Taking Teenage Girls’ Smartphones Away Won’t Reduce Suicides

A look at basic numbers shows that suicides won’t be reduced by restricting teen girls’ access to smartphones.

After the Facebook whistleblower Frances Haugen testified before a Senate subcommittee, Americans panicked about smartphones and social media destroying the lives of our teens. The alarm Haugen sounded resonated with lots of concerns that the public already had about this twin threat.

Many parents were already putting the blame on their children’s ready access to smartphones for instances of bullying and sexting, as well as their children more generally having less interest in interacting with them.

Jonathan Haidt and Jean Twenge were well-positioned to weigh in. They are psychologists with bestselling books. They write convincingly for laypersons, citing studies and claiming that social science supports what they say. The two are sought by journalists to provide pithy and provocative authoritative quotes. They are thus superb expert influencers of public opinion for concerns about smartphones and mental health. Haidt is active on Twitter ramping up his call for action, much more than Twenge.

Separately and together, Haidt and Jean Twenge passionately argue that a major crisis became evident around 2012. The spread of smartphones and social media into the lives of teenage girls caused a significant rise in teenage girls’ mental health problems, including death by suicide.

Haidt and Jean Twenge are escalating from their vague calls to action to specific, drastic measures, such as the government prohibiting teens from possessing smartphones and imposing big fines on social media companies that do not restrict services to children and teens.

See popular articles like

Haidt’s Op-Ed in The New York Times, This is a chance to pull teenage girls out of the smartphone trap.

Haidt’s Facebook’s dangerous experiment on teenage girls in The Atlantic.

Or Twenge’s Has the smartphone destroyed a generation? It is an article that is controversial in its claims as well as the title the magazine gave to it.

These articles may be suitable for the outlets that published them. But as pitches for major policy changes, the articles were disappointing.

The articles relied too heavily on emotion and anecdotes in which people said and did just the right thing to reinforce Haidt and Twenge’s point. There were some scientific studies being cited, but only very selectively and not convincingly. These studies could not serve as the backbone of the case that smartphones are so bad for teenage girls’ mental health that they should be taken away. These were breathless advocacy pieces pushing a particular point of view and inciting public action consistent with this point of view.

I was amused and then annoyed by the articles. I wanted to argue with Haidt and Twenge first about how it is difficult to turn these sentiments –if these feelings are really based on best evidence — into effective action that really does anything to alleviate a complex problem.

Second, my mongrel progressive-libertarian sense of American freedoms was offended by any proposal for the government to restrict one group from doing something other groups in society are doing, particularly when that group is defined by the intersection of gender and arbitrary age.

Were Haidt and Twenge really prepared to restrict girls’ access but not boys? Haidt and Twenge’s admission that adverse effects on boys were not of public health significance? If boys were to be included, how could that be justified?

The public may become alarmed by what experts say about smartphones being dangerous to some people. Are they willing to take the drastic measures these experts propose? I doubt that they would, once the sense of urgency fades a bit. I think it would be a sufficient lesson for parents to attempt to implement these views in their own families. They will get a sense of how arbitrary this take on smartphones is and how difficult to enforce the prohibition of their use.

Most people vitally depend on their smartphones. Their lives would become hopelessly chaotic and terribly lonely if their smartphones or Apple watches were suddenly seized. It would be difficult or impossible for them to insist on their teen girls giving up their usage, while the adults and brothers maintained their use, without apparent damage to their mental health or becoming suicidal.

We should expect a lot of backlash if Haidt and Twenge’s proposals are taken seriously and acted upon.

Imagine someone being interviewed on CNN and defiantly shouting into the camera:

“The US government will have to peel my dead fingers from my iPhone if they want to take it from me.”

Or on Fox News:

“First they come for our cigarettes, now they are taking our smartphones, next our guns… and then they will require us to be vaccinated and to get regular testing for us to earn a living.”

Actually, resentment over public health measures needed to control COVID probably makes this a particularly bad time to attempt restrictions on smartphones.

Let’s imagine a skeptical journalist looking for something disastrous happening and coming up with the headline.

“Teen girl poisons family after stepfather announces he is taking away her smartphone because it is making her depressed and suicidal.”

