THE FORM COMPARABILITY OF THE WAIS AND THE WAIS-R: PRACTICE EFFECTS AND THE EQUALITY OF MEANS
Abstract
Ninety-six (48 males; 48 females) undergraduate, graduate, and professional student volunteers were randomly assigned to six groups and tested on the WAIS or WAIS-R under unpracticed, practiced (same test), and practiced (other test) conditions. Testing was carried out by trained student examiners at the doctoral level. Data were analyzed using a 2 x 3 x 2 complete factorial analysis of variance with gender as the third independent variable. No significant differences were found between WAIS and WAIS-R mean Full Scale IQs (121.60 and 118.69, respectively) although this is thought to be due in large part to the 2:1 ratio of practiced to unpracticed subjects in the presence of a strong test by practice interaction. Evidence is presented to suggest that the WAIS-R may, in fact, yield significantly lower Full Scale IQ scores. Estimates of the magnitude of that difference are also provided. Significant practice effects were found with the largest gains made by subjects practiced on the other test. The test by practice interaction, thought to confound Latin Square designs in many previous studies, was also found statistically significant for the Full Scale IQ scores. The WAIS Full Scale IQ appeared to be practice-resistant, while that of the WAIS-R was not. At the scaled score level of analysis, the test by practice interaction decremented considerably, and several scaled scores demonstrated significant test effects. Significant gender differences were found, but due to the voluntary nature of the sample, were not believed to be interpretable as effects of interest. According to several other criteria, the two tests appeared to be surprisingly comparable. They demonstrated parallel profiles, particularly among verbal subscales. The mean raw content mastery scores were virtually identical. All IQ and scaled scores demonstrated homogeneity of variances, and a multivariate test demonstrated homogeneous variance-covariance matrices across the two tests as well.
This paper has been withdrawn.