Internalizing Weight Stigma: Prevalence and Sociodemographic Considerations in US Adults
Funding: This research was supported from a donation by the Rudd Foundation.
Disclosure: The authors declared no conflicts of interest.
Abstract
Objective
This study aimed to conduct a comprehensive assessment of the presence, severity, and sociodemographic correlates of weight bias internalization (WBI) across three distinct samples of US adults.
Methods
Levels of WBI were compared in (1) a sample of adults with obesity and heightened risk of weight stigma (N = 456), (2) an online community sample (N = 519), and (3) a national online panel (N = 2,529). Samples 2 and 3 comprised adults with and without obesity. Participants completed identical self-report measures, including demographic variables and weight-related factors, to determine their relationship with low, mean, and high levels of WBI.
Results
At least 44% of adults across samples endorsed mean levels of WBI (as determined by sample 3). The highest levels of WBI were endorsed by approximately one in five adults in the general population samples and by 52% in the sample of adults with obesity. Individuals with the highest WBI were white, had less education and income, were currently trying to lose weight, and had higher BMIs, higher self-perceived weight, and previous experiences of weight stigma (especially teasing).
Conclusions
Internalized weight bias is prevalent among women and men and across body weight categories. Findings provide a foundation to better understand characteristics of individuals who are at risk for internalizing weight bias.
Introduction
As many as 40% of US adults have been a target of weight-based teasing, unfair treatment, or discrimination (1, 2). These experiences, commonly referred to as “weight stigma,” are associated with numerous health consequences. Weight stigma induces physiological stress, increased caloric consumption, and depleted dietary self-efficacy, and it is associated with increased emotional distress (depression, low self-esteem, and body dissatisfaction) as well as disordered eating and lower physical activity (3, 4). Prospective associations link weight stigma with long-term declines in self-reported health, increased maladaptive eating behaviors, and weight gain (5-7). These associations remain independent of sociodemographic characteristics and body mass index (BMI), highlighting weight stigma as a unique contributor to poor health and a barrier to effective obesity prevention and treatment.
Distinct from stigmatizing experiences, many individuals internalize negative societal stigma directed toward them (8, 9). Internalized stigma occurs when a person is aware that he or she has a stigmatized identity and applies negative societal stereotypes to oneself (10). Emerging evidence indicates that this is also true of individuals who are stigmatized because of their weight (11). This form of self-stigma, or “weight bias internalization” (WBI), involves applying negative weight-based stereotypes to oneself and engaging in self-blame for one's weight status (11, 12).
WBI is a significant and independent contributor to poor emotional and physical health (including depression, anxiety, stress, clinical eating pathology, binge eating, health-related quality of life, poor weight loss maintenance, and metabolic syndrome) beyond the effect of BMI (13-20) and, in some cases, independent of experienced weight stigma (12, 21, 22). Some studies have pointed to WBI as a mechanism through which experienced weight stigma exacerbates adverse health behaviors (23), potentially through low self-esteem and basing one's self-evaluation on body weight (24, 25). While longitudinal research in this area has been limited, existing evidence has documented poorer weight loss outcomes in individuals with high baseline levels of WBI (26, 27). Despite increasing evidence of WBI as a contributor to health, there is not yet a comprehensive understanding of factors that characterize individuals who internalize weight bias. Most studies have relied on homogeneous samples (19, 21, 27-29) composed primarily of white women, making it unclear how WBI occurs in different segments of the population. An exception to this literature is a recent systematic examination of WBI in a national sample of US adults (N = 956), which found that WBI, but not experienced stigma, varied across race and gender; women reported higher WBI than men, and black participants reported less WBI than white participants (2). Thus, determining the extent of WBI across different sociodemographic characteristics in larger, diverse samples is an important next step to clarify who is at heightened risk and may benefit from supportive clinical intervention.
In addition, we need to obtain a clearer picture of how levels of WBI differ among people with heightened vulnerability to weight stigma compared with the general population. Some studies have documented higher WBI in people with high BMI versus low BMI (15, 16, 19, 21) and among individuals with binge eating symptoms compared with those without (17, 30). However, less is known about other weight-related variables that may affect WBI, such as self-perceived weight status, different types of experienced weight stigma, or weight loss efforts. Examining these variables in the context of WBI can advance our understanding of potential risks that WBI may pose for weight management, obesity, and related health behaviors.
To begin to address these gaps in the field, we conducted a comprehensive systematic assessment of the presence, severity, and sociodemographic correlates of internalized weight bias in three distinct samples of adults. Levels of WBI were compared in (1) a sample of adults with obesity, (2) an online community sample, and (3) a national online panel. To identify characteristics that describe individuals who internalize weight bias, we assessed the predictive value of key demographic variables and weight-related factors in relation to different levels of WBI.
