Can an upward social media comparison lower self-esteem?

Three minutes with one fictional profile was enough to move one measure of how people felt about themselves. But the effect was small, specific and easy to overstate.

In a randomized experiment, undergraduates who saw a profile with heavy social activity reported lower state self-esteem than those who saw the same person with little activity. The gap was 0.21 points on a five-point scale.

That is evidence that an upward social-media comparison can change an immediate self-evaluation. It is not evidence that every polished post causes depression, that all scrolling is harmful or that one platform moves every user in the same direction.

The more useful question is not simply how much social media someone uses. It is what kind of comparison a particular feed makes possible—and whether the person sees a path forward or only a widening gap.

One profile created a small self-esteem difference

The anchor experiment came from a 2014 paper by Erin Vogel and colleagues. The researchers recruited 128 undergraduates, 94 of them women, with a median age of 19. Each participant spent three minutes viewing a fictional Facebook-style profile that supposedly belonged to another student of the same sex at the same university.

The profile changed along two dimensions. Personal posts portrayed healthy or unhealthy behavior. Network activity showed either many or few likes and comments. The name, interests, friend count, profile picture and neutral filler posts stayed the same.

Participants were randomly assigned to one of the four combinations and then completed a 20-item state-self-esteem scale.

The high-activity network condition produced lower mean state self-esteem than the low-activity condition: 3.53 versus 3.74 on a five-point scale. The standardized difference was small, d = .34.

That is a controlled contrast. It is more informative than a survey asking whether frequent users also feel worse, because random assignment makes the profile difference a credible cause of the immediate difference between the groups.

It is still not a before-and-after decline. The experiment did not measure each participant before the profile, and it had no neutral condition in which people saw no comparison at all.

The obvious positive posts did not move the main outcome

The study’s most important null result complicates the familiar highlight-reel story.

Changing the target’s personal posts from unhealthy behavior to healthy food, exercise, volunteering and an active holiday did not have a statistically significant main effect on state self-esteem. The interaction between personal content and network activity was also not significant.

What moved the main outcome was the surrounding social proof: the amount of visible activity from other people. The authors proposed several possible reasons. Likes and comments may have been more salient, more quantitative or more credible than self-authored posts. Those explanations were not separately tested.

This distinction matters for product design. A polished image and a visible response from a crowd are not the same signal. Only the network-activity manipulation yielded a statistically significant state-self-esteem contrast in this study. The study did not directly establish that its effect exceeded the personal-content effect.

Daily-life studies point to the comparison mechanism

A newer study followed 200 children and adolescents aged 10 to 14 for 14 days. Every evening, participants reported their use of Instagram, TikTok and YouTube; whether other people seemed better off; and their positive and negative self-worth and affect.

On days with more social-media use than usual, participants also reported more extreme upward comparisons. Those were days when other people appeared to have better lives, do cooler things, or look happier, prettier or more popular.

Upward comparison was associated with worse self-worth and affective wellbeing both within people over time and between people. The statistical models were consistent with upward comparison carrying part of the association between use and positive or negative self-worth.

But the result split by outcome. The researchers did not find a statistically significant daily association between social-media use and affective wellbeing, either directly or indirectly through upward comparison.

This was an intensive diary study, not a randomized feed experiment. A bad day could make other lives look better, just as a comparison-heavy feed could make a day feel worse. The authors explicitly describe reciprocal effects as a live possibility.

New adult evidence makes the answer less absolute

Two studies published in 2025 examined 139 and 413 young adults. They linked Instagram and Facebook use with global and physical self-esteem, depressive symptoms, exposure to upward comparison and the perceived size of the gap to a comparison target.

The models again placed upward comparison in the middle, but the results were mixed. Study 1 supported the indirect pattern for global, but not physical, self-esteem. Study 2’s preregistered self-esteem models and exploratory depressive-symptom models produced mixed indirect paths. Across the reported outcome models, explained variance was about 5% to 9%.

The overall use result was not consistently negative. In some analyses, more frequent social-network use was associated with higher self-esteem. Frequent users also reported less extreme upward comparisons, which partly buffered the negative pattern tied to large perceived gaps.

These studies were cross-sectional. They can show that use, comparison and outcomes travel together in a particular model. They cannot establish which one moved first.

Upward comparison has two different endings

Comparison with someone doing better can create contrast: their success makes your own position feel worse. It can also create assimilation: their success feels relevant, attainable or instructive.

A 2025 measurement study made that difference explicit. In 462 social-media-using undergraduates from one US West Coast university—mostly women—assimilative upward comparison was associated with more favorable affect, more benign envy and higher life satisfaction. Contrastive upward comparison showed the opposite pattern.

That study validated measures rather than randomly assigning a feed. It does not prove that making a target seem attainable will improve wellbeing. It does show why upward is not a complete diagnosis.

The same destination can look like a route or a verdict. The viewer’s relationship to the target, the relevance of the domain, perceived distance, current mood and surrounding feedback may all change which one it becomes.

Why “use social media less” is not the evidence-based answer

The strongest broad synthesis in this set covered 141 studies, roughly 145,000 participants and 897 effect sizes for active and passive social-media use. Most associations with mental health, wellbeing and social support were negligible.

Passive use was associated with slightly more illbeing in general social-media contexts, r = .11, but not inside social-media groups, r = −.01. All pooled within-person effects were negligible at the population level.

That meta-analysis did not isolate upward comparison. Its value here is a boundary: neither scrolling nor posting is a sufficient explanation for an individual outcome. Context and content do more analytical work than a raw minute count.

A practical response should therefore begin with the experience:

• Which account or topic repeatedly creates a contrast rather than a route?
• Is visible popularity doing more work than the actual content?
• Does the comparison change after rest, distance or a different reference group?
• Is the feed also providing information, support or belonging that a simple time limit would erase?

These are diagnostic prompts, not clinical advice. Persistent low mood, distress or depressive symptoms require a broader assessment than any feed audit can provide.

Primary sources

• Vogel et al., randomized social-media profile experiment and correlational study
• Irmer and Schmiedek, 14-day youth diary study
• Le Blanc-Brillon et al., 2025 young-adult studies
• Godard and Holtzman, meta-analysis of 141 active/passive-use studies
• Yang et al., meta-analysis of 73 social-network-use samples
• Huang and colleagues, assimilative and contrastive upward-comparison scale validation

Independent editorial summary. The authors are not affiliated with LifeScore.