Identity Importance and the Overreporting of Religious Service Attendance: Multiple Imputation of Religious Attendance Using the American Time Use Study and the General Social Survey
The difference between religious service attendance measured using conventional surveys and time diaries has been attributed to identity processes; a high level of religious identity importance may prompt overreporting on a survey question. This article tests the hypothesized role of identity importance as an individual determinant of overreporting and the result of socially desirable behavior. A time diary measure of attendance (from the American Time Use Study 2003–2008) is imputed for conventional survey data (from the General Social Survey 2002–2008) using the multiple imputation for multiple studies procedure ( Gelman, King, and Liu 1998a ). Logistic regression models predicting self-reported attendance and overreported attendance are estimated using identity importance as a key covariate and controlling for demographic variables associated with attendance. Identity importance is a strong predictor of both self-reported and overreported attendance. Attendance, while a biased measure of actual behavior, may be a good indicator of religiosity.