Methods
Participants and procedure
Across three samples, 3,504 adults (56.4% women) completed online self-report surveys. Online survey software (Qualtrics, Provo, Utah) prevented participants from completing the survey more than once. Participation was anonymous and voluntary. All procedures were approved by the appropriate institutional review board. Table 1 shows characteristics of the full sample and each subsample.
|
Full Sample N = 3504 |
Sample 1 n= 456 |
Sample 2 n=519 |
Sample 3 n = 2529 |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| M | SD | M | SD | M | SD | M | SD | df | F | p | |
| Age | 44.01 | 15.81 | 51.00 | 11.08 | 37.71 | 11.38 | 44.05 | 16.77 | 2, 3485 | 89.91 | 0.000 |
| BMI | 28.11 | 7.33 | 36.41 | 9.01 | 27.62 | 7.47 | 26.64 | 5.74 | 2, 3386 | 428.29 | 0.000 |
| Highest BMI | 32.51 | 11.08 | 46.66 | 10.95 | 31.64 | 9.15 | 30.01 | 9.36 | 2, 3268 | 563.52 | 0.000 |
| N | % | N | % | N | % | N | % | df | Χ2 | p | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Sex | 2 | 283.29 | 0.000 | ||||||||
| Male | 1527 | 43.6 | 34 | 7.5 | 231 | 44.5 | 1262 | 49.9 | |||
| Female | 1977 | 56.4 | 422 | 92.5 | 288 | 55.5 | 1267 | 50.1 | |||
| Race/ethnicity | 8 | 161.47 | 0.000 | ||||||||
| White, non-Hispanic, non-Latino | 2438 | 69.6 | 393 | 86.2 | 425 | 81.9 | 1620 | 64.1 | |||
| Latino, Hispanic, or Mexican-American | 436 | 12.4 | 19 | 4.2 | 20 | 3.9 | 397 | 15.7 | |||
| Black or African American | 375 | 10.7 | 24 | 5.3 | 35 | 6.7 | 316 | 12.5 | |||
| Asian or Pacific Islander | 175 | 5.0 | 6 | 1.3 | 25 | 4.8 | 144 | 5.7 | |||
| Other | 72 | 2.1 | 11 | 2.4 | 12 | 2.3 | 49 | 1.9 | |||
| Education | 6 | 62.29 | 0.000 | ||||||||
| Less than high school or high school | 500 | 14.3 | 40 | 8.8 | 43 | 8.3 | 417 | 16.6 | |||
| Some college | 1114 | 31.9 | 137 | 30.1 | 188 | 36.4 | 789 | 31.3 | |||
| College graduate | 1273 | 36.5 | 171 | 37.6 | 223 | 43.2 | 879 | 34.9 | |||
| Postgraduate degree or higher | 602 | 17.3 | 107 | 23.5 | 62 | 12.0 | 433 | 17.2 | |||
| Income | 8 | 45.37 | 0.000 | ||||||||
| Under $25,000 | 536 | 15.4 | 63 | 14.0 | 96 | 18.5 | 377 | 15.0 | |||
| $25,000-$49,999 | 873 | 25.1 | 100 | 22.3 | 145 | 28.0 | 628 | 24.9 | |||
| $50,000-$74,999 | 746 | 21.4 | 85 | 18.9 | 138 | 26.6 | 523 | 20.8 | |||
| $75,000-$99,999 | 615 | 17.6 | 83 | 18.5 | 78 | 15.1 | 454 | 18.0 | |||
| $100,000 or more | 715 | 20.5 | 118 | 26.3 | 61 | 11.8 | 536 | 21.3 | |||
| Current BMI Category | 6 | 450.35 | 0.000 | ||||||||
| Underweight | 188 | 5.4 | 1 | .2 | 10 | 1.9 | 177 | 7.0 | |||
| Normal Weight | 1160 | 33.1 | 33 | 7.2 | 217 | 41.8 | 910 | 36.0 | |||
| Overweight | 1018 | 29.1 | 86 | 18.9 | 151 | 29.1 | 781 | 30.9 | |||
| Obese | 1138 | 32.5 | 336 | 73.7 | 141 | 27.2 | 661 | 26.1 | |||
| Highest Ever BMI Category | 6 | 417.86 | 0.000 | ||||||||
| Underweight | 70 | 2.0 | 0 | 0 | 6 | 1.2 | 64 | 2.5 | |||
| Normal Weight | 684 | 19.5 | 5 | 1.1 | 111 | 21.4 | 568 | 22.5 | |||
| Overweight | 865 | 24.7 | 17 | 3.7 | 146 | 28.1 | 702 | 27.8 | |||
| Obese | 1645 | 46.9 | 418 | 91.7 | 256 | 49.3 | 971 | 38.4 |
Sample 1
Sample 1 comprised 456 members of the Obesity Action Coalition (OAC). The OAC is a national nonprofit organization of 54,000 + adults who support individuals affected by obesity through education and advocacy. Participants who reported struggling with their weight on OAC's internal membership survey (N = 2,663) were invited to participate in a survey about their experiences related to body weight and opinions about weight stigma, advertised via emails and announcements from the OAC's membership newsletter or website in September and October 2015. Upon survey completion, participants could choose to be entered into a random drawing for a one-in-ten chance of winning a $25 gift certificate to a national online retailer. Details about this sample and recruitment were reported previously (31).
