Simple Summation vs Differential Weighting in the Construction of WISC Short Forms

Abstract

WISC short forms that entail the differential weighting of subtest scores may be expected to afford something of an advantage over those that require only simple summation. Finley and Thompson (195s) have constructed such a short form especially for use with the mentally retarded, but in another context, these same investigators observed that if the correlations with the Full Scale are comparable and the standard errors of estimate not greatly different, it is easier to calculate one weighted score rather than five individual weighted scores (Thompson & Finley, 1963). The purpose of the analysis reported here was to assess the advantage of differential weighcing. Finley and Thompson used the Wherry-Doolittle method to select the best short form and found that a combination of Information, Picture Arrangement, Picnire Completion, Coding, and Block Design had a m~iltiple correlation of .896 with the Full Scale, with a standard error of 3.12 IQ points. The corresponding simple correlation, calculated by the present writer from their data, is .883, with a standard error of 3.31 IQ points. Moreover, the correlation between the short form entailing differential weighting and the one requiring only simple summation is ,983. The weighting problem is actually quite complex, and the potential advantage of differential weighting depends on a number of factors, e.g., the number of subtescs that comprise the short form, their average intercorrelation, and the range of their standard deviations, which serve as weights in simple summation. In a previous study, based on portions of the WISC standardization data, Silverstein (1967) concl~~ded that the advantage of differential weighting was minimal. This conclusion appears to hold for the Finley-Thompson short form as well.