Social media and phones are not the major cause of the mental health crisis (at least not yet)

My column on a recent study was invited and killed at the last minute

As I have written about here and in Science, I believe that the debate on social media and phones and their role in the mental health of young people is far from resolved. It’s a fascinating situation for me as one side of the debate (documented in the blockbuster bestseller The Anxious Generation by Jon Haidt) that claims that phones and social media are the major cause of the adolescent mental health crisis has a massive megaphone and the other side are scholars mostly publishing in peer-reviewed journals. This kind of asymmetry in scientific debates has played a major role in many interactions of science and the public over the decades.

When a recent study showing that locking phones away at school for three years had no or very modest effect on positive outcomes in the classroom was released, I sent it to an editor at a major national outlet who asked me if I wanted to write about it for them. I interviewed the authors of the study and got quotes from Haidt and one of the main critics who wasn’t on the study, Candice Odgers. I wrote the piece up and it was all set to run and then killed at the last minute the night before it was supposed to appear by the top editor.

I’m running the piece below. In the next three posts, I’ll run fuller Q&A’s with the two authors I interviewed and then something about what I think would make this debate more generative. Here is the first Q&A with Tom Dee.

I am not a fan of social media companies, in fact, I have tangled publicly with Facebook over their optimistic interpretation of papers we published and then over whether they were honest with the researchers. I do believe that phones and social media are partial contributors to poor mental health and that there may even be a causative element to it. But Haidt has claimed it was the “major cause.” We would never allow a claim like that in a scientific paper unless there was substantial evidence for it. Is it OK to put that claim in a bestselling book that has captured the public zeitgeist? Interesting question. Our friends over at Nature have argued that “public-facing science-communication work should adhere to the same research-integrity principles that are used for scholarly publications.” That is certainly not happening here.

The best deeply reported piece on this so far is from Stephanie Lee at the Chronicle of Higher Education. A quote from one of the academics in the piece that sums up the situation is that Haidt is “telling a story with Google Docs. Hundreds of academics who aren’t being listened to are doing science.” At least, he means, doing science by publishing peer-reviewed research articles and not Google Docs, commentary, and op-eds.

Here’s the piece:

Caveats don’t launch movements

It feels good to give up your phone or cut back on social media. Two years ago, I got off Twitter (now X) and Bluesky and haven’t looked back. More recently, I took my work email off of my phone and deleted the other social media apps (it’s not like my laptop is that far away). Also an improvement.

Anecdotes like this are real and plentiful, and they fuel support for asking young people to get off of their phones so that they can flourish, too. This idea is everywhere in the last two years thanks to Jonathan Haidt’s book The Anxious Generation, which has captivated the media and political discourse. But systematic studies that show benefits of giving up phones on youth mental health are hard to come by. So what does it mean if it feels good to give up your phone in the moment, but that never shows up in the data? And what if all the work to stop these evil devices and platforms won’t actually fix what ails America’s youth?

Because Haidt’s book and media appearances are so widespread, many in the public could understandably believe that the science showing the damage created by phones and social media is resolved. But it’s not. Other researchers have been measuring the changes on adolescent mental health parameters and failing to observe statistically significant changes when phones were restricted. Many of these researchers believe that Haidt has prematurely created a moral panic without sufficient evidence—and that the decline in adolescent mental health is more likely caused by a lack of parental support, economic distress, a decline in parental mental health, or many other possibilities.

I’m not an expert on this kind of research. But in my day job as the editor of Science magazine, I’m constantly watching scientists argue with each other—and then seeing how these debates play out in the public eye. In this case, one side has a massive megaphone and has dominated the public discourse while the other side is mostly putting their findings in scientific journals. What’s more, much of the public and the political class has embraced a simplified version of the evils of phones and social media when, in fact, the reality is much more complicated. What happens to public confidence if it turns out that the supposed gains from eliminating phones and social media aren’t enough?

