Concurrent and Predictive Validity of the MacArthur-Bates Communicative Development Inventory for Children with Autism

Background: The MacArthur-Bates Communicative Development Inventory (CDI) is often used as a measure of early language development for children with autism spectrum disorders (ASD). However, there is a need for more information on both the concurrent and the predictive validity of the CDI, especially for children who exceed the upper CA limits of the test norms. Objectives: To investigate the relationship between scores on two versions of the CDI (Words & Gestures, WG; and Words & Sentences, WS) and standardized measures of language development in children with ASD. Methods: Participants in Study 1 were 34 children with autism (79% males, mean CA 45 months), whose mean MA on the Mullen Scales of Early Learning was 16.25 months at the first assessment. They were assessed shortly before the onset of early intervention (T1) and an average of 12 and 24 months thereafter (T2-T3), with the CDI:WG, Preschool Language Scale-3 (PLS), Expressive One Word Picture Vocabulary Test, and the Peabody Picture Vocabulary Test-IIIA (PPVT). Spearman rank order correlations were calculated to examine associations between the total number of words understood, words produced, early gestures, and late gestures on the CDI:WG and raw scores on the language measures or subscales thereof, for T1:T1 (concurrent validity) and T1:T2-T3 (predictive validity). Participants in Study 2 were 29 children with autism (86% males, mean CA 61 months), whose mean composite AE on the Vineland was 30.95 months at first assessment. They were assessed 6 months after the initiation of early intervention (T1), and an average of 6 months and 18 months thereafter (T2-T3), using the CDI:WS and the same language measures used in Study 1. Data analysis paralleled that of Study 1, except that CDI:WS subscale scores were used for the total number of words produced, mean words in the child's three longest sentences, and sentence complexity. Results: In Study 1, significant correlations were found between T1 CDI:WG scores for words produced, early gestures, and late gestures and all T1 language measures (r=.34-.60); and between T1 words understood and both PLS receptive language subscale scores and PPVT scores (r=.35-.42). Significant correlations were also found between T1-T2 scores on all CDI:WG subscales except for late gestures and all language measures (r=.38-.69). However, from T1-T3, only words produced and early gestures on the CDI:WG were correlated with all language measures (r=.50-.69). In Study 2, correlations between all T1 CDI:WS scores and all T1 language measures were significant (r=.54-.88). Similarly, there were significant correlations between all CDI:WS scores and all language measures from T1-T2 (r=.51-.73) and from T1-T3 (r=.48-.73). Conclusions: The results provide support for both the concurrent and predictive validity of both the CDI:WG and the CDI:WS for young children with ASD whose chronological ages exceed the test norms. The validity of the CDI for children who are older than the normative group is important information for clinicians in particular, who often experience the limitations of formal tests of language for children with ASD and other significant language impairments.