Sample 2
Sample 2 comprised 519 individuals from Mechanical Turk (MTurk), a reliable and psychometrically sound online data source shown to be similar in diversity relative to other panels and adequate for measuring attitudes related to body weight and shape (32, 33). Participants responded to a posting for a survey on attitudes about body weight. Data were collected in September and October 2016. Participants were compensated $0.75 (credited to their MTurk account), which is consistent with compensation practices for short survey research with MTurk (32). A total of 618 individuals entered the survey, but 99 were excluded for missing demographic or anthropometric questions (e.g., sex, height, weight).
Sample 3
Sample 3 comprised 2,529 adults drawn from a national online survey panel administered by Survey Sampling International LLC (SSI) (Shelton, Connecticut). SSI's panel includes more than 2 million active research respondents derived from 3,400 + sources. SSI provides a variety of incentives for participation, including research feedback, charitable donations, and monetary and points rewards. Quotas were established for sex, income groups, and race to approximate US Census characteristics. Data collection occurred in July 2015 as part of a larger online study on weight and health (advertised to participants as a survey on attitudes related to body weight and health behaviors) reported elsewhere (20).
Measures
In all three samples, participants completed self-report measures to assess demographic characteristics, weight status, dieting behavior, history of experienced weight stigma, and internalization of weight bias. These measures are described below.
WBI
WBI was measured by using the Modified Weight Bias Internalization Scale (WBIS-M) (11, 12, 29), which assessed the extent to which participants blame themselves for stigma, apply negative weight-based stereotypes to themselves, and judge themselves negatively because of their body weight (combined sample α = 0.89; sample 1 α = 0.85; sample 2 α = 0.88; sample 3 α = 0.90). Participants responded to 10 items on a scale of one (strongly disagree) to seven (strongly agree). To examine differences in WBI, we constructed a variable to identify participants who scored (1) 1 standard deviation (SD) below the mean on internalization, (2) at the mean on internalization, and (3) 1 SD above the mean on internalization. We used the mean and SD from sample 3 (M = 3.36; SD = 1.51) to construct this variable given that this sample was a large, diverse national sample of the general population of US adults. Low internalization (1 SD below the mean) corresponded to WBIS-M scores ≤ 1.85, and high internalization (1 SD above the mean) corresponded to WBIS-M scores ≥ 4.87.
Demographics
Participants in all samples answered questions regarding their age, sex, race, education, income, and marital status.
Anthropometrics
Participants self-reported their height, weight, and highest-ever weight (excluding pregnancy). BMI was calculated from self-reported height and weight. For subjective weight status, participants indicated how they would describe themselves (“underweight,” “about the right weight,” “overweight,” or “obese”). While continuous BMI was used in regression analyses, BMI status was stratified into weight categories by using clinical guidelines from the Centers for Disease Control and Prevention (Tables 1 and 3) (34).
Weight stigma
The history of experienced weight stigma was measured by using three (yes/no) questions in which participants indicated whether they had ever been teased, treated unfairly, or discriminated against because of their weight (35). Participants were coded as having experienced weight stigma if they answered “yes” to any of these questions.
Dieting behavior
Participants indicated whether they had tried to intentionally lose weight by dieting in the past year (yes/no) and what their current weight goals were (“lose weight,” “gain weight,” “stay the same weight,” and “I am not trying to do anything about my weight”) (36).
Statistical analysis
All analyses were performed using SPSS Statistics version 22.0 (IBM Corp., Armonk, New York). One-way analyses of variance (ANOVAs) and χ2 tests compared sample characteristics (Table 1) and weight variables (Table 2) across the three samples. One-way ANOVAs compared these variables at low versus high levels of WBI (Table 3). Linear regression was used to assess predictors of WBI (Table 4).
