Social Media and the Youth Mental Health Crisis
Is social media having a "tobacco moment"?
In 1964, the Surgeon General's report Smoking and Health touched off a transformation in public consciousness. Although tobacco was already considered harmful, this rigorous, methodologically innovative document spurred legislative action and a national media frenzy. Within a year, federal law required cigarette packages to display that famous warning "Caution: Cigarette Smoking May Be Hazardous to Your Health" and soon the majority of Americans, whether they smoked or not, were more or less convinced of the hazards.
Eight decades later, social media is facing a similar intensification of concern. On January 31, the Senate Judiciary Committee grilled the leaders of Meta, Tik Tok, and other social media giants for not doing enough to protect kids against sexual predators, bullying, eating disorders, and other harms. Since late March, The Anxious Generation, Jonathan Haidt's condemnation of adolescent social media use, has remained near the top of the New York Times Best Seller list and is lighting up the blogosphere. Coincidentally, today, May 23, is the first anniversary of the Surgeon General's influential advisory report describing, among other things, the potential harms of social media use among young people.
The perniciousness of social media is among the few issues that tend to unite Americans across the political spectrum. In a Pew Research Center poll conducted this February, adults were asked what sort of impact social media has on the U.S. today. Here are the findings by political affiliation:
As you can see, partisan differences are outweighed by the prevalence of negative responses compared to the rest.
After tobacco had its "moment", data continued to accumulate, and nobody doubts anymore that prolonged use is harmful. In contrast, the data on social media use remains inconclusive, although some themes have emerged. As far as we can tell, social media is a mixed blessing for young people: It's harmful in some ways and beneficial in others, depending on platform, user, frequency of use, and the way it's used. As the American Psychological Association (APA) concludes, "Using social media is not inherently beneficial or harmful to young people." Thus, if social media is experiencing a "tobacco moment", we should be concerned, because the data doesn't show that it's categorically unhealthy in the way that tobacco products are.
In this two-part series I'll be sharing some of the recent data. My starting point will be controversies around The Anxious Generation. I believe that much can be learned from the way Haidt handles – and mishandles – the data in this book. Next week, I'll be discussing reports published this month, including a new survey that explores the topic from teens' perspective, as well as a new study suggesting that social media use neither helps nor hinders adolescents' social skills, but may foster offline friendships. (I want to thank the author of that study, Dr. Silje Steinsbekk at the Norwegian University of Science and Technology, for her perspectives, and for guiding me to some key meta-analyses on this topic.)
These newsletters are dedicated to a friend of mine whose daughter, an 11th grader, has had an intense, love-hate relationship with Discord, where she finds both community and, at times, a source of despair. (Who advises a 16-year-old to cut herself when she's frustrated?) Stories like hers are a reminder not to lose sight of individuals when exploring the data. As the APA suggests, few if any generalizations about social media will apply to everyone. This week's newsletter focuses on young people in general, only because this is Haidt's focus and, as academics go, he's unusually influential. Next week I'll say more about groups, such as LGBTQ+ teens, as well as individuals.
(I'll be using the phrase "young people", because age ranges are not always the same from one study to the next. Assume something like 13 – the age of internet adulthood – to 24, with some data on younger kids, and a lot of data on the teen years.)
Although we need to know young peoples' stories, aggregate data are important too. Thanks to statistical methods developed in the early 20th century (supported by computing power that began to grow exponentially in the latter part of the century), we have a historically unprecedented ability to study large groups of people, understand their experiences, and draw conclusions that might lead to changes in platforms like Discord, or at least help users engage with those platforms in ways that are more consistently positive. What we learn from groups of people, if interpreted carefully, can benefit individuals.
The Anxious Generation
Jonathan Haidt (pronounced "height") is a social psychologist at NYU and author of the new bestseller The Anxious Generation.
Haidt argues that during the time period 2010-2015, young people began to experience a "tidal wave" of mental health problems (mainly anxiety, depression, and self-harm) that has grown worse every year up through the present. The cause is social media use combined with increasing restrictions on unsupervised play. As Haidt puts it, childhood has been transformed from a play-based to a phone-based experience, and we should be deeply alarmed by the social media platforms young people access via smart phones.
