Purpose
Studies showing religiosity is inversely associated with substance use tend to rely on cross-sectional designs and the few longitudinal studies fail to account for conceptually important variables such as peers and parental influences. Failure to control for well-established confounds may hinder the conclusions one can draw about the protective effect of religiosity on substance use. We examined the effect of adolescents' religiosity on substance use with longitudinal data, and controlling for numerous conceptually and empirically important variables such as parental monitoring, substance use, and religiosity, peer influences, and adolescent mental health.
Methods
Data are from two waves of a NIDA-funded longitudinal study of substance use among 800 youth-mother pairs in Santiago, Chile. Longitudinal information from youth and mothers were utilized to group the youth according to their propensity to being religious - low to high propensity. We used propensity score stratification (PSS) to investigate the role of adolescents' religiosity on their use of five substances (binary outcome of lifetime experience of drunkenness, nicotine dependence, and marijuana-use; count outcome of number of drinks and cigarettes consumed in the past 30 days). The PSS method with treatment heterogeneity differs from other regression-based models in that regression methods generally estimate a single coefficient measuring the average relationship between the independent and dependent variables under the unrealistic assumption that all youth respond similarly to religiosity. Recognizing the possibility of heterogeneity across different groups of youth who share similar observed (and potentially unobserved) characteristics, we used the PSS method to compute several estimates of the religiosity-substance use relationship.
Results
Approximately 57.0%, 18.5%, and 4.9% of adolescents indicated they were Catholic, Evangelical, or belonged to other religions. About 27.9% considered religious faith as a crucial determinant in making daily and major life decisions. A total of 13.4%, 17.8%, and 12.1% of adolescents had experienced drunkenness, nicotine dependence, and marijuana use. Mean past 30-day consumption of alcohol and cigarettes was 1.4 and 10.7, respectively. Results of the logistic and negative binomial regression models showed no significant associations between religiosity and adolescents' substance use. However, results from the PSS model indicated that, among adolescents with a low propensity of being religious, even with extensive controls, religiosity was a significant predictor of their lifetime experiences of getting drunk (p<0.05) and the number of alcoholic drinks consumed in the past 30 days (p<0.0001). No such associations were found among adolescents with a mid or higher propensity to being religious.
Conclusions
There are two mayor findings. First, the religiosity-substance use association became non-significant when controlling for well-established protective and risk factors were included. Second, when adolescents were stratified into several groups according to their observed tendencies of being religious and data analyzed with the PSS method, the religiosity-substance use association became significant for those adolescents with the lowest propensity to being religious. These latter findings point to the more specific adolescent populations that could benefit from interventions that encourage adolescents to examine and increase their religiosity, namely heavy-alcohol consuming adolescents with low propensity to be religious.
