Associations and Potential Pathways Linking Excessive Short-Video Use to Mental Health Among Vocational High School Students in China: An Integrated Analysis Based on Machine Learning and Path Analysis
Excessive short-video consumption is a global concern, yet its association with mental health among vocational high school students in China remains under-researched. Among many candidate individual, psychological, and behavioral factors, it remains unclear which are the most robust predictors of mental health. This study develops a comprehensive framework to identify core predictors and examine potential pathways linking short-video use to mental health. A survey was administered to 8346 vocational high school students recruited from vocational high schools in Tianjin, China, yielding 6096 valid responses for multi-stage analysis. First, Random Forest (RF) and linear Support Vector Regression (SVR) screened 13 individual, psychological, and behavioral features to identify core predictors. Subsequently, a Path Analysis (PA) with the Bootstrap method was used to examine statistical mediating pathways involving these core features. Across RF and SVR, social support, excessive short-video use, positive coping, and negative coping emerged as the four most important predictors of mental health. PA showed generally good fit (χ2/df = 5.86, CFI = 0.998, TLI = 0.99, RMSEA = 0.03, SRMR = 0.01). Excessive short-video use was significantly and directly associated with mental health problems (effect = 0.221, 95% CI = [0.189, 0.253], 76.7% of the total association) and showed small indirect associations through three statistical pathways: (1) lower social support (effect = 0.023, 8.1%); (2) higher negative coping (effect = 0.035, 12.0%); and (3) a small serial indirect association from reduced social support to decreased positive coping (effect = 0.007, 2.6%). Findings suggest that potential intervention efforts may consider regulating short-video use while simultaneously enhancing social support and adaptive coping mechanisms as part of strategies to reduce psychological risks.