A new review pulls together seven lines of evidence — randomized detox trials, natural experiments and Meta's own internal research — and argues the link between social media and teenage mental illness is causal, not merely correlational.
Every time someone says social media harms kids, an expert appears to recite the incantation: that is correlation, not causation. This essay takes the incantation apart.
Quit for a month and the symptoms drop
The hardest evidence comes from detox experiments: participants are randomly assigned to quit social media for a week or more, then their mental state is measured. Pooled across more than 30 studies and 7,000 people, depression and anxiety symptoms fell.
Those drops are often waved away as too small to matter. The authors do the arithmetic: the effect is comparable to school-based depression prevention programmes and to youth psychotherapy, and it needs no extra staff to deliver.
At population scale, halving heavy use could prevent nearly nine per cent of depressive cases.
Where the fibre went, the diagnoses followed
Another line comes from natural experiments. In Spain, fibre rolled out province by province between 2007 and 2019. Provinces with more fibre saw bigger rises in mental health diagnoses, self-harm and suicide attempts, concentrated in girls aged 15 to 19.
Studies in Germany, Italy and the United States found something similar. If gaming or streaming were the culprit, the sex split would not look like this.
The neatest blow comes from Facebook itself: as it spread campus by campus in 2004-2006, the campuses that got it first saw mental health worsen.
The company knew
The ugliest part is Meta's internal research. Documents say things like 'teens are hooked', 'Facebook is net bad for well-being' and 'we have to assume we are contributors, amplifiers'.
In 2019 Meta ran its own deactivation study. A week in, depression, anxiety, loneliness and social comparison had all fallen — and the study was shut down.
A 2021 internal survey found 14 per cent of Instagram users aged 13 to 15 had received unwanted sexual advances in a single week. In 2023, the 'Accounts You May Follow' feature recommended nearly two million minors to adult groomers in three months; a fifth of those recommendations produced a follow request.
One quieter harm is short-form video. Meta-analyses tie it to worse attention, weaker working memory and poorer impulse control, with associations to depression and anxiety roughly double those of social media in general.
The plain version: this is not another moral panic about new technology. It is people reading the company's own books back to it.
Why it matters
The argument has dragged on for a decade because 'you cannot prove causation' was the platform's best shield. Once the causal evidence starts coming from the platforms' own experiments and sealed files, the question shifts: how long did they know, and what did they do about it.
Curated from high-quality sources, with concise summaries and key takeaways.
