Thirteen Reasons Why Not

The anxious about kids these days generation

it’s gonna make sense i promise

A friend celebrated she graduated last week and we all got together to celebrate; one of the guests was a former close friend who talked about her experiences as a schoolteacher - kids were all disrespectful, cruel, and made crude jokes about each other. Whether or not this was the case when we were in school (it was) and why was a topic of conversation for a long time (most of my participation was about how Meghan Thee Stallion did not deserve any blame).

Social media usage and teen behavior, particularly concerning mental health, has also been a persistent topic of social media discourse. I don’t think that testimonials by high-powered social media users are of any use because, well, the people at the casino all say they’re having fun.

But it’s a very interesting question because I agree with the fundamental premise, that teen mental health is worse and that social media plays a role, but also I think that most of the evidence in favor of the narrative is garbage and that it’s being pushed by people with an extremely specific agenda in mind. So let’s see.

The Anxious About Kids These Days Generation

The main purveyor of narrative in this area is the American journalist Jonathan Haidt, author of a bunch of books and articles with titles like The Wokening of the American Mind about how college students complaining about the bahn mi at Oberlin are just like the Gestapo or whatever. In all seriousness, his recent book titled The Anxious Generation is about teen mental health and social media use - specifically, that the mental health of teenagers is much worse, and that this is due to social media usage.

I have not read the book (disclaimer) but I did listen to a podcast reviewing it, and have read a few pieces by Jonathan Haidt preceding and postceding (?) his book’s release. Here’s a brief overview, and he’s spoken in the US Senate and done an interview with Yascha Mounk explaining his argument in some detail:

There's only this theory. It's the only one that can explain the timing of it. We have the correlation of evidence, we have the longitude of evidence, and we have the experiments. What more can I do to persuade you that this is what's happening? Oh, I have an idea. Let's ask the kids. Let's ask the teenagers. What do you think it is? And what do you think they say? Do they say it's school shootings or global warming? No, they say it's the phones. You know, they might enjoy their particular phone because they're in a trap. (…) But the teenagers themselves think that the phones are messing them up. And this is the most amazing thing to me.

When you look at what Gen Z is writing about this, they're all telling the same story of devastation. If this was a murder case, you have a body, you have a guy holding the gun, you have eyewitnesses, you have ballistics tests—at some point you have to think maybe this guy did it.

The core facts are as follows:

In sum, all five Anglosphere countries exhibit the same basic pattern:³

A) A substantial increase in adolescent anxiety and depression rates begins in the early 2010s.

B) A substantial increase in adolescent self-harm rates or psychiatric hospitalizations begins in the early 2010s.

C) The increases are larger for girls than for boys (in absolute terms).

D) The increases are larger for Gen Z than for older generations (in absolute terms). 

Haidt’s view (I can’t find it but here’s Twenge on it) is that phones harm teen mental health through multiple main channels: first, that kids are using their phones instead of going outside to play. Second, that kids aren’t getting enough sleep and that’s worsening their mental health (fact check: true). Third, kids are getting bad ideas from social media that are harming their mental health (the “reverse CBT” theory that is kind of crotchety). Fourth, that kids are getting cyberbullied or exposed to unhealthy but preexisting social dynamics. Fifth, that kids are becoming addicted to their phones (a bit superflouous), mainly through the fact that social media started being on your phone in 2011ish and that specific technologies are embedded in modern social media (infinite scrolling, algorythmic feeds, etc). And Haidt thinks that, for adults, it either doesn’t have much of an impact or at least the impact is their own fault - you’re a grown ass person and have to own up to your social media use decisions.

Now, unlike many other cases, Haidt is an actual psychologist and I think he gets several things right. First, I like that he’s using actual evidence and research and not just “many people are sayign” and like two graphs (looking at you, David Leonhardt). The facts, I think, are quite straightforward: in the last 15-ish years, social media use has become widespread, especially in the English-speaking world. At the same time, many other things happened, one of them being that teen mental health has gotten substantially worse (particularly for women).

