Cannabis Abuse Screening Test (CAST): Psychometric properties and measurement invariance in a national sample of...

Background: Cannabis use is prevalent in young adults, and earlier onset and frequent use increase risk for problematic use. In Chile, the Cannabis Abuse Screening Test (CAST) is widely used, but evidence in higher education students remains limited.Objectives: To evaluate the psychometric properties of the CAST in a national sample of Chilean higher education students.Methods: Secondary analyses of the Third Study of Drug Use in Higher Education (EDES) included cannabis users (N = 6,521; 54.9% female). Confirmatory factor analysis (WLSMV) and McDonald's ω assessed structure and reliability; multigroup CFA tested invariance; structural equation modeling regressed the latent CAST factor on age of onset and past-year use frequency.Results: A revised one-factor model with a correlated residual between Items 1 and 2 fit well (CFI = .996, RMSEA = .033) and was reliable (ω = .843). Scalar invariance held across gender (partial residual invariance after freeing Item 6 residual variance), full residual invariance held across macrozones, and metric plus partial scalar invariance held across institution types after freeing the Item 6 intercept. Earlier onset (β = -.229, p < .001) and higher past-year use frequency (β = .221, p < .001) were associated with higher latent problematic use.Conclusion: The CAST showed a coherent unidimensional structure, adequate score reliability, and generally strong, although not uniform, measurement equivalence across relevant groups. These findings support its use for research and surveillance within Chilean higher education students.