The eponymous "anxious generation" refers to anyone born after 1995. For Haidt, this is Gen-Z, and it's their misfortune to have entered puberty right around the time smart phones and phone-based social media came into being (iPhones in 2007, social media phone apps in 2008, forward-facing phone cameras and Instagram in 2010, etc.). Social media "rewires" childhood, according to Haidt, undermining mental health by depriving young people of social interactions and sleep, fragmenting their attention, and turning them into phone addicts.
(From here on, you'll notice that restrictions on play drop out of the discussion. Much of the controversy around Haidt's book concerns the influence of social media. This is his emphasis in the book and popular newsletter After Babel.)
Psychologists tend to be skeptical that anything as complicated as mental health trends can be attributed to one or two causes. Much of the academic community either disagrees with Haidt or argues that he's merely captured part of what has driven the youth mental health crisis.
To be clear, critics tend to agree (at least in part) on the historical record: Teen mental health began to worsen at roughly the same time that phone-based social media apps became widely available. However, this may be (at least in part) a coincidence. After all, lots of important, distressing things occurred just before or during the time period 2010-2015, such as the 2008 recession and its aftermath. The fact that two historical events occur at around the same time doesn't prove they're related. Donald Trump was elected president the same year as a major outbreak of Zika virus, but I don't think anyone suspects that one caused the other.
I see four problems with the way Haidt treats the scientific evidence.
1. Cherrypicking.
Haidt is often accused of cherrypicking data that fits his thesis. Whatever his motives, this is a hard criticism to dodge.
In fairness, some of the data are clearly supportive. A good example is a 2022 American Economic Review study showing that as Facebook was rolled out across American colleges, rates of depression increased among student populations, while self-reported academic performance declined.
What's particularly strong about this study is that the researchers recorded exactly when Facebook came online at each college between 2004 and 2006, so that changes in mental health and academics could be linked to the introduction of the platform at each institution. This is stronger evidence than what Haidt sometimes presents (i.e., a spike in some mental health problem around the time phone-based social media became widely available). All the same, the Facebook study isn't slam-dunk evidence, because the researchers didn't track individual student use.
Even when the data support Haidt's view, it tends to offer mixed rather than unqualified support. In a 2022 review of 226 studies, Jeff Hancock and colleagues at Stanford found a sort of trade-off: More frequent social media use was associated with more anxiety and depression, but also with with greater social well-being (better relationships, a greater sense of acceptance from others, etc.).
Other studies show no overall relationship between social media use and mental health. An umbrella review of 25 reviews published between 2019 and 2021 found that in most cases, the association is weak or inconsistent. (Even studies that find an association may not indicate the direction of causality Haidt had in mind. Some data suggests that social media use doesn't cause depression; rather, teens who are already depressed begin to use social media more frequently.)
In short, Haidt fails to acknowledge much of the evidence that contradicts his thesis, and, except for a few brief passages in his book, he ignores data on how social media benefits young people. Social support, connectedness, opportunities for self-expression...there's a long list of positives that I'll touch on next week.
Bottom line? As one team of reviewers concluded recently, the literature consists of "a mix of often conflicting small positive, negative and null associations." Most experts seem to agree. Although we can point to individuals and individual platforms and speak confidently of benefits or harms (or, in the case of my friend's 11th grader, both), when we step back, it's hard to say whether social media use, on the whole, is a positive, negative, or neutral experience.
2. Small effects.
"The Anxious Generation" feels like the work of someone on a mission, if not a crusade. I admire Haidt's passionate commitment to supporting mental health, but I think efforts to be supportive will be hampered when the data are mishandled.
Associations between social media use and mental health are exceedingly small. I have not seen a meta-analysis in which key correlations, for instance, exceed .20. (Much of the time, the significant values fall in the range of .10 to .15.)
What does that mean in plain English? Let's take the example of depressive symptoms – sadness, loss of motivation, fatigue, etc. Researchers might ask use a measure that, in essence, asks a person how much they're experiencing each symptom, tallies up responses, and then assigns the person a depression score. Within any large group of people, depression scores will vary widely. Statistically speaking, a .20 correlation between social media use and depression scores means that 4% of the variability in depression among people can be explained by how frequently they use social media. Put another way, 96% of that variability can be attributed to other causes – stress, relationship difficulties, health problems, etc.
