โ Scribed by Alan Feingold
No coin nor oath required. For personal study only.
Recent research (Feingold, 1982) on the incremental validity of the WAIS has found that the Vocabulary and Information subtests are each as valid as the Full Scale IQ. The present WAIS review demonstrated that weighting subtests that differ in validity equally in the computation of the IQ is responsible for reducing potential validity. Two main conclusions emerged from these data. One, Vocabulary and Information in concert yield a measure that has greater validity than either subtest by itself. Two, the addition of a third test in a stepwise multiple regression in which Vocabulary and Information already have been added will not result in a consistent increase in validity. These findings must be considered tentative, but can serve as a guide for devising research designs to validate the recent revision of the WAIS (WAIS-R).
Recent research (Feingold, 1982) on the incremental validity of the Wechsler Adult Intelligence Scale (WAIS) (Wechsler, 1955) has found that the Vocabulary (Voc) and Information (Info) subtests are each as valid as the Full Scale IQ (FIQ). The only conceivable explanation for this disconcerting conclusion is that WAIS subtests have differential validities. Weighting subscores that differ in validity equally in the computation of the FIQ as Wechsler (1955) specifies is the most plausible interpretation of Feingold's (1982) results.
A survey of the studies reviewed by Feingold (1982) supports this interpretation. The verbal subtests, for example, are better predictors of academic achievement than the performance measures (Conry & Plant, 1965). There is also evidence that some verbal tests (e.g., Voc) are more valid than others (e.g., Digit Span) (Conry & Plant, 1965).
It appears, then, that to obtain maximum validity from the WAIS, subtests must be differentially weighted. The most sensible technique would be the use of a stepwise multiple regression in which the subtests are weighted according to their validity coefficients and their correlations with each other. Such a procedure would result in a streamlined WAIS that is psychometrically superior (i.e., less redundant), more valid, and requires less time for administration. This paper will review the sparse literature on WAIS subtest validity with the intention of selecting a battery of subtests that yield maximal validity. Although it will be assumed that these results are generalizable to the newly revised WAIS (WAIS-R), it is expected that validity studies on the new scale will be conducted over the next few years. Any conclusions drawn in this paper should be considered tentative until they are replicated, although findings reported here may serve as a guide for developing designs to validate the WAIS-R. It is important that future investigators abandon the prevalent practice of reporting validity coefficients for the IQ scores alone. Subtest correlations with a validating criterion always should be reported (Feingold, 1982).
METHOD
Ss consisted of several groups who had participated in published WAIS research, including three age categories of the WAIS standardization sample. As Feingold (1982) reviewed WAIS research to validate Voc and Info, the same groups were used for the
๐ SIMILAR VOLUMES