A Causal Link Between Facebook and Mental Health

Audio of this conversation is available via your favorite podcast service.

In 2004, Mark Zuckerberg launched “TheFacebook” at Harvard University before rolling the social networking site out to other students at Dartmouth, Columbia, and Yale. Soon, it was available on hundreds of college and university campuses, and thereafter the rollout included high schools.

Now, there are nearly 3 billion monthly active users of the site, and it is readily apparent that it has had a significant impact on society in a variety of ways. One such impact is on mental health. Researchers have found that Facebook use is associated with multiple mental health issues, ranging from anxiety, insomnia, depression and addiction to body image and eating disorders, alcohol use, and more. But while much of the evidence collected is concerning, most such studies have not identified a solid causal connection between Facebook and negative mental health, and many skeptics remain.

But in today’s episode, we’re going to discuss one study that does appear to draw a causal connection between the use of Facebook and negative impact on mental health with two its authors: Luca Braghieri, an Assistant Professor in the department of Decision Sciences at Bocconi University in Italy; and Alexey Makarin, an Assistant Professor in the Applied Economics group at the MIT Sloan School of Management.

What follows is a lightly edited transcript of the discussion.

Justin Hendrix:

I first encountered this new paper that you've written recently in a Twitter thread from Sinal Aral, who said that this was an important and amazing paper on the causal link between social media use and mental health. We're going to talk a little bit about what you did in this research and what your findings were. But first, can we just speak a little bit about what your research area is generally? Luca, perhaps we'll start with you.

Luca Braghieri:

Most of my research is in behavioral economics and political economy, and I started thinking about the effects of social media primarily from a political economy standpoint. There's a lot of talk about social media having large effects of producing large externalities on democracies, and also large political effects in, say, authoritarian countries. Alexey has done some work on that. That's where I approached the problem of social media to begin with, but once I started thinking about it, then a lot of other interesting questions cropped up, and in particular potential negative effects of social media on mental health, especially among teenagers.

Justin Hendrix:

Alexey, what about you?

Alexey Makarin:

Luca and I, we have a relatively similar background. Most of my research is in a subfield of economics called political economy, which is sort of the intersection between political science and economics. We study how politics affect economics and vice versa. Part of my research was on the impact of social media. That's how I got into that. As Luca mentioned, we had this paper on the causal impact of social media on anti-authoritarian protest participation in Russia, and that got us thinking about what else can social media affect? Of course, social media is on top of everybody's minds, so eventually we were hoping to write something about that, and I'm very happy that we were able to do it.

Justin Hendrix:

Let's talk about this July 2022 paper, Social Media and Mental Health, which you two co-authored with a third author, Ro'ee Levy. Social media, mental health, particularly mental health among youth, is of course a hot button issue. It is the kind of underlying issue in a lot of legislation focused on online harms. How did you get into this topic? What were the big themes that led you to do this research?

Luca Braghieri:

Yeah, so we started off by noticing that the mental health trends of adolescents and young adults in the United States and in various other countries have worsened considerably over the last 10 years. Since the point in time which those trends began to worsen is around the time in which social media was introduced, then that led many to speculate that social media might be at least a contributing factor to these worsening of mental health trends. What we did was try to expand on the existing literature which documents a correlation between a use of social media and negative mental health outcomes and try to come in with something that can more plausibly establish causation, more plausibly say that social media causes, in fact, negative effects on mental health.

Justin Hendrix:

Speak to me a little bit about the methodology for this particular research. What did you set out to do, and what was the sort of experimental setup?

Alexey Makarin:

We rely on the empirical strategy called the difference-in-differences empirical framework. When we applied it to our empirical case, we are basically comparing the changes in the mental health of college students in colleges that just received Facebook versus changes in mental health conditions of college students in colleges that did not receive Facebook at the time. Under the assumption that at least before the introduction of Facebook, the mental health conditions were evolving along parallel trends among different colleges, we are able to establish the causality.

Justin Hendrix:

Can we talk a little bit about the data sources? I know that you have essentially two specific data sources. They're fairly large data sources, for that matter.

Luca Braghieri:

Sure. There are two key components in this empirical strategy, so there are two sources of data then we need. The first one is we need to know when Facebook was rolled out at various colleges in the United States. One of our datasets simply tells us the date in which Facebook was rolled out at each of 775 colleges. The second dataset is from the American College Health Association, and it is a dataset that contains the answers to a survey called the National College Health Assessment. That is a survey that inquires about students' mental and physical health. We have data going between the spring of 2000 and the spring of 2008. This is a survey that is administered twice a year, and then there are different universities that drop in and out of each wave. So, we have all this, basically, the universe of answers between 2000 and 2008.