So what’s going on? I’ll get into the research in a bit, but to give credit where credit’s due, I think Haidt & co are fundamentally right that there aren’t really many plausible explanations: the Great Recession doesn’t account for differentials in trends between boys and girls or for most parts of the timing, political developments have not been uniform enough between countries (and also the timing is just not right - liberalism was dominant in 2012 through 2016 in the Anglosphere). Plus, fundamentally, it can’t really be that the “outside world” is so traumatic when, like, you don’t see the kind of spike we’ve been seeing recently in events like 9/11, and it doesn’t really stand to reason that “things getting worse” would hit teen girls that badly.

Mental case

Fundamentally, the problems to (dis)prove the Haidt view are threefold: first, there has been a lot of growing awareness of mental health, especially for kids, since at least the late 1980s - the movie Heathers shows a world with enormous concern about rising teenage suicides, you had the eating disorder worries of the 1990s, teen pregnancy in the 2000s, bullying prevention campaigns in the 2010s, etc. So that means people’s mental health can be as good or bad as ever, but they just have actual vocabulary to express it (I think a big part of the differences in self reports between liberals and conservatives are this). Secondly, a lot of the data is very bad: people are notoriously bad at estimating their own mental health (women in particular), how much time they spend online, and spending 5 hours online a day can mean anything from texting with friends (good) to reading articles to watching porn to, IDK, getting brainwashed into Woke or whatever. Thirdly, many studies find a correlation between mental health and social media use, but that’s not indicative of causality, in fact quite the opposite - some research points to low mental health causing intensive social media use and not the other way around. That’s why authors tend to try to stick to longitudinal studies (tracking the same people over time) and causal studies (where people who did not use social media before start using it, but some other group does not use it).

So what’s the evidence for the Haidt view? Well, he has a frequent research partner called Jean Twenge (very scientist sounding name ngl), and they have compiled a behemoth spreadsheet of the evidence in their favor. There’s evidence that earlier use of social media is linked to posterior life unhappiness, teen loneliness has drastically increased and is positively correlated with smartphone use, and a bunch of other stuff (the Haidt/Twenge review doc is around 66 pages long). Howevere, there are also a few studies in direct contradiction: some longitudinal studies don’t find a clear association, and some causal studies also don’t find a clear association, plus meta analyses also don’t 100% line up with the Haidt/Twenge explanation. There was also an Instagram data leak saying that they blame themselves for bad mental health, but the data quality is also abysmal, so I think it’s kind of an own goal to rely on it. Fundamentally the data is quite scattered around and not very consistent, and some of it may not even be replicable from public sources.

Fortunately for us, there’s a bunch of papers by economists on the topic (for some reason), two of which are summarized here (in Spanish). Firstly, let’s look at this paper by David G. Blanchflower, Alex Bryson & Xiaowei Xu - they find that there has been a worsening of unhappiness in the UK and US, and that it has gotten more acute in the last five years or so (which is kind of compatible with political explanations of any kind, to be fair). A similar subset of authors (at least Bryson and Blanchflower) also do similar studies on the United Kingdom, and France, Italy, Germany, Spain, and Sweden, all with similar enough findings. Blanchflower & Bryson also find that (cyber) bullying and other such maladies affect mental wellbeing, and other studies find that childhood depression is really damaging for future prospects and that sleep deprivation negatively affects the performance of children in the future.

So this is serious stuff. But there’s two main economics papers about the Haidt/Twenge view, that seem (in principle) confirmatory. Firstly, there’s “Social Media and Mental Health” by Luca Braghieri, Ro'ee Levy, and Alexey Makarin. The paper utilizes quite a clever methodology: back in 2004, Facebook (still called “The Facebook”) wasn’t available everywhere, but rather, on select college campuses. This means that there were a lot of places that got The Facebook at similar-ish times and with clear comparison groups - so you can subtract the specific effects of “bad mental health schools” (somewhat infamously, Cornell), demanding schools (MIT), and “party schools” from the comparison. Before The Facebook got implemented, students had similar mental health outcome patterns (that is, it wasn’t increasing or decreasing before), but after, their mental health unwellness indicators spiked - a sign that The Facebook made things way worse. A different paper, by Hunt Allcott, Luca Braghieri, Sarah Eichmeyer, and Matthew Gentzkow, finds that a group that voluntarily took a break off Facebook for a short period (in exchange for money) had better self-reported mental health outcomes - particularly, because they spent more time outside. Fundamentally, each study has huge caveats (at least for Jonathan Haidt and Jean Twenge): the The Facebook study happened well before The Facebook was Facebook (and thus didn’t have basically any of the features that they thought cause harm), and the voluntary Facebook break study does not really have the same magnitude of an effect that correlational studies find, and is also heavily based on self-reports and other such rubbish.

