What the researchers argue about, and what they don't
Two camps, the same data, opposite conclusions about screens and children's health. We read the studies from both sides. The fight is real, but five things everyone agrees on, and they are enough to wait.
By Barnluren
In March 2024 Jonathan Haidt published The Anxious Generation and said that the smartphone and social media had "rewired" a generation's childhood and caused an epidemic of mental illness. Two weeks later the psychologist Candice Odgers wrote in Nature that the claim was "not supported by science". Both are senior researchers. Both have read the same studies.
What is a parent supposed to do with that?
We went through the studies the two camps lean on. It turned out that the fight is real, but smaller than it looks. The camps partly measure different things, and there is a core neither of them questions.
What the fight is about
The sceptics are right about one thing: the link between total screen time and how young people feel is small. In the most cited study, Orben and Przybylski 2019, the researchers went through three datasets with 355,000 teenagers and let every reasonable way of counting speak. Screen time explained at most 0.4 percent of the variation in well-being. A Stanford meta-analysis of 226 studies landed on a correlation of 0.01. That is zero in practice. And when Vuorre and Przybylski compared two million people in 168 countries with their countries' internet rollout, they found no smoking gun.
Haidt's side is right about another thing: social media has demonstrated effects on some groups, and some of the evidence is causal. The strongest comes from economists, not psychologists. When Facebook was rolled out college by college in the United States between 2004 and 2006, Braghieri, Levy and Makarin could compare students who got access early with those who had to wait. Those who got Facebook reported more depression, and the mechanism was social comparison. In a randomised experiment, Allcott and colleagues had 2,700 people switch off Facebook for four weeks. Their well-being went up, they saw family and friends more, and many kept using Facebook less afterwards. A pooled analysis of ten such experiments from 2025 gives a small but real effect on depression.
So: a small effect of screen time on average, a real effect of social media on some. The two claims do not contradict each other. They only look like they do when the debate squeezes them into one.
It also means that whoever says "the research shows screens harm children" is wrong, and whoever says "the research shows it is harmless" is wrong too. The honest wording is longer: what is shown, for whom, and by what method.
Five things they don't argue about
1. Sleep. A screen in bed means shorter and worse sleep. Carter and colleagues pooled 20 studies with 125,000 children: a device in the bedroom more than doubled the risk of too little sleep, and it was enough that the device was there, it did not need to be used. This is the most robust finding in the whole field, and the Swedish Public Health Agency's advice "the phone outside the bedroom at night" rests on it.
2. The school day. The phone in the classroom lowers results, most for those already struggling. When English schools banned phones, test scores rose by six percent of a standard deviation, and by fourteen for low achievers, while high achievers were not affected at all (Beland and Murphy). In Norway, Sara Abrahamsson followed 400 lower secondary schools that introduced bans: girls' grades went up, their visits to psychologists for mental health problems fell sharply, and bullying fell for both girls and boys. No effect showed on boys' grades. UNESCO reckons it takes up to 20 minutes to regain concentration after an interruption.
There is a counter-image here worth taking seriously. An English study in the Lancet from 2025 compared 30 schools with different rules and found no difference in pupils' well-being or total screen time. The ban did not move the use, it just waited until three o'clock. That does not contradict the Norwegian and English results, it explains them: school governs six hours of the day. The rest belongs to parents.
3. The sensitive years. The effect of social media is not spread evenly across ages. Orben and colleagues followed 17,000 young people in Britain over time and found two windows where heavy use predicts lower life satisfaction a year later: girls at 11 to 13, boys at 14 to 15. That is the sceptic camp's own study, and on age it points the same way as the Nordics.
4. Displacement. Time on the screen is time not spent on sleep, movement or being with others. You do not have to believe in "rewiring the brain" to see it. Allcott's participants who switched off Facebook filled the time with family and friends. The Swedish Public Health Agency's evidence review from 2024 says exactly this: screens must not crowd out other important things.
5. The gap. Almost all the research concerns teenagers and adults. Odgers points out in her review that the experiments Haidt cites almost exclusively involve 18 to 35-year-olds, and that no study involves children under 13. Haidt's side does not dispute that. The consequence is uncomfortable for everyone: nobody can say today "the research shows nine-year-olds are harmed by a smartphone", and nobody can say the opposite.
What Sweden says
The Public Health Agency uses the word "association", not "cause", when it describes the research behind its recommendations from September 2024: under two, preferably no screens at all; two to five, at most one hour a day; six to twelve, one to two hours; thirteen to eighteen, two to three hours, all counted as leisure time. The research behind them speaks of associations with worse sleep, depressive symptoms, dissatisfaction with one's body, less physical activity. That is a correct reading of the field. In June 2026 the agency added the advice to wait with a smartphone of one's own until the year the child turns 13, and Finland's THL had said the same six months earlier.
Swedish young people's psychosomatic complaints, headaches, low mood, stomach aches more than once a week, have doubled among 13 and 15-year-olds since the mid-1980s. That is the Swedish counterpart of the American curve where the share of girls with persistent sadness went from 35 to 57 percent in ten years. But the Swedish rise began long before the smartphone, and that is the sceptics' best argument on Swedish soil.
At the same time, something happened that no authority ordered. Between 2022 and 2024 the share of nine-year-olds with their own mobile fell from 82 to 67 percent, and 9 to 12-year-olds' screen time fell by 40 minutes a day. Parents turned before the law. The phone ban in compulsory school only came in August 2026.
Why the Nordics set 13 without proof
Sweden, Finland and Denmark have all landed on 13 for the first smartphone. None of them claims the number is proven. It rests on three things: Orben's sensitive window that opens at 11, a global study of 100,000 young adults in which well-being was better for every year the first smartphone was delayed, and the precautionary principle. The US Surgeon General put it plainly: we cannot conclude that social media is sufficiently safe for children.
That is a reasonable decision under uncertainty. It is not a proven one. The difference matters, because whoever promises more than the research holds loses the parent who reads on.
What it means for us
We sell a phone, so read this section with that in mind. None of the researchers above recommends us, and none of them has commented on products like ours.
But we stand in the gap. The child who is six to twelve and waiting is the group nobody has researched. The only things shown for that age are sleep and the school day, and a landline without a screen touches neither: it does not lie in the bed, and it does not go to school. What is unclear, whether social media harms a ten-year-old, can be postponed to an age where the evidence is better. Meanwhile the child calls grandma.
We wrote about the parents organising that wait in Mind the gap before the smartphone. This post is the answer to the question they get most often: is it really that dangerous? The answer is that nobody knows for the youngest, that several things are shown for the older ones, and that what is shown is enough.
Method and sources
Every link in the text goes to the study or the university's own page, not to press coverage. The studies are ranked by method: randomised experiments and quasi-experiments weigh more than cross-sectional studies, regardless of camp. Haidt's list of experiments lives in his and Zach Rausch's own working document, not in a peer-reviewed journal, and is therefore not quoted with numbers. Ferguson's meta-analysis finding a null effect of reducing social media exists, as does the criticism of it. Ward and colleagues' experiment showing that the mere presence of a phone lowers working memory is not in the text but belongs to the picture.