Social media is not “rewiring childhood.” Smartphones have not “destroyed a generation.”

Those who blame teenagers’ depression on social media issue new analyses admitting the effects (if any) are tiny

In recent meta-analyses, psychology professor Christopher Ferguson and colleagues conclude that studies claiming social media harms, and desistance from social media benefits, mental health show only small, insignificant effects.

Psychologists Jonathan Haidt and Zach Rausch, along with David Stein, who argue social media seriously damages teens’ mental health, just issued challenges to Ferguson’s findings.

Statistical arguments over issues like study inclusion, modeling, effect sizes, etc., can wade deep into weeds. Fortunately, these three analyses, as well as Haidt, Rausch, and Jean Twenge’s past one, share a common finding relevant to discussion and policy.

So, let’s talk about the Big Picture – the forest, not the weeds.

The bottom line the three analyses argue over is simple: How tiny is the tiny effect social media has on users’ mental health? Is it next-to-nothing insignificant, or truly-nothing insignificant?

Compare Ferguson’s conclusion about the effects of giving up social media to the most favorable findings for their case Haidt & Rausch, and Stein, make from their choices of selected studies, using Cohen’s d-statistic as the measure of the effect desistance from social media has on mental health:

·         Ferguson: d = +0.08 (from all 27 studies, which translates into a real-world positive effect on less than 3% of users)

·         Stein: d = +0.17 (from his selection of 10 “higher powered” studies, or an effect on less than 7% of users)

·         Haidt & Rausch: d = +0.20 (from their selection of 10 long-term studies, an effect on 8% of users).

The three analyses agree: statistically, social media influences lie at the very smallest end of the Cohen-d distribution’s already “small” effect size category. All, in fact, find non-significant results, according to the minimum d = 0.21 threshold Ferguson pre-registered before commencing his analyses (pre-registration of study specifications is important to prevent researchers from lowering their standards when findings prove disappointing).

It should also be noted that the studies analyzed employ weak, single factor, before-after designs that are vulnerable to “demand characteristics” – that is, subjects can easily figure out the study’s purpose and conform their responses to what researchers expect. That is why strong designs, such as for medical studies, follow “double blind” standards in which neither subjects nor experimenters know what intervention is expected to cause what effect.

Even given these caveats, and even if only the maximum effects of social media are accepted and all other factors affecting mental health are ignored, Haidt’s & Rausch’s and Stein’s re-analyses confirm Ferguson’s larger point applicable to policy: social media is a trivial influence on mental health. Banning or curbing all teenagers’ social media use would have scant, if any, good effects.

Dealing with the smallness of social media’s effects in reality – after publicly hyping social media as the catastrophic destroyer of childhood – is where Rausch & Haidt’s logic goes badly wrong.

They acknowledge that social media’s effect on mental health may be “small,” but that doesn’t matter. Even if very few teens would benefit from banning or severely curbing social media for all teens, that is still a “positive” benefit. They cite a medical analogy: even if a drug benefits only a small fraction of the population, we still include that drug in treatment regimens.

The obvious problem with medical analogies: they’re not analogous. Drugs and treatments (again, evaluated using rigorous double-blind, not weak before-after experimental designs) are prescribed only to the fraction of individuals diagnosed to benefit, not to the whole population. Doctors don’t take out everyone’s gall bladder.

An intervention like a vaccine is applied to the whole population only if its benefits provably and greatly outweigh its costs. For the COVID or flu vaccine, preventing hundreds of thousands of deaths and tens of millions of illnesses outweighs the negative side effects on 0.001% of those vaccinated.

In the case of social media, the harms caused by mass bans or curbs to large majorities of teens greatly outweigh the benefits to a small fraction, who would better be served by individual help.

How would banning social media harm teens? For two examples, the CDC’s massive 2021 survey numbers show that compared to teens who rarely use screens (less than 1 hour a day), teens who frequently use screens (5+ hours per day) report less self harm, fewer suicide attempts, more sleep, less hard-drug use, less violence, lesser increases in drinking and drug use, and smaller overall risks. Moderate screen users (2-4 hours a day) generally fare as well or better. The 2023 CDC survey, to be released in the fall, will refine these questions.

Pew Research’s 2022 survey found that while 9% of teens say social media is harmful for them, 58% said that social media helps them feel “more accepted,” 67% said it connects them to “people who can help them get through tough times,” and 80% said it keeps them up with “what’s going on in their friends’ lives”.

A barrage of recent, large-scale, longer-term, better designed studies are finding teens use social media in far more varied and positive ways than previously thought. These will be discussed in a later post.