Online Program Home
My Program

Abstract Details

Activity Number: 475 - SPEED: Predictive Analytics with Social/Behavioral Science Applications: Spatial Modeling, Education Assessment, Population Behavior, and the Use of Multiple Data Sources
Type: Contributed
Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
Sponsor: Social Statistics Section
Abstract #330124
Title: A Multidimensional Array Model for Religiosity
Author(s): Guangyu Tong*
Companies: Duke University
Keywords: separable factor analysis; array data; religiosity; cohort effect; multiway data; Bayesian estimation

Secularization (i.e., whether there is a decline of religiosity across time) has been one of the most debatable topics in the sociological study of religion. Empirical analysis often utilizes the individual-level service attendance frequency obtained in the large-scale cross-sectional surveys across time to characterize secularization, and such measure is often correlated with age, gender, cohort and geographical regions. Existing regression models for such data often allows for limited correlation structure to model the attendance frequency, leading to inefficient estimates and potentially misleading image of secularization. This study treats the religiosity data as a multi-dimensional array and utilizes the separable covariance model proposed by Fosdick and Hoff (2014) to model the service attendance. Using data from multiple surveys of European and North American countries, this study shows the rich structure of changes in religiosity. A cross-validation check is also implemented to demonstrate the superior performance of the separable covariance model over simpler methods.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program