For years, the debate over social media and mental health followed a frustrating pattern. Observational studies would show correlations between heavy social media use and poor wellbeing. Critics would point out that correlation is not causation — maybe unhappy people simply use more social media. Advocates would counter with case studies and anecdotes. Neither side could offer what the other demanded: experimental proof.
That has changed. As of early 2025, we now have three independent meta-analyses of randomized controlled trials — studies in which researchers randomly assigned people to reduce or quit social media and measured what happened. The combined evidence base includes over 30 experiments involving thousands of participants. The results are real, statistically significant, and more nuanced than headlines in either direction suggest.
This post examines what those experiments actually found — not the correlational studies that dominate media coverage, but the controlled experiments that come closest to answering a causal question: if you quit social media, will you feel better?
The Three Meta-Analyses: What 32 Experiments Show
In 2025, Burnell and colleagues at RTI International published a comprehensive meta-analysis in SSM — Mental Health that synthesized 32 randomized controlled trials involving 5,544 participants and 91 effect sizes. All included studies instructed participants to either limit or entirely abstain from social media for a discrete period, with at least one subjective wellbeing outcome measured. The average participant was 23 years old, and the samples skewed female (70%).
Their headline finding: social media restriction yielded a statistically significant but small and heterogeneous effect on wellbeing. In other words, quitting social media does help — but the effect is modest, and it varies considerably from person to person and study to study.
Two other 2025 meta-analyses arrived at compatible conclusions. May, Malouff, and Meynadier synthesized 10 RCTs with 1,491 participants and found that reducing social media use significantly decreased depressive symptoms, with an effect size of g = 0.25 (a small-to-moderate effect) after adjusting for publication bias. Liu and colleagues analyzed 20 RCTs with 56 effect sizes and similarly found a small positive effect on wellbeing — both for positive indicators (life satisfaction, positive affect) and negative indicators (depression, anxiety).
Taken together, these three independent analyses converge on the same conclusion: reducing social media use produces real but modest improvements in mental health for the average person. The effect is not zero, as some skeptics have argued. But it is not the dramatic transformation that quitting-social-media narratives often imply.
The Landmark Individual Experiments
The meta-analyses gain their power from aggregation. But several individual experiments are worth examining in detail, because they reveal how the effects work — and for whom.
Hunt et al. (2018): The "No More FOMO" Study
Melissa Hunt at the University of Pennsylvania conducted one of the first well-controlled experiments on social media limitation. She randomly assigned 143 undergraduates to either limit their use of Facebook, Instagram, and Snapchat to 10 minutes per platform per day (30 minutes total) or continue using social media as usual — for three weeks.
The results, published in the Journal of Social and Clinical Psychology, showed that the limited-use group experienced significant reductions in both loneliness and depression compared to controls. Interestingly, both groups showed decreases in anxiety and fear of missing out — an effect Hunt attributed to the self-monitoring that both groups engaged in (tracking their usage made even the control group more conscious of their habits).
The mechanism Hunt identified was specific: reduced FOMO (fear of missing out) mediated the reduction in loneliness. When participants spent less time watching others' curated lives, they stopped feeling as excluded from social activities they were not part of.
Lambert et al. (2022): The One-Week Break
Jeffrey Lambert and colleagues at the University of Bath tested a more dramatic intervention: complete abstinence from Facebook, Instagram, Twitter, and TikTok for seven days. Their randomized controlled trial, published in Cyberpsychology, Behavior, and Social Networking, assigned 154 participants (mean age 29.6) to either quit entirely or continue as normal.
At one-week follow-up, the abstinence group showed significant improvements across all three outcome measures: wellbeing (mean difference 4.9 points), depression (mean difference -2.2), and anxiety (mean difference -1.7). The intervention group reported using social media for an average of 21 minutes across the entire week, compared to approximately seven hours for controls.
A particularly interesting finding emerged from the mediation analysis: the reduction in depression and anxiety was partially mediated specifically by reduced time on TikTok and Twitter — not Facebook or Instagram. This suggests that not all platforms affect mental health equally, and that the short-form, algorithmically driven content on TikTok and the conflict-heavy environment of Twitter may be more detrimental than platforms where users primarily interact with people they know.
Allcott, Braghieri, Eichmeyer & Gentzkow (2020): The Facebook Deactivation Experiment
The largest and most methodologically rigorous experiment to date was published in the American Economic Review. Hunt Allcott and colleagues recruited 2,743 Facebook users through display ads and paid them to deactivate their accounts for four weeks before the 2018 US midterm election. Sixty-one percent of participants with a willingness-to-accept below $102 were randomly assigned to treatment (deactivation) or control.
The results revealed a complex picture. Deactivation increased subjective wellbeing by approximately 0.09 standard deviations — a small but statistically significant effect, comparable to about 25-40% of the impact of psychological interventions for depression. It also increased offline socializing with family and friends, reduced political polarization, and — notably — caused a persistent reduction in Facebook use even after the experiment ended.
But it also increased time spent watching television alone and reduced factual news knowledge. The wellbeing gains were not simply a matter of replacing bad activity with good activity. Participants redistributed their time in both beneficial and neutral ways.
The Complications: Why "Just Quit" Is Too Simple
If the experiments consistently show benefits from reducing social media, why not simply recommend that everyone quit? Several findings complicate this straightforward prescription.
