LAUSR

Purpose: An empirical examination of the long-term association, disaggregated by gender, between religiosity and self-rated health with reference to demographic shifts in labor force participation, education, and income. Design: General Social Survey data. Setting: United States, 1974 to 2012. Participants: A total of 23 353 respondents. Measures: Self-assessed health; 2 key religiosity variables: attendance and intensity of belief; income, labor market variables, education, standard demographic variables, household size, region, and time dummies. Analysis: Probit estimation conducted for the aggregate sample by gender as well as by decades to examine possible gender differential changes over time. Results: Attendance has declined overall with a much greater decline for women. The overall positive association between religiosity and health masks considerable heterogeneity across gender and time; higher and stable for males, there is no longer a significant association for females. Increased education, income, and labor force participation can explain only part of this association. Education is the strongest mediator. Conclusion: The way women and men benefit from religious attendance has changed, suggesting that some pathways may be working differently for women now, especially those with less education. Moving away from church networks could be due to a perceived lack of support or substitution by other social networks. Ceteris paribus, since religious participation has been shown to weaken preference for risky consumption, declining participation, especially for women, may show up as an increase in risky behavior.