| Full sample, N = 3,504 | Sample 1, n = 456 | Sample 2, n = 519 | Sample 3, n = 2,529 | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| M | SD | M | SD | M | SD | M | SD | df | F | P | |
| Weight bias internalization (WBI) | 3.53 | 1.59 | 4.72 | 1.45 | 3.31 | 1.66 | 3.36 | 1.51 | 2, 3,501 | 160.46 | 0.000 |
| n | % | n | % | n | % | n | % | df | Χ2 | P | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Subjective weight status | 6 | 666.6 | 0.000 | ||||||||
| Underweight | 156 | 4.5 | 1 | .2 | 24 | 4.6 | 131 | 5.2 | |||
| About the right weight | 1,338 | 38.2 | 53 | 11.6 | 215 | 41.4 | 1,070 | 42.3 | |||
| Overweight | 1,513 | 43.2 | 165 | 36.2 | 215 | 41.4 | 1,133 | 44.8 | |||
| Obesity | 487 | 13.9 | 235 | 51.5 | 62 | 11.9 | 190 | 7.5 | |||
| Experienced weight stigma | 2 | 362.51 | 0.000 | ||||||||
| Any | 1,735 | 49.5 | 410 | 89.9 | 270 | 52.0 | 1,055 | 41.7 | |||
| None | 1,767 | 50.4 | 45 | 9.9 | 249 | 48.0 | 1,473 | 58.2 | |||
| Teased about weight | 2 | 392.02 | 0.000 | ||||||||
| Yes | 1,592 | 45.4 | 397 | 87.1 | 255 | 49.1 | 940 | 37.2 | |||
| No | 1,906 | 54.4 | 58 | 12.7 | 264 | 50.9 | 1,584 | 62.6 | |||
| Treated unfairly because of weight | 2 | 476.15 | 0.000 | ||||||||
| Yes | 1,151 | 32.8 | 351 | 77.0 | 167 | 32.2 | 633 | 25.0 | |||
| No | 2,348 | 67.0 | 103 | 22.6 | 352 | 67.8 | 1,893 | 74.9 | |||
| Discrimination because of weight | 2 | 422.66 | 0.000 | ||||||||
| Yes | 923 | 26.3 | 298 | 65.4 | 136 | 26.2 | 489 | 19.3 | |||
| No | 2,577 | 73.5 | 157 | 34.4 | 383 | 73.8 | 2,037 | 80.5 | |||
| Tried to lose weight in last year | 2 | 224.87 | 0.000 | ||||||||
| Yes | 2,514 | 71.7 | 434 | 95.2 | 350 | 67.4 | 1,730 | 68.4 | |||
| No | 377 | 10.8 | 17 | 3.7 | 169 | 32.6 | 191 | 7.6 | |||
| Current weight goals | 6 | 363.41 | 0.000 | ||||||||
| Lose weight | 2,087 | 59.6 | 409 | 89.7 | 277 | 53.4 | 1,401 | 55.4 | |||
| Gain weight | 560 | 16.0 | 3 | .7 | 37 | 7.1 | 520 | 20.6 | |||
| Stay the same weight | 610 | 17.4 | 25 | 5.5 | 113 | 21.8 | 472 | 18.7 | |||
| I am not trying to do anything about my weight | 228 | 6.5 | 11 | 2.4 | 89 | 17.1 | 128 | 5.1 |
| BMI | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Underweight | Normal weight | Overweight | Obesity | ||||||||
| n | % | n | % | n | % | n | % | df | Χ2 | P | |
| BMI & subjective weight status concordance | 9 | 2,292.31 | 0.000 | ||||||||
| Underweight | 56 | 29.9 | 80 | 6.9 | 9 | 0.9 | 11 | 1.0 | |||
| About the right weight | 97 | 51.9 | 855 | 73.9 | 344 | 33.8 | 42 | 3.7 | |||
| Overweight | 32 | 17.1 | 220 | 19.0 | 645 | 63.4 | 616 | 54.4 | |||
| Obesity | 2 | 1.1 | 2 | 0.2 | 19 | 1.9 | 464 | 41.0 | |||
| −1SD, WBI ≤ 1.85 | M, WBI: 1.86-4.87 | +1SD. WBI ≥ 4.88 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| M | SD | M | SD | M | SD | df | F | p | |
| Age | 48.12a | 17.19 | 43.73b | 15.97 | 41.47c | 13.47 | 2, 3483 | 32.73 | 0.000 |
| BMI | 24.27a | 3.92 | 27.73b | 6.47 | 32.34c | 9.36 | 2, 3384 | 243.60 | 0.000 |
| Highest BMI | 27.11a | 6.09 | 31.69b | 9.64 | 39.08c | 14.21 | 2, 3268 | 238.27 | 0.000 |
| N | % | N | % | N | % | df | χ2 | p | |
|---|---|---|---|---|---|---|---|---|---|
| Sex | |||||||||
| Male | 336 | 52.6 | 968 | 46.9 | 223 | 27.9 | 2 | 110.44 | 0.000 |
| Female | 303 | 47.4 | 1097 | 53.1 | 577 | 72.1 | |||
| Race | 8 | 38.37 | 0.000 | ||||||
| White | 409 | 64.1 | 1408 | 68.3 | 621 | 77.8 | |||
| Black | 87 | 13.6 | 224 | 10.9 | 64 | 8.0 | |||