There’s also one final controversial possibility: that the whole post-2012 impact was due to… Obamacare, of all things. In 2009, Barack HUSSEIN Obama signed the Affordable Care Act (Obamacare) into law, which famously gave health insurance to a lot of poor people, but also did a lot of tinkering around the health system itself - including demanding a lot of changes to mental health reporting. In fact, most data finds two important inflection points: 2012 (strengthening claims that it’s the damn phones with the damn Facebooks) and 2016 (strengthening the claim that it’s Donald Trump). But in fact, according to Adriana Corredor-Waldron & Janet Currie’s paper on the subject, it’s caused by something else: Obamacare reporting requirements that started when changes in mental health screening recommendations got into place in 2012, and changes in diagnostic criteria for suicidal ideation in 2016. In fact, medical training into spotting mental healthcare is a huge deal: doctors with higher diagnostic skill are more capable of spotting and treating mental health disorders. Fundamentally, the rise of teen mental health unwellness (including hospitalizations) is not separable with increasing interest in the phenomenon by the medical profession that is contemporary to it - for example, my girlfriend mentioned how an acquaintance had died, and by the obituary, it was not clear if it was an (accidental) overdose or an (intentional) suicide. Such distinctions are hard to make and depend heavily on the diagnostic criteria, training, and capabilities of doctors.

So the evidence, as a whole, is intensely muddy and fairly contradictory - a lot of very objective indicators, like hospitalizations and suicides, are up (but also there’s a lot of changing guidelines in the medical profession), and a lot of subjective indicators, that are fairly unreliable, show also a significant worsening trend. In a rare instance of a good take, Jesse Singal has it right: we just have to take it seriously because something is clearly up but to not get tangled up in “kids these days” discourse.

Why Nations Post

So, why do kids use social media? It’s an important question, but not really one that gets featured a lot in this debate. Why do people use social media? Well, they produce and they consume content, aka “posts”. Users value producing posts (because they get attention), but they also value higher quality posts more than lower quality ones, which means that some users have more followers than others. This means that people have incentives to post, but also to agree to see other people’s posts - follow them back, for instance. This means that people follow individuals with “posting ability” equal or greater to their own, and are followed by people with “posting quality” lesser or equal to their own. I have 16,000 and change followers on Twitter, and many people say my posts are quite good. If you look at Econ Twitter, you can see that in action: people with few followers follow most of them back, and people with many followers follow fewer other people, most of which have many followers themselves. You can also model social media as a form of consumption, either of information or of entertainment. If it’s information, it can motivate people towards action; if it’s entertainment, it provides them empty joy. It can also brign similar or different people together, and incentivize more or less hostile relationships.

The important thing is that being popular online can provide a form of intangible benefit (social capital) that is quite valuable and quite immaterial - people want to be popular online, especially among other people similar to them (not everyone is as contrarian as Matt Yglesias). I feel like this provides a fairly stable framework to kick off discussion, but it’s also quite, well, robotic - it would explain influencers, but not really “friends”. I follow quite a few accounts with very few followers, who are mostly people I’m (IRL or online) friends with. I made a friend with whom I talk about movies and tv a lot, and my girlfriend and I mostly talked on Twitter on the early stages of our relationship (now it’s primarily Whatsapp and Instagram). I feel like the social aspect of social media is quite lost on researchers, who are mostly focused on the information aspect - i.e. how they use social media (mostly to find other researchers¹).

In fact, the main reason why people use social media is because other people use it. The Spanish word for social media is not “medios sociales” but rather “redes sociales” - social networks. The most important part, to most people, is talking on social media with other people. This can be good (keeping up with your friends over Instagram) or bad - a large number of people who use social media would rather not use certain or any social media sites, and a lack of ability to control time spent on social media seems to make up an important share of use time (another Haidt/Twenge win).

