Screen Time Umbrella Review: 2.8 Million Kids, Small Effects

The largest screen time umbrella review yet, published in JAMA Pediatrics on September 28, 2026, pooled 23 meta-analyses covering 2,817,252 children in 64 countries. Its verdict on screens and child development: the links are real, and most of them are small. Screen use was associated with lower academic achievement (r = −0.15), weaker language skills (r = −0.14) and more internalizing and externalizing behavior problems (r ≈ 0.10). It was not associated with executive function, or clearly with math, science or reading scores. For parents, the useful question is what “small” means in a real family, and why a small average can hide a large effect on a particular child. This review answers the first question well. The second needs other studies.

Key Takeaways

• The review pooled 23 meta-analyses, covering 2.8 million children ages 18 and under from 64 countries, published through November 2025.
• Screen use was linked to lower academic achievement (r = −0.15) and language skills (r = −0.14), and to more internalizing (r = 0.10) and externalizing (r = 0.11) behavior.
• Two outcomes had larger but much less precise estimates: emotional problems (r = 0.30) and peer problems (r = 0.28), each with very wide confidence intervals.
• No association was found with executive function, and results for math, science and reading were too uncertain to call.
• A correlation of 0.10 to 0.15 is small for any one child on average but can matter across millions. Other research shows effects vary widely between children, so the average is not your child.

What the Screen Time Umbrella Review Is

An umbrella review is a review of reviews: instead of pooling individual studies, it pools meta-analyses, each of which has already pooled many studies. The September 2026 umbrella review, “Screen Use and Child and Adolescent Development and Mental Health Outcomes,” was led by Elizabeth Al-Jbouri with colleagues including Sheri Madigan, Pasco Fearon and Dimitri Christakis.

The team searched Embase, MEDLINE and PsycINFO through November 25, 2025, for meta-analyses of population-based samples of children 18 or younger. They found 23. Two reviewers independently extracted data, and all effect sizes were converted to Pearson correlations (r) so they could be compared. The four outcome domains were academic functioning, cognitive skills, socioemotional development and mental health.

Two design facts shape how to read it. First, the underlying studies are population-based and report associations. The review describes “associations,” not effects, and most screen-time research cannot separate cause from correlation. Second, the authors found that screen type, the subtype of behavior problem, publication year and child age moderated some results. “Screen time” is a bundle of very different activities.

What the Review Found, Outcome by Outcome

Outcome	Correlation (r)	95% confidence interval	Plain reading
Academic achievement	−0.15	−0.23 to −0.07	Small negative link
Language skills	−0.14	−0.18 to −0.10	Small negative link, precisely estimated
Internalizing behavior (anxiety, low mood), general screen use	0.10	0.07 to 0.13	Small positive link
Internalizing behavior, social media use	0.09	0.06 to 0.12	Small positive link
Externalizing behavior (acting out)	0.11	0.08 to 0.14	Small positive link
Emotional problems	0.30	0.08 to 0.50	Possibly large, but very imprecise
Peer problems	0.28	0.04 to 0.49	Possibly large, but very imprecise
Executive functioning	0.01	−0.09 to 0.11	No association
Prosociality	−0.22	−0.43 to 0.01	Not statistically significant
Mathematics	−0.10	−0.23 to 0.03	Not statistically significant
Science	−0.03	−0.15 to 0.11	Not statistically significant
Reading	−0.02	−0.12 to 0.09	Not statistically significant

The emotional and peer-problem rows deserve care. A correlation of 0.30 would be meaningful, but the interval running from 0.08 to 0.50 means the true value could be small or large. Those rows signal where more research is needed, not where the answer is settled.

What “Small” Actually Means

A correlation measures how closely two things move together, from 0 (no relationship) to 1 (perfect). The number is easy to misread, so here are three ways to translate it.

1. Benchmarks from psychology. David Funder and Daniel Ozer, writing in Advances in Methods and Practices in Psychological Science in 2019, proposed practical benchmarks. An r of 0.10 is “still small at the level of single events but potentially more ultimately consequential.” An r of 0.20 is medium, and 0.30 is large. They also warn that squaring r (to get “percent of variance explained”) makes effects sound smaller than their real-world consequences. By that standard, most of this review’s findings sit at small-but-potentially-consequential.

2. A two-group translation. Robert Rosenthal and Donald Rubin’s binomial effect size display, from a 1982 paper in the Journal of Educational Psychology, converts r into an intuitive comparison. Imagine splitting children into higher and lower screen-use halves, and splitting outcomes into above and below average. An r of 0.15 corresponds to roughly 57.5% of the lower-screen group landing above average in achievement, versus 42.5% of the higher-screen group. That is a visible difference across a classroom, though nowhere near destiny for any individual child.

3. Comparison with earlier work. In 2019, Amy Orben and Andrew Przybylski, in Nature Human Behaviour, analyzed three large datasets totaling 355,358 adolescents and found digital technology use explained at most 0.4% of the variation in wellbeing. The new review’s numbers are in the same family: real, consistent, and modest.

One more piece of honest framing. Correlations this size can be produced by other factors that travel with screen use, such as family stress, sleep, income or a child’s existing difficulties. The review can’t rule those out. It shows where screens and outcomes move together, not that one causes the other.

Why the Average Hides the Kids Most Affected

The phrase “small effect” is about the average child. Two lines of research show why that average can mislead.

