Get ready for the next round of lying about teen suicide

“Teen suicide” has been falling since 2017. You never heard about it because anti-social-media and other interests profit from spreading fear. Now, they’ll start grabbing undeserved credit.

Psychologist Jonathan Haidt excitedly proclaims 2024 “the year… the adolescent mental health crisis began to reverse… a tipping point, a culture change, a vibe shift about the dangers of a phone-based childhood.”

So, expect a 180-degree tone reversal as well. No more gloom and doom about “teenage mental health.” Happy days are here, Haidt announces, now that “phone-based childhood is beginning to reverse.” Soon, we’ll start hearing about how much better teenagers’ mental health and suicide trends are getting. Interests that treat teenagers as a mere commodity, acceptable to lie about, will start grabbing credit for miraculous “improvements.”

Reality means nothing, of course

Haidt and his followers, including major media and political leaders, have never displayed any caring about what was actually going on among teenagers – only the narrow, selective framings that fed their crusade to blame social media.

For example, suicide rates among U.S. teens have been falling since 2017, reaching a lower rate in 2022 than in the 1980s and 1990s eras Haidt and his After Babel substackers sanctify as the Garden of Eden of childhood before the social-media serpent intruded. Teenaged girls’ suicide rates have been stable for 8 years, while boys’ rates have fallen. In a country that cared in the slightest about teenagers, the teen mental-health industry would be loudly demanding study of why post-2017 teens were killing themselves less.

This is not that country. We heard nothing of the kind. Why not? Because the 2017-2022 teen suicide drop was damned inconvenient.

During a time when teens’ social media and smartphone use were increasing rapidly, teenagers were obstinately committing less suicide. Cyberphobe dogma holds that can’t possibly happen. Predatory interests had not properly positioned themselves to claim unwarranted credit for such an unwelcome improvement. Just the opposite. They needed to fan more fear of teens’ self-destruction to push their proprietary remedies and ban-kids agendas – not a soothing message that in a rising teen population using social media more, several hundred fewer teens committed suicide in 2022 than five years earlier.

So, all we’ve been hearing is that teen suicide is soaring. Haidt and colleague Zach Rausch have sensationalized girls’ suicide trends as proof that “phone-based childhood” is driving an Anglo-world pandemic – all without presenting a shred of evidence that smartphones and social media drive girls’ suicide.

It gets worse, as next week’s analysis of Haidt’s political agenda will show. Haidt and many leaders have exploited fears of social media to create an image that “liberal girls” and campus protesters he disagrees with politically have too much freedom to communicate and access information.

In fact, the best evidence indicates the opposite. As psychologist Chris Ferguson’s new meta-analysis shows, Monitoring the Future data connecting girls’ liberalism to greater depression rests on tiny “effect sizes” (r-values of 0.04 to 0.10) that “cannot be reliably distinguished from statistical noise.” Those who make loud, sweeping claims connecting adolescent social media use and thinking to depression based on miniscule, barely significant associations keep “social science … within the realm of pseudoscience,” Ferguson warns.

Further, abused and depressed girls use social media more to connect with other people and services. The best, largest survey of American teenagers, the Centers for Disease Control’s 2023 Youth Risk Behavior Surveillance, shows the younger girls Haidt crusades to ban from social media actually attempt suicide and harm themselves LESS (14.5% and 4.0%, respectively) if they use social media frequently compared to those who never use social media (15.0% and 6.6%).

Again, one would expect those like Haidt, Rausch, and others who incessantly deplore girls’ depression and suicide would jump at analyzing these surprising results.

Of course not. This isn’t about girls or suicide

Instead of buying Haidt’s and After Babel’s non-science about how liberal ideology combined with social-media messaging promote misery and victim-mentality in girls, let’s look at actual trends provided in the CDC’s mortality files – the definitive ones everyone uses.

First, teenaged girls are very unlikely to commit suicide, far less likely (711 suicides among girls age 10-19 in 2022) than (say) Haidt’s smaller cohort of 60-aged men (5,310) – yet another reality you’d never guess from following After Babel.

In fact, the suicide rate among White men age 60-69 is 10 times higher than among White girls and 12 times higher than among girls of Color age 10-19. You’d think that instead of champing to cut teens off from social media, Haidt would express humility about the depressive conditions his older, vastly more suicidal demographic is causing its teenaged granddaughters.

Again, of course not. Haidt’s movement and substack following is not about humility and scientific introspection, but invoking self-superiority and suppression.

