Assessment Across Wide Age and Ability Ranges: Use of the Peabody Picture Vocabulary Test As a Proxy for Verbal IQ

Background:  

Children with Autism Spectrum Disorder (ASD) span the full range of verbal and nonverbal cognitive ability. A single study may implement multiple different cognitive instruments, including use of tests outside the standardized age range to derive estimates for children with Intellectual Disability (ID). Completing full cognitive batteries requires significant time and expense that can be prohibitive to studies ascertaining large samples (e.g., genetic consortiums). 

The Peabody Picture Vocabulary Test – 4^thEdition (PPVT4) is a measure of receptive vocabulary designed for use with children and adults (Dunn & Dunn, 2007). It takes 10-15 minutes to administer and can be used as a proxy for verbal IQ in individuals who cannot complete full IQ assessments (Karns et al., 2011). Recent studies have demonstrated strong associations between PPVT4 and tests of nonverbal cognitive abilities in youth with Down Syndrome, ID or typical development (Phillips et al., 2014) and minimally verbal children with ASD (Plesa-Skwerer, 2015). No studies have yet investigated the convergent validity of the PPVT4 and other cognitive measures in children with ASD across the full range of abilities.

Objectives:  

To compare the PPVT4 to measures of verbal (VIQ) and nonverbal IQ (NVIQ) in children and adolescents with ASD.

Methods:  

Participants were 4-17 year-olds with ASD from the Simons Simplex Collection who completed the PPVT4 and a full cognitive battery (N=2420; 86% male). Verbal and nonverbal estimates were derived from a hierarchy of tests, most often the Differential Abilities Scale or Mullen Scales of Early Learning. Pearson correlations were calculated to examine associations between PPVT4 and VIQ and NVIQ. Children were divided by age (4-5, 6-12, 13-17), language (ADOS Module), cognitive level (VIQ<70 vs. >70; NVIQ<70 vs. >70) and test type (e.g., DAS, Mullen). Fisher r-to-z transformation was used to compare correlations across groups. PPVT4-VIQ differences were computed; differences greater than 15 points were considered discrepancies.

Results:  

PPVT4:VIQ correlations were very strong (r=0.70-0.95, p<0.0001) across age, language, verbal and nonverbal abilities, and test type. The majority of children (77%) had PPVT4 scores within 15 points of their VIQ estimate. Only 4% had PPVT4<VIQ, whereas the remaining 20% had PPVT4 that exceeded VIQ by more than 15 points. A somewhat higher proportion of minimally verbal children (i.e., ADOS Module 1, 26%) had PPVT4>VIQ profiles compared to verbal children (19%, ADOS Module 2-4; X2=9.587, p=0.008). Notably, PPVT4:NVIQ correlations were also strong (r=0.80), though somewhat reduced in children with VIQ>=70 (r=0.54) compared to VIQ<70 (r=0.68, z=5.02, p<0.0001).

Conclusions:  

Across children of varying ages, language and ability levels, the PPVT4 showed good convergent validity with VIQ estimates derived from multiple instruments. Further analyses are needed to understand the 20% for which the PPVT4 may overestimate verbal abilities. Nonetheless, these results suggest that the PPVT4 provides a good proxy for VIQ. This is particularly relevant for genetic studies ascertaining large samples across the full range of abilities, as using the PPVT can significantly reduce costs and allow time spent with participants to be dedicated to collection of other phenotypic measures.