Much of the research specific to the ASD population with regard to use of cognitive assessment instruments focuses on comparability of scores across tests. The DAS-II has been found to yield significantly higher scores than WISC-IV scores, which were attributed to relative weaknesses in processing speed among the ASD group (Kuriakose, 2014); processing speed is measured on Wechsler tests but not on the DAS-II. Similarly, Bardikoff and McGonigle-Chalmers (2014) underscored the need to isolate timing/processing speed criteria when using the WISC-IV in ASD. These researchers compared the KABC-II NVI to the nonverbal components of the WISC-IV’s PRI and PSI, and though they found no significant group differences for the PRI subscale of the WISC-IV nor for the NVI subscale of the KABC-II, significantly lower scores for the PRI vs. NVI for the ASD group were found. Moreover, these researchers found that the ASD group scored significantly lower on the PSI of the WISC-IV. Overall, processing speed is likely to be impacted in persons with ASD and examiners should consider this when selecting and interpreting assessment results.
In addition, DAS-II and Mullen Scales of Early Learning (see Developmental Assessment section for summary on this test) are highly correlated, suggesting these measure a similar construct (Farmer, Golden, & Thurm, 2015). Notably, however, even when cognitive tests scores are highly correlated between groups of individuals with ASD, variability among individual children must be considered (i.e., comparability of test scores still cannot be assumed) (Farmer et al., 2015; Kuriakose, 2014). In other words, correlation of test scores does not mean an absence of significant differences on obtained scores. For example, scores of students on the SB-5 and WISC-IV were found to be correlated, yet the obtained FSIQ and VIQ scores differed significantly between the two tests (Baum, Shear, Howe, & Bishop, 2015).
Moreover, clinicians should not assume complete comparability of scores obtained on tests from revision to revision. Specifically, Tsatsanis et al.’s work (2003) resulted in their conclusion that the Leiter-R was useful for children with autism but noted that greater clinical success may be achieved using the original Leiter for very low functioning and severely affected children, particularly young children. Also with regard to assessment of assessing cognitive ability among lower functioning individuals (i.e., those with Intellectual Disability [ID]) examiners should be aware that scores on the SB-5, specifically, may be affected by floor effects, leading to erroneously flat profiles (Sansone, Schneider, Bickel, Berry-Kravis, Prescott, & Hessl, 2014). Even in a comparison of two different nonverbal intelligence tests, strong and positive correlations were found scores on the UNIT and the Test of Nonverbal Intelligence-Third Edition (TONI-3; Brown, Sherbenou, & Johnsen, 1997), but performance of children with ASD was nuanced in that they scored better on the TONI-3 (an abstract measure of intelligence) compared with the Abstract Reasoning subscale of the UNIT (a measure of both abstract reasoning and real-world knowledge).
Relationships between nonverbal measures and traditional cognitive assessment instruments have also been investigated in the ASD population. Scores from nonverbal measures generally tend to be higher than from traditional measures. More specifically, scores from the Leiter-R tend to be significantly higher than SB-5 scores (Grondhuis et al., 2018), particularly among young children (Grondhuis & Mulick, 2013). These authors concluded that the Leiter-R and SB-5 are not equivalent measures of intellectual functioning in children with ASD and using only one could result in misclassification of intellectual ability. Similarly, Aiello (2013) found UNIT scores to be significantly higher (i.e., on average, more than 10 points) than WISC-IV scores. Using both traditional and nonverbal IQ tests might better capture the level of functioning and needed educational supports among children with ASD.
Some researchers have investigated use of neuropsychological measures within the ASD population. With regard to the NEPSY-II, Barron-Linnankoski et al. (2015) found that compared with typically developing peers, elementary-age children with high-functioning ASD had higher verbal reasoning skills but performed significantly worse than TD children in set-shifting, verbal fluency, and narrative memory. Interestingly, no differences were found in terms of social perception, including both the Theory of Mind and Affect Recognition subtests. Moreover, no correlation was found between FSIQ and impaired neurocognitive functioning on the NEPSY-II in the HFASD group, which “supports the usefulness of the NEPSY-II for different clinical groups” (p. 68).
The NEPSY-II measures multiple cognitive processes, including executive function, in a direct (i.e., clinician-administered format). However, informant-based measures of executive function (e.g., BRIEF) have been found to be sensitive for detecting executive control impairments in real-world settings for preschoolers with ASD (Smithson et al., 2013) and also to differentiate between children with ASD and typically developing controls (Demetriou et al., 2018). Such ecologically valid measures may be most appropriate for practice, including diagnostic and intervention frameworks.
Regarding abbreviated battery cognitive scores, Twomey and colleagues (2018) found that the SB-5’s ABIQ correctly identified overall level of cognitive functioning for approximately 80% of preschoolers with ASD. However, clinicians should use cognitive scores generated on the basis of abbreviated batteries as screeners and not as a substitute for a comprehensive score, as these scores may misrepresent the cognitive ability of some persons with ASD (Coolican et al., 2008), particularly those with lower cognitive functioning (Twomey et al., 2018).