If we really want to deter suicide and self-harm, we should prescribe social media to teenagers, not ban it

The anti-teen-social-media case keeps falling apart even as it peaks politically.

Girls report feeling more “sad or hopeless” (53%) than boys (28%) do, the 2021 and 2023 Centers for Disease Control surveys show. Authorities and media commentators, in psychologist Jean Twenge’s typical claim, declare girls’ rising depression, suicides, and self-harm are just because…

“…social media provides an endless way for other kids to be cruel, they can never achieve the perfect bodies they see on Instagram, they are constantly judged for their appearance in the endless selfies they are compelled to post, unknown adults can sexualize them, they are continually stressed about how many likes they’re going to get, and some social media accounts glorify (and even instruct about) self-harm.”

We get it. Twenge thinks teenage girls are stupid, vain, mean, and shallow. That is not at all my experience working with teenagers for 30 years in family, community, and wilderness programs and school/university settings, but Twenge’s stereotype is common. Her rant offers zero evidence, which seems to enhance, not dampen, the media’s adoration.

Let’s consider its basic statistical case:

· 59% of girls who use social media frequently daily or more report chronic sadness, compared to 49% of those who use social media less often (boys’ comparable figures are 31% and 27%).

· Of the 22% of girls who are bullied online, 77% report persistent sadness, 29% report a suicide attempt, and 13% report self-harming, much higher percentages of mental distress than reported by the unbullied.

Twenge and others insist that mere “positive correlation” no matter how limited, selective, and “small” is all that’s required; social media must be the problem. Don’t look any further. What else could girls possibly be sad about?

The huge factor Twenge leaves out

Here’s one hint from the same girls on the same survey. Of the 22% of girls who report being bullied online (cyberbullied):

· 84% ALSO report being bullied (emotionally abused) at home by parents and household grownups,

· 59% have parents with “severe” mental health problems,

· 50% have parents who abuse drugs/alcohol, and

· 48% have violent parents who hit, beat, slapped, kicked, etc., their kids and/or each other.

Of the 12% of boys who report being cyberbullied:

· 78% ALSO report being emotionally abused at home by parents and household grownups,

· 45% have parents with “severe” mental health problems,

· 42% have parents who abuse drugs/alcohol, and

· 46% have violent parents.

Funny, Twenge and others who incessantly deplore social media, cyberbullying, and girls’ mental troubles never mention these crucial contexts.

This dereliction is unconscionable, since the 22% of girls who are bullied online who also are among the 70% who are abused by grownups at home suffer the most severe mental health problems themselves (84% of these multi-bullied girls report chronic sadness, one-third have attempted suicide, 14% self-harm… can you blame them?). How can authorities simply dismiss this multiple tragedy?

I keep talking about this because others won’t

As dissected repeatedly on this substack, the statistical flaws in social-media-blamers’ simple “correlation equals causation” argument are threefold: (a) their correlations of social media use and mental health are woefully weak, barely “small” in effect; (b) it is confounded by a reverse correlation; that is, depressed teens use social media more; and, worst of all, (c) it suppresses far more important factors in teens’ lives that cause sadness.

Twenge and others’ negligible social-media effect sizes result from their exclusion of a broad set of factors – parents’ and adults’ abuses, violence, drug/alcohol problems, poor mental health, and jailing – the 2023 CDC survey shows are absolutely critical to those seeking a true teen mental health picture.

My last few postings analyze the effects of parents’ drug/alcohol abuse, which multiple measures show has soared over the last 15 years to a staggering 5.5 million hospital overdose emergencies and deaths among parent-aged adults in 2024. The CDC’s survey found 34% of girls and 25% of boys reported drug/alcohol abusing parents.

Suddenly, the whole perspective shifts

Tables 1-6 do something no other analysis does: they divide the effects of teens’ social media use (as defined by the CDC) into two categories: those with parents who abuse drugs and/or alcohol, and those whose parents don’t. The first 3 tables show raw percentages.

Source: CDC 2024.

Three-fourths of girls and nearly half of boys whose parents abuse drugs/alcohol suffer serious sadness, compared to fewer than half of girls and one-fourth of boys whose parents don’t abuse drugs/alcohol (Table 1). Sadness is somewhat more prevalent among teens who use social media more.

Complications follow. When it comes to suicide attempts and self-harm requiring medical attention – much worse than simply being sad – parents’ drug/alcohol abuse is a major associate. Girls who use social media often are at much LESS risk than girls who rarely use social media. Boys and teens whose parents don’t abuse drugs/alcohol show more ambiguous results for suicide attempt, but the self-harm pattern is clear: both sexes show substantially LESS risk the MORE teens use social media, regardless of whether their parents abuse drug/alcohol (Tables 2, 3).

