Exercise Frequency and Subjective Well-Being Among Older Adults in China: Health-Related and Relational Pathways in Two National Surveys

Abstract Background Exercise is associated with well-being in later life, but it remains unclear whether health-related and relational pathways are reproducible across different surveys. We examined associations between exercise frequency and subjective well-being among older adults in China, distinguishing perceived health from contact and trust. Methods We analysed adults aged 60 years or older in the 2023 Chinese General Social Survey (CGSS; N = 2108) and the 2022 China Family Panel Studies (CFPS; N = 4672). Subjective well-being was measured by happiness in the CGSS and life satisfaction in the CFPS. Self-rated health represented the health-related pathway; social-contact frequency and generalized trust in the CGSS and neighbor trust in the CFPS represented distinct relational pathways. Adjusted ordinary least-squares models, bootstrap indirect-association estimates, exploratory interactions, survey weighting, and additional sensitivity analyses were used. Results Exercise frequency was positively associated with subjective well-being in both primary analyses. Self-rated health showed a positive indirect association in both surveys and was the only proposed pathway reproduced across datasets. Relational findings were indicator-specific: social-contact frequency did not show a supported indirect association in the CGSS, generalized trust showed a small indirect association in the CGSS, and neighbor trust did not show an indirect association in the CFPS. A gender interaction appeared only in the CGSS, while age and residential-background interactions were not significant in either survey. In weighted analyses, the CGSS association remained statistically significant; the weighted CFPS estimate remained positive but was weaker statistically (b = 0.0123, p=.064). Conclusions More frequent exercise was associated with higher subjective well-being among older adults in two Chinese national surveys. The most reproducible indirect pattern involved self-rated health, whereas relational pathways depended on the indicator measured. These findings support distinguishing health-related from relational processes and caution against interpreting cross-sectional indirect associations as causal mechanisms.