| Summary: | This thesis examined the predictive validity of different strategies for constructing personality inventory scales. The ACL was used as a common item pool to construct sets of 8 scales by each of 6 strategies. A "prototype strategy" was introduced in this study and its validity was compared with 4 traditional strategies and 1 random strategy in predicting 21 criterion measures. The 4 traditional strategies compared were: empirical, factor analytic, rational, and internal consistency. The subjects were 234 paid University of British Columbia fraternity males ranging in age from 19-28 who supplied ACL responses and the 21 criterion measures. Included in the 21 criteria were peer ratings on the traits of achievement, dominance, nurturance, affiliation, exhibition, autonomy, aggression, and deference; university GPA; high and low peer nominations on the dimensions of leadership, likeability, academic potential, athletic potential, heterosexuality, and success. Multiple regression procedures were used in a double cross-validation design. Results indicated that the prototype strategy did better than all of the other strategies and significantly out-performed the rational strategy which it was designed to improve upon. === Arts, Faculty of === Psychology, Department of === Graduate
|
|---|
