Does WISC-IV Underestimate the Intelligence of Autistic Children?

Background:  Wechsler Intelligence Scale for Children (WISC) remains a widely used measure of intellectual functioning in autism. However, previous findings comparing autistic performance on WISC-III versus Raven’s Progressive Matrices (RPM) suggest that while both tests provide similar estimates of nonautistic intelligence, autistics perform significantly, and sometimes dramatically, better on RPM. Furthermore, in contrast to autistic children’s uneven WISC-III subtest profile, item by item RPM performance in autistics and nonautistics is highly correlated. These results suggest that RPM, a durable and important marker of fluid intelligence, better represents autistic intelligence than does Wechsler FSIQ. WISC-IV, released in 2004, introduced significant changes in Wechsler subtests and in the structure of different index scores. Indeed, the new perceptual reasoning index (PRI) has only one timed visuo-motor subtest and includes the new Matrix Reasoning subtest, which is similar to some aspects of RPM.

Objectives:  We aimed to determine whether the latest WISC version continues to underestimate autistic intelligence.

Methods:  26 autistic and 22 typically developing children (age 6-16 years) completed WISC-IV and RPM at two different times. Levels of performance were compared through inspecting percentiles derived from mean standard scores (WISC-IV) and from mean raw scores and ages (RPM) for each group. Parametric and nonparametric statistical comparisons were then conducted with individual percentiles. Similar results were obtained with all procedures.

Results:  Typical children achieved similar percentile values on RPM (73^rd percentile) and WISC-IV FSIQ (75^th percentile). This was not the case for autistic children, who scored at the 61^st percentile on RPM but only at the 21^st percentile on WISC-IV. A significant difference between RPM and WISC-IV scores was found in the autistic group (p<.0005 in both parametric and nonparametric tests) but not in the typical group (p=.57 in parametric and p=.67 in nonparametric tests). In line with these results, the advantage of RPM over WISC-IV FSIQ was significantly greater for autistic than for typical children. Five autistic children, but no typical children, obtained RPM scores over 50 percentile points higher than their WISC-IV scores. While three autistic children had standard scores below 70 on WISC-IV, the lowest autistic RPM score was at the 10^th percentile (IQ-equivalent estimate 81). The rest of the autistic children achieved RPM percentile scores of 18 or higher, with estimated RPM IQ-equivalents over 85. With respect to WISC-IV index scores, autistic children attained a mean Verbal Comprehension Index standard score of 84.6, compared to a mean PRI score of 104, which in turn was similar to their estimated RPM IQ-equivalent score.

Conclusions:  Our results are consistent with and add to existing findings that Wechsler FSIQ significantly underestimates autistic intelligence. Given what is known about RPM as a complex test of fluid and general intelligence, our results also further challenge the notion that autistic strengths are at best a collection of simple, isolated, and low-level perceptual abilities. Finally, our results provide preliminary evidence that the WISC-IV PRI index score may better estimate autistic intelligence than WISC-IV FSIQ. These findings merit attention in both research and clinical practice.