Psychometric properties and invariance testing of the Cannabis Refusal Self-Efficacy Scale in a Chilean sample

Background: Cannabis refusal self-efficacy (CRSE), as the people´s belief about their capacity to resist cannabis, has been probed to predict cannabis use and related behaviors. The CRSE Questionnaire (CRSEQ) has 14 items grouped into Emotional Relief Self-Efficacy (6 items), Opportunistic Self-Efficacy (5 items), and Social Facilitation (3 items), forming a CRSE higher-order factor.Objective: To validate the CRSEQ for its use in the Chilean population.Method: The CRSEQ was administered to 1,275 individuals aged 12 to 77 (835 females). Confirmatory Factor Analysis was used to evaluate 14-item and 19-item models. Estimates of internal consistency, temporal stability, and convergent validity with cannabis use behaviors and reward sensitivity were obtained. Additionally, three multigroup invariance tests were conducted.Results: Both models exhibited good fit indices. The 14-item model showed χ² (74) = 245.662, p < .001, CFI = .974, RMSEA = .047, indicating slightly better fit compared to the 19-item model, which showed χ² (149) = 559.596, p < .001, CFI = .958, RMSEA = .051. Both models showed strong internal consistency (α = .80 to .96 for the 14-item model, α = .80 to .96 for the 19-item model), moderate to high temporal stability (ICCs 14-item model/ 19-item model: ERSE = .77/.78, OSE = .88/.89, SF = .82/.82), and significant convergent validity (correlations with cannabis use behaviors: .22 to .58).Conclusion: The better fit of the 14-item model makes it suitable for most applications. Convergent validity and multigroup invariance analyses confirmed the questionnaire's equivalence across sex, age groups, and cannabis use problem status. This allows for meaningful comparisons of cannabis refusal self-efficacy between different demographic groups, facilitating its applicability in diverse settings such as educational institutions and substance use treatment centers.