February 4, 2019   4 mins

There’s this trick that climate deniers used to use. They used to say “there’s been no warming since 1998”. And in a weird way they were right: looking at global atmospheric surface temperatures, none of the years that followed was as hot as 1998.

But they were cheating. They picked 1998 deliberately since it was an outlier – an El Niño year much hotter than the years around it. If you were, on the other hand, to measure from 1997 or 1999, then there were lots of much hotter years on record; and the clear trend was that later years, on average, were hotter than earlier ones. It was a wobbly, noisy line, with some outliers, but the average temperature really was going up, and the only way you could hide that trend was by cherry-picking statistics.

I was thinking about this as I read the Sunday Times splash this week, which (using as-yet unavailable data from the Office for National Statistics) claimed that the “suicide rate among teenagers has nearly doubled in eight years”. It expressed concerns that we are raising “a suicidal generation”.

They said that the suicide rate among 15- to 19-year-olds in 2010 was just over three per 100,000. The ONS figures, due out in September, will (apparently) show that it is now over five per 100,000. Inevitably enough, the piece links the purported rise to the growth of social media since 2010.

But this is – and I don’t want to get too technical here, but bear with me – absolute bollocks from top to bottom. It’s a masterclass in what scientists call “hypothesising after results are known”, or HARKing. If you have the data in front of you, then you can make it say almost anything you like.

First, it’s worth noting that very few teenagers kill themselves. The total number of suicide deaths among 15- to 19-year-olds in 2017 in England and Wales was 177, out of about 3.25 million. That means that small changes can look like big percentage swings. More important, though, the Sunday Times story did exactly what the climate deniers did. The year 2010 had the lowest rate of teen suicides of any year since at least 1981, when the ONS records begin. You could compare it with literally any other year and you’d see a rise.

Added to which, picking social media as your reason for is completely arbitrary. Social media did not start in 2010. The BBC TV series Sherlock, starring Benedict Cumberbatch and Martin Freeman, did, though. Maybe we should blame that.

You could, if you wanted to, use the same trick to tell the exact opposite story. Facebook was first released in 2004, when the suicide rate among 15- to 19-year-olds in England and Wales was 4.7. But after six years of social media being available, it had dropped to 3.1! It’s a life-saver, no?

Or, you could use the trick to tell other, more political stories. For instance, one of the highest rates of suicide on record was 1998, shortly after the election of Tony Blair’s New Labour, when 6.1 out of every 100,000 teenagers killed themselves. But by the time they left office in 2010, it was – as we’ve seen! – an amazing 3.1; New Labour almost halved teen suicide! And with the election of three Tory-led governments in a row, it’s crept back up almost to where it was.

I have no idea if that’s true – and I should admit, I’ve done some shameless cherry-picking of my own, there, choosing 1998 instead of the year Blair actually swept to power, 1997, which was a less spectacular 5.5 per 100,000. You could tell a plausible-sounding story about cuts to mental health provision and social care and housing support and so on. But if you did so, you’d also have to explain why suicide in general, including among young people outside this very specific bracket, is on the decline.

The trouble is, for people who want to say that social media is causing an epidemic of suicide, is that there’s just no good evidence for it. For a start, as I just mentioned, there isn’t really an epidemic of suicide anyway. If you look at 10- to 29-year-olds, it’s gone from a consistent plateau of about 15 per 100,000 from the late 1980s to the early 2000s, down to about nine per 100,000 by about 2004, and that’s where it’s stayed, pretty much. There’s a similar story among the population as a whole.

What’s more, when you look at the links between social media and mental wellbeing, they are deeply ambiguous and extremely weak – and the data doesn’t show causation, anyway. An enormous analysis of three major datasets by Oxford University scientists, published in Nature Human Behaviour last month, found a tiny, uncertain correlation between the two. But that correlation about the same as the correlation between well-being and regularly eating potatoes, or well-being and wearing glasses. We do not talk of potatoes as having “destroyed a generation”, or worry about bespectacled teens and suicide. The Human Behaviour study was also interesting because it showed how you could find scarier results – by doing exactly what the Sunday Times story did, in essence: by HARKing, getting your data first and then choosing the bits to display to make it look scary.

The sadness of it all is that this comes on the back of some extremely sensible advice being issued by the Royal College of Paediatrics and Child Health, and a surprisingly even-handed report by the House of Commons Science and Technology Committee. Both acknowledged the shortage of good evidence, and made common-sense suggestions for parents – in particular, rather than imposing hard limits, the RCPCH guidelines simply said that if you find that social media or screen time gets in the way of family time or things you want to do, then you should reduce it. It felt like a little turning point in the endless barrage of scaremongering. Hopefully it really was.

The climate deniers, by the way, appear to have largely stopped using the 1998 trick, after 2014, 2015 and 2016 were all hotter than 1998 and, in fact, each hotter than the last. The trick, in essence, stopped working. We don’t need extra reasons to hope that teen suicides go back down once more, but making it harder to scaremonger about social media would certainly be a bonus.


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

TomChivers