***

The Anxious Generation was certain to be a bestseller. Haidt had already seized the public’s attention with his 2018 anti-woke barnburner Coddling of the American Mind, and his legions of fans were sure to buy this one. It claimed that the replacement of play-based childhood with phone-based childhood was the “major cause” of the international epidemic of anxiety and depression in teens. After all, social media and phones took hold right at the time that large increases in adolescent anxiety and depression occurred. And the increase occurred simultaneously for adolescent girls around the world, suggesting that locally different effects could be ruled out as potential causes. The four-fold solution to the problem, Haidt argued, was (1) no phones before high school, (2) no social media before age 16, (3) phone-free schools, and (4) more time playing outside.

Around the same time, though, a National Academy of Sciences, Engineering, and Medicine study came to a less exciting conclusion: “available research that links social media to health shows small effects and weak associations…. Contrary to the current cultural narrative that social media is universally harmful to adolescents, the reality is more complicated.”

The Anxious Generation was also met with howls from a number of academic practitioners, especially Candice Odgers, Andrew Przybylski, and Chris Ferguson. Odgers, for instance, wrote a scathing review in Nature. “I will probably use The Anxious Generation in my classes as a terrific example of how not to interpret correlational data and infer causation,” she wrote.

Nonetheless, The Anxious Generation has been on the New York Times bestseller list for 106 weeks, and Haidt has spread his message far and wide. Politicians on both sides of the aisle have embraced his framing enthusiastically, and phone and social media bans are beginning to be rolled out around the world. In 2024, President Joe Biden’s surgeon general, Vivek Murthy, called for a warning label on social media, calling it an “important contributor” to the youth mental health crisis. (Note that he didn’t say it was the major cause.)

In the words of Arlo Guthrie, “And friends, they may think it’s a movement.”

***

Since the publication of Haidt’s book, the two camps have traded studies and analyses. On one side, more studies have emerged showing negligible to small changes in well being for the time spent on social media. Meanwhile, Haidt has controversially cited surveys of researchers generally supporting the idea that social media and phones have a negative (but not necessarily major) effect on mental health.

A couple of weeks ago, a study by researchers who are generally hopeful about phone bans but not strictly in either camp was posted as a preprint, meaning it has not yet been submitted to a journal and peer reviewed. In this study, the researchers were able to get three years of data on the use of pouches made by a company called Yondr that lock phones away for the school day. The researchers set out to determine —and publicly stated the methods—whether locking phones during school would lead to improvements in test scores, attendance, and discipline. The researchers also collected student-reported data on their well-being, classroom attention, and extent of online bullying.

Instead, the study showed no detectable effect on test scores or attendance. Although discipline problems increased in the first year they recovered back to the baseline after that. Student well-being also went down in the first year and then improved modestly above the baseline. There was no statistically significant change in perceived online bullying or, more strikingly, in classroom attention. If taking phones away made it easier to pay attention, it didn’t register.

While advocates of phone bans might latch onto the improvement in self-reported student well-being, one of the authors, Tom Dee of Stanford University, told me the study had failed to realize the hypothesized benefits: he called the results “sobering.” Still, Dee hopes this isn’t the final word. He is hoping for more experiments to nail down whether looking over periods longer than three years or at other measures would support the idea that phone bans improve student learning.

Another of the co-authors, Hunt Allcott, also of Stanford, told me he was very appreciative of the spotlight that Haidt has put on the field. Still, he characterized the effects measured in the study as “broadly small.” Allcott’s other studies are among those showing an improvement in reported adult well-being when social media is restricted. He’s generally on board with phone bans. But when I asked him if phones and social media were the major cause of mental health declines, he said, “I don’t know.”

Allcott said that his findings alone were not enough to justify a book like Haidt’s. What he finds compelling about Haidt’s argument is that he has produced a “collage of evidence” that includes some strong anecdotal and descriptive evidence from surveys along with some forensic evidence, which—despite effects that some other researchers may see as underwhelming— is supportive. However, that forensic evidence remains hotly contested in academic circles, and even a fan like Allcott concedes that “the causal evidence is never going to be perfect.” Rather, he says, we have to rely on various types of evidence, including people's reports of what they perceive the effects to be, to complete the collage.