Justin Hendrix:

Can you tell us a little bit more about this National College Health Assessment? What is it for, and how does it work?

Alexey Makarin:

It's the most comprehensive survey about student mental health and physical health that is available at the time around Facebook expansion. That's why we use it. The survey is available for more than 350 colleges and more than 300,000 college students.

Justin Hendrix:

You've got a lot of data about the mental health and general health of these students. Can you talk a little bit about the, I guess, quantitative approach that you had to meshing that information and then comparing it to the presence of Facebook on those campuses?

Luca Braghieri:

Our empirical strategy, like Alexey had said, is a difference-in-differences. The idea there is you have something like a control group, so these are, say, students attending colleges that do not have Facebook access, and you look at how their mental health evolves over time. Then, you compare that evolution to the evolution of the mental health of students attending colleges that just received Facebook access. All you need to do in order to run that empirical strategy, you need to know the college that a student attends, you need to know the survey wave that the student participated in, which tells you... The two of them tell you whether Facebook was available at the college at the time. You need to know the students' answers to the survey, so those are the outcome variables, and then you can basically carry out your empirical strategy. Again, you can compare the mental health of students that take the survey right after Facebook was released at their colleges, to the mental health of students that attend colleges where Facebook had not yet been released.

Justin Hendrix:

How did you deal with the possibility of there being some other external factor? How did you pinpoint Facebook?

Alexey Makarin:

Yeah, so whatever external factors you'll be thinking of, they would have to follow the same staggered rollout in the same way as Facebook did, which, we think that it's implausible. It would also have to appear exactly at the same time when Facebook started to roll out and have no influence before. Because before 2004, we observed that mental health of college students in colleges in different Facebook expansion waves, as we call them, those mental health conditions were evolving on parallel trends before then. That external factor needed to switch on precisely when Facebook was starting to roll out, and it needed to switch on precisely in the same staggered manner, which is difficult to think of. It's difficult to think of such external factor.

Justin Hendrix:

You did address this idea of stigma, though, as an alternative explanation.

Luca Braghieri:

Sure. Let me maybe take a step back and say one thing about what Alexey said. One thing that Alexey mentioned is that our setting lends itself particularly well to the analysis we're trying to carry out, because the rollout of Facebook was staggered. Just Facebook was rolled out at different colleges at different points in time. Imagine Facebook had been rolled out everywhere at the same time. They would be more worried about confounding factors. Maybe there could have been one confounding factor that occurred precisely at that time when Facebook was rolled out. That would be definitely a concern. It seems less likely that there would be the same confounding factor that occurs at the same time precisely in the colleges where Facebook is rolled out in the staggered adoption setting. The confound seemed to be less likely, because of the staggered adoption nature of Facebook.

As far as stigma is concerned, yeah, so we were worried, I think this is a general worry about the increase in this mental health or the worsening of mental health trends. One important question in general, is this a worsening of real baseline mental health? Or is this an increased willingness to discuss mental health, increased willingness to maybe go to the doctor and get a mental healthcare diagnosis? We were worried about that in the context of our paper, as well, because you can imagine a world in which Facebook gets rolled out, students go on, they participate in whatever Facebook discussions about mental health and then that reduces stigma. We have a variety of strategies to try to counter that. Maybe I can let Alexey discuss those.

Alexey Makarin:

Yeah. One of the strategies we use is just to see whether on Facebook there were any discussions about mental health. It seems like it was not really, according to the leadership that we found it was not really a big deal on Facebook at the time. Another thing that we do is we show that, so in our survey we also have questions about the take-up of anti depression therapy, of anti depression medication and actual depression diagnosis. We also find effects for those variables. If you had to believe that it's only about and how I feel about myself now suddenly after Facebook rolled out, I feel sad and nothing else is going on. Not only I feel sad, but also for the most predisposed students based on their individual characteristics, you find that they also actually do something about it. They go to the doctor and potentially take anti depression therapy, take up anti depression therapy and medication.

Justin Hendrix:

There are a variety of different mental health conditions that are described by the responses to the NCHA survey, as I understand it, attention deficit disorder, depression, anxiety disorder, seasonal effect disorder, eating disorder, stress, sleep difficulties. Was there any kind of granularity in the connection to any of these specific issues with regard to the rollout of Facebook?

Luca Braghieri:

What do you mean by granularity?

Justin Hendrix:

I suppose you've made a connection between poor mental health, generally, in the rollout of Facebook. Is there any kind of, I guess more specific connection or maybe more pronounced connection between one of those conditions versus the other?

