A randomized experiment with 1,313 participants tested how people judge someone who lost weight on a GLP-1, someone who lost it by dieting, and someone who didn’t lose any. The drug user came out worst. Here is what the study shows, and where its numbers are thinner than the headlines.
There is a promise buried inside every weight-loss conversation: do this, and people will stop treating you that way. A study published in the International Journal of Obesity on 3 April 2026 put that promise into an experiment, and the promise did not survive.
Erin Standen (Rice University), Sean Phelan (Mayo Clinic) and Janet Tomiyama (UCLA) ran two randomized, between-subjects experiments. Participants read a short description of a fictional person and then rated them. The only thing that changed between groups was how that person’s weight history was described.
What the participants actually read
In both studies the fictional target was the same person: 38 years old, college-educated, gender deliberately left unspecified, described as having had obesity since puberty, currently around 220 pounds (roughly 100 kg). The authors say they built these details around an average-aged, average-weight U.S. adult with the most common age of onset.
Study 1 (N = 607, recruited on Prolific in October 2024) split participants into three groups. The target had either lost 35 pounds (about 16 kg) over the past year using a GLP-1 medication, lost the same 35 pounds through diet and exercise, or had not lost weight at all. In both weight-loss conditions the description added that the person was no longer in the “obese” BMI category.
Study 2 (N = 706, February 2025, no overlap with Study 1) had four groups: regained the weight after stopping a GLP-1, regained it after stopping a diet-and-exercise plan, never lost weight, or lost weight and kept it off with the method unspecified.
Participants then rated the person on ten positive traits (honest, generous, sociable, productive, organized, friendly, outgoing, intelligent, warm, humorous) and eight negative ones (lazy, undisciplined, self-indulgent, unclean, weak-willed, sloppy, insecure, sluggish) on a 1-to-7 scale, rated them separately on “unhealthy”, and answered six questions about whether they would want anything to do with them, adapted from the Interpersonal Attraction Questionnaire. The study materials and the de-identified data are public on OSF.
What they found
Against the diet-and-exercise version of the same person, the GLP-1 version lost on every measure: fewer positive traits, more negative traits, rated more unhealthy, and less socially wanted (all Tukey HSD p < 0.001).
Then came the finding that made the press releases. The authors had predicted that the person who did not lose weight would be judged most harshly of all. That is not what happened. Compared to the GLP-1 user, the person who never lost weight was rated higher in positive traits (p = 0.003), and people were more willing to spend time with them. On willingness to affiliate, all three groups differed from each other: diet and exercise 5.51, no weight loss 5.25, GLP-1 4.99, on a seven-point scale (F(2, 603) = 12.54, p < 0.001).
In plain terms: in this experiment, taking the drug cost more social standing than staying fat.
Where the numbers are thinner than the headline
Two things get lost when this study is summarized in a sentence, and both matter.
First, the “worse than not losing weight at all” result is narrower than it sounds. It holds for positive traits and for willingness to affiliate. It does not hold for the negative traits, and it does not hold for “unhealthy”: on both of those, the GLP-1 user and the person who never lost weight were rated the same (p = 0.26 for each). So people did not call the drug user lazier or unhealthier than a fat person who did nothing. They just liked them less and granted them fewer good qualities.
Second, the effects are small. The gap in willingness to affiliate between the GLP-1 user and the person who never lost weight was 0.26 points on a 1-to-7 scale, with a confidence interval running from 0.02 to 0.51 — that is, just barely clear of zero. The effect size for the whole model was ηp² = 0.04 in Study 1 and 0.01 in Study 2. These are real, statistically significant differences, and they are not large ones.
Third, and this is a limitation the authors state themselves: participants were rating a paragraph about a person who does not exist. That design buys clean causal inference and pays for it with artificiality. All participants were U.S. adults recruited on an online survey platform. Nothing here has been shown for Germany, and nothing here has been measured on actual behaviour towards actual people.
The part nobody quoted
Study 2 is the more interesting half, and it got almost no coverage.
When the weight came back, the difference between the drug and the diet disappeared entirely. Participants rated the person who regained after a GLP-1 and the person who regained after dieting the same way — no difference in positive traits, negative traits or unhealthiness (p > 0.55). What did produce a large difference was regain itself: both regain groups were rated significantly higher in negative traits and more unhealthy than the person who lost weight and kept it off (p < 0.001).
So the “easy way out” penalty is attached to the ongoing state of having taken a shortcut. Once you are fat again, nobody cares how you briefly stopped being fat. You are simply back to being blamed for the weight.
Two words we are not going to use
The Rice press release calls this “the GLP-1 paradox”. The paper does not use the word “paradox” once. It is a framing added afterwards, and it is the wrong frame, because it implies something contradictory happened. Nothing contradictory happened.
Look at what the trait list is actually measuring: lazy, undisciplined, self-indulgent, weak-willed. That is not a list of things people believe about fat bodies. It is a list of things people believe about fat character. The judgement was never really about the number on the scale. It was about whether you are the kind of person who takes the hard road. A drug removes the weight without removing that suspicion — it confirms it. Which is exactly why the diet-and-exercise version of the same person, with the same body, scored best of all three.
There is no paradox in a moral system that rewards suffering. There is only consistency.
What follows from it
We have written before about how the injection moves the shame rather than dissolving it, and about how access to these drugs splits along class lines. This study adds the piece that was missing: a direct comparison with the option of not losing weight at all. And on the measures where it holds, that option came out ahead.
This is not an argument for or against anyone’s medication. It is nobody’s business what you put in your body, and the authors’ own conclusion is that stigma reduction, not behaviour change, is the thing that needs work. It is an argument against the promise. If you are being told that the judgement stops once the weight does, the best available experimental evidence says otherwise. It does not stop. It re-attaches.
Which means the thing worth working on was never the body. It is the standard being applied to it.
Sources
- Standen EC, Phelan SM, Tomiyama AJ. An experimental investigation of the stigmatization of weight loss and regain from GLP-1 receptor agonist use and cessation. International Journal of Obesity 50, 1478–1485 (2026), published 3 April 2026, open access. https://www.nature.com/articles/s41366-026-02061-y
- Study materials and de-identified data: https://doi.org/10.17605/OSF.IO/7FM2R
- Rice University press release, 4 May 2026: https://news.rice.edu/news/2026/glp-1-paradox-rice-study-finds-weight-loss-drugs-may-carry-unexpected-stigma
A note on language: the paper uses “higher-weight” and BMI categories throughout, and treats weight loss as a health goal. We quote its findings, not its framing.