Joking aside, I am taking Haidt and Twenge seriously. I assume that they are not just trying to sell books. In good faith, they are making a serious proposal based on their assessment of the public health consequences of teen girls having ready access to smartphones AND their conviction that restricting access to smartphones would save lives.

I am writing this article to convince Haidt and Twenge (with dim prospects of succeeding) and anyone who wants to listen that they should abandon this effort because the numbers are so stacked against them. Indeed, we need to stop making fluctuations in the low rates of suicides in girls a focus of moral outrage. Actually, we should stop predicting tsunamis of mental health problems are hitting our shores.

I bring a perspective to the table that has been missing. I have more than two decades of involvement in various capacities in implementing and evaluating ambitious multilevel community-based programs to improve care for depression and reduce suicide.

I moved from single-country projects in the United States and Germany to multisite programs across Europe. The investigators went big in their vision, in part because funders were convinced their goals were worthy of bigness. Honestly, the investigators also went big because they knew huge samples were needed to detect any effect on deaths by suicide.

The last article from these projects on which I am a co-author reported results of coordinated efforts in four European countries, each with an intervention region in which a lot of resources were poured in and a control region where care for depression and efforts to prevent suicide went on as usual. When the results were in, we agonized but felt compelled to report transparently that we failed to observe a statistically significant reduction in the designated primary outcome of deaths by suicide plus serious attempts at suicide.

We can come with reasons why we failed to find an effect in the different countries or overall, assuming that the intervention would otherwise be effective. It certainly did not help that the intervention region in Portugal had serious flooding in the area when the intervention was being implemented or that Portugal was facing a major downturn in its economy during the follow-up period.

I want to preserve hope that some specific interventions or a combination of them could substantially reduce deaths by suicide. Or that some components would be justified in their dissemination and implementation because they achieved other valuable goals in some outcome of interest other than reduced suicides.

I hope we learn all we can by past efforts with qualitative process analyses and post hoc analyses of particular components within and across the four countries. These studies and the ones that proceeded it did not fail as scientific evaluations. They just failed to find a reduction in deaths by suicide.

However, the numbers of suicides in many countries are just too small for an intervention to show an effect, no matter how simple or complex. That is something we can readily establish by checking some numbers before we get enthusiastic about intervening.

Here are some basic numbers I assembled for the Canadian government. I wanted to underscore that it would be difficult for a Canadian national framework for preventing suicide to show a measurable reduction in suicide in a reasonable follow-up period. Politicians should be prepared for that outcome if anyone was watching and demanding evidence that the program reduced suicide.

A further complication is that Canada is now one of the most ethnically diverse countries in the world. Presumably, interventions would have to be tailored to regional differences in current services, great ethnic and racial differences (White vs First Nation), and that ethnic/racial minorities are quite diverse themselves and small groups are scattered across one of the largest countries in the world in terms of landmass.

Every life is precious and every premature and preventable ending of a life is a tragedy for someone, whether at the individual, family, or community level. That observation quickly becomes an emotional issue, but it should not lead to a squandering of scarce resources on interventions that we have no reasonable expectation that will make a difference. If we still want suicide prevention with no evidence that reductions in suicide will result, we should call them something else.

With all the emotion attached to suicide in teenage girls and the whipping up of outrage attached to their having smartphones, I am surprised that neither the outraged nor those calling for calm give much attention to readily available numbers.

Suicide in pre-teen and teen girls: The numbers

Preteen and teen girls are among the lowest risk groups for deaths by suicide in the US and elsewhere. I cannot think of a worse group to expect an effect of the intervention.

Researchers have a good reason for expressing rates of suicide in children and teens in terms of rates per million. It allows statistical comparisons in samples and time periods in which the denominator (number of children or teens overall) changes. It is too confusing to talk about raw numbers when making comparisons.

The problem is that we stop thinking in terms of actual numbers: How many kids an intervention would have to target to reduce the number who died. Here comes a shocker:

Between 1993 and 2012, a total of 657 children aged 5 to 11 years died by suicide in the United States, with 553 (84%) who were boys and 104 (16%) who were girls.

And

Between 1975 and 2016, a total of 85 051 suicide deaths were identified for youth aged 10 to 19 years in the United States (68 085 male [80.1%] and 16 966 female [19.9%]), with a male to female IRR of 3.82 (95% CI, 3.35–4.35). Beginning in 1975, the trend in suicide rates for youth aged 10 to 14 years increased 5.4% annually for female individuals until 1992, and 4.5% annually for male individuals until 1993.

