AUTHOR=Li Chao , Ning Guangjie , Xia Yuxin , Liu Qianqian TITLE=Health benefits of physical activity for people with mental disorders: From the perspective of multidimensional subjective wellbeing JOURNAL=Frontiers in Psychiatry VOLUME=13 YEAR=2022 URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2022.1050208 DOI=10.3389/fpsyt.2022.1050208 ISSN=1664-0640 ABSTRACT=

This paper uses a large scale and nationally representative dataset, Chinese General Social Survey, to empirically examine the role of physical activity in reducing the negative effects of depression among people with mental disorders. Empirical results demonstrate that physical exercise could help to alleviate depression's adverse consequences on work and life for depressed individuals. The impact mechanism is that physical activity may decrease the severity of depression, enhance life satisfaction, improve mood, and make people have a better sense of purpose and meaning in life. Therefore, from the perspective of multidimensional subjective wellbeing, evaluative wellbeing, experienced wellbeing and eudaimonic wellbeing all play mediating roles in the reduction of depression's adverse effects. Heterogeneity analysis shows that there are no significant gender differences in the health benefits of physical exercise, but its impact tends to be more prominent for depressed individuals who are younger and higher educated, with better health status, and live in urban areas. It is also found that socioeconomic status may play an important moderating role. The health benefits of physical activity seem to be greater for depressed people who have lower income, work in the secondary labor market, and have lower levels of social capital and assets. In addition, the instrumental variable approach is used to identify the causal impact of physical activity, which further proves a significant effect of it based on tackling the endogeneity problem. Meanwhile, this paper uses different explanatory and explained variables, different statistical models, as well as machine learning and placebo techniques to conduct robustness tests, all of which lend credence to above findings.