JMIR Publications

ABSTRACT

Background:

Perceived stress and internet gaming disorder (IGD) risk often emerge during late adolescence, when digital environments are central to learning, recreation, and social interaction. How specific stress symptoms connect with IGD risk symptoms remains unclear.

Objective:

We examined symptom-level associations between perceived stress and IGD risk among Chinese late adolescents and identified model-derived symptom candidates for future longitudinal and experimental research.

Methods:

We conducted a cross-sectional survey between September 2024 and September 2025 using convenience cluster sampling. The analytic sample included 1212 Chinese late adolescents. We estimated symptom-level associations using Ising networks and Gaussian graphical models. We then used exploratory Bayesian networks and model-based simulations to generate hypotheses about directional dependencies and candidate symptom targets.

Results:

Overall, 9.74% of participants met the screening threshold for high IGD risk, and 7.51% met thresholds for both high IGD risk and elevated perceived stress. Male participants reported higher IGD scores (t=10.02, P<.001, Cohen d=0.57), whereas female participants reported higher perceived stress (t=3.18, P=.002, Cohen d=0.18). Across network specifications, gaming tolerance, especially the perception that a whole day was insufficient for in-game activities, was the most consistent bridge symptom (bridge expected influence=2.29). Exploratory Bayesian analysis placed perceived difficulty managing daily hassles upstream of gaming-related emotion-regulation symptoms. In model-based simulations, this stress-control symptom produced the largest modeled reduction in combined symptom scores (meanSumscore=10.843).

Conclusions:

Perceived stress and IGD risk were connected through specific symptom-level pathways in this late-adolescent sample. Gaming tolerance may be a robust bridge feature, whereas difficulty managing daily hassles may warrant further study as a stress-regulation target. Because the study was cross-sectional and simulation-based, the findings should be interpreted as hypothesis-generating rather than causal or clinical evidence.