SEARCH

SEARCH BY CITATION

ABSTRACT The intent of the present study was to determine whether the selective aggregation of background data subgroups could be used to identify and control for differences across persons in the meaning of religious involvement. After background data subgroups had been clustered based on similarity in the variables related to religious involvement, two clusters were identified. One cluster contained individuals whose religious involvement was characterized by a functional, instrumental orientation, while the other was composed of individuals whose religious involvement was characterized by an affective, expressive orientation. Subsequently, causal models intended to describe continuity and change in religious involvement were constructed using individuals assigned to each cluster. It was found that different models were required to describe continuity and change in religious involvement in the instrumental and expressive clusters.