Longitudinal effects of physical activity on university students’ mental health via self-efficacy and the moderating role of peer support

To investigate the longitudinal mechanisms through which physical activity (PA) impacts students’ mental health (well-being and emotional problems), and to systematically test the mediating role of self-efficacy alongside the moderating role of peer support. A two-wave longitudinal panel design with a 5-month time lag was conducted among 847 Chinese university students. Validated and culturally adapted scales were utilized to precisely assess physical activity (incorporating PE course exertion), well-being, emotional problems, general self-efficacy, and perceived peer support. Data were analyzed using Mplus 8.3 to construct cross-lagged panel models and longitudinal moderated mediation models, rigorously controlling for baseline variables and demographic covariates. (1) Cross-lagged analyses demonstrated that T1 physical activity significantly and positively predicted T2 well-being (β = 0.15, p < 0.01) and negatively predicted T2 emotional problems (β = -0.12, p < 0.01), with no significant reverse causality paths identified. (2) Longitudinal mediation analyses revealed that T2 self-efficacy fully mediated the pathways from physical activity to both mental health indicators, accounting for 61.8% and 64.3% of the total effects, respectively. Upon the inclusion of this mediator, the direct effects of physical activity on mental health were rendered non-significant. (3) Moderated mediation models confirmed that peer support significantly moderated the initial segment of the mediational chain (Physical Activity → Self-efficacy; interaction term β = 0.09, p < 0.01). Simple slope and conditional indirect effect analyses indicated that a high-level peer network synergistically amplified the efficiency with which physical activity translated into self-efficacy. Physical activity exerts a robust temporal protective effect on university students’ mental health, a macro-level benefit realized primarily through the cultivation of deep-seated self-efficacy. Furthermore, a high-quality peer support environment serves as a critical socio-ecological catalyst that amplifies these psychological dividends. By partially addressing the temporal limitations of cross-sectional designs, this study provides empirical foundations for a psychologically-oriented paradigm shift in PE curricula and the development of cross-departmental synergistic interventions.