Girls’ and other teens’ real suicide trends

The next four charts detail teen suicide trends during the smartphone era, 2010 through 2022, broken down into the staunchest conservative counties (those voting Republican in states with Republican statehouse and legislative dominance in 2020, the central year of the analysis) versus the most liberal counties (those voting Democratic in Democratic-run states). Analyses using 2016 or 2024 breakdowns show the same results.

According to Haidt’s claim that liberal cultural values, less religion, and greater social media use promote more teen (particularly girls’) depression and (presumably) suicide, we should see suicide rates rise the fastest to the highest levels among the most liberal populations in liberal bastions while remaining lower among the most conservative populations in conservative ones.

The reality is dramatically the opposite. Contrary to what Haidt predicts, girls’ suicide rates and trends are much worse in solidly conservative than solidly liberal areas. Conservative areas saw explosions in teenagers’ suicide rates, including among girls, Whites, and youth of Color. On a per-youth basis, conservative areas’ leap in suicides among teens was two to three times worse than in liberal areas.

Girls’ age 10-19 suicide rates rose much faster during the 2010-2022 era (up 82%) in strongly conservative areas to much higher (+43%) levels than in strongly liberal areas (up 58%) (Figure 1).

Figure 1. Girls’ suicide rates rose much faster to higher levels in conservative areas

Sources for all charts: CDC 2024; Census 2024. Presentation is mine. Suicide rates are per 100,000 population age 10-19 by year, and rate changes are from regression trendline (dotted).*

This isn’t a demographic quirk (Figures 2, 3, 4). Among the most conservative cohort, White teens in Republican counties in Republican states, suicide rates rose by 52% to levels 65% higher than among White teens in Democratic counties in Democratic states (up 23%). Among teens of Color, who have lower suicide rates, suicide also rose much faster in conservative areas (up 128%) to levels 64% higher than in liberal areas (up 63%).

Figure 2. White teens in conservative areas showed the worst suicide rates of all

Figure 3. Teens of Color in conservative areas also fared badly

Conservative areas tend to be more rural and perhaps choose to offer fewer mental health services than urban areas. Figure 4 shows teen suicide trends in urban (areas encompassing metropolitan populations) versus rural (small-town and rural populations) by political governance.

Figure 4. Rural teens in liberal areas show plummeting teen suicide while conservative areas show rising rates. Liberal urban areas show low rates.

Figure 4 is complicated but worth analysis. It basically shows that the teen suicide increase during the smartphone era (2010-22) was very uneven. Most areas showed increases in suicide rates from 2010 to 2014-2018, then declines through 2022. The exception was the 3 million teens in conservative rural areas, who showed the worst trends and highest rates.

The more liberal and urban, the lower the teen suicide rate and the better trends are; the more conservative and rural, the higher teen suicide rates and the worse trends are. That’s the opposite of the result Haidt’s liberal-girl disparagement would predict. Whatever depression liberal teens, girls in particular, suffer, it is not the kind that promotes suicide.

The whole picture

For all 41 million teens, suicides rose by 3.0 suicides per 100,000 population in conservative areas to levels 58% higher than in liberal areas (up 1.9) (Figure 5).

Figure 5. Overall, conservative areas saw worse teen suicide rates and trends

Again… do we care enough about teen suicide to analyze it carefully even if it doesn’t validate our pet prejudices. Or are teenagers just a commodity to exploit so adults can feel good about themselves, push more dumbass culture warring, and promote interests’ political popularity and profits?

Fraud is easy. Does your agency’s funding or book sales’ profit require teen suicide to be seen as rising?* Pick, say, 2021 and 2010 to compare (up 51%!). Or, if your agency implemented a new policy or program and your substack declares teenage mental health is improving, your new bottom line now might require teen suicide to be seen as falling. Pick 2017 and 2022 to compare – teen suicide down by 350 per year!

That’s the kind of misinformation we’re going to see continued, grossly disrupting the vital research task of understanding and preventing suicide tragedies. That’s a task Haidt’s suicide-prone 60-age male cohort in particular should welcome, not sabotage. If the grownups start fixing themselves and societal ills instead of inventing more culture-war distractions, we’re going to see a lot fewer depressed teenagers.

*These results, unlike selective presentations rampant in media and political forums, rely not on cherry-picked numbers chosen to validate one’s pet hypothesis. Note from the full sets of numbers presented that an interest can select any two years to show anything wanted. To avoid this statistical misuse, the analyses in this substack use regression equations that incorporate all years’ data – not just a selected base year and selected comparison year – to compute the trend.