| Asian | 35 | 5.5 | 114 | 5.5 | 26 | 3.3 | |||
| Latino | 90 | 14.1 | 270 | 13.1 | 76 | 9.5 | |||
| Other | 17 | 2.7 | 44 | 2.1 | 11 | 1.4 | |||
| Education | 6 | 14.56 | 0.024 | ||||||
| Less than high school or high school | 99 | 15.6 | 290 | 14.1 | 111 | 13.9 | |||
| Some college | 182 | 28.6 | 652 | 31.7 | 280 | 35.1 | |||
| College graduate | 221 | 34.7 | 763 | 37.1 | 289 | 36.2 | |||
| Postgraduate degree or higher | 134 | 21.1 | 350 | 17.0 | 118 | 14.8 | |||
| Income | 8 | 26.88 | 0.001 | ||||||
| Under $25,000 | 74 | 11.7 | 312 | 15.2 | 150 | 18.8 | |||
| $25,000-$49,999 | 163 | 25.7 | 488 | 23.8 | 222 | 27.9 | |||
| $50,000-$74,999 | 139 | 21.9 | 453 | 22.1 | 154 | 19.3 | |||
| $75,000-$99,999 | 114 | 18.0 | 361 | 17.6 | 140 | 17.6 | |||
| $125,000 or more | 145 | 22.8 | 439 | 21.4 | 131 | 16.4 | |||
| Current BMI Category | 6 | 449.45 | 0.000 | ||||||
| Underweight | 37 | 5.8 | 110 | 5.3 | 41 | 5.1 | |||
| Normal Weight | 351 | 54.9 | 670 | 32.4 | 139 | 17.4 | |||
| Overweight | 205 | 32.1 | 647 | 31.3 | 166 | 20.8 | |||
| Obese | 46 | 7.2 | 638 | 30.9 | 454 | 56.8 | |||
| Highest BMI Category | 6 | 403.18 | 0.000 | ||||||
| Underweight | 10 | 1.6 | 42 | 2.2 | 18 | 2.4 | |||
| Normal Weight | 234 | 38.4 | 385 | 20.1 | 65 | 8.8 | |||
| Highest BMI Category | |||||||||
| Overweight | 225 | 36.9 | 543 | 28.4 | 97 | 13.1 | |||
| Obese | 141 | 23.1 | 945 | 49.3 | 559 | 75.6 | |||
| Subjective Weight | 6 | 872.67 | 0.000 | ||||||
| Underweight | 30 | 4.7 | 107 | 5.2 | 19 | 2.4 | |||
| About the right weight | 474 | 74.3 | 750 | 36.4 | 114 | 14.3 | |||
| Overweight | 126 | 19.7 | 1020 | 49.5 | 367 | 46.0 | |||
| Obese | 8 | 1.3 | 182 | 8.8 | 297 | 37.3 | |||
| Experienced Weight Stigma | 2 | 635.1 | 0.000 | ||||||
| Any | 119 | 18.6 | 945 | 45.8 | 671 | 84.0 | |||
| None | 520 | 81.4 | 1119 | 54.2 | 128 | 16.0 | |||
| Teased about weight | 2 | 558.74 | 0.000 | ||||||
| Yes | 112 | 17.6 | 855 | 41.5 | 625 | 78.2 | |||
| No | 525 | 82.4 | 1207 | 58.5 | 174 | 21.8 | |||
| Treated unfairly about weight | 2 | 683.95 | 0.000 | ||||||
| Yes | 44 | 6.9 | 560 | 27.2 | 547 | 68.5 | |||
| No | 594 | 93.1 | 1502 | 72.8 | 252 | 31.5 | |||
| Discrimination due to weight | 2 | 611.92 | 0.000 | ||||||
| Yes | 28 | 4.4 | 429 | 20.8 | 466 | 58.3 | |||
| No | 610 | 95.6 | 1634 | 79.2 | 333 | 41.7 | |||
| Tried to Lose Weight in Last Year | 2 | 126.54 | 0.000 | ||||||
| Yes | 297 | 71.7 | 1496 | 87.1 | 721 | 94.9 | |||
| No | 117 | 28.3 | 221 | 12.9 | 39 | 5.1 | |||
| Current Weight Goals | 6 | 646.99 | 0.000 | ||||||
| Lose weight | 132 | 20.7 | 1272 | 62.0 | 683 | 85.7 | |||
| Gain weight | 201 | 31.5 | 316 | 15.4 | 43 | 5.4 | |||
| Stay the same weight | 240 | 37.6 | 319 | 15.6 | 51 | 6.4 | |||
| I am not trying to do anything about my weight | 65 | 10.2 | 143 | 7.0 | 20 | 2.5 |
| B | β | P | R2 | F | df | P | |
|---|---|---|---|---|---|---|---|
| 0.38 | 113.24 | 14, 2,555 | 0.000 | ||||
| Sample (ref sample 3) | |||||||
| Sample 1 | 0.30 | 0.07 | 0.001 | ||||
| Sample 2 | −0.31 | −0.08 | 0.000 | ||||
| Age | −0.02 | −0.19 | 0.000 | ||||
| Female | 0.10 | 0.03 | 0.062 | ||||
| Race (ref white) | |||||||
| Asian | −0.01 | 0.00 | 0.928 | ||||
| Black | −0.29 | −0.05 | 0.001 | ||||
| Latino | −0.20 | −0.04 | 0.015 | ||||
| Income | −0.06 | −0.05 | 0.003 | ||||
| Education | −0.07 | −0.04 | 0.019 | ||||
| BMI | 0.02 | 0.09 | 0.003 | ||||
