Big picture: The researchers followed a large group of adolescents over several years. They found that higher levels of mental health difficulties were linked to a slightly greater chance of later gaming disorder. They did not find evidence that gaming disorder predicted later increases in mental health difficulties.
Who took part?
- 4,289 adolescents were included. Their average age was 168.8 months, which is about 14 years old. The number in brackets, 8.2 months, is the standard deviation: it describes how much participants’ ages typically varied around the average.
- 2,391 of 4,288 adolescents (56%) were male. The total is 4,288 rather than 4,289 because sex information was missing for one participant.
- Income information was available for 3,877 households. Of those, 1,374 (35%) reported an income between $100,000 and $199,000. Income information was not available for every household.
What did the main statistical results mean?
The researchers used cross-lagged panel models, or CLPMs. In simple terms, these models look at whether one thing measured earlier is connected to another thing measured later, while also considering how both change over time.
- From year 2 to year 3: Higher mental health difficulty scores at the earlier measurement were linked to a greater likelihood of gaming disorder later. The reported value was β = 0.03, a small association.
- From year 3 to year 4: The same pattern appeared, with a somewhat stronger association: β = 0.07.
What is β (beta)? It is a statistical estimate showing the direction and size of a relationship. Here, positive values mean that higher earlier psychopathology scores were associated with more gaming disorder later. These numbers are not percentages, and they do not mean that 3% or 7% of adolescents developed gaming disorder.
Confidence intervals and P values
A 95% confidence interval (CI) gives a range of values that are reasonably compatible with the data. For the year 2-to-3 result, the range was 0.002 to 0.06; for the year 3-to-4 result, it was 0.04 to 0.10. Both ranges are above zero, supporting the idea that the relationships were positive.
The P value helps show how surprising the results would be if there were actually no relationship. P = .003 and P < .001 are considered strong statistical evidence against no relationship. A low P value does not tell us how important an effect is, or prove that one factor caused the other.
What happened after accounting for other factors?
The researchers also adjusted for other personal characteristics that might be related to risk. The link between earlier psychopathology and later gaming disorder remained from year 3 to year 4, but it was smaller: β = 0.04, with a 95% CI of 0.002 to 0.07 and P = .04. This suggests a small remaining association after those other factors were considered.
What did the researchers not find?
They did not find that gaming disorder was associated with later increases in psychopathology. In other words, the results showed a link in one direction—earlier mental health difficulties were associated with later gaming disorder—but not in the opposite direction.
A second type of analysis, called a hierarchical mixed-effects model, took into account that measurements came from the same people over time and that participants were grouped in the data. It produced results that supported the CLPM findings.
Important takeaway: The study found an association, not proof of cause and effect. It suggests that adolescents with more mental health difficulties may be at somewhat higher risk of later gaming disorder, but it does not show that mental health difficulties directly cause gaming disorder.