Luca Braghieri:

Yeah, so a bit, yeah. The NCHA survey has many questions about depression, fewer questions about other conditions. As far as depression is concerned, we have a host of depression symptoms. As Alexey mentioned, we have self-reported take-up of depression therapy, use of antidepressants, et cetera. We find definitely on all the symptoms of depression, we find that the introduction of Facebook had worsened them and in many cases, significantly so. There were a few conditions where we expected to see some movement, but we didn't detect much movement and those were eating disorders, so things like anorexia and bulimia.

This was somewhat surprising to me, especially given the fact that the mental health crisis among adolescents is primarily related to teenage girls and has to do at least in part with their body image, or at least people think that that's the case. I expected to see more on obesity disorders, but one possibility for why we might not see much on them is that we're just looking at a slightly different population from the population that might be more severely affected. Here we're looking at 18 to 22 year olds, and it is possible that the population that could be more or severely affected by social media as far as eating disorders are concerned, would be maybe a population of younger teenagers, so maybe 13 to 15, 17 year olds.

Justin Hendrix:

Let's talk about some of the mechanisms where you feel that this depression is perhaps coming from, or these effects of mental health are coming from. You name a handful of them and essentially present evidence of a connection to all of them.

Alexey Makarin:

We can see there several mechanisms. One is social comparisons and favorable social comparisons. The second one is disruptive internet use. The third one is sort of indirect. It could be that social media affects some other first variables in behaviors, for example, alcohol use or drug use. That in turn, could indirectly affect mental health. These are the free mechanisms that we consider. In the end, we find most support for the first one, which is the favorable social comparisons. We do not find much support for the other two.

Justin Hendrix:

Let's talk about what unfavorable social comparison is.

Luca Braghieri:

Yeah, so unfavorable social comparisons is simply the idea that you might go on social media. You might be a social media user, and go on Facebook. Then once you're there, you might compare yourself to others, to the pictures that you see of your friends and acquaintances. To the extent that those comparisons are unfavorable, they might be detrimental to your self-esteem and mental health. The thing that I think is possibly particularly pernicious about social media is the fact that what people put on social media, the way they portray their lives on social media is very inflated.

Rarely do you see pictures in which people don't look very good. Rarely do you see pictures of people doing boring activities. Generally everybody portrays a very exciting version of their lives, a highly curated version of their lives. If you're a social media user and you're a bit naive, you might not understand how curated other people's lives, at least the portrayals of their lives are. I think that in fact, it may be the case that you might consider many more social comparisons as unfavorable than there actually would be if you had full information or if you were able to process information correct.

Justin Hendrix:

Are you able to tell something about the kind of degree of this effect of the, I suppose, causes of depression or inputs on depression? What is the degree of effect that you're able to observe of the introduction of Facebook on these figures?

Alexey Makarin:

We aggregate all of our mental health related questions in one big index called index of poor mental health. We find that the introduction of Facebook led to the worsening of this index of poor mental health by about 8.5% of deviation. That's a reasonably sizable impact. For instance, as a point of comparison, we find that it's about one fifth of the impact of losing one's job on mental health as compared to the effects found in the meta study that we cite in the paper. We also do a background level or the level calculation where we think, under relatively strong assumptions, how much Facebook can explain in this big trend of worsening of mental health conditions of the young adults in the United States over the past 15 years. We find it's about 25%, but of course that can be taken with a greater amount of salt.

Justin Hendrix:

That's a pretty sizable impact. It means that, essentially, you are observing that Facebook is contributing to the decline in mental health in this youth population fairly significantly. It's not the only cause, perhaps, certainly perhaps not the greatest cause, but appears to be exacerbating the mental health situation.

Luca Braghieri:

Definitely so. As Alexey said, I would take with a grain of salt, the part of the paper in which we try to quantify what fraction of the worsening in mental health in young adults over the last 10, 15 years can be attributed to Facebook. But still, the effects we find, which are in the ballpark of oh 0.1 standard deviation units are effects that are non-negligible. I think those are effects that we want to definitely think about and possibly be worried about, from a policy standpoint.

Justin Hendrix:

This was looking at data, mental health data now, from 20 years ago. A critic might come and say, "Oh, Facebook's completely changed and the teens are on Instagram and TikTok these days anyway." How would you address that on some level, especially the kind of concern that perhaps Facebook itself has changed its features? We know that it's doing internal research on this stuff. Perhaps they've addressed this already.

