Risk for marijuana-related problems among college students: an application of zero-inflated negative binomial regression.

METHOD: This study examined the association between marijuana-related problems and social norms, impulsivity, and perceived use utility among 292 college students. Zero-inflated negative binomial regression was used to simultaneously predict expected nonusers as well as predict counts of reported marijuana-related problems among expected users. Gender, social norms, impulsivity, and perceived use utility were used to predict expected nonusers as well as number of marijuana-related problems among expected users.
RESULTS: Only social norms were associated with the prediction of zero-values. In contrast, only perceived use utility was associated with the prediction of number of marijuana-related problems.
CONCLUSIONS: Results generally are consistent with theories of the differential association of social-environmental and biopsychological variables with use and problems, respectively. Zero-inflated regression models are a useful strategy to examine risk behaviors with low base rates.