Weight Stigma, Depression and the Role of Stress

Eine dicke Frau sitzt laechelnd in einer sonnigen Wohnkueche am Tisch und fuellt an einem Laptop ein Online-Formular aus, im Hintergrund macht ein Mann Kaffee. KI-generiertes Motiv, keine realen Personen.

Weight stigma goes with more depressive and anxiety symptoms. A 2025 US survey of 1,993 adults asks how: does perceived stress sit in between? The authors find that it carries about 37 to 38 percent of the statistical link. The study is cross-sectional, so “carries” describes a model, not an observed sequence of events.

The paper is “Weight stigma and mental health symptoms: mediation by perceived stress” by David G. Figueroa, William D. Murley, Jordan E. Parker, Jeffrey M. Hunger and A. Janet Tomiyama (UCLA and Miami University), a Brief Research Report published on 2 July 2025 in Frontiers in Psychiatry (open access). The authors state that the study was preregistered.

How the data were collected

An online survey ran on Qualtrics between December 2019 and January 2020. Recruitment used quotas matched to US Census benchmarks for age, gender, race and ethnicity, income and region. Of 2,022 respondents, 1,993 remained after exclusions: failed attention checks, implausible height or weight, or a BMI below 12 or above 70. Mean age was 47.2 years (SD 17.3), 51 percent women, 63 percent White, 15.5 percent Hispanic or Latino, 13 percent Black or African American.

Chart: mediation paths from weight stigma via perceived stress to depressive and anxiety symptoms (Figueroa et al. 2025, Table 3)
Chart: Fatosphere Editorial. Source: Figueroa et al., Front Psychiatry 2025;16:1587105, Tabelle 3. Read on 07.10.2026.

Weight stigma was measured with two single items, averaged: how often respondents were treated with less respect, harassed or discriminated against because of their weight (experienced), and how often they worried about being stereotyped or mistreated because of their weight (anticipated). Both used a 4-point scale from “not at all” to “often”. Perceived stress came from a 4-item version of the Perceived Stress Scale (past month, 5-point scale). Depressive and anxiety symptoms came from 4-item PROMIS short forms (past seven days). Missing data (up to 11.8 percent per variable) were handled with Bayesian multiple imputation. Age, BMI, region, gender, income, education and race and ethnicity were covariates.

The results

Weight stigma was linked to more stress, depression and anxiety. Partial correlations (age, BMI and income held constant, non-imputed data): weight stigma with perceived stress .25, with depressive symptoms .38, with anxiety symptoms .37. Perceived stress correlated .66 with depressive and .62 with anxiety symptoms.

Stress accounted for about 37 to 38 percent of the link. The table shows the two mediation models as reported (unstandardised coefficients, 95 percent credible intervals in brackets).

PathDepressive symptomsAnxiety symptoms
Stigma to stress (a)0.24 (0.20 to 0.29)0.24 (0.20 to 0.29)
Stress to symptoms (b)0.79 (0.75 to 0.84)0.70 (0.65 to 0.74)
Indirect effect0.19 (0.16 to 0.23)0.17 (0.14 to 0.20)
Direct effect (c’)0.31 (0.26 to 0.35)0.29 (0.25 to 0.34)
Total effect (c)0.50 (0.44 to 0.56)0.46 (0.41 to 0.52)
Share mediated38 percent37 percent

I checked the arithmetic: 0.24 times 0.79 gives 0.19, 0.24 times 0.70 gives 0.17, and 0.19 divided by 0.50 and 0.17 divided by 0.46 give the stated shares. The authors call the indirect effect small by Cohen’s benchmark (υ = .02).

Both direct effects stayed after adjusting for BMI. The authors read this as weight stigma being more than a marker of body size. In an exploratory regression, younger age and higher BMI went with more reported stigma, and Hispanic or Latino and Asian respondents reported less than Black respondents (the reference group).

Reading it carefully

The design cannot order the variables. The authors say so: no causal conclusions are possible, and a mediation model across one time point needs a compelling case for the order. They argue from other, experimental work that stigma raises stress and stress raises symptoms, and call their result preliminary. In my reading, the data alone would fit other orders as well, for example people with more depressive symptoms reporting more stress and more stigma. The paper does not test that reversed model.

Stress and anxiety are measured close together. One of the stress items reads “How often have you felt nervous and stressed?” and an example item of the anxiety scale reads “I found it hard to focus on anything other than my anxiety”. The b path for anxiety (0.70) and the correlation of .62 are large. In my reading, content overlap between the two scales can inflate the mediated share. That is my observation, not the authors’.

“Weight stigma” here is two single items. The composite mixes experienced stigma with worry about future stigma, and each is one question. The authors cite earlier work showing it correlates with validated questionnaires, but that is a different publication on the same dataset. Mental health was also measured with brief 4-item forms, and everything is self-report. The authors name both limits.

The data are from before 2020 and were used before. Collection ended in January 2020, before the pandemic. The covariates were chosen based on earlier work with the same dataset, so this is a preregistered analysis of existing data, not a fresh sample.

Quota sampling is not random sampling. The discussion says the census-matched sample gives “high generalizability” to the US population. Quotas match margins such as age and income, and the text I read mentions no reweighting after the exclusions. I would read that claim as plausible, not shown. The text I read does not report the BMI distribution of the sample, so how many respondents were in larger bodies is unknown.

Framing. The paper builds on the COBWEBS model, which proposes that stigma-related stress leads to weight gain through eating and cortisol. This study measures none of that, and I report the framing, not the claim. The authors conclude that stress-reduction strategies may be needed in interventions “if replicated”, and mention mindfulness work as promising, citing other studies. No intervention was tested here, and this article does not evaluate any treatment or programme.

The authors report no conflicts of interest and funding from the National Science Foundation and the NIH. They state that no generative AI was used.

Why it belongs here

The study tests a mechanism, not an outcome rate: stigma as a stressor. That fits a plain reading of the data. Being treated badly, or expecting to be, goes with feeling more stressed, and stress goes with more symptoms. Whether it works in that order, in people in larger bodies specifically, needs longitudinal data. For related studies in this series see the Gerend survey on coping and depression in 2,632 adults, which asks what may buffer the link, and the Hong Kong longitudinal study, which follows people over time and finds no vicious cycle. The survey is US-only and says nothing about Germany or the UK.

Source: Figueroa DG, Murley WD, Parker JE, Hunger JM, Tomiyama AJ. “Weight stigma and mental health symptoms: mediation by perceived stress.” Frontiers in Psychiatry 2025;16:1587105, published 2 July 2025, doi.org/10.3389/fpsyt.2025.1587105. Full text read on 7 October 2026.