Longitudinal relationships between physical exercise, psychological resilience, and academic engagement in Chinese university students: a random-intercept cross-lagged panel study

Abstract

The relationships among physical exercise, psychological resilience, and academic engagement have predominantly been examined using cross-sectional designs, which preclude inference about directionality. This study examined their longitudinal relationships in a two-wave panel of Chinese university students. Of 1,024 students who provided valid baseline data (October 2024), 876 (Mage = 20.34, SD = 1.67; 58.3% female) provided valid follow-up data 6 months later (April 2025) on exercise volume (PARS-3), psychological resilience (CD-RISC-10), and academic engagement (UWES-S). Cross-lagged panel models (CLPM) and random-intercept cross-lagged panel models (RI-CLPM) were estimated in Mplus 8.8, with longitudinal measurement invariance established for all instruments. In the CLPM, exercise volume at T1 predicted resilience at T2 (beta = 0.17, p < 0.001) and resilience predicted engagement at T2 (beta = 0.14, p = 0.003), whereas the reverse paths and the direct exercise-to-engagement path were not significant. In the RI-CLPM, which separates between-person from within-person variance, the within-person association of exercise volume with subsequent resilience remained significant (beta = 0.11, p = 0.009) and was unaffected by assumptions about the reliability of the single-indicator exercise measure, whereas the within-person resilience-to-engagement path was attenuated (beta = 0.07, p = 0.074). Neither gender nor age moderated the relationships. Exercise volume is thus prospectively associated with subsequent resilience at the within-person level, whereas the resilience-engagement link appears to operate primarily between persons. Because the two-wave design and the single-indicator exercise measure require identifying constraints, the within-person decomposition is model-implied, and these findings are best interpreted as temporally ordered associations rather than confirmed within-person causal effects; exercise promotion stands as a candidate strategy for resilience support in university populations, pending experimental confirmation.