Most experts recognize that small associations mean that we only know part of the story. A correlation of .20 (it's not actually that high) would suggest that social media only has a tiny influence on how depressed people feel. This is not to deny that any one person can be strongly influenced. The point is simply that among large groups of people, differences from one person to the next in how depressed they feel are influenced by many things, and social media is not a major influence.
Setting aside the fact that in some studies, depression may spur more social media use rather than the other way around, along with the fact that some studies don't even find a relationship, what Haidt should be saying is: I've captured part of the story. I've found a contributor to depression that has at least a small influence overall.
Instead, in his book and in After Babel, Haidt makes a technical misstep. He argues that these small effects are meaningful – and worrisome – because they're cumulative. In other words, the effects will eventually snowball and become large. The mental health of young people who continue to use social media will deteriorate, as if they were taking small doses of poison each time they went online.
The logic here is questionable. As time goes on, anything that contributes to depression can have cumulative effects. If the correlation between social media use and depression were .20 (again, it's not that high), then as much as 96% of the variance in future depression might be attributable to influences other than social media use – conflict with family and friends, loneliness, stress, and, of course, a preexisting disposition to being depressed.
In short, the effects of social media use may accumulate, but so may the effects of many other things. A teen whose family conflicts worsen around the same time their academic stress increases may experience more depression, regardless of how much or little social media they use, because family dynamics and stress more strongly impact mental health. (By the same logic, that same teen might actually become less depressed during the same time period if they develop a close friendship with an amazing, extremely supportive person. Benefits can accumulate too.)
Bottom line? At worst, social media use only has a small negative impact on mental health. (Again, this is a statement about overall effects rather than individuals.)
3. Winner-take-all mindset.
Many if not most experts view the youth mental health crisis as complex and attributable to more than one cause. Even when they advocate for their pet theory (e.g., the economic downturn after 2008; climate change; the increasing prevalence of school shootings; skyrocketing drug abuse among parents) they acknowledge that the literature is mixed. As Amy Orben at University of Cambridge notes,
"Some reviews find that social media increases well-being, social communication, social support, social capital, authentic self-presentation and social connectedness while decreasing loneliness—even though these reviews routinely note that other studies have found exactly the opposite."
Haidt, in contrast, argues in effect that he's correct and everyone else is wrong. In the book he only acknowledges conflicting data in order to dismiss it, while in his newsletter After Babel he more systematically deconstructs alternative views.
I emailed Dr. Haidt to ask about the notion that his thesis is basically accurate but not the only cause of the youth mental health crisis. His response was curiously evasive. He wrote that he could only refer me to a 2023 Generation Tech post written by Jean Twenge (author of the 2017 bestseller iGen and another prominent critic of social media use). In this post, Twenge addresses 13 alternative explanations for the teen mental health crisis, arguing that they're all wrong – except for the one that's consistent with Haidt's view.
The problem here is not that Haidt (and Twenge) believe that Haidt is right. We all tend to feel that way about our beliefs. The problem is that what's being debated is not a clear-cut question such as who's going to win the NBA championship this year. Haidt tends to reduce a complex issue – one that his thesis might help illuminate – to a simplistic me-right/you-wrong dichotomy. If me right and you wrong, there's no room for what you say to help understand what's happening to young people.
Bottom line? As I'll discuss next week, a winner-take-all attitude isn't helpful for understanding or addressing the youth mental health crisis.
4. Interpretive bias.
One explanation for the crisis is that as stigmas about mental illness decline in countries like the U.S., young people are opening up more about their struggles. In other words, since the early 2010s, the actual incidence of mental health issues hasn't increased, or at least not by as much as we think. Rather, young people are just disclosing more.
One might add that young people are also getting better at articulating mental health concerns, and in some settings, they're increasingly encouraged to do so (e.g., by teachers and school counselors). They may even sense that doing so is expected. My friend's 11th grader told him that in one of her Discord groups, a couple of people not only pushed her to explore her feelings, but also encouraged her to reframe minor frustrations as signs of being angry and depressed.
To the extent that the openness hypotheses is accurate, social media use may have helped (or encouraged) young people to be more forthcoming about their problems, without necessarily causing declines in mental health.
