Polish teens who spend 5+ hours a day on social media are three times as likely to show problem use

Twelve point seven percent. That is the share of 9,073 Polish schoolchildren who, asked nine plain questions about the past year, said yes to at least six of them. Have you regularly found yourself unable to think about anything other than the moment you could next use social media? Have you lied to your parents about how much time you spend on it? Have you had serious conflict with family over it? Six yeses out of nine is the threshold the scale's authors set for problematic social media use, and roughly one student in eight cleared it.

Katarzyna Porwit, Monika Szkultecka-Dębek and Zbigniew Izdebski ran the survey between March and May 2024 across the Lubuskie Voivodeship in western Poland. They drew 88 schools at random from the education ministry's database, half primary and half secondary, and surveyed roughly the 13, 15 and 17 year old grades. About 30% of the region's students in those grades took part. The questions came mostly from the international Health Behaviour in School-aged Children protocol, which means the answers can be lined up against surveys elsewhere.

The strongest single risk factor was the obvious one, but the size of it is worth sitting with. Students who said they spent five or more hours a day on social media were, after the team adjusted for everything else they measured, about 2.7 times as likely to meet the problem-use threshold as students who spent an hour or less. In raw terms, 22.3% of the heaviest users qualified, against 7.5% of the lightest. The two to four hour group sat in between at 11.3%.

Age ran the other way. Seventh graders, around 13 years old, showed the highest rates at 14.8%; by eleventh grade, around 17, that had fallen to 8.9%. The authors read this as a developmental signal rather than a generational one: younger adolescents lean harder on their peer group and have less mature self-regulation, so the pattern may loosen as they get older rather than lock in.

The gender result that flipped

The most peculiar finding in the paper is one the authors flag as new. Looking at the raw numbers, girls came out worse: 14.1% met the threshold against 10.9% of boys, matching what most previous studies have found. But the team built their statistical model in blocks, adding one group of variables at a time. When they added hours spent on social media, the gender difference vanished. When they added health and well-being measures, it came back, pointing the other way. In the final model, girls were about 24% less likely than boys to show problem use.

This is what statisticians call suppression: girls spend more time on social media and report more psychosomatic complaints, and those two things had been carrying the apparent gender effect. Hold them constant and the sign reverses. The authors treat it as a real and novel observation, though it is worth reading as a claim about comparison rather than about lived experience: among boys and girls with the same screen hours and the same symptom load, the boys were doing worse.

Among the health measures, physical symptoms mattered most. Students reporting at least two psychosomatic complaints (headaches, stomach aches, trouble sleeping, irritability and the like) more than once a week were about 2.2 times as likely to show problem use. Depression risk, measured separately, dropped out of the final model once relationships were accounted for; only mildly reduced mood stayed significant.

Protection came from three places: getting older, having a strong sense of future orientation, and getting on well with parents. Future orientation, essentially the capacity to delay a reward and stick to a goal, produced the sharpest split in the whole table. Among students scoring low on it, 21.5% met the threshold; among those scoring high, 6.9%. Students who described good relationships with their parents were roughly 43% less likely to show problem use than those in the middle group.

Two results resisted a tidy story. Peer support and school climate both behaved non-linearly: students with low peer support showed elevated risk, but so did students with high peer support. A supportive school climate was, in the final model, associated with more problem use, not less. The authors are careful here. They suggest heavy messaging and fear of missing out might drive the high-support pattern, and note that supportive schools may have looser phone rules, but they also concede the three social variables share so much variance that they cannot cleanly separate them.