Multivariate analyses of the profile stability of intelligence tests: high for IQs, low to very low for subtest analyses.

Profile stability involves the consistency of a set of scores over time. That is, does a profile of scores change on retesting and does this change affect clinical decisions? While psychologists routinely examine the reliability of individual scores, little research has examined the stability of a profile or set of scores. The first study described in this paper examined potential measures of profile stability using a simulation computer program. The results suggest that several measures show promise in this context, particularly Cattell's coefficient of pattern similarity (r(p)), salient variable similarity index (S), and the D(2) coefficient. In the second study, selected measures of profile stability were applied to Wechsler test-retest data. The results suggest that profiles composed of IQ and index scores demonstrate acceptable stability and can be usefully interpreted in clinical and research situations. However, subtest score profiles are inherently less stable and provide little useful clinical information.