| Highest BMI | −0.01 | −0.04 | 0.250 | ||||
| Subjective weight status | 0.63 | 0.29 | 0.000 | ||||
| Experienced stigma | 0.94 | 0.30 | 0.000 | ||||
| Tried to lose weight in the past year | 0.38 | 0.08 | 0.000 |
Results
Sample characteristics
Tables 1 and 2 display characteristics for the full sample and differences by sample for demographics, anthropometrics, WBIS-M scores, experienced weight stigma, and dieting behavior. On average, participants were 44.01 (SD = 15.81) years old. The mean BMI was 36.41 in sample 1 (73.7% with obesity), 27.62 in sample 2 (27.2% with obesity), and 26.64 in sample 3 (26.1% with obesity). Participants identified as white (69.6%), Hispanic or Latino (12.4%), black (10.7%), Asian (5.0%), or other (2.1%). More than 80% of participants reported educational attainment of some college or higher. In total, 49.5% of participants reported that they had experienced weight stigma.
Age, BMI, and highest BMI were significantly higher in the OAC sample (sample 1) compared with sample 2 (MTurk) and sample 3 (SSI). Experienced weight stigma was higher in sample 1 (89.9% experienced stigma) relative to samples 2 (52%) and 3 (41.7%). Similarly, more participants in sample 1 endorsed high WBIS-M scores (52.2%; M = 4.72) compared with sample 2 (20%; M = 3.31) and sample 3 (18%; M = 3.36). No differences in WBIS-M scores emerged between samples 2 and 3; approximately 60% of participants in these samples endorsed the mean level of internalization. We compared subjective weight status with BMI categories to examine concordance rates (Table 2). Participants with an underweight or normal-weight BMI were more likely to perceive their subjective weight status as “about right.” Participants with a BMI in the overweight or obesity range were likely to perceive their subjective weight as overweight or obesity.
More participants in sample 1 reported dieting in the past year (95.2%) relative to samples 2 (67.4%) and 3 (68.4%), and participants in sample 1 were more likely to be currently trying to lose weight (89.7%) relative to samples 2 (53.4%) and 3 (55.4%). These results suggest that participants in sample 1 had elevated characteristics relevant to WBI (higher BMI, more dieting, and more weight stigma) compared with samples 2 and 3.
Differences in WBI across sociodemographic characteristics
Table 3 shows differences in sample demographics, anthropometrics, experienced stigma, and dieting behavior among individuals scoring 1 SD below the mean (low internalization), at the mean, or 1 SD above the mean (high internalization) on WBIS-M. Age differed linearly across internalization categories with younger adults (M = 41.47 years) endorsing high internalization and older participants (M = 48.12 years) endorsing low internalization. Men and women were similarly distributed in both the low internalization (52.6% men, 47.4% women) and mean internalization categories (46.9% men, 53.1% women), but among individuals with the highest WBIS-M scores, 72.1% were women. Regarding racial differences, fewer black (8%) and Latino (9.5%) participants endorsed high internalization relative to low internalization (black = 13.6%, Latino = 14.1%) or the mean level of internalization (black = 10.9%, Latino = 13.1%). The proportion of white participants increased at each level of internalization (low WBIS-M = 64.1%, mean WBIS-M = 68.3%, high WBIS-M = 77.8%).
