top of page

Better Social Media

How To Improve Social Media (for teen mental health, and for democracy) 


This page has not been updated since 2020 and is out-of-date. See Jon's most recent work here.

In the book we noted two ways that social media contributes to the problems besetting Gen Z, universities, and many democracies. In chapter 7 we asserted that social media is a likely contributor to the sudden rise in rates of anxiety, depression and self harm that hit Gen Z around 2012. We cover additional research on this claim in section 1 below (and see also our page on Better Mental Health). In chapter 6 we asserted that social media (along with cable television) amplifies moral outrage and fuels the "polarization cycle" that makes America's politics so bitter, and that has made political conflict on campus so much more intense since 2015. We cover newer research on that question in section 2 below. In section 3 we offer suggestions for improving social media to make it a more constructive force in society, and in the lives of teenagers.

1) Is there evidence that social media contributes to teen anxiety and depression?

[Update Nov. 2021: See Jon’s article in The Atlantic, The Dangerous Experiment on Teen Girls]

We focused our analysis on social media, but in reviewing the published research in 2017 we sometimes expanded our focus to "screen time" or "digital media." We now believe that this was a mistake. The evidence we have gathered since the book was published shows that our concerns about social media were justified, as we'll show below. But from engaging in debates with our critics we have learned that "screen time" is not a very useful construct. Studies that merge all screen time together, including those that facilitate social interaction (such as talking with a friend on FaceTime, playing a video game with a friend, or watching TV with a sibling) find only very small and inconsistent correlations with poor mental health.

However, when we look only at social media, it is now increasingly clear (though still contested by some) that heavy use of social media is bad for teen girls, overall. If you want to dive in to the details of the debate and see the studies for yourself,  please see this Google Doc, where Jon and Jean Twenge have laid out the various kinds of evidence, including all the studies we can find that run counter to our hypothesis. You'll find comments added by researchers on both sides of the debate. If you are a researcher, please request access to the doc and add your comments.

The basic pattern found in many correlational studies is this, from Kelly, Zilanawala et al. (2019)


So far the conclusions emerging from the debate seems to be that

  • Links between social media use and bad mental health are more consistently found for girls than for boys.

  • Correlational studies generally show curvilinear relationships with measures of anxiety and depression, meaning that light to moderate use of social media is not associated with bad outcomes, but heavy use is. (The percentage of variance accounted for is usually small, and there is some debate about how to interpret that.) 

  • Most time lag and longitudinal studies show links to bad outcomes, but a large minority do not (especially those that use short time intervals like one day). 

  • Of the 14 true experiments we have found that were published since 2014 and that used random assignment, ten of them found that reducing or eliminating social media time caused improvements in variables related to mental health. 

  • Meta-analyses that claim to find no links are almost always mostly about "digital media" or total screen time. When meta analyses or review studies allows us to zoom in on social media, they almost always find an effect equivalent to a correlation coefficient between .10 and .22, which about as large as public health findings get.

In early 2020, a few new studies came out that seem to offer much greater clarity by distinguishing between social media versus "screen time," and looking at girls separately from boys:


  1. Twenge, Haidt, Joiner, and Campbell (2020) re-analyzed the 3 large datasets used in the most widely cited study that claims to exonerate "screen time" (Orben & Przybylski, 2019).  That study claimed to show that time spent on "digital media" was harmless -- linked to mental health problems at the same super low level (correlations below .05) as "eating potatoes." But when we limited the analysis to social media (rather than all screens) and when we used proper controls (rather than controlling for factors associated with depression, such as happiness at school), we found much larger links between social media use and depression. For girls, the correlation rose to around r=.17, which is not small potatoes.

  2. Orben (2020) reviews a variety of review papers and concludes that the findings for "screen time" are too small and erratic to justify concern, but the findings for social media consistently show larger effects, with correlation coefficients in the range of .10 to .15. (That's for boys and girls together, so it is consistent with our finding of .17 for girls only.) 

  3. You can read an account of the debate written for a non-academic audience in this essay on Medium by journalist Markham Heid.

