More screen use, slightly lower school achievement and more behavior problems, across 2.8 million children

Children and teenagers who spend more time with screens do slightly worse at school, have slightly weaker language skills and show somewhat more behavior and emotional problems. That is the conclusion of a new attempt to add up the research, covering about 2.8 million young people in 64 countries.

The word that matters is slightly. The links are real in the statistical sense and small in size, and several things parents worry about, including reading and mathematics scores, showed no reliable link. The review was published on September 28 in JAMA Pediatrics.

It also cannot say what causes what. Most of the studies behind it took one snapshot of each child.

What kind of review of screen use this is

Research on children and screens is vast and often contradictory. A single study surveys a few thousand families. A meta-analysis pools many such studies on one question to get a steadier answer. This paper goes one level higher. It is an umbrella review, which means a review of meta-analyses.

The authors set out to synthesize and quantify meta-analytic associations between screen-based media exposure and developmental outcomes among individuals aged 18 years or younger. Three research databases were systematically searched from inception to November 25, 2025, inception meaning the earliest records they hold. They kept meta-analyses of population-based samples of children aged 18 years or younger, meaning ordinary children and not groups chosen because they already had a diagnosis.

In the end, data from 23 meta-analyses were extracted and synthesized. The primary outcomes were academic functioning, cognitive skills, socioemotional development, and mental health.

An independent researcher who read the paper praised its method. “This is a methodologically very sound article. It is a review of reviews,” said Tatiana Íñiguez Berrozpe, a sociologist at the University of Zaragoza in Spain who was not involved. The authors removed studies that appeared in more than one meta-analysis, so that no child was counted twice.

What the review found on screen time and children’s development

Results are given as correlations. A correlation is a number between minus 1 and 1 that says how closely two things move together. Zero means no relationship. In this field, 0.1 is usually called small and 0.3 moderate.

Screen-based media exposure was associated with lower academic achievement and language skills, with correlations of minus 0.15 and minus 0.14.

On the mental health side, the review separates two kinds of difficulty. Internalizing behavior means problems turned inward, such as anxiety, sadness and withdrawal. Externalizing behavior means problems turned outward, such as aggression and rule-breaking. Both were more common with more screen use.

Outcome	Correlation with screen exposure	Plausible range
Academic achievement	-0.15	-0.23 to -0.07
Language skills	-0.14	-0.18 to -0.10
Internalizing behavior, any screen use	0.10	0.07 to 0.13
Internalizing behavior, social media	0.09	0.06 to 0.12
Externalizing behavior	0.11	0.08 to 0.14
Emotional problems	0.30	0.08 to 0.50
Peer problems	0.28	0.04 to 0.49

The last two rows stand out, and they are the shakiest. Their ranges run from almost nothing to a strong link. Íñiguez Berrozpe noted that the estimates for emotional problems and difficulties with peers are based on very few studies.

The picture was not uniform. Screen type, subtype of behavior problems, publication year, and age moderated some of these associations, moderated meaning that the size of the link changed with those factors. According to an independent commentary on the paper gathered by Spain’s Science Media Centre, the link with distress was stronger for mobile phones than for television or video games, and stronger in adolescents than in younger children.

What screen use was not linked to

Several outcomes showed nothing reliable. Screen-based media exposure was not associated with executive functioning, the set of mental skills used for planning, holding attention and self-control. Nor was it associated with prosocial behavior, which means helping and cooperating with others.

Scores in individual school subjects also showed no clear link. The correlations were minus 0.10 for mathematics, minus 0.03 for science and minus 0.02 for reading, each with a range that included zero.

That leaves a puzzle inside the school results: overall achievement was lower with more screen use, while the three subjects measured separately were not.

How small the screen time links are

The federal government in the United States has treated the subject as urgent. Scripps News reported that earlier this summer a federal advisory warned that harmful screen use among children and adolescents has become a growing public health concern.

The outside experts read the new numbers more coolly. “There is an association between screen time and poorer indicators of mental health and academic performance, but it is very small,” Íñiguez Berrozpe said. The expert put it at around 1% of the differences between one group of adolescents and another, meaning screen time accounts for about a hundredth of the variation among them.

A small average link across millions of children can still matter if it reflects a real effect. It can also hide a large effect in a few children and none in most. A correlation cannot tell these apart.

The same expert argued that the count of hours is the wrong thing to focus on. “It matters less how many hours devices are used for and more which device is used, for what purpose, with whom, and at what stage of development,” Íñiguez Berrozpe said.

Where this fits: earlier screen use reviews and one experiment

This is not the first umbrella review of the question. In 2023 an Australian-led team, independent of the new authors, harmonized effects from 102 meta-analyses (2,451 primary studies; 1,937,501 participants), harmonized meaning converted to a common scale. Their results were similar in size and more mixed in direction. Screen use was negatively associated with literacy, but this effect was positive when parents watched with their children. Social media was associated with depression, with a correlation of 0.12.

That team was blunt about the quality of what they were summarizing. They found medium-to-high risk of bias in 95 out of 102 included meta-analyses. Their advice to parents and governments was to carefully weigh the evidence for potential harms and benefits of specific types of screen use.

A second independent line of work asked how much weight such small links can bear. Researchers at the University of Oxford analyzed three large-scale social datasets (total n = 355,358), or 355,358 young people in all. The association they found between digital technology use and adolescent well-being was negative but small, explaining at most 0.4% of the variation in well-being. They concluded that these effects are too small to warrant policy change.

What nearly all of this lacks is an experiment. One exists. A Danish team noted that, for the idea that screens harm mental health, experimental evidence supporting this hypothesis is lacking, and ran a trial in 89 families (with 181 children and adolescents) from 10 Danish municipalities. Families in the intervention group had to reduce their leisure-time screen media use to 3 hours per week or less per person and hand over smartphones and tablets for two weeks.

At the end, children’s scores on a standard questionnaire of behavioral difficulties differed between the groups, favoring the screen media reduction intervention. The greatest improvements were observed for internalizing symptoms. The trial was small, lasted two weeks, and the parents who filled in the questionnaire knew which group they were in.

So the evidence lines up loosely. Surveys find small links, one experiment finds a short-term benefit from cutting back, and nobody has shown what happens over years.

What the screen use review cannot show

Direction is the main gap. Esther van Sluijs, a professor at Loughborough University in England who was not involved, noted that most of the data come from studies in which screen use and the outcome are measured at the same time. “This means that it cannot say much about whether higher screen use leads to poorer outcomes,” van Sluijs said. A child who is anxious or struggling at school may turn to screens more, and that would produce the same numbers.

Other influences are tangled in. Family income, parents’ education and how much time adults have for their children all shape both screen habits and school results. Íñiguez Berrozpe cautioned that studies of screen time typically do not adjust for family circumstances.

The measure is crude. Most studies ask about hours and not about what is on the screen. Víctor González Calatayud of the University of Murcia, who also reviewed the paper independently, said the purpose of use, the content and the level of interaction are aspects that are not usually taken into account.

And the technology keeps moving. Some of the pooled studies date from the era of television. The review gives one figure for social media, but van Sluijs said there was not enough data to draw conclusions about specific types of use such as social media or gaming.

Across 2.8 million children, more screen exposure went with slightly lower achievement and somewhat more behavior and emotional problems, in studies that mostly cannot say whether the screens came first.