Confirmatory Factor Analysis of the WPPSI, WPPSI-R, and WISC-R: Evaluation of a Model Based on Knowledge-Dependent and Processing-Dependent Subtests

Abstract

Previous factor analyses of the Wechsler scales have supported a two-factor (WPPSI and WPPSI-R) and three-factor solution (WISC-R). The present paper explores the validity of a four-factor model across these three instruments. The four-factor model maintains Wechsler's original distinction between verbal and nonverbal domains but distinguishes between “knowledge dependent” and “processing dependent” factors within each domain. Four separate LISREL maximum likelihood confirmatory factor analyses were performed on the WPPSI, WPPSI-R, and WISC-R standardization samples. A best-fitting model was determined by comparing the present four-factor model to the traditional two-factor model that distinguished between Verbal and Performance and a three-factor model that parallels the WISCR factor structure. The new four-factor model resulted in a significant improvement of fit compared to both the two-factor and three-factor models across the three samples. The clinical validity of the model was examined by analyzing the profile patterns of language-impaired children (N = 198) tested with the WPPSI and reading-impaired children (N= 230) tested with the WISC-R. Because processing-dependent tests require more mapping than knowledge-dependent tests, the pattern “knowledge dependent” > “processing dependent” is predicted. This pattern was confirmed in both samples.