Factor analysis of the Woodcock-Johnson with conduct disordered adolescents

Abstract

A factor analysis of the 12 subtests of the Woodcock-Johnson Tests of Cognitive Ability was performed, using 286 conduct disordered male adolescents as subjects. A principal factor analysis procedure was utilized. Factors to be retained were confirmed using the scree test. The factor analysis procedure yielded three factors. These were interpreted as being a primary verbal ability factor, a secondary reasoning factor, and a secondary numerical processing factor. The perceptual speed and memory groupings identified by Woodcock were not duplicated. Implications of these results suggest the cautious use of the Cognitive Factor cluster scores for diagnosis of strengths and weaknesses for the behavior disordered population.Questions have been raised regarding the usefulness and validity of the various clusters of the Woodcock-Johnson Tests of Cognitive Ability (W JTCA) (1977). Woodcock (1984) discussed at length the rationale behind both the development and use of the Scholastic Aptitude clusters (Reading Aptitude, Math Aptitude, Written Language Aptitude, and Knowledge Aptitude). While these clusters were designed to provide differential expectancy information, their applicability is directly related to their degree of discriminant validity. Cummings and Moscato (1 984) demonstrated equivalent correlations between the Broad Cognitive Ability score to each of the four areas of achievement and between the four Scholastic Aptitude clusters to their respective areas of achievement on the W-J Achievement clusters. Finding results comparable to those of Cummings and Moscato, Phelps and Rosso (1985) concluded that the Scholastic Aptitude clusters lacked sufficient differentiation to justify computation when used with behavior disordered (ED) youth. Woodcock (1984) justified the continued use of these clusters by stating that they were not confounded with achievement, that they reduced testing time, that they provided discrepancy norms for each area of achievement, and that they provided higher clinical validity. Such reasons are quite warranted if the clusters are distinct and separate functions. However, for the BD population, the lack of discriminant validity between the various Aptitude-Achievement pairings would lend doubt to that premise (Phelps & Rosso, 1985).It would appear that the four Cognitive Ability clusters of the W-J might also lack specificity. These clusters (Verbal Ability, Reasoning, Perceptual Speed, and Memory) were designed to "provide information regarding special intellectual abilities that may influence school learning and certain other kinds of performance" (Woodcock, 1984, p.357). The clusters were developed using both multiple regression and factor analyses. Results from these statistical procedures were then used to determine the differential weightings of the various subtests within each cluster. However, Woodcock has clearly stated that the development of the clusters was "aided by, but not limited to, the application of cluster and factor analysis procedures to the final set of 12 cognitive ...