Measurement of Nonverbal and Verbal Abilities in Minimally Verbal Children with Autism Spectrum Disorders

Background: Cognitive abilities in children with autism spectrum disorders (ASD) vary widely (Fombonne, 2005). More advanced nonverbal and verbal cognitive abilities have been shown to predict positive response to intervention (Harris & Handleman, 2000) and overall outcomes (Howlin, Goode, Hutton & Rutter, 2004). However, many assessments of cognitive and language abilities are not standardized using children with developmental delays. These tests may not capture abilities in the same way for children with developmental disabilities as they do with typically developing children. In particular, minimally verbal children with ASD represent a group for which the validity of testing may be compromised (i.e. difficulty with language demands, atypical skill development, attention difficulties). 

Objectives: The aim of this study is to test whether different assessments of nonverbal and verbal abilities, specifically the Mullen Scales of Early Learning (MSEL), Leiter-R, Preschool Language Scales (PLS-5), and Vineland Adaptive Behavior Scales (VABS-II), demonstrate convergent validity in minimally verbal children with ASD.

Methods: Participants currently include 20 children (4-8 years of age), with 26 children expected by May 2015, who were assessed prior to the start of a randomized intervention trial. All children were classified as minimally verbal through parent report and a language sample (<20 spontaneous words in 20 minutes). ASD diagnoses were confirmed with the ADI-R and ADOS-2. Nonverbal abilities were measured with the MSEL and Leiter-R. Six children were excluded from the nonverbal analyses because they did not achieve a ceiling on a subscale of the MSEL. Verbal abilities were measured with the MSEL, PLS-5, and VABS-II. Ratio IQs were calculated for the MSEL (nonverbal estimate comprised of the Fine Motor and Visual Reception domains, verbal estimate comprised of the Receptive and Expressive Language domains), PLS-5, and VABS-II (verbal estimate comprised of the Receptive and Expressive Language subdomains). Standardized scores from the Leiter-R Brief IQ screener were used. Paired samples t-tests were used to compare means and examine relationships between measures.  

Results: In comparing NVIQ estimates from the MSEL and the Leiter-R, results indicate that scores from the two tests were correlated at r=.79, p=.001. Notably, scores on the Leiter-R (M=54.79, SD=17.91) were significantly higher than scores on the MSEL (M=38.29, SD=10.06), p<.001. With respect to language abilities, scores from the MSEL and the PLS-5 were correlated at r=.65, p=.002. VABS-II language scores and the MSEL language scores were correlated at r=.41, showing a trend towards significance, p=.088. Scores on the VABS-II were significantly correlated with scores on the PLS-5 (r=.60, p=.009). MSEL language scores (M=25.75, SD=7.73) were significantly lower than scores on both the PLS-5 (M=29.85, SD=8.73), p=.016, and the VABS-II (M=31.5, SD=11.79), p=.036. 

Conclusions: These findings demonstrate moderate correlations across multiple measures of nonverbal and verbal ability. Despite these correlations, significant mean differences in scores, particularly the large difference between estimates of NVIQ from the MSEL and Leiter-R, have implications for researchers and clinicians using these assessments. More research is needed to understand these differences and to examine the validity of individual measures for use with specific subgroups of children with ASD.