An application of a multitrait-multimethod test to validity data of a social learning treatment for aggressive children

Campbell and Fiske's (1959) multitmit-multimethod matrix was applied to data from two independent social learning measures designed to evaluate procedures used to aid aggressive children. Six parallel or like-named variables — noncomplying, crying, whining, negativism, destructiveness, and yelling — from both a home observation assessment system and a parent telephone data system were used. The data from 47 cases indicated that the categories of whining, crying, and destructiveness demonstrated significant convergent validity, but only whining demonstrated discriminant validity. The situation-specific versus cross-situational consistency aspects of the behaviors studied, and the implications for future research, were discussed.This article is based on a doctoral dissertation submitted to the Department of Educational Psychology of the University of Oregon. The author wishes to express his appreciation to his dissertation advisor, Lloyd Lovell, and to Gerald Patterson for his generous offer to use data that he had spent years collecting.