Does Social Media Harm Teen Mental Health? What 355,358 Adolescents Showed

The short answer: a real link, a tiny size, and no proof of cause

Ask whether social media harms teen mental health and you will get two confident answers. One says the phones did it — a generation's anxiety and depression rose with the smartphone, case closed. The other says the panic is overblown and the research is flimsy. A 2019 paper in Nature Human Behaviour by Amy Orben and Andrew Przybylski, at the University of Oxford, is the rare study both camps have to reckon with, because it was built to answer a prior question: how much of what earlier studies found was the data, and how much was the way researchers chose to analyse it?

Their answer is uncomfortable for the loudest version of the harm claim. Across more than a third of a million adolescents, more digital technology use went with slightly lower well-being — but the link explained at most 0.4% of the variation between teenagers. The authors' own conclusion was that effects this size are "too small to warrant policy change." One scope warning belongs up front: the study measured digital technology use broadly, with social media inside it, not any single app.

Social media and teen mental health statistics: the numbers in this study

The study's scale is the first thing to understand, because scale is what makes a tiny effect believable rather than dismissible. Orben and Przybylski did not collect a new sample. They took three existing large-scale social datasets — the kind governments and universities gather on thousands of families — covering 355,358 adolescents in total, and re-analysed them.

Then they did the thing that makes this paper different. In a dataset with hundreds of variables, a researcher can make dozens of reasonable choices: which well-being measure to use, which controls to include, how to score technology use. Each combination is a defensible analysis — and earlier work had shown that different defensible choices could produce headlines in opposite directions. So instead of picking one combination, the authors ran the full range of them, a method called specification curve analysis, and reported the overall picture rather than the single most dramatic result.

That method matters for how you should read every other headline in this debate. When a study reports that social media use is significantly linked to depression, "significant" can mean nothing more than: in a sample of tens of thousands, a very small effect cleared a statistical threshold. This paper's contribution is to put the size back into the sentence.

How does social media affect teenagers' mental health? The 0.4% finding

With every specification laid side by side, the pattern was consistent: the association between digital technology use and adolescent well-being was negative, but small — explaining at most 0.4% of the variation in well-being. In plain terms, if you lined up teenagers by how much they used digital technology, that line-up would tell you almost nothing about where each of them falls on well-being. 99.6% of what separates a thriving teenager from a struggling one, in these datasets, was something else.

Two cautions belong inside the finding itself, not buried below. First, the 0.4% is a ceiling, not a typical value — the authors' wording is "at most." Second, the exposure here is digital technology use as the datasets measured it, which lumps social media together with other screen activities. A parent asking specifically about one app will not find that app's number in this paper, because the datasets never separated it out. Anyone who quotes this study as "social media explains only 0.4%" is already one step beyond what it measured.

Does social media cause depression in teens? Why this study can't say that

No — and the paper never claims otherwise. The authors describe what they examined as "correlational evidence," and correlation is where the study starts and ends. Teenagers who used more digital technology reported slightly lower well-being. The data cannot say whether the use lowered the well-being, whether teenagers who were already struggling reached for their phones more, or whether something else in their lives drove both.

This is not a technicality. It is the exact point where the public argument usually goes wrong. A correlation this small, in data this tangled, cannot carry a claim that social media causes depression — and it equally cannot carry the opposite claim that a struggling teenager's phone use is definitely innocent. What the study rules on is size, not direction: whatever the average relationship is, it is not large.

It also rules on something subtler. The authors argue that the research literature itself helped manufacture the alarm: with huge samples and flexible analyses, small effects kept being reported as significant findings, and different teams analysing similar data reached contradictory conclusions. Specification curve analysis was their attempt to show every result at once instead of the single most dramatic one.

Does social media affect teen mental health at all? Small, negative — and averaged

Yes, on average, a little — and the word "average" is doing heavy work. An average across 355,358 teenagers can be near zero while still containing groups for whom the relationship is stronger. A teenager being bullied through an app at 2 a.m. is not reassured by a population average, and this study does not identify which teenagers, if any, sit in a higher-risk corner of the data. It was not designed to. Its datasets asked about overall use and overall well-being, not about what a teenager was doing online, with whom, or what it displaced.

So the honest reading cuts both ways. The study is strong evidence against the claim of a large, general harm sweeping the whole generation — an effect that size would have shown up in this analysis, and it did not. It is not evidence that social media is harmless for every teenager, and the authors do not present it as a clean bill of health. It is a measurement of the average, done unusually carefully.

What this research can't tell you

It cannot name a culprit app. The exposure was digital technology use overall. Nothing in this paper separates social media from messaging, video, gaming or homework screens, and nothing ranks platforms against each other.

It measured well-being, not illness. The outcomes were well-being measures in population datasets — not diagnoses of depression or anxiety made by clinicians. A small shift in average well-being and a rise in clinical cases are different claims, and this study speaks only to the first.

It cannot set a safe amount. There is no hour threshold in this paper — no line below which use is fine and above which it harms. An average association of at most 0.4% cannot be converted into a family rule, however it is spun.

Its data are from before today's platforms. The datasets were gathered in the years before this 2019 analysis. Recommendation-driven feeds have changed since; the study measures the relationship in the data it had, not in the apps a teenager opens now.

For context, official guidance reads the same evidence more cautiously than this paper's headline. The US Surgeon General's Advisory on Social Media and Youth Mental Health (2023) states that "we cannot conclude social media is sufficiently safe for children and adolescents." That is not a contradiction of this study: the Advisory asks whether safety can be assured — including for vulnerable adolescents — while Orben and Przybylski ask how big the average association is. A tiny average and an unproven safety case can both be true at once, and parents are left holding both.

Frequently Asked Questions

The takeaway

The most careful re-analysis of the social media question found a link that is real, negative — and explains at most 0.4% of how teenagers differ in well-being. A number that small does not support the claim that phones are wrecking a generation. It also says nothing about the teenager who is awake at 2 a.m. being targeted in a group chat. Watch the average for what it is worth, and watch your own teenager for everything else.