage psychological process underlying people's immediate responses to ambiguous weight stigma. Attributing signs of social bias or mistreatment to one's body weight could promote a cascade of responses that ultimately lead to poorer physical and psychological health. As such, attributional processes may be a key psychological factor that confers risk for or protection from the negative effects of weight-based discrimination and may serve as a useful target for future intervention efforts.