Effects differ by child. Ina Beyens and colleagues, in Scientific Reports in 2020, sampled adolescents’ feelings six times a day for a week. After passive social media use, 44% felt no different, 46% felt better, and 10% felt worse. If one in ten children is meaningfully harmed while others are unaffected or slightly helped, the average looks small even though some children are genuinely affected.

Effects differ by age. Orben and colleagues, in Nature Communications in 2022, found “windows of developmental sensitivity” when heavier social media use predicted lower life satisfaction a year later: ages 11 to 13 for girls and 14 to 15 for boys. A review that pools all ages will dilute an effect concentrated in a few years. We cover that study in windows of sensitivity for girls online.

The umbrella review itself found that age and screen type moderated results, which points the same way. The practical upshot: the average says screens are not a catastrophe for most children. It does not say your child is the average.

What to Do at Home With a Small-Average Finding

Look at what screens are replacing

A small average link often reflects displacement. If screen time is crowding out sleep, conversation, reading or outdoor play, the trade-off matters more than the minutes. The language finding (r = −0.14), estimated precisely, is a nudge to protect talk and reading time, especially for younger children.

Watch your child, not the population

Use the Beyens finding as a lens. Notice how your child acts after specific kinds of screen use. Calmer and more connected after a video call with a cousin? Irritable after an hour of scrolling? That pattern is more informative than any population correlation.

Pay extra attention during sensitive windows

If the strongest effects concentrate in early adolescence, the years around 11 to 15 are when to be most deliberate about social media in particular: when accounts start, what feeds look like, and whether use is pushing out sleep.

Don’t panic over the large-but-imprecise numbers

The emotional and peer-problem correlations near 0.30 sound alarming but come with intervals wide enough to include small effects. Treat them as a reason to keep an eye on mood and friendships, not as evidence that screens cause emotional disorders.

What not to do

Don’t conclude “screens don’t matter” because effects are small, and don’t conclude “screens are poison” because they’re consistent. Both misread the data. For a balanced look at another heavily debated outcome, see our review of screen time and emotional regulation research.

What to Watch For Over the Next 3 Months

• Week 4: Track one week of your child’s screen use by type (school, video, games, social, calls). The review found type mattered. Most families have never looked at the mix.
• Month 2 red flags: Signs that screens are displacing something important: later bedtimes, less talking at dinner, dropped activities, or a mood change after particular apps. Those are the mechanisms behind small average links.
• Month 3 self-check: Has anything changed after you adjusted the mix, not just the total? If your child is sleeping and talking more, the change is working, whatever the minute count says.

Frequently Asked Questions

What did the 2026 screen time umbrella review find?

Pooling 23 meta-analyses of 2.8 million children, it found screen use linked to lower academic achievement (r = −0.15) and language skills (r = −0.14), and to more behavior problems (r ≈ 0.10). It found no link with executive function, and no clear link with math, science or reading scores.

Does this review prove screens harm children?

No. It reports associations from population-based studies. Screen use may contribute to these outcomes, or children with existing difficulties may use screens more, or other factors such as sleep and family stress may drive both. The review cannot separate those explanations.

How big is a correlation of 0.15?

Small but noticeable at scale. Using a standard two-group translation, it corresponds to about 57.5% versus 42.5% of children landing above average when split by screen use. Effect-size researchers describe correlations around 0.10 as small for single cases but potentially consequential over time.

Why do some studies find bigger screen effects than others?

Because effects vary by child, age and type of screen use. Studies of specific ages, platforms or vulnerable groups can find larger effects that disappear when pooled across all children. The umbrella review found that age and screen type moderated several results.

Should I change my child’s screen limits because of this review?

The review supports paying attention to what screens displace, such as sleep, conversation and reading, more than fixating on a single daily number. If your child seems fine and is sleeping, learning and connecting, small average associations are not a reason to overhaul your rules.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Al-Jbouri, E., Nivison, M. D., Neville, R. D., Marsh, A., McArthur, B. A., Christakis, D. A., Fearon, P., & Madigan, S. (2026). “Screen Use and Child and Adolescent Development and Mental Health Outcomes: A Quantitative Umbrella Review.” JAMA Pediatrics. https://doi.org/10.1001/jamapediatrics.2026.4339
2. Funder, D. C., & Ozer, D. J. (2019). “Evaluating Effect Size in Psychological Research: Sense and Nonsense.” Advances in Methods and Practices in Psychological Science. https://doi.org/10.1177/2515245919847202
3. Rosenthal, R., & Rubin, D. B. (1982). “A simple, general purpose display of magnitude of experimental effect.” Journal of Educational Psychology, 74(2). https://doi.org/10.1037/0022-0663.74.2.166
4. Orben, A., & Przybylski, A. K. (2019). “The association between adolescent well-being and digital technology use.” Nature Human Behaviour. https://doi.org/10.1038/s41562-018-0506-1
5. Beyens, I., Pouwels, J. L., van Driel, I. I., Keijsers, L., & Valkenburg, P. M. (2020). “The effect of social media on well-being differs from adolescent to adolescent.” Scientific Reports. https://doi.org/10.1038/s41598-020-67727-7
6. Orben, A., Przybylski, A. K., Blakemore, S.-J., & Kievit, R. A. (2022). “Windows of developmental sensitivity to social media.” Nature Communications. https://doi.org/10.1038/s41467-022-29296-3