Prospective associations between perceived school climate and psychological distress among Chinese adolescents: a three-wave longitudinal study of school belonging, social avoidance, and psychological resilience

Perceived school climate has been associated with adolescent mental health, but the prospective processes linking school experience with later psychological distress remain incompletely understood. This study examined school belonging, social avoidance, and psychological resilience as relational, behavioral, and adaptive-resource correlates. A three-wave school-based survey was conducted at approximately six-month intervals among 1,016 Chinese students in Grades 5–9 who provided valid data at all waves. Three separate traditional cross-lagged panel models (CLPMs) were the primary analyses. Measurement validity, common-method diagnostics, theoretically informed sensitivity analyses, exploratory cascade models, and random-intercept CLPM robustness checks were also conducted. The 14-item Social Avoidance subscale was the primary social-withdrawal measure. Measurement analyses supported the empirical distinctiveness of the focal constructs; all 30 mixed-indicator HTMT estimates were below 0.85 (range = 0.307–0.674). In the selected CLPMs, perceived school climate showed significant T1→T2→T3 indirect associations with Time 3 psychological distress through Time 2 school belonging (B = − 0.116, 95% CI [− 0.141, − 0.090]), social avoidance (B = − 0.077, 95% CI [− 0.092, − 0.063]), and psychological resilience (B = − 0.092, 95% CI [− 0.108, − 0.077]). Direct and reciprocal prospective associations were also observed. Because the random-intercept models did not yield admissible solutions, the traditional CLPM estimates cannot be interpreted as pure within-person change over time and may partly reflect stable differences between students who consistently reported different levels of school climate, the candidate variables, and psychological distress. More favorable perceived school climate was prospectively associated with lower later psychological distress through relational, behavioral, and adaptive-resource correlates. Because admissible RI-CLPM solutions were not obtained, these estimates cannot be interpreted as changes occurring within an individual student over time and may partly reflect stable between-student differences. The observational and primarily separate-model design also precludes causal or relative-importance claims.