Temporal and gender heterogeneity in the association between physical exercise and subjective wellbeing in China: an age–period–cohort analysis

Background Although the positive association between physical exercise (PE) and subjective wellbeing (SWB) is well documented, little is known about whether its magnitude varies across the life course, birth cohorts, survey periods, and gender groups. Methods This study analyzed pooled repeated cross-sectional data from five waves of the China General Social Survey (2012, 2015, 2018, 2021, and 2023), comprising 30,247 adults aged 18–97 years. A Hierarchical Age–Period–Cohort Cross-Classified Random Effects Model (HAPC-CCREM) was used to examine age-, period-, cohort-, and gender-related heterogeneity. Robustness was assessed using 5-year cohorts and the original five-category PE measure. Results Exercisers consistently exhibited higher predicted SWB than non-exercisers across the age range. The PE-related advantage followed a nonlinear age pattern, being pronounced in young adulthood, attenuating through middle adulthood, and strengthening again in later life. The cohort-level PE random-slope variance was statistically significant, whereas no statistically significant period-level variation in the PE slope was detected under the current model specification. Cohort-specific estimates descriptively indicated that the PE-related advantage was relatively large among earlier-born cohorts, narrowed across successive cohorts, reached a comparatively low level among those born during 1990–1999, and recovered modestly among those born in 2000 or later. Gender-related interaction terms were jointly significant. The PE–SWB association was stronger among men at the mean age, and the gender difference varied with age. The main findings remained stable across the sensitivity analyses. Conclusion The positive association between PE and SWB varies across age, birth cohorts, and gender. These findings highlight the importance of considering life-course position, generational experience, and gendered circumstances when examining heterogeneity in this association. Because the data are repeated cross-sectional and the proposed mechanisms were not directly measured, the results should not be interpreted causally. Differentiated PE and wellbeing promotion strategies may therefore be appropriate for different population groups.