Duration matters — in unexpected ways
Burnell's meta-analysis found that the majority of studies (22 out of 32, or 69%) tested restrictions lasting only one to two weeks. This is a significant limitation. We do not know whether the modest benefits observed at one week persist, grow, or fade at one month, three months, or a year. The few studies testing longer durations show mixed results — some show adaptation (the benefits plateau), while others suggest cumulative improvement.
More surprisingly, some evidence suggests that very short restrictions (under one week) may actually worsen mood. This could reflect withdrawal effects — the initial discomfort of losing a habitual activity before the benefits of its absence emerge. It parallels findings in other behavioral change literature: the first days of quitting any habitual behavior are often the worst.
Individual differences are enormous
The word "heterogeneous" appears repeatedly across all three meta-analyses. The average effect of quitting social media is small and positive. But this average conceals wide variation. For some participants, quitting produced substantial improvements. For others, it made no difference. For a minority, it may have made things slightly worse — particularly those whose primary social connections were maintained through these platforms.
Who benefits most? The evidence points to heavy users and those who use social media passively (scrolling and consuming rather than posting and interacting). Light users and those who primarily use social media to coordinate real-world social activities may see little benefit from reduction — and could lose a genuine coordination tool.
The "what instead" question
Perhaps the most important finding from the Allcott study is that what people do with their reclaimed time matters as much as the quitting itself. Participants who replaced social media with face-to-face socializing showed the largest wellbeing gains. Those who replaced it with passive media consumption (television) showed smaller gains. The benefit is not simply in the absence of social media but in the presence of something better.
This aligns with a broader principle in loneliness research: reducing a negative is not the same as building a positive. Removing social media removes a source of unfavorable social comparison and passive consumption. But it does not automatically create the meaningful social connection that humans need. That requires active effort — initiating conversations, showing up to shared activities, maintaining relationships through sustained contact.
The Natural Experiment: What Facebook's College Rollout Revealed
While the randomized experiments tell us about short-term effects of individual quitting, a different kind of evidence addresses longer-term, population-level impacts. Braghieri, Levy, and Makarin published a study in the American Economic Review in 2022 that exploited a natural experiment: Facebook's staggered rollout across US colleges between 2004 and 2006.
Because Facebook became available at different colleges at different times, the researchers could compare mental health outcomes at colleges before and after Facebook arrived, using colleges that had not yet received access as controls. This difference-in-differences design approximates the causal power of a randomized experiment at population scale.
The results showed that Facebook's introduction at a college was associated with a significant negative impact on student mental health, including increased reports of impaired academic performance due to poor mental health. The mechanism, according to additional evidence in the paper, appeared to be unfavorable social comparisons — students comparing their own lives to the curated presentations of their peers.
This study is important because it addresses a different question than the quit-experiments. The quit-experiments ask: "Does reducing social media help people who already use it?" The college rollout study asks: "Did introducing social media harm people who previously did not have it?" Both point in the same direction, strengthening the causal interpretation.
What the Evidence Does and Does Not Support
Synthesizing across all available experimental evidence, several conclusions appear well-supported:
- Reducing social media use produces real, statistically significant improvements in wellbeing and depressive symptoms for the average person. This is no longer a matter of debate — three independent meta-analyses converge on this finding.
- The effects are small. An effect size of g = 0.25 means that social media reduction helps, but it is not a cure for depression or loneliness. It is one factor among many.
- The effects vary enormously between individuals. Heavy, passive users benefit most. Light, active users may benefit little or not at all.
- What replaces social media matters. The gains come not just from removing a negative but from the opportunity to engage in something more nourishing — particularly face-to-face social interaction.
- Not all platforms are equal. Algorithmic, short-form content (TikTok) and conflict-driven platforms (Twitter/X) may be more harmful than platforms focused on personal connection.
- Long-term effects remain unknown. Most experiments last one to four weeks. We have very limited experimental evidence about what happens over months or years of reduced use.
What the evidence does not support is the claim that social media is the primary driver of the loneliness or mental health crisis. The effect sizes are too small for that. Social media restriction is one tool — a modest one — in a much larger picture that includes economic precarity, weakened social infrastructure, rising housing costs that separate friends, and the broader erosion of "third places" where people once gathered without a screen.
From Restriction to Replacement
The experimental literature suggests a shift in how we should think about social media and loneliness. The question is not "should I quit social media?" but rather: "What am I replacing it with?"
A person who quits Instagram but spends the reclaimed hours watching television alone has not solved a loneliness problem. A person who reduces social media by 30 minutes a day and uses that time to text a friend, attend a group activity, or have a phone conversation has done something meaningfully different — and the experimental evidence suggests this is where the wellbeing gains concentrate.
This reframing aligns with what the broader loneliness intervention literature shows. As we have previously discussed on YaraLoop, the most effective interventions for loneliness are not those that simply remove isolation but those that create opportunities for repeated, low-pressure social interaction around shared activities — what researchers call "structured contact opportunities."
The experimental evidence on social media restriction, taken as a whole, tells us something both modest and important: reducing passive consumption of curated social content frees up time and psychological bandwidth. But the bandwidth is only valuable if it gets filled with genuine human connection. The absence of a negative is not the same as the presence of a positive. Building real friendship requires showing up — repeatedly, imperfectly, and in person or in shared experience — in ways that no amount of digital passivity can replace.
See the Research in Action
YaraCircle is built on the friendship science we cover here.