  4. You can read Haidt's account of the recent studies and the state of the debate in this Twitter thread.

  5. The next phase of the debate is beginning: it is about emergent or group-level effects, rather than "dose-response" effects that treat social media use like sugar consumption -- OK in small quantities, and bad for the person who consumes it in large quantities. Jon and Jean Twenge found a correlational link across nations between rising smart-phone adoption and rising loneliness in school, as reported in the PISA survey. See our NYT essay on the research here, and see comments from critics of the research here. 

Summary of the empirical evidence: multiple kinds of evidence suggest that there is a causal connection between heavy social media use and bad mental health, for girls, but the size of the effect is debated. We also don't know how social media affects pre-teens -- the group that shows the biggest percentage-wise increases in self harm and suicide. Light to moderate daily "screen time" (in contrast to social media) does not seem to be associated with harmful mental health outcomes. We did not know that when we wrote the book.

Breaking news, Sept. 17, 2021: Facebook’s own internal research--leaked to journalists at the Wall Street Journal--shows that they have known since 2018 that Instagram harms large numbers of teen girls, including the finding that it makes body image problems worse for ⅓ of teen girls. See the Wall Street Journal’s multi-part expose how Facebook knows that it is damaging teen health, and democracy.

Summary of advice to parents: 

  1. Don't worry so much about "screen time" causing depression and anxiety. Especially during this period of reduced social contact because of coronavirus, it is OK to allow much more "screen time" than you did before, especially on apps that promote direct synchronous interaction, such as Zoom, Facetime, and multi-player video games.

  2. Do continue to worry about social media, especially for your daughters, and especially before high school. We continue to believe that the rapid transition of social life onto social media platforms, which happened between 2009 and 2012, is likely to have contributed to the sudden surge in depression and anxiety that began among teen girls around 2012.

  3. Don't take this finding, about mental health, as a reason to remove all limits on device use. Even if "screen time" does not cause depression, it still imposes "opportunity costs" on your child. Remember that many of the apps are designed by psychologists to take control of your child's mind and schedule, like a cuckoo bird that pushes all competitors out of the nest. Kids need vast quantities of unsupervised free play with friends, and since the 1990s American kids have been severely play-deprived. Talk with your child about how much time she'd like to spend on her phone (or other devices) each day, about what non-phone activities she'd like to do, and about which apps make her life or relationships better, which ones just drain her time. Then consider using Apple's Screen Time app for imposing limits (hard or soft) on some activities, or on total time. Ultimately we want our kids to be self-regulating. But even many adults have trouble with this; high schoolers and especially middle schoolers will need some support or structure to prevent digital media from taking over most of their day. 


2) Is there evidence that social media contributes to political polarization?

Jon teamed up with Tobias Rose-Stockwell to tell the story of how social media (particularly Facebook and Twitter) changed, between 2009 and 2012, to become a much more effective "outrage machine." See The Dark Psychology of Social Networks, in The Atlantic (2019).

As for whether the new outrage machine can be shown to have actually amplified political polarization and contributed to political dysfunction in the years after 2102, this is a very difficult question to adjudicate, because it is difficult to conduct controlled experiments. This is a question that political scientists are currently debating, and there are arguments on both sides.


Here is Jon's open source google doc where he and Chris Bail (author of Breaking the Social Media Prism) are collecting empirical studies on both sides. If you are a researcher in this area please add your comments.

3) How to make social media less harmful, more helpful

To come. Major ideas will include:

  • Raise the age of internet adulthood to 16. It never should have been set to 13

  • Enforce age restrictions. Eleven year olds should not be able to get into everything just by lying about their birthday.

  • Major platforms should have "know your user" requirements before users can post. That means some kind of identity verification, to keep out trolls and manipulators. It does not mean that people must post using their real names.

  • Empower a government regulator in each country. In the UK it is Ofcom. In the USA it is ??

  • Regulation should focus more on the dynamics of virality and the authenticity of participants; focus less on content moderation (which is a hopeless whack-a-mole game that is sure to trigger charges of censorship and ideological bias).

bottom of page