Alexey Makarin:

To the extent that social media is still having some core features that relate to social comparisons. To the extent that social comparisons is the driving force behind our results that we find the negative impact of Facebook and mental health. We believe that our results very easily extend to other social media platforms and to Facebook right now. The second point I want to make is that actually, if you look at public media discourse right now about Facebook and other social media platforms, they typically talk a lot about features that did not even exist back in the day in 2004, 2005 to 2006. They're talking about political news being the source of depression. They're talking about algorithms, live button, various features that actually did not exist back in the day. Because we find negative effects, even in that environment, one might be concerned that potentially something in the nature of social media is driving negative effects on mental health.

Justin Hendrix:

You write that the mechanisms whereby Facebook use can affect mental health might have been exacerbated rather than mitigated by many of the technological changes undergone by Facebook and related platforms in the last 15 years. There's at least some reason to believe it's probably a worse effect at present than it was in the past.

Luca Braghieri:

Well, I think that the closest piece of actual evidence that we have on that is from a paper that I wrote with Sarah Eichmeyer, Matt Gentzkow, and Hunt Allcott back in, I guess maybe we ran an experiment in 2018. In that, what we did there is we recruited about 1,600 Facebook users and then we randomized them into a treatment and a control group. We paid the treatment group to deactivate our Facebook accounts for a month, and then we did nothing to the control group. Now, we could study the effects of the activation on a variety of outcomes, including things like subjective wellbeing. We had one question about depression.

It turns out that if you spend a month, if you are paid to... You're in the treatment group, so you're paid to deactivate your Facebook account for a month, turns out to be the case that you report a decrease in symptoms of depression. That decrease, if you look at the magnitude of that effect, that is actually pretty close to the magnitude of the effect that we find in our new paper. I think we have fairly direct evidence that the effects have not changed dramatically over time, even though these are, I just got to say that these are very different studies done, different samples with different methodologies, but even despite all that the coefficients are fairly close to one another.

Justin Hendrix:

Let's talk about the final two kind of areas. One is what open questions this research leaves that you intend to try to address in the future. Then second, have you had any response either from the company or any external party on this research?

Alexey Makarin:

I think in terms of open questions, two come to mind. The first one is, okay, so we're talking about the negative impacts of social media. What are the potential positive impacts of social media? I think the positive, the benefits of social media actually are relatively understudied, surprisingly. The second question is, okay, so you find this negative impact of social media on mental health, so what can be done about it? What are the interventions that might help? That's also a relatively understudied area. I, myself do not have any work in that area right now, but I don't know, maybe look at us. But these are the questions that come to mind.

Luca Braghieri:

In terms of the response, so I talked to the Office of Communication in the UK, who is an entity that is at least in charge of studying the media landscape in the UK and suggesting regulations. They were very interested in this kind of research and they are trying to replicate, not maybe this particular study, but some of the studies in this literature in the context of the UK and that track suggests policy reforms to the government.

Justin Hendrix:

Of course Ofcom is the regulator of record, I suppose, designated in the online safety bill. I suppose similarly, we have the Kids Online Safety Act under consideration here in the United States, which I suppose to some extent could be informed by research like this.

Alexey Makarin:

I also wanted to mention that Luca presented this work in the Italian National Health Service back in the day, and recently I presented this paper to staff at MIT Medical. It seems the medical community is pretty much interested, is very much interested. On the side of companies, we have not heard from Facebook directly. One thing that we found interesting is that we also observed some downstream effects of social media on academic performance. We got a little bit lucky that in the survey there is a specific question about whether you feel that your academic performance got worse because of reason X. One of the reasons at least that was mental health, mental health related conditions. We do indeed find that after Facebook rolls out at a college, students report that their academic performance got worse because of mental health issues.

Justin Hendrix:

You're actually able to see that kind of roll on effect or extract from mathematically backwards to see the effect on literally their grades.

Alexey Makarin:

We do not have grades themselves, unfortunately. We only have the survey.

Justin Hendrix:

Their reported academic performance.

Alexey Makarin:

Right, exactly.

Luca Braghieri:

Their subjective assessment of their academic performance.

Justin Hendrix:

Anything else that you all want to add?

Luca Braghieri:

We're very happy with the paper we wrote, but I think in general it's very good to look at literatures more than individual papers. I think there is a literature on the effects of social media on mental health that encompasses correlational studies. It encompasses field experiments, it encompasses laboratory experiments and then cause experimental papers like the one we just wrote. I think that that literature paints a fairly coherent picture. Above and beyond what anybody thinks about this particular paper, I think it would be good to grapple with that entire literature, that I think paints the picture that social media is detriment to the mental health of especially adolescents and young adults.

Justin Hendrix:

Certainly a lot of work for everyone in this field to do to try to isolate and address that problem. Luca, Alexey, thank you so much for joining me today.

Luca Braghieri:

Thank you.

Alexey Makarin:

Thank you. It was a pleasure.