One thing is for sure, though: adults really love phone bans and restricting social media. The teachers in the Yondr study overwhelmingly supported the policies. Parents were enthusiastic about the bans and expected significant gains in test scores and mental health. I can relate to this. After 40 years of teaching college students, I’m acutely aware of the changes to the classroom resulting from devices and am much more satisfied teaching when they are put away. When I told Haidt that I benefited from getting devices out of the classroom, he said “most people seem to share your experience— that reducing the interruptions and distractions of our phones makes us feel better and be better able to concentrate. And teachers widely report seeing it in their students; they seem almost universally to love phone-free school policies.”

***

For someone who writes about disagreement in science, this has been an irresistible drama. When I first started writing about it two years ago, Odgers told me that “the most shocking thing is the distance between what people believe and what you can see in the data.” The Yondr study is the latest to bear it out. Parents and teachers believed that the phone bans would produce significant gains that ultimately weren’t there. Similarly, when Australia announced a nationwide ban on social media, economist Tyler Cowen called out the same problem in The Free Press, writing, “After much rigorous investigation, the harms are relatively small. Yet in headlines they are reported as major negative effects.” As Cowen predicted, Australia has had major challenges implementing the ban.

When I contacted her for this column, Odgers told me her view hadn’t changed, and she was emboldened by the Yondr findings. “What is clear is that adults feel good about banning phones and politicians know that supporting bans is a political win,” she told me. “What is not clear is that spending all this energy, money, and time on banning and blaming digital technology does anything to benefit our kids.” Indeed, as I was writing this, another preprint appeared showing “no clear evidence that the school ban policy reduced screentime or improved psychological wellbeing.”

Haidt’s “collage of evidence” includes both anecdotal and forensic evidence. The anecdotal findings—individual stories of shocking harms, tech companies behaving badly, or personal satisfaction with putting the phones away—are not particularly controversial. But Haidt’s forensic evidence is hotly contested. For example, critics say that when asking teens whether they had positive experiences giving up social media, they know the researcher wants a “yes.” And the effects measured are all either undetectable or much smaller than expected. These lines of evidence are likely to remain contested; even ban-friendly researcher Allcott is cautioning that convincing causal evidence is unlikely to emerge.

When I asked Haidt if he had changed his mind about phones and social media being the major cause of the adolescent mental health crisis, he stood firm. He certainly has been consistent: when I asked him two years ago if he should have included in his book more of the research that argued against his position, he said, “I don’t think so.” It’s a strange position for someone whose previous crusade was for viewpoint diversity. But then, caveats don’t launch movements: a recent study shows that public health messaging with too many disclaimers is largely ineffective.

***

To my mind, it is unlikely that the status of this field will change much. We will continue to drown in anecdotal information suggesting that phones and social media are dangerous, and that people feel good giving them up. Meanwhile, causal evidence that is convincing enough to create a scientific consensus will remain elusive. However, the phone-free movement has so much momentum that bans will continue to spread despite little demonstrable and measurable benefit any time soon. “No study is definitive, nor should it be,” Odgers said about the Yondr study, “but these findings should cause us all to pause and ask ourselves whether trying to ban our way out of a youth mental health crisis and learning loss is the best use of our limited resources and time.”

I worry more that the public will eventually lose confidence due to the overselling of the bans as the benefits continue to be unconvincing. The history of science is riddled with episodes of scientists ultimately undermining trust by enthusiastically overpromising benefits that never appear. A recent example is the error made during the pandemic of creating the impression that the Covid vaccines would completely stop transmission rather than just reducing it. (Despite what you read about me, I was one of the first people to call this out.) When that was walked back, the public understandably became skeptical of the other real benefits of the vaccines. Will that happen here if studies continue to show none-to-meager benefits of banning phones and social media? We don’t know yet, but it’s not a pleasant thought to contemplate.

Note: A few minor changes were made to this after the original posting. The most important is that the original piece characterized the pre-registration in the Yondr study as publicly hypothesizing that an effect would be observed instead of saying that the authors were just making a commitment that certain things would be measured. That has been corrected.