Haidt argues against this explanation, because we also see "objective" evidence of the growing crisis. His main evidence is an increase in rates of suicide among young people since 2010-2015.
Unfortunately, the data that Haidt relies on here is messier than he acknowledges. (I assume unconscious bias here rather than a deliberate attempt to mislead.)
(a) Haidt focuses almost exclusively on percentage changes in suicide rates, while failing to add some important context: The actual rates are low, statistically speaking. For instance, among 10 to 14 year old girls, there were 0.84 suicides per 100,000 individuals in 2012. By 2020 that figure had risen to 2.01 per 100,000, an increase of 139%. Haidt bludgeons the reader with percentages like this but fails to note that the difference is roughly 1 girl out of 100,000.
My point here is not that such a difference is trivial. The loss of even one young person's life is a tragedy that exists apart from whatever statistics have to say. My point is simply that the extent of the problem, society-wide, looks quite different when you consider absolute rates rather than percentages.
I'm also not arguing that Haidt is completely wrong. I'm just suggesting that we don't know to what extent the growing mental health crisis can be attributed to greater openness about disclosing problems. Haidt and his colleague Zach Rausch also point to increases in hospitalizations and other therapeutic interventions for mental health issues, but these too could partly reflect a society-wide decline in stigmas around seeking mental health support, since parents are often the ones who press for, or at least support, such interventions. (Parents of "the anxious generation" might be growing more open about mental illness too, as some data hints.)
(b) Haidt's contention that suicide rates among teens have increased since 2010-2015 is too simplistic. Rates have increased among other groups as well, the timing of the shift is slightly off, and some trends are not accounted for. For instance, consider following graphic presented by Haidt and his closest collaborator Zach Rausch:
Haidt and Rausch add a vertical line for 2012, in order to emphasize when suicide rates began to rise for this age group. But looking at the entire figure, you can see that suicide rates for boys increased sharply from roughly the mid-1980s to the mid-1990s, then fell, then rose again after 2006, but never surpassed the peak value in 1990. Haidt's theory is unable to explain these rises and falls, and why they're specific to boys of this age group. Perhaps he's right about what happens after 2012. Alternatively – and more plausibly, in my view – there's no single historical event, such as the rise of social media, that's solely responsible any major fluctuations, even if such events influence certain individuals at certain times.
There are other reasons to question exactly how well we understand the extent of the youth mental health crisis (and, by extension, how much we can blame social media). For instance, as David Wallace-Wells has noted, in 2011, near the outset of the crisis, the Department of Health and Human Services issued new guidelines recommending that adolescent girls be screened annually for depression and mandating that insurance companies cover the screenings, while in 2015, H.H.S. required hospitals to record whether injuries are accidental or self-inflicted. These changes might allow more mental health problems to be revealed even if the actual rate of problems had not increased. (They might also be among the many contributors to greater openness about mental health that could be enumerated.) Wallace-Wells doesn't deny that adolescent mental health has been worsening in recent years. His point is simply that the extent of change may be less than we realize.
Bottom line? Young people are experiencing a mental health crisis, but we may not know exactly how much worse it has grown in recent years.
Conclusions
1. Young people have experienced declining mental health during the past 10 to 15 years. The true extent of decline is probably unknown.
2. Hundreds of studies have examined whether social media use has contributed to the youth mental health crisis. The data are incomplete and inconsistent. Some studies link social media use to poorer mental health, some studies document benefits, and some show no association. The majority of studies indicate no overall relationship, or small negative effects.
In this newsletter I've focused on the big question of whether social media, in general, is harmful to young people, in general, because Haidt makes the simplistic claim that it is, and this is an influential perspective, whether or not people have read his book. (When I emailed Haidt a thank you note for his message, the auto-reply indicated that he'd just left town for a month-long book tour. So, we might expect more media coverage added to the dozens – if not hundreds – of articles and reviews that have already featured The Anxious Generation.)
Next week, I'll show that the impact of social media use depends on frequency and type of use, as well as on the characteristics of users, and I'll discuss what can be done to promote more positive social media experiences. As you can probably guess, I don't think the best approach will be to demonize social media as the new tobacco.
Thanks for reading!