Fig. 6: Coefficients of midpoints in the cross-sectional and longitudinal models for participation, importance of religion, and belonging.
From: The three stages of religious decline around the world
Based on all countries. With Pew data and WVS/EVS 7 data, the cross-sectional cases, we visualize the midpoints for older (40 + ) and younger ( < 40) individuals for the sigmoid models predicting percentage of weekly participation, high importance of religion, and religious belonging. With WVS/EVS 1-7 data, the longitudinal case, we visualize the sigmoid models predicting the three variables without distinguishing between older and younger cohorts. The lines represent highest (posterior) density intervals at the 95% level. Since the models are not linear, the intervals do not have to be symmetric. We control for continent, government support, government regulation, government discrimination, and pct migrants. In the longitudinal case, we omit the control for continent.
