The Development of A New Brief Measure of Social and Nonsocial Autistic-Like Traits In Young Children

Background: The prodrome of autism spectrum disorders (ASDs) has received considerable attention over the last twenty years, in terms of both theoretical interest but also because of the potential for early intervention (Yirmiya & Charman, 2010) A brief measure of social and nonsocial autistic-like traits for early childhood would be valuable in epidemiological research, for detecting subtle manifestations of the broader autism phenotype in at risk sibling studies, and in molecular genetic research of population-based samples. 

Objectives:  The aim of this study was to develop a new brief parent report measure of social and nonsocial autistic-like traits (also known as restricted repetitive behaviours and interests) and to investigate the factor structure, validity, reliability and specificity of the measure.

Methods:  A sample of eighty-six 24-month-old children, half of whom were young siblings of children with autism, was employed from the British Autism Study of Infant Siblings (BASIS).  Items were designed to capture the range of autistic-like social and nonsocial traits that have been reported in this age group. The variance shown by individual items was assessed and principal components analysis was conducted.  The validity, reliability and specificity of the scales were studied.

Results: The principal components analysis suggested a two-factor solution, which formed the basis of the social and nonsocial autistic-like trait scales.  Social and nonsocial autistic-like trait scales correlated weakly in 24-month-olds (r = 0.11, p = 0.35, n = 83).  Children with high scores on the Autism Diagnostic Observational Scale--Generic (ADOS-G; Lord et al., 2000) had higher mean scores on social and nonsocial autistic-like trait