A cohort study of more than 100,000 Australian adolescents found a U-shaped association between after-school social media use and well-being, with moderate users consistently demonstrating the most favorable outcomes. Both nonusers and the highest users faced elevated odds of low well-being—but the magnitude of risk varied substantially by sex and developmental stage.
The study followed students in grades 4 through 12 for up to 3 years. Among girls in grades 7 through 9, the highest social media users (12.5 or more hours per week) had approximately three times the odds of low well-being compared with moderate users. Boys in the same grades showed a similar but attenuated pattern, with about twice the odds. By late adolescence (grades 10-12), however, the picture shifted dramatically: nonuse became increasingly problematic, with boys who abstained from social media having three times the odds of low well-being compared with moderate users and girls having nearly twice the odds.
Researchers analyzed data from the Wellbeing and Engagement Collection, an annual census conducted in South Australian government schools. The analytic sample comprised 173,533 observations from 100,991 participants (50% female; mean age, 14 years), wrote lead study author Ben Singh, PhD, of the University of South Australia, and colleagues. Well-being was operationalized as the mean score across eight validated indicators: happiness, optimism, life satisfaction, worry, sadness, perseverance, emotional regulation, and cognitive engagement.
Social media use was classified into 3 categories based on self-reported after-school engagement (3 PM to 6 PM on weekdays): none (0 hours per week), moderate (greater than 0 to less than 12.5 hours per week), and highest (12.5 or more hours per week). The upper threshold corresponds to approximately 2 or more hours per weekday, consistent with prior research identifying this level as a potential inflection point for psychosocial risk.
Developmental trajectory analyses revealed sex-specific patterns. Among girls, all social media use categories were associated with increasing probabilities of poor well-being as grade level advanced, with trajectories plateauing in senior secondary school. The highest users consistently had the highest probability of low well-being across all grades, peaking at 51% in grade 9. From grade 10, nonusers showed similarly elevated probabilities, though uncertainty was high due to the small number of female nonusers in later grades.
Boys exhibited a different pattern. Those with no or moderate use had the lowest probabilities of low well-being in grades 4 and 5. However, nonusers experienced a steep increase from grade 7 onward, reaching 36% by grade 11. Moderate users, by contrast, increased only from 12% in grade 4 to 21% in grade 12. Among boys, the highest social media users had the highest probability of low well-being in early grades, but this was overtaken by nonusers from grade 9 onward.
The researchers situate these findings within the context of adolescent development: “In early adolescence, offline interactions may sufficiently meet social needs, explaining why nonuse was not detrimental. By middle adolescence, however, social media becomes a key setting for maintaining peer connections, identity expression, and belonging. At this stage, moderate engagement may foster connectedness and social support, whereas both very high use and complete nonuse may increase vulnerability, through either social comparison or social isolation.”
The findings align with the Goldilocks hypothesis, which posits that moderate digital media use is least risky. However, the researchers emphasize that their data reveal dynamic, developmentally contingent associations rather than static thresholds. “By late adolescence, as self-identity and emotional regulation mature, the risks of high use may diminish somewhat, while abstinence may increasingly limit opportunities for social participation and peer belonging,” the researchers wrote.
The study design combined elements of cross-sectional and longitudinal approaches, using repeated annual surveys with some participants contributing up to 3 years of data. Mixed-effects logistic models adjusted for demographic covariates (school grade, parental education, region of residence, calendar year) and included random intercepts at school and participant levels. Sensitivity analyses examining each well-being indicator separately, stratifying by individual school grade and calendar year, and following specific longitudinal cohorts all yielded consistent results.
Regarding clinical and policy implications, the researchers caution against simplistic screen-time guidelines: “These findings call for a balanced approach that combines reasonable time-based limits to encourage moderate (but not excessive) use with a more nuanced focus on the quality, purpose, and context of digital engagement.” They further note that “overly restrictive or abstinence-based strategies may isolate adolescents from important peer networks, especially in later adolescence, while excessive use remains associated with emotional dysregulation and social comparison.”
Several limitations warrant consideration. The analysis captured only after-school weekday use and did not account for weekend engagement or platform-specific behaviors. Self-reported data may be subject to recall or social desirability bias. The study lacked clinical mental health diagnoses and could not assess compulsive or addiction-like use patterns, which recent research has linked to suicidality. Additionally, as the researchers acknowledge, “after-school social media use may act as a proxy for differences in after-school activities, homework support, or family supervision rather than representing social media behavior alone.”
Disclosures can be found in the study.
Source: JAMA Pediatrics.