Those are much larger numbers, but they are from a span of 41 years. So we are talking about an average of 414 teenage girls dying by suicide on average (noting that number is trending upward).

Here are overall numbers of young persons in these age categories in 2016

The ratio of girls to boys is 50.1: 49.9. and so we are talking about roughly 425 girls aged 10 to 19 dying by suicide out of a sample from a population of 20.54 million. Put differently outcomes would be 425 deaths by suicide versus 20.49 million teen girls who did not die by suicide.

Now think of these figures as the basis for calculating the likelihood of observing an effect of a mass intervention that was not due to chance or other factors. Think of a population-based intervention for 20.5 million people expected to produce a reduction from 425 suicides of public health significance. Consider all the things that could go wrong, including unanticipated consequences of such an extreme intervention carried out on a population basis or the more likely inability to implement the intervention as planned. Anyone who has watched their first major implementation wobbly stumble into a region or country can smile.

Of course, the relatively infrequent event of individual suicide is determined by multiple causes. Even the most wild-eyed advocates of taking away smartphones would concede that at best, we would expect access to smartphones would play a decisive role in only a minority of cases. And that denial of access to smartphones being a decisive factor would greatly across regional and sociodemographic variables.

For comparison purposes, we are talking about differences in groups or cells in a table that would be tiny, with lots of cells (like deaths by suicide among girls in Utah being allowed to keep their smartphones versus those whose use was restricted as part of a study).

We could quit there. But let’s pile on some other realities instead. Individual suicides are not predictable from single risk factors nor any combination of known risk factors. Almost half of all deaths by suicide are not associated with an elevation in any established risk factor.

Lots of risk factors can be shown to predict death by suicide in research studies. Investigators commonly report suicidal thoughts or depression as having a statistical association. But the performance of a risk factor in predicting suicide is not a matter of correlation, but of the balance of true and false positives and true and false negatives. Neither suicidal ideation nor clinical depression performs very well. Go back to our 425 deaths by suicide versus 20.49 million young girls who do not die by suicide. We could confidently predict that most girls with elevations in suicidal ideation would not take their own lives and that we could not predict who did without a huge number of false positives.

You don’t need to be a statistician or an experienced gambler to see that you should bet that a given girl won’t die by suicide in a year, regardless of whether she scores high on suicidal ideation or is clinically depressed or both.

There are lots of things that we can do in the service of preventing suicide that are beneficial in themselves, even if we cannot dream that they actually reduce deaths by suicide in an available sample in a reasonable follow-up period.

My favorites include improving the quality of care for depression and restricting the quantity of prescription drugs available that could be used in a deliberate overdose. I think I could get a major consensus from experts on the prevention of suicide for these selections. I have no hope of winning such a vote with restricting smartphones. I would be embarrassed to ask.

That is why I am dismissing the idea of banning smartphones with the goal of measurably reducing deaths. I do not even need to imagine all of the chaos and unintended consequences of attempting a ban. I do not have to invoke a vision of the kind of authoritarian government that would even contemplate such a measure.

I think this is evidence-based common sense.

I know that in this article, I have been unfair to Jonathan Haidt and Jean Twenge, but I can fix this in my next article. I have cast them as advice-givers in the popular media with previous books for sale and maybe a forthcoming book in the works. Anyone who knows much about me knows my aversion –call it bias — against such activities done in the name of promoting health and well-being with weak evidence.

Yet, Jonathan Haidt and Jean Twenge have done something amazing that most such advice merchants do not do. They maintain an ongoing open-source literature review to which researchers are invited to add studies or offer a critique of what is already there. I checked with Jon and he indicated this includes a personal invitation to me and any other researcher already has an invitation.

I am working on a commentary that is responsive to that invitation and I will be posting it shortly. This article is being posted first because I strongly believe my criticism is pre-emptive. We won’t change or postpone the verdict with an appeal to “Further research is needed.”

Regardless of any conceivable research that can be conducted or assembled in a review, do not expect restricting smartphones will reduce suicides. It is more plausible than restricting exposure to smartphones to reduce brain cancer, but still well outside the realm of possibilities in accomplishing its aim.