Using invariance testing to study religiosity: A within culture comparison of the bases of religiosity

• Exploration and replication of invariance parameters of the bases of religiosity across six Brazilian religious groups showed partial scalar invariance. • The cognitive Believing and the Behavioral component of religiosity showed the fewest invariance issues, whereas the Bonding component of religiosity revealed most group-specific variability in the interpretation of ritual and symbols. • We demonstrate the utility of within-cultural invariance analysis, controlling for language and shared cultural history to provide new insights into the underlying psychological meaning of constructs of religiosity. • We strongly advocate for replication of invariance studies to attempt to identify core parameters that systematically vary between groups (vs sample specific fluctuations). We test Saroglou’s four-dimensional framework of religiosity in Brazil. The presence of religious traditions that vary in their formalization of institutional structures offers an important context for testing the scale. Using recent advances in invariance testing, we conducted two studies (total N = 3091) with adherents from five religious groupings as well as spiritual but not religious individuals compared to non-religious individuals. Overall, the four-factor scale showed good psychometric properties in the total sample. However, for followers of Afro-Brazilian religions as an example of non-institutionalized traditions a simpler single-factor structure was found. We obtained partial scalar invariance across a subset of religious groups. Explorations of the noninvariant item intercepts suggested that some items carried group-specific interpretations, especially those related to ritual practices and emotional experiences (Bonding dimension). Latent means were overall highest for Believing and lowest for Belonging, with Evangelicals scoring highest on all dimensions. Both the invariance and latent mean patterns were replicated and further differentiated in a replication study. Our study demonstrates the utility of within-cultural invariance analysis and highlights the importance of statistical replications for exploring cognitive dimensions underlying survey responses. Considering bases of religiosity, our invariance results suggest both a shared cognitive structure of religiosity and group-specific interpretations of motivations around ritual practices and emotions across religions varying in their level of institutionalization.