There’s nothing quite like a graph to capture the public imagination. Ever since the Covid era, in which viral dashboards featuring flawed cross-country comparisons were all the rage, charts have taken on a life of their own. The most recent is a striking visual from John Burn-Murdoch of the Financial Times. It appears to show trends in fertility rates in several countries plunging as “smartphones take off”, and was featured as part of a broader article on birth rates.

One glance at the chart and the conclusion seems incontrovertible: the human race is destined to die out as we’re all too busy looking at our phones. Even cautious observers concluded that the “preponderance of evidence” points to smartphones, and that in “country after country” birth rates plunged regardless of previous trends. But the graph does not in fact prove that births have been declining because of smartphone use.

The trigger for the piece seems to be a non-peer-reviewed preprint published in April on 4G rollout and the drop in teenage pregnancy in the US and UK. Together, they propose that increasing smartphone use increases our access to social media and hence reduces our time for other things, such as making babies.

Firstly, to state the obvious, correlation is not the same as causation. But the specific details also raise further questions. The FT piece talks about sudden fertility drops, but zoom out and, for most countries, the picture looks more like a long-running decline than a dramatic drop. Part of this confusion comes from the statistic used, which measures fertility relative to the pre-smartphone trend, stripping away the background decline and making the remaining dip appear more sudden.

More broadly, as the pandemic taught us, international comparisons are often misleading. Every country has its own mix of economics, culture, politics and social change, which makes neat cause-and-effect stories almost impossible to prove. In the time period where the drops occurred across parts of continents, there was the 2008 financial crisis and a pandemic where people were locked in their homes, which surely affected these figures.

In cross-country comparisons such as this, there’s the temptation to pick the nations that fit best for each argument (without a clear explanation of why those countries have been picked), tweak the metric, and smooth out the mess to produce a straightforward story. But comparing half a dozen countries is hardly definitive. After all, we wouldn’t think much of a study of half a dozen people.

So how would we know if smartphones are fast-tracking a drop in fertility? Proving any causal link is extraordinarily difficult, not least because smartphones are now so universal that there’s barely a control group left. To move beyond seductive graphs, researchers would need far more rigorous evidence. Individual-level data could be used rather than sweeping country comparisons, as could transparent, pre-registered datasets which avoid cherry-picking. It would also be crucial to carefully account for the many other forces shaping fertility, from housing costs and economic insecurity to Covid itself. Most importantly, a claim this big would need a broad, peer-reviewed body of evidence pointing in the same direction, not a viral graph.