March 5, 2021 - 8:50am

If you’re a parent, imagine that someone comes up to you and offers you two choices. One, your daughter will become a regular user of social media. Or two, she will become a regular user of heroin. Which would you prefer?

Obviously you answered “social media”. Many millions of girls around the world use social media, and while it would be silly to say that there aren’t negative consequences to it, most of those girls seem to have grown up all right. Some people have been using Instagram for 11 years; there will be people, just about, who started in their teens and are now in their 30s. Twitter and Facebook have been around longer still, and while they were not at their present level of ubiquity 15 years ago, millions have used them for years. 

Of course, lots of people take heroin too, and some of them have happy productive lives, but a large fraction of them suffer severe negative consequences. (At least partly as a result of prohibition, I’d say, but still.) If social media had large negative impacts with anything like the regularity heroin does, it would be staggeringly obvious.

Which is why I was startled to see the New Statesman tweeting that “heavy use of social media, for adolescent girls, is correlated more strongly with depression than [is] the use of heroin”.

The piece is by Louise Perry, an esteemed UnHerd colleague, and I don’t want to suggest for a moment that she’s wrong to be concerned about social media. There probably are real things to worry about for a subset of users, and yes, it does seem to be girls who suffer the most.

But the “heroin” claim is unfortunate. It comes from a study by the psychologist Jean Twenge, itself a response to an earlier piece of research by Amy Orben and Andy Przybylski which I’ve written about here (I won a Royal Statistical Society award for that piece, if I’m allowed to brag). Orben and Przybylski showed that the correlation between social media use and mental wellbeing was extremely small and uncertain; negative, but only as negative as the correlation between well-being and wearing glasses. They also showed that you could find more negative correlations, but only by chopping the data up to cherry-pick bits out.

I won’t go into the full back-and-forth. But Twenge et al, as far as I can tell, did indeed take the most dramatic correlation; she denies cherry-picking it, but certainly she seems to have misunderstood what Orben and Przybylski were trying to do; they respond here.

In fairness to Twenge, the “heroin” comparison seems to have been a response to the other scientists’ comparing social media’s effects with “wearing glasses” and “eating potatoes”. But it is, I think, a really unwise comparison. We use terms like “internet addiction”, “social media addiction”, “gaming addiction” as though these are real things, but they are not: they simply do not exist in the same way as drug addictions do; there is no well-defined definition and the attempts to create one do not find very much. Comparing social media to drugs is entirely unhelpful. Yes, this is specifically (and I think wrongly) comparing the effect on depression, but people will read it and see “social media as bad as heroin”, without the nuance.

And right now, especially, we all rely on social media, on screen time, on these little glowing portals to other people, to get us through the bloody day. I can’t do my work without it, my children can’t do their schoolwork, none of us can talk to our friends, unless we use the internet. Social media is vital at the moment. Telling people that it’s as bad as heroin – that they might, in essence, be killing their daughters by letting them use it – seems cruel, especially since it’s almost certainly not true.


Tom Chivers is a science writer. His second book, How to Read Numbers, is out now.

TomChivers