Date of this Version
Journal for the Scientific Study of Religion (2010) 49(2), pp. 311-319.
Period and cohort effects on reports of no religious preference and religious disaffiliation are estimated. Results show that by 2006, the probability of reporting no religious preference had risen to almost 0.16. Additionally, the growing proportion of Americans raised with no religious preference substantially influences the increase in reporting no religious preference. Two recently developed analysis techniques for disentangling period and cohort effects are used: intrinsic estimator models and cross-classified, random-effects models. Results show (1) period-based increases in the probability of reporting no religious preference from 1990 to 2006, (2) across-cohort increases in no religious preference beginning with those born in the 1940s, and (3) a decline in the likelihood of religious disaffiliation (being raised with a religious preference but currently reporting no religious preference) for those born in the 1960s and 1970s compared to those born between 1945 and 1959.