With all the attention currently focused on AI, we should not forget the previous debate. The relationship between social media and mental health remains one of those topics where the temperature of the debate is often higher than the available evidence. On one side are those who see social media as a major cause of a mental health crisis among young people, with Jonathan Haidt as one of the most prominent voices. On the other side are those who dismiss any possible negative influence. As is so often the case, reality probably lies somewhere between these two extremes, and scientists like me, who tend to look for nuance, often find themselves placed in one camp or the other.

Through the Human Progress newsletter, I came across a new study by Christopher Ferguson in Current Psychology. Ferguson has for years been one of the most outspoken critics of what he sees as a moral panic surrounding social media. That does not mean he is wrong, but I do think it is relevant context when reading his conclusions.

The study draws on data from more than 15,000 British adolescents in the Brainwaves Project. Ferguson began by examining the relationship between daily social media use and a range of well-being indicators, including depression, anxiety, self-esteem, and quality of life. As many previous studies have done, he found small associations. Young people who spent more time on social media reported slightly more mental health problems and slightly lower levels of well-being. Correlations, in other words. And as always with correlations, the direction of causality remains unclear.

Ferguson then introduced several control variables, including emotional regulation, resilience, school connectedness, and a sense of belonging. Once he accounted for these factors, the associations between social media use and virtually all mental health outcomes disappeared.

In other words, young people who struggle may simply spend more time on social media. That does not necessarily mean that social media causes their problems. The same underlying factors may well drive both.

This finding is hardly unprecedented. Over the past few years, several meta-analyses have reported similar patterns. Researchers often detect small effects when they rely on simple correlations. Once they account for relevant background factors, those effects typically shrink further or disappear altogether.

That brings us to an important methodological point. Many public debates treat correlations as if they demonstrate causation. Yet the fact that young people who use more social media report more anxiety symptoms on average does not automatically mean that social media causes those symptoms.

Researchers therefore try to rule out alternative explanations. At the same time, we should avoid swinging to the opposite extreme. This study does not show that social media has no influence whatsoever. Rather, it suggests that the number of hours spent on social media each day adds little explanatory power once other factors are included in the model. That is not the same as showing that social media is irrelevant. Research on gaming, for example, has often pointed in a similar direction: context, motivation, and patterns of use frequently matter more than the number of hours alone.

That distinction matters because “hours per day” is a remarkably blunt measure. Young people do not all use social media in the same way. Staying in touch with friends differs from endlessly doomscrolling. Participating in an online community differs from constantly comparing yourself to others. Anyone who focuses only on screen time collapses these very different experiences into a single number.

One important limitation remains. This study is cross-sectional. The researchers measured everything at the same point in time, which means they cannot establish causal processes or long-term effects.

What struck me most is where the strongest effects appeared. Social media use did not emerge as the strongest predictor of well-being. Resilience, emotional regulation, connectedness, and social support did. We have known for years that these factors play a substantial role in young people’s lives, and this study points in the same direction once again.

So what do I take away from this study? Even if social media plays some role, the evidence increasingly suggests that we cannot reduce the story to a simple formula such as “less screen time equals better mental health.” The growing literature on smartphone bans points in much the same direction. Once again, reality proves more complicated than slogans, bestselling books, or political proposals often suggest.

At the same time, this study remains observational. It shows that the associations disappear once control variables are included in the model, but it cannot rule out the possibility that some of those variables themselves are part of the causal pathway. That is precisely why this debate is unlikely to disappear any time soon. And perhaps that is no bad thing. The evidence continues to evolve, whereas simple answers remain less convincing than many would like.