BMI and highest-ever BMI differed linearly across levels of WBI; individuals with higher BMIs endorsed high internalization on the WBIS-M (BMI: M = 32.34; highest-ever BMI: M = 39.08), and individuals with lower BMIs endorsed low internalization (BMI: M = 24.27; highest-ever BMI: M = 27.11). Among individuals with the highest WBIS-M scores, 56.8% had a BMI in the obesity range, 20.8% had an overweight BMI, and 17.4% had a normal BMI. Subjective weight status followed a similar trajectory; individuals with low internalization were primarily those who felt they were “about the right weight” (74.3%), while participants endorsing high internalization primarily perceived themselves as having overweight (46%) or obesity (37.3%). While most participants indicated they had dieted in the past year, this percentage increased linearly across WBI levels; 71.7% of individuals with low WBIS-M scores reported dieting in the past year relative to 94.9% who endorsed high WBI. Similarly, nearly all participants with high internalization (85.7%) indicated they were currently trying to lose weight versus 20.7% with low internalization.
Experienced weight stigma followed a linear pattern; 18.6% of participants reported experienced weight stigma at the low level of internalization, 45.8% experienced stigma at the mean for internalization, and 84% at the high level of internalization. Teasing was the most common type of experienced stigma, followed by unfair treatment and discrimination. Among individuals who endorsed low internalization, 17.6% reported experiencing weight-based teasing, compared with 78.2% of individuals who reported teasing in the high internalization category. A similar pattern emerged for unfair treatment (6.9% of the low internalization category vs. 68.5% of the high internalization category) and discrimination (4.4% vs. 58.3%).
Predicting WBI
By using a linear regression, we predicted WBI by sample, demographics, anthropometrics, and dieting behavior. Sample 3 was used as the reference group for samples 1 and 2 because it was a large and diverse national sample. Men were the reference group for women, and white was the reference group for race/ethnicity. Table 4 displays the coefficients for each predictor of WBI for the combined samples. Accounting for sample, age, sex, race, income, education, BMI, highest-ever BMI, subjective weight status, experienced stigma, and dieting behavior in the past year explained 38% of the variance in WBIS-M: F (14, 2,555) = 113.24; P < 0.001. Results showed that being in sample 1 (relative to sample 3) predicted WBIS-M scores (B = 0.30; P = 0.001), with higher scores present in the OAC sample versus the national SSI sample. Internalization was lower in the MTurk sample relative to the SSI sample (B = −0.31; P < 0.001).
Gender was not associated with WBIS-M scores (B = 0.10; P = 0.062). Being black or Latino (relative to white) was negatively associated with internalization (B = −0.29; P = 0.001; B = −0.20; P = 0.015), suggesting that black participants and Latino participants had lower internalization than white participants. Income (B = −0.06; P = 0.003) and education (B = −0.07; P = 0.019) were negatively associated with internalization. BMI (B = 0.02; P = 0.003) and subjective weight status (B = 0.63; P < 0.001) were both positively predictive of higher internalization, although highest-ever BMI did not predict WBIS-M scores. Experienced stigma (B = 0.94; P < 0.001) and dieting behavior in the past year (B = 0.38; P < 0.001) predicted higher WBIS-M scores. Collectively, these results indicate increased vulnerability to high WBI among white adults with lower education or income, high BMI and self-perceived high body weight, active engagement in weight loss efforts, and a significant history of experienced weight stigma. By using the standardized regression weight (beta) as a measure of comparative effect size, subjective weight status (β = 0.29) and experienced stigma (β = 0.30) were the strongest predictors of WBI (Table 4).
Discussion
Our findings show that internalized weight bias is prevalent among US adults. At least 44% of adults across all three samples endorsed the mean level of WBI. High levels of WBI (corresponding to 1 SD above the WBIS-M mean) were endorsed by approximately one in five adults in the general population samples and by 52% in the OAC sample, suggesting that a considerable portion of the population may be at risk for internalization, especially if they share characteristics with the OAC sample, such as high BMI, considerable exposure to weight stigma, and trying to lose weight.
Compared with previous research, the mean level of WBI in sample 1 (M = 4.72) was similar to treatment-seeking samples of adults with obesity (M = 4.6) (37), binge eating disorder (M = 4.75) (30), and bariatric surgery candidates (M = 4.54) (27) but higher than the mean level of WBI observed in a recent study of adults with obesity enrolled in a weight loss trial (M = 3.6) (18). Because of the lack of previous research assessing WBI in general population samples, it is difficult to compare the mean WBIS-M scores observed in samples 2 (M = 3.31) and 3 (M = 3.36) with similarly diverse populations. As expected, these means were lower than levels of WBI in treatment-seeking samples of individuals with obesity noted earlier. However, comparisons with previously published community samples are less clear; WBIS-M means in our samples were slightly lower compared with levels of WBI reported in a community sample of 198 adults with obesity (M = 3.95) (11), slightly higher than mean scores in an MTurk sample of 148 adults (M = 3.27) (12), and identical to mean levels of WBI in a community sample of 81 women with obesity (M = 3.31) (21). As these studies reflect smaller, more homogeneous samples, it will be important for additional research to assess levels of WBI in larger, diverse populations comparable with the present study.
