We use data from a nationally representative, longitudinal survey of older Taiwanese to examine the relationship between religious involvement--including religious affiliation, religious attendance, beliefs, and religious practices--and self-reported measures of overall health status, mobility limitations, depressive symptoms, and cognitive function; clinical measures of systolic and diastolic blood pressure, serum interleukin-6, and 12-h urinary cortisol; and 4-year mortality. Frequency of religious attendance shows the strongest, most consistent association with health outcomes. But, with only one exception, this relationship disappears in the presence of controls for health behaviors, social networks, and prior health status. Religious attendance remains significantly associated with lower mortality even after controlling for prior self-assessed health status, but the coefficient is substantially reduced. Other aspects of religiosity are only sporadically associated with health and, in all cases, private religious practices and stronger beliefs are associated with worse health; again, this relationship disappears after controlling for prior health status. These results suggest that reverse causality may partly account for both the positive and negative correlations between religiosity and health. We find no significant associations between religious involvement and biological markers. Notably, even after controlling for prior health, participation in social activities has a more robust effect on health than religious attendance. Consequently, we question whether the purported health benefits are attributable to religion or to social activity in general.
