External Validity of the Profile Variability Index for the K-ABC, Stanford-Binet, and WISC-R: Another Cul-de-Sac Rex B. Kline, Joseph Snyder, Sylvie Guilmette, and Maria Castellanos Clinicians often attach interpretive significance to high variability among subtest scores in IQ profiles of adults and children. In this study we evaluated the extemal validity of a relatively new measure of IQ subtest variability, the profile variability index (PVl). Within a sample of referred children, we administered the Kaufman Assessment Battery for Children, the Fourth Edition Stanford-Binet, and the Wechsler Intelligence Scale for Children-Revised. We calcu- lated PVIs for each child from each test and correlated these values with achievement scores and indexes of discrepancy between actual and predicted scholastic achievement. Against these extemal criteria, PVl information from all three cognitive-ability batteries had essentially nil validity. We discuss implications of these findings for future research in this area. C linicians have long been inter- ested in the interpretive signifi- cance of score variability in cognitive-ability profiles of children and adults. The large body of literatxire in this area can be divided into two broad areas of study: (a) discrepancies between test summary scores, such as Verbal-Performance IQ differences on Wechsler scales (e.g., Kaufman, 1976b; Matarazzo & Herman, 1985; Reynolds & Gutkin, 1981); and (b) variability of subtest scores, either within specific scales—for example, Wechsler Adult Intelligence Scale-Revised (WAIS-R) (Wechsler, 1981) Verbal scale—or across a whole test. Results of many studies from the first area suggest that large summary-score discrepancies often re- late to performance on nontest criteria. For example, large Wechsler Verbal- Performance IQ differences covary with school achievement for children (e.g., Longman, Inglis, & Lawson, 1991; Richman & Lindgren, 1980) and certain t3rpes of brain damage for adults (e.g., Kaufman, 1990; Lezak, 1983; Matarazzo, 1972). Results of studies from the second area of study—variability among sub- test scores—have been much more equivocal. At one time, clinical lore held that high intersubtest variability indicated possible cognitive dysfunc- tion (e.g., learning disabilities, neuro- logical damage) or psychopathology. Researchers who studied base rates of intersubtest variability found, how- ever, that high amounts of variability are present in IQ test profiles of non- disabled subjects. For example, aver- age subtest-score ranges (highest score minus lowest score) in the normative samples of the WAIS-R, the Wechsler Intelligence Scale for Children-Revised (WISC-R) (Wechsler, 1974), and the Kaufman Assessment Battery for Chil- dren (K-ABC) (Kaufman & Kaufman, 1983a, 1983b) are all about 7 scaled score points (Chatman, Reynolds, & Willson, 1984; Kaufman, 1976b; Matarazzo, Daniel, Prifitera, & Her- man, 1988). Although clinicians have had to ad- just their expectations upwards about how much intersubtest variability is "normal" since publication of the above types of studies, there is actually little empirical evidence about the diag- nostic significance of subtest variability. Among children, for example, the in- cremental validity of profile variability over simple average subtest scores (i.e., elevation) in predicting children's scholastic skills has often been found to be negligible (e.g.. Hale & Saxe, 1983; Kline, Snyder, Guilmette, & Castellanos, 1992). Results of other re- cent studies with adults have indicated that the WAIS-R profiles of brain- damaged patients, nondisabled con- trols, and psychiatric patients (e.g., with affective disorder, schizophre- nia, psychosis) (Piedmont, Sokolove, & Fleming, 1989; Ryan, Paolo, & Smith, 1992) have comparable amounts and types of intersubtest variability. In fact, the overall paucity of positive findings in this area has led many re- viewers to caution clinicians against JOURNAL OF LEARNING DISABILITIES VOLUME 26, NUMBER 8, OCTOBER 1993 PAGES 557-567