Initial Psychometric Evaluation of the Problematic Use of Generative Artificial Intelligence Scale Among Chinese School Students

Key Findings

• — The 8-item PUGAIS-8 showed a dominant one-factor structure with mixed confirmatory fit, Cronbach's alpha 0.940, and invariant thresholds and loadings across sex, school type and age group
• — PUGAIS-8 scores correlated with depressive symptoms (PHQ-9), anxiety (GAD-7), insomnia (ISI) and short-video addiction (r=0.365-0.413, all p<0.01)
• — Latent class analysis identified a high-score subgroup of 7.2%; an exploratory ROC analysis gave an internally derived descriptive threshold of 20.5
• — A substantial floor effect limited differentiation at the low end, and the item-reduction result was partly sensitive to analytic treatment
• — The analytic sample was 93.0% of students reporting prior generative AI use, after excluding respondents with more than 20% missing data on key measures

Methodology Notes

Single-city school survey (Wenzhou), grades 4-9; random split-half exploratory and confirmatory factor analysis; no test-retest data; the threshold is internally derived and not clinically validated; the authors note the SAS-SV adaptation may miss features specific to generative AI and that the cross-sectional design rules out causal inference. 'Generative AI use' is not limited to chatbots. Published 2026-09-28. mdpi.com returned 403 directly; the full text was read through a reader proxy, and the abstract is also in Crossref and OpenAlex. The author list above is the first eight in Crossref; the full text lists further authors, including a corresponding author at The Chinese University of Hong Kong.

Sources