nonacademicresearch.org Editorial
Public concern that social media use damages mental health — particularly among adolescents — has grown substantially since the mid-2010s. The empirical evidence is more mixed than public discourse suggests. While some longitudinal studies find associations between heavy social media use and depression or anxiety, effect sizes are typically small, causality is difficult to establish, and experimental studies have produced inconsistent results. Understanding what the evidence does and does not support is essential for sound policy.
Concern about the effects of social media on mental health — particularly among adolescents — has intensified as rates of depression, anxiety, and self-harm have risen in many high-income countries since approximately 2012. Whether social media use is a cause of these trends is contested. The evidence from experimental and longitudinal studies is mixed, and the magnitude of associations found in large observational studies is often smaller than public discussion implies. Understanding what is and is not known is important for parents, clinicians, and policymakers alike.
In 2011 and 2012, survey data from multiple countries began showing rising rates of depression, loneliness, and anxiety among adolescents — particularly girls — after years of relative stability. The timing coincided with the rapid adoption of smartphones and the rise of Instagram, Snapchat, and other image-centric social platforms. Researchers and commentators drew the connection, and a substantial public discourse emerged claiming that social media was causing a mental health crisis in young people.
The causal claim has proven difficult to establish. Social media adoption happened simultaneously with many other changes in adolescent life — shifts in time use away from in-person socializing, changes in sleep patterns, economic stress following the 2008 financial crisis — making it hard to isolate the specific contribution of any single factor.
Several large-scale observational studies have examined the association between social media use and wellbeing. A frequently cited analysis by Twenge et al. (2018, Clinical Psychological Science) used data from the Monitoring the Future survey and found negative associations between screen time (including social media) and psychological wellbeing across several measures. Heavy users (five or more hours per day) were 66% more likely to have at least one risk factor for suicide compared to non-users.
However, a rigorous re-analysis of the same datasets by Orben and Przybylski (2019, Nature Human Behaviour) applied a multiverse analysis — systematically testing all defensible analytical specifications rather than a single pre-selected one — and found that social media use explained only 0.4% of the variance in adolescent wellbeing, an effect size smaller than wearing glasses. The association was also frequently non-linear and inconsistent across measures.
A separate analysis by Orben and Przybylski (2019, PNAS) used experience sampling methods (real-time ecological momentary assessment) rather than retrospective self-report and found weak or null associations between social media use and momentary wellbeing, in contrast to the stronger effects found in retrospective survey studies.
Randomized experiments provide stronger causal evidence. Hunt et al. (2018, Journal of Social and Clinical Psychology) randomly assigned college students to limit all social media use to 10 minutes per platform per day for three weeks or to use it normally. The restricted group showed significantly lower loneliness and depression at follow-up. Shakya and Christakis (2017, American Journal of Epidemiology) found in a longitudinal study that Facebook use predicted subsequent decreases in wellbeing, while in-person social interaction predicted increases.
However, the experimental literature is not uniform. A pre-registered study by Yuen et al. (2019, PNAS) deactivated Facebook accounts for roughly 2,800 users for four weeks prior to the 2018 U.S. midterm elections and found increases in subjective wellbeing, reduced political polarization, and more time spent offline — but effects on depression and anxiety, while positive, were modest in absolute terms.
One of the most important questions is whether effects are specific to particular populations. Several studies find that associations between social media and poor mental health are stronger for girls than boys. Twenge et al. (2020, PNAS) analyzed 500,000 adolescents and found that the strongest negative associations were for girls in the 10-15 age range. The mechanism proposed — comparisons with idealized body images, cyberbullying disproportionately targeting girls — has some evidential support, but the picture remains incomplete.
Haidt and Allen (2020, reviewing multiple studies in Social Psychology) argue that the aggregated cross-national timing evidence — multiple countries showing parallel adolescent mental health declines beginning around 2012 — is hard to explain by factors other than a shared global shift in technology use. Critics of this view note that the timing correspondence is imprecise and that other explanations (economic insecurity, school pressure) have not been ruled out.
A potentially important distinction is between passive social media use (scrolling, viewing others' content) and active use (posting, messaging, interacting). Verduyn et al. (2015, Journal of Experimental Psychology: General) found in a randomized experiment that passive Facebook use increased envy and decreased affective wellbeing, while active use had no such effect. This distinction may explain some inconsistencies in the literature and suggests that the specific type of engagement matters, not merely the total hours spent.
Some researchers — notably Candice Odgers — have argued that correlational data consistently show that the most distressed adolescents use social media most heavily, but that this may reflect pre-existing vulnerability rather than harm caused by social media itself. Odgers and Jensen (2020, Journal of Child Psychology and Psychiatry) reviewed the preregistered longitudinal studies and found that most of them show small or inconsistent effects, and that the media narrative has outrun the evidence.
There is also a contested question about the relevance of population-level associations to individual clinical practice. Even if social media has a small average effect, it may have a larger effect on vulnerable individuals — those already experiencing depression or with lower social support — which may not show up clearly in population averages.
The evidence on social media and adolescent mental health is genuinely contested. Large observational studies find an association, but effect sizes are typically small and methodological choices substantially influence results. Experimental studies suggest that reducing social media use can improve wellbeing, but effects are modest. The cross-national timing evidence raises the probability of a genuine contribution from social media to adolescent mental health trends, but does not establish it conclusively.
The honest summary is: social media likely has a real negative effect on wellbeing for some adolescents, particularly girls, particularly through passive image consumption and social comparison, and particularly at very high use levels. The size of this effect for the average adolescent is smaller than public discourse implies, and it is embedded in a more complex set of causes for the mental health trends observed.
nonacademicresearch.org Editorial (2026). Social Media and Mental Health: What the Evidence Actually Shows. nonacademicresearch.org. nar:c1dkuupp6lq5fkqxy3
@misc{2p43b5kt,
title = {Social Media and Mental Health: What the Evidence Actually Shows},
author = {nonacademicresearch.org Editorial},
year = {2026},
howpublished = {nonacademicresearch.org},
note = {nar:c1dkuupp6lq5fkqxy3},
}Temporary identifier. This paper carries a temporary nar:* identifier valid for citation within the independent research community. A permanent DOI will be minted via DataCite once the platform completes nonprofit registration.
Log in to join the discussion.
Loading…