Screen Time and Kids: No Signal at Ages 5 or 10, Heavy Gaming at 15

The argument in most houses is not really about research. It is about a Tuesday: a ten-year-old who has been on the tablet since dinner, a parent doing the arithmetic on how many hours that makes, and a decade of guidance in the background implying that the number itself is the problem. Then a week of coverage arrives saying the number may not matter much after all, which lands as something stranger than reassurance.

The study underneath that coverage was published on 3 August in JAMA Network Open by Fiona McQuaid, Valeria Skafida, Ruth McQuillan and Bohee Lee. It is a secondary analysis of a large Scottish birth cohort, following the same children at ages 5, 10 and 15 between 1 January 2010 and 31 December 2020, with the data obtained and analysed between April 2025 and March 2026. Screen use during non-school hours was reported by parents, and by the young people themselves as they got older.

The outcome was the Strengths and Difficulties Questionnaire, completed by a parent at each time point. The authors read it two ways: as a continuous score, and split at the established cutoffs, where a total difficulties score of 14 or above counts as above average and a prosocial score below 8 counts as below average. Three groups of children sit behind the numbers that follow: 3,786 at age 5 (median age 58 months, 51.0% male), 3,126 at age 10 (median 122 months) and 2,598 at age 15 (median 175 months). The authors' stated reason for the exercise is a gap in what came before, since few studies, as they put it, account for longitudinal associations or fully adjust for key covariates such as family functioning.

Cross-sectionally, with screen use and score measured at the same age, ages 5 and 10 came back empty. Neither showed a significant association between screen use and total difficulties. Daily screen use rose as the children got older and the pattern varied by screen type, but neither fact translated into a difficulties score that moved with the hours.

At 15 it moved, in one place. More than 5 hours a day of gaming was associated with 2.61 times the odds of an above-average total difficulties score compared with less than an hour a day (95% CI, 1.49 to 4.57). That is the study's strongest and most quotable number, and it repays being read precisely: it compares the heaviest gamers with the lightest, at one age, in one cohort, at a single moment in time.

The longitudinal arm asked whether screen use at 5 forecast anything later. More than 5 hours a day at age 5 was associated with increased odds of an above-average difficulties score at age 10 (OR, 2.68; 95% CI, 1.25 to 5.76). By 15 that association was gone, and it also did not hold up in the authors' own sensitivity analyses, which is the more consequential half of the sentence. The prosocial score showed no associations anywhere.

The finding least likely to reach a headline is the one that complicates the story most. At 15, more than an hour a day of social media and messaging was associated with lower odds of an above-average difficulties score than less than an hour a day, with odds ratios running from 0.57 (95% CI, 0.39 to 0.84) to 0.61 (95% CI, 0.38 to 1.00) for one to two hours a day.

There are at least two honest readings of that, and this design cannot choose between them. One is that ordinary sociable messaging at 15 is simply what ordinary sociable fifteen-year-olds do, so a teenager doing almost none of it may register on a difficulties questionnaire for reasons that have nothing to do with a phone. The other is that the association reflects what parents notice and report rather than what is happening. The second interval is worth its own sentence: 0.38 to 1.00 reaches no effect at its upper bound, so that particular estimate is compatible with nothing happening at all.

The same caution applies at the gaming end, in reverse. A cross-sectional association at 15 cannot say whether long gaming sessions raise difficulty scores or whether a teenager who is already struggling spends longer inside a game. Both readings fit these data, and the study was not built to separate them.

The authors' conclusion is deliberately flat: screen use was not consistently associated with poorer outcomes for children. Their next sentence is the one with consequences, since guidelines focusing solely on screen time, they write, may have limited impact on psychosocial outcomes.

That is a claim about policy instruments rather than a verdict that screens are harmless. It says a rule specified in hours is aiming at a variable this cohort could barely detect, while the variable that did separate groups was what the screen was being used for. Five hours of gaming and five hours of messaging did not behave alike here; at 15 they pointed in opposite directions.

What the study does not supply is a replacement number. It contains no threshold a family could adopt, no evidence at all about content quality or co-viewing, since neither was measured, and nothing at ages 5 and 10 beyond an absence. An absence measured across 3,786 children carries more weight than an absence measured across 300, and it remains an absence rather than a clearance.

The honest version of this study is duller than either version now in circulation. It is not evidence that screens are fine, and it is not the collapse of the case against them; it is evidence that hours-per-day is a weak instrument for the thing everyone is trying to measure. What persuades here is not the gaming result, which is a single cross-sectional comparison at one age and needs replication before it carries weight, but the shape of the whole table: three ages, two outcomes, several screen types, and almost nothing moving with duration, including a longitudinal association that folded under the authors' own sensitivity checks. A cohort this size failing to find a consistent signal is not proof of absence, and the authors do not claim that it is. What it does is move the burden somewhere more useful, onto what the hours are made of rather than how many there are, and that is a harder question than any guideline written in hours currently answers.