That’s startling

Both girls and boys who use social media more are somewhat sadder. Yet, these sadder, social-media-frequenting girls are much less likely to attempt suicide and to harm themselves than social-media-avoiding girls. A similar, weaker pattern is found for boys.

You can do the math from the table. Among sadness-prone girls with drug/alcohol abusing parents and who rarely use social media, 45% go on to attempt suicide and 1 in 6 self-harm. Among corresponding girls who often use social media, just 29% go on to attempt suicide and 8% self-harm. Boys and teens whose parents don’t abuse drugs/alcohol show a similar progression – more social media use seems to deter really bad outcomes.

You’d think authorities would be falling over themselves to understand what mechanism connected to social media use is associated with deterring girls in particular from going on to rash acts. We want to prevent suicide and self-destructive behaviors, right? (…right?)

That not one major authority mentions this startling fact shown their own official CDC survey documents tells us how dishonest the entire teens-and-social-media discussion has become.

How important are these factors?

It isn’t just raw percentages or “statistical significance” that matter; far more important is how much relative influence different factors have on teens’ mental health.

Tables 4-6 show the odds ratios (which compare the odds of an event happening to the odds of that event not happening) for 3 outcomes – teens’ sadness, suicidality, and self-harm – and 2 potential causes: social media use, and parents’ drug/alcohol abuse.

Source: CDC 2024.

An odds ratio of 1 denotes no effect; below 1, a reverse effect. For most social science work, weaker odds ratios of 0.7 to 1.4 shouldn’t be taken seriously.

Even odds ratios of 1.4 up to 2.5, or 0.7 down to 0.4, indicate only small effects. That smallness, not the mere fact of statistical significance (that is, the 95% confidence intervals in parentheses are both higher or both lower than 1.00), ethically should be reported as the main finding.

At odds ratios of 2.5, and especially 4.0 or higher (or below 0.4, and especially 0.25), we sit up and take notice. These are medium and strong effects. You can start to claim a real finding (always couched as, “merits further investigation…”).

Odds ratios can be used to calculate Cohen’s d, the standard statistic of “effect size.” It has its own distribution. Let me speak plainly:

· d below 0.20, you got nothin’, shut the hell up;

· d = 0.20 to 0.50, small, you ain’t found much;

· d = 0.50 to 0.80, medium, use your indoor voice;

· d = 0.80 or above, start shouting.

In the upper lefthand corner of Table 4, girls who use social media frequently are 1.56 times more likely to suffer serious sadness compared to girls who use social media less. The odds that the true proportion falls between 1.35 and 1.81 are 95%, which is “significant.” However, this translates into a d value of 0.25, barely above “small.”

You’ve just seen the entire statistical case that social media harms teens’ mental health; in fact, my numbers are more generous than most studies find.

It’s downhill from here

The blame-social-media mob might be able to whisper that social media might be weakly associated with more sadness in some girls, pending investigation into multiple factors, but is far below the level needed to recommend policy.

Otherwise, the blame-social-media endeavor is a hoax – especially when we get to the important stuff like suicide and self-harm. Social media use has no effect on suicide attempt and actually appears to help deter self-harm.

However… parents’ drug/alcohol abuse? Consistently, the negative effects on teenagers’ mental health are worth talking, and sometimes shouting, about.

Usually, a regression analysis just confirms odds-ratio findings. Here, stepwise regression comparing the effects of two independent causes (social media use and parents’ drug/alcohol abuse) on three dependent outcomes (teens’ sadness, suicide attempt, and self-harm) rejects social media use as a significant factor in all 3 cases and leaves only parents’ drug/alcohol abuse (p=0.000 for all outcomes).

It’s no contest

The CDC’s results, and mine here using the CDC’s definition of social media use, are very similar. Parents’ drug-alcohol abuse is so dramatically more important in influencing teens’ sadness, suicide attempt, and self-harm that blaming social media is a waste of time.

It’s important to recognize the CDC survey further shows the teens most at risk of poor mental health suffer not just one family issue, but an average of 2-3 serious parental and family risks, which diminishes social media below “nothing” as a cause of poor mental health.

The only way to elevate social media use even to minimal importance as a factor in teens’ mental health is by arbitrarily excluding parents’ troubles from analysis. Even then, social-media-blamers typically produce anemic d values of around 0.20 at best, which they wildly ballyhoo to permissive journal editors and gullible politicians and media editors as apocalyptic proof that “social media is destroying a generation.” Except for a few media-vulnerable teens and adults who do need help, the social-media panic is descending into destructive fraud.