In fact, this duality of being upset about social media but using it anyways to keep up with others, is present even on the depressed teenagers themselves. A study that interviewed a group of 30 teenagers who had attempted suicide found that they saw good and bad uses of social media. The bad uses were seeing things that were “triggering” to their mental health, trouble controlling time using social media, people being hostile or rude, their friends expecting a lot from them, and negative comparisons between themselves and each other, particularly concerning metrics (likes, retweets, followers, etc.). The positive uses were meeting affirming or similarly interested users, connection and support from their social group, and resources for mental health. One of the more notable findings is that a frequent cause of mental health issues from social media is feeling left out - knowing which of your friends are hanging out without you, for example.

I’ve had the same friend group since high school, so brief example: three times, it’s come out that there were subgroups that made plans without each other. The first time, it was when we were 15 or 16, and it was fairly uncomfortable, even if none of the smaller groups I was in ever talked about any of the other ones unfavorably. The second time, at 19 or 20, was far less tense, and it was simply noted that “of course we do things without each other”. And recently, we’re all fairly open about hanging out with subsets of people - for instance, I have one friend who watches the same tv shows as I do, so we debrief House of the Dragon or discuss Succession or Yellowjackets, one friend who uses Twitter as intensively as me; one friend who liked the same video games I did a hwile ago, and one friend who listens to similar music as I do, and those are topics of conversation between each pair of us. Likewise, some subgroups of my friends like movies, some like KPop, or some like going out partying, and it’s fairly obvious that we don’t all have to do or talk about those things together. And that’s okay to basically all of us, because we’re adult. But to teenagers, “FOMO” is serious - kids are just figuring out their place, and not getting invited to a hangout feels bad, even if it’s something that they clearly don’t like and clearly wouldn’t have wanted to go to anyways. Telling teen Maia that not everyone dislikes Marvel movies.

Posting Alone, Golfing for Trump

I’ve mentioned how social media confers a form of social capital, and I honestly don’t think that it’s a coincidence how the impact of that on self perception is very muddled, because social capital itself is fairly muddled.

There’s a different ongoing debate about whether people with high or low social capital voted for Donald Trump in 2016 and 2020 (deaths of despair are not real and I’m not talking about them), which is also linked to similar debates about whether the biggest supporters of the the Nazis were low or high in social capital. So is Trump just a more uncouth reedition of Mitt Romney, is he the tribune of economically anxious White Trash, or is he the voice of America’s aggrieved boat-owning Mittelstand?

A common theory, known as the “Bowling Alone” thesis, states that the disintegration of social groups and civil society organizations led people to become extremists - basically, that declining civic engagement and rising interpersonal inequality was creating a quasi underclass of people who didn’t have any meaningful social ties, so they became nazis (give or take). The term came from the fact that people didn’t belong in social groups, like bowling clubs, anymore, so they just went and did it alone. One example is that areas which had big declines in the membership of social organizations (churches, clubs, etc.) had much higher shares of Trump support than similar areas without those declines. Areas with lower social trust (as measured by online surveys) also have higher Trump vote shares - very Bowling Alone.

The sociologist Arlie Russell Hochschild proposes an alternative interpretation: that Trumpism is not driven by the rage of those stuck in drying out ponds, but rather, by the big fish who inhabit them - Trump is not the candidate of the victims of globalization, but rather, the candidate of the local elites of the places where those people live. If you analyze the voting behavior of people who are not particularly well off but who are much wealthier than their neighbors, they are the footsoldiers of Trumpism in small communities. Trump was the candidate of hillbillies and rednecks like the first paper suggests, but he’s also the candidate of car dealership owners with weekend boats, like the second paper suggests.

So what’s going on? Well, what even is social capital - it’s membership in organizations, like churches, clubs, sports leagues, etc. The problem is that all these organizations are not created equally: I’m sure that Trump lost voters who attend ceramics classes decisively, but he also won voters who go to church decisively. In fact, the support for Trump grew very strongly in declining locations with very strong social capital networks - the first paper omits areas that never had high social capital, so membership never declined, and areas that never lost their social capital.

