A Multivariate Analysis of the Wechsler/Woodcock-Johnson Discrepancy Controversy

Abstract

The Woodcock-Johnson Tests of Cognitive Ability (WJ-COG) is a relatively new measure of intellec tual ability that is being used by psychoeducational practitioners. Research comparisons suggest strong concurrent and superior predictive validity when compared with the WISC-R. However, early research reports suggested that for some children the WJ-COG produced "lower" scores than the WISC-R, a finding that produced much controver sy. The current study examined the validity of three content difference hypotheses (viz., differences in g, verbal, or achievement content) that have been advanced to explain WISC-R/WJ- COG discrepancies. A series of canonical correla tional analyses were conducted with an elemen tary school sample of 167 children to explore the redundancy between the WISC-R, WJ-COG, and achievement as measured by the Woodcock- Johnson Tests of Achievement. Inspection of the canonical variates, redundancy coefficients, and canonical loadings on the two significant variates extracted in each analysis did not support any of the content difference hypotheses. The findings suggest that the WISC-R/WJ-COG mean score discrepancy may be due to different factor struc tures that have not been appreciated fully in prior research.