Our findings highlight a pattern of sociodemographic characteristics among individuals who internalize weight bias. Similar to some previous research (11), women and men in our samples were similarly distributed at low and mean levels of internalization. However, among adults with the highest levels of internalization, 72% were women, supporting other studies that have shown an increased vulnerability to WBI among women compared with men (15, 38). Women's vulnerability to higher WBI may be attributed to a potentially increased risk of exposure to weight stigmatization relative to men (7, 39). Compared with gender, little work has examined WBI in relation to socioeconomic status or race/ethnicity. In our study, income and education were negatively associated with internalization, similar to sociodemographic correlates of WBI documented in a national German sample (15). Regarding race, black participants and Latino participants had lower WBI than white participants. As few studies have systematically compared exposure to weight stigma in white participants compared with racial minorities (40), more research is needed to examine WBI across different racial/ethnic groups.
Similar to previous research (15, 16, 41), our study found BMI and subjective views about body weight to be positively associated with WBI. Most individuals with high WBIS-M scores had a BMI in the obesity range (56.8%). However, it is noteworthy that among individuals reporting high WBI, 20.8% had an overweight BMI and 17.4% had a normal BMI. Thus, people across body weight categories may be vulnerable to high WBI. These findings support recent evidence of WBI observed among women with normal weight and underweight (16, 41) and highlight the importance of studying WBI in populations with diverse body weights.
Finally, our study provides new insights about links between experienced and internalized weight stigma. Our findings show that 84% of adults with high WBIS-M scores reported a history of experienced weight stigma. While previous work has observed positive associations between experienced and internalized stigma (16), our findings point to the importance of considering different types of experienced stigma. Specifically, severe forms of weight stigma were experienced by a considerable percentage of individuals with high WBI (e.g., 68.5% reported unfair treatment because of their weight, 58.3% reported weight discrimination); however, weight-based teasing was the most commonly reported form of experienced stigma (78.2%) among individuals with high levels of internalization. This finding suggests that seemingly less severe forms of experienced stigma (e.g., teasing vs. discrimination) are pronounced in people who internalize weight bias. It may be that individuals who are teased about their weight from peers or family members have a heightened risk for blaming themselves compared with those who face weight discrimination in less personal scenarios (e.g., employment). Alternatively, individuals high in WBI may have heightened awareness of stigma and, in turn, be more likely to perceive weight stigma from others. Given that subjective weight status and experienced stigma were the strongest predictors of WBI, it will be important for future research to examine these relationships and determine whether WBI incurs different health consequences depending on personal views about weight as well as the type, severity, or frequency of experienced stigma.
Several limitations of this study should be noted. First, all measures were self-reported, including weight and height. However, research has demonstrated good concordance rates between online self-reported weight and measured weight (42). Second, although the identical measure of internalized weight bias (WBIS-M) was used across samples, it will be important for future work to compare different measures of this construct. As research on WBI is relatively new, very few measures have been developed or validated. Preliminary comparisons of the WBIS to another measure, the Weight Self-Stigma Questionnaire (43), are favorable, but longitudinal studies on their stability and predictive validity are needed (44). Third, our samples are not nationally representative and are limited to individuals who use the internet. The OAC sample, while similar in demographic composition to previous research on WBI with treatment-seeking samples, lacks generalizability to more diverse groups who may be at heightened risk for WBI. Finally, the three samples completed the survey at different time points from 2015 to 2016, introducing possible cohort effects.
Strengths of our study include the comparison of identical measures across three large and distinct samples, allowing for novel comparisons of WBI that are currently absent in research. The diverse national panel in sample 3 approximates US estimates for characteristics such as race, sex, and body weight distributions, allowing for a comprehensive analysis of WBI with improved generalizability to broader groups. The inclusion of weight-related variables, such as type of experienced stigma, provide novel insights on factors relevant to WBI that have previously been unstudied.
Conclusion
This study provides a comprehensive analysis and comparison of internalized weight bias in US adults. Findings indicate that internalized weight bias is prevalent in the general population, present in both women and men, and occurs across diverse body weight categories. Individuals with high levels of WBI are more likely to be white, have a higher BMI and self-perceived higher body weight, have previous experiences of weight stigma (especially teasing), and be actively engaged in efforts to lose weight. This evidence provides an initial foundation to better understand characteristics of individuals with high WBI. Given recent calls for efforts to address weight stigma in weight loss treatment (31, 45), our findings have implications for interventions targeting obesity and weight management. Efforts to improve weight-related health are likely to be more successful with a clearer understanding of who is at risk for high WBI and who may benefit most from clinical support to prevent internalization from contributing to emotional distress or interfering with healthy behaviors.
Acknowledgments
The authors would like to thank Joseph Nadglowski, James Zervios, and Lindsay McGhay at the OAC for their assistance in distributing the survey and compiling internal membership data.
