It actually does. Religious affinity can absolutely be useful for longer trend studies, and census data is usually of much, much higher quality than other random sample studies.