Fundamentally, this points to the fact that Trumpism is driven less by material concerns (economic anxiety is largely bogus) and more by symbolic and cultural concerns - largely “are people like me being sidelined”. While the question of whether Trump is a fascist is fraught, I feel like his movement is made up of angry enough people who hate democracy and want a dictator that at least some analogies are merited. In this sense, not even the original example of Bowling Alone was true: places with high social capital had higher, not lower, support for the NSDAP - and the authors explicitly mention bowling clubs as a Nazi-leaning organization. If you understand fascism as some weird disease of advanced societies, it makes sense: some people are very angry about their loss of status, so they form an angry, hateful, dictatorial mob that wants to put people like them in charge. The nature of social capital, thus, depends on the nature of the associations and of political orders: some social groups expose people to others different from them, or to people who are nice, and others expose them to hateful mobs. Homeowners associations, country clubs, golf clubs, and (some) churches are all very pro-Trump groups. And support for the Nazis was actually concentrated in the “left behind” areas of Germany, squaring the circle here: people who belonged to organizations like veterans’ clubs became Nazis, while the flock of tolerant Catholic bishops opposed them.

In sum, there’s “good civil society” that exposes people to tolerant Catholic sermons, there’s also “bad civil society” that turns them into screaming bigots. Similarly, and I’ve already made this comparison before, there’s “good social media” that connects teens to their friends and other nice, pleasant individuals, and “bad social media” that connects them to Andrew Tate or self harm. People who retire early are substantially less happy than they expect, mostly because they get lonely while all their (few) friends are off at work and their kids aren’t around at home. And economics PhD students are significantly more prone to depression than others, which many have pinpointed as resulting from the fact they’re constantly comparing themselves to others and having their shortcomings pointed to them.

I think that this is what’s happening to teenagers - some are immersed in “good social media”, where they get to catch up with their friends after class because of declining outside time (largely because of parenting decisions), or “bad social media”, where they develop various antisocial attitudes as a result of experiencing comparisons that starkly highlight their own lack of popularity, standing, or larger cultural cache. Teens who self harm are just the “your friends are posting bad pictures of you on purpose” version of adults who support Donald Trump. Remember the paper about people who turned off their Facebook? It showed that active users were in fact happier than doomscrollers, mainly because they actually used it for connecting with others and not for passively ingesting upsetting information. And the The Facebook paper doesn't really match basically any of the Haidt/Twenge claims because The Facebook didn't have any real features besides a chronological feed, but does match the idea that people were getting factual comparisons to other people - particularly their social life. Good social media and bad social media at play. Bad social media is when I had just had surgery and spent days on end watching Tiktok in bed. Good social media is when I use my Instagram to keep up with people, or Twitter to make friends and hang out with them in DC.

Conclusion

So, all in all, I think that there is a teen mental health crisis, and I think that it’s probably linked to social media, but I don’t appreciate the lack of nuance. This is a big, serious issue, and I don’t like how it’s quietly turning into some kind of weird debate about “woke” - which, to be fair, is how Jonathan Haidt sees it himself: he came into it from his Wokening of the American Mind projects, and he thinks that a substantial component is progressive politics teach children “reverse CBT” where tenderqueer theyfabs yell at you about climate change, Palestine, and to hate yourself because you’re White. While this world where trans is the norm and straight is a minority is almost certainly imaginary (and it’s fairly funny that the reverse CBT guys are all in on constantly pissing themselves off by ingesting copious misinformation about transgender sports), it’s also derailing an important conversation about whether kids and families should change their attitudes significantly - and I think they should

1. Which to be fine, is a very good use of social media. It’s where I found a bunch of papers for this blog, and I am “oomfs” with a ton of really cool and interesting people, some of them even actual economists. A few people I know have listened to specific musicians because I retweeted stuff about them (mainly Mitski tbf), I am pretty sure that my posts about the album “brat” have gotten to the eyes of prominent people, and most hilariously of all, a few normie late-30s-to-early-40s economists mentioned they’d watched Bottoms, a very Gen Z lesbian sex comedy, because I really liked it.