Convergent and discriminant validity for assessment of skill in resisting a role play alcohol offer

Drug abuse prevention researchers are increasingly concerned with validating assessments of adolescent behavioral resistance skills. This paper describes a procedure in which five people representing three distinct perspectives observed a student's behavior in a role play alcohol offer situation. Each rater then evaluated the student's skill at refusing the offer and intentions to drink alcohol in the future. The three distinct types of raters were "methods," while skill and intentions were "traits", producing the familiar multitrait-multimethod matrix (MTMM; Campbell & Fiske, 1959). We used an analytic procedure (following Widaman, 1985) that compares hierarchically nested LISREL models to assess convergent and discriminant validity. The results showed good validity for the behavioral skills and intentions constructs. A final LISREL model showed a significant, positive relationship between resistance training and the MTMM behavioral skills factor, providing evidence that resistance training programs have the desired short-term effect on adolescents' behavioral skills.