Another massive study finds social media impacts even on younger teens’ mental health are minimal at worst

Wait… didn’t its authors say that “higher…social media use” is associated “with greater depressive symptoms” among 9-12-year-olds? Let’s examine what that means.

A just-published 3-year study of 11,876 youngsters ages 9-12 in the prestigious Journal of the American Medical Association illustrates the vast gap between what authors say they find versus what their study actually showed. This study, acronymed “ABCD”, again illustrates a mammoth problem in a debate over social media and mental health that loathes complications.

The authors’ quotable conclusion that “more time spent on social media during early adolescence may contribute to increased depressive symptoms over time” dodges the fact that their own results are both tiny and contradictory.

The authors first measured social media effects longitudinally: that is, did a 9-year-old who reported more time on social media later report being more depressed at ages 10, 11, and 12 than a less social-media using 9-year-old? Not really. The study’s effect sizes predicting an individual’s later depressive symptoms from their social media use (presented as standardized betas, β) are tiny: 0.07 to 0.09. That is, near-nothing; an effect barely worth calling “small.”

That these authors should lead with such a trivial finding and cite it as a basis for parental or policy action aimed at millions of children epitomizes the scourge of social scientists vastly overstating the importance of near-nothing results.

The large-sample, low-effect-size fallacy

Here’s the first problem. If your sample size is large enough, lots of things become “statistically significant.” For example, in the 2023 Centers for Disease Control survey of 20,000 teens, 19 independent factors were significantly associated with teenagers’ poor mental health. Emotional abuse by parents/adults (β = 0.28) was by far the most significant, followed by parents’ severe depression (β = 0.15), lack of school friends (β = 0.14), and lack of sleep (β = 0.13). After these, no other factor comes close.

Note that the two most critical factors associated with teens’ poor mental health are never cited by anyone, except me.

What factor(s) will wind up being wildly hyped in media and political forums? Not the most significant ones, of course, but the most popular. For example, blame-the-screens advocates can cite low-level online factors like social media use (β = 0.05) or cyberbullying (β = 0.04). If we favored the indoors or held stock in sunscreen, we’d blame sunburn (β = 0.08) for teens’ depression. If we hated school, we’d blame the six school-related factors, ranging from no friends to school racism (β = 0.03), school injuries (β = 0.02), etc. Just keep teens indoors, in bed, and away from school and their parents, and they won’t be depressed.

Then, if you start comparing factors to each other in a multiple-regression analysis, these factors start to combine. The online factors, for example, are closely tied to each other, as are the school factors. Lack of sleep is closely related to parental abuses. 84% of teens who are cyberbullied are also parentally abused much more often. By the time the statistical bloodbath is over, just a few variables show any real importance – in this case, family and grownup troubles… exactly the ones no one wants to hear about.

The second problem with the JAMA study

Further, even the authors’ tiny longitudinal effect is complicated by the study’s contradictory finding when authors analyzed youths’ social media use cross-sectionally. When 9-10-year-olds’ social media use versus later depression was compared over time across the entire sample – controlling for demographic and family-conflict measures – their previous individual finding disappeared. No effect at all was found (β = -0.01).

“This suggests that adolescents with consistently high (or low) social media use were not necessarily the same adolescents with consistently high (or low) depressive symptoms after accounting for demographic and familial factors and within-person estimates,” the authors concluded. Wow. Why wasn’t that the main conclusion?

That is, the study found no social media effects even on younger children. If only the authors of this large JAMA-published study had delineated the three things of profound importance they did find:

1. Even if measured assuming a child’s individual social media use were the main factor determining their individual mental health, the results were trivial – barely any effect was found.

2. Then, when more important factors potentially affecting mental health (i.e., family conflict) were included across the entire sample, even that tiny effect disappeared.

3. That is, “residual and unmeasured confounders” – not social media use – might well be the real culprits in children’s depression.

These findings add to the growing mountain of studies showing social media effects on teenagers, even when measured in isolation, are non-existent or tiny – a reality on both skeptics and fervent anti-social-media advocates agree. This study also reinforces the CDC’s analysis showing that the effects on teenage mental health, opiate misuse, and suicidality associated with parent- and household-grownup-inflicted abuses, drug/alcohol problems, and mental health troubles are huge, ranging from two-thirds to over 90%, compared to only tiny effects attributable to social media.

Social science researchers, reviewers, and editors need to impose drastically more rigorous standards if the field’s research – unfortunately ripe for political and media sensationalism but lacking scientific utility – is to be taken seriously. Weak, near-nothing effect sizes produced by huge samples and omission of relevant factors cannot continue to win reviewer and editor approval until authors are willing to candidly state uptop: “We got nothin’” – which is a valuable finding.

Unfortunately, “social media” has become the default culprit in young people’s mental health discourse, the purported causal factor endlessly presented to the exclusion of all others – even those factors researchers find are important. It is long past time to feature the most important causal factors in mental health upfront and downgrade social media to its scientifically-determined trivial status.