I have counted the hours. I bet you have too: that weekly screen-time report that pops up on the phone, the number you quietly grade yourself by, the figure every parenting article tells you to push down. For years I treated it as the whole ballgame. Fewer hours, healthier kid; more hours, trouble coming. So when I finally sat with the data that was supposed to prove all of that, I was not ready for the answer, which lands somewhere around: the number on the clock barely predicts a thing.
That is not the story we have all absorbed. The loud version says the smartphone broke a generation. Jonathan Haidt’s The Anxious Generation turned it into a bestseller and a movement, and governments took it literally, with Australia passing a world-first ban on social media for under-16s. The whole approach treats the hours themselves as the poison. Here is the question it never asks: what if the metric is wrong, and counting minutes is exactly the frame that lets the people who engineered the compulsion off the hook?
A cohort study published in JAMA in June 2025 followed 4,285 children from the Adolescent Brain Cognitive Development study across 21 US sites, from about age 10 through four years. It did the thing the panic usually skips: it pulled apart how much time kids spent from how they used it, mapping trajectories of addictive use, the can’t-stop, feel-sick-without-it, sneak-it-under-the-desk kind, for social media, phones, and video games.
The split is stark. Baseline total screen time was not associated with suicidal behavior, suicidal thinking, or the mental-health outcomes, full stop. Children on a high or increasing addictive-use trajectory carried a very different risk. Increasing addictive social-media use came with a risk ratio of 2.14 for suicidal behavior (95 percent CI, 1.61 to 2.85); the high-peaking group sat at 2.39 (1.66 to 3.43), a rough doubling to tripling against the low-use kids. And this was no fringe sliver: 31 percent of these children, 1,342 of 4,285, were on an increasing addictive social-media trajectory that started around age 11.
| Result | Value (RR) | Lower bound | Upper bound |
|---|---|---|---|
| Increasing addictive social-media use | 2.14 | 1.61 | 2.85 |
| High-peaking addictive social-media use | 2.39 | 1.66 | 3.43 |
| Measure | Share |
|---|---|
| increasing addictive social-media trajectory that started around age 11 | 31 percent |
The commentary running alongside it, from pediatricians at UC San Francisco, put it plainly: the compulsive pattern of use, not the raw dose, is what tracked with harm.
Then a report released October 1 by Australia’s National Mental Health Commission, drawing on the Longitudinal Study of Australian Children, added the part that stopped me. Existing mental-health problems predicted greater later technology use, especially social media, and that link ran stronger than heavy use predicting later mental-health problems. The direction a lot of us took for granted runs, at least partly, the other way.
The feed is where the distress goes
So why would the arrow point that way? Sit with it, because it flips the parenting script. A child who is already anxious, already low, already a little isolated does not get that way because she scrolled. She scrolls because the feed is the one thing that reliably answers when she reaches for it. The distress comes first. The phone is where it goes to get soothed, and the product is built so the soothing never quite finishes.
Which is why the design, not the duration, is where the Australian analysis points. Its authors single out the features built to keep a kid reaching back: algorithmically recommended content, beauty filters, the engagement tricks that pay out on an unpredictable schedule, the same intermittent-reward wiring a slot machine runs on. And the experiences stack. Each additional type of negative online encounter a child ran into was tied to a 1.5- to 2-fold rise in the odds of developing clinically significant symptoms.
per added type of negative online experience
- Lower bound
- 1.5
- Upper bound
- 2
Notice who the just-count-the-hours story is convenient for. If the problem is a number, the fix is a parental-control app and a weekly report to feel bad about, and the engineers who built the variable-reward machine get to walk away clean. A blanket under-16 ban runs on the same logic: treat the clock as the hazard, treat the design as a given, which is backwards from what two large datasets just suggested. I am not soft on Big Tech here. The harm sits in the architecture of compulsion they chose to build, and “screen time” is the frame that lets that choice disappear.
Let me be straight about what these studies can carry. Both are observational, so neither proves compulsive use causes the harm rather than riding alongside a deeper vulnerability, and the addictive-use pattern came from self-report questionnaires, which are as rough as they sound. There is no randomized trial here, and on a question like this there probably never cleanly will be. What the two datasets do give us is a direction: the variable everyone polices barely moved, and the pattern almost nobody measures moved a lot.
So I am retiring the weekly-hours number as my read on whether my people are okay. What I watch for now is the pattern. Can she put it down. Does she get agitated when she cannot. Is she reaching for the feed to fix a feeling. That is the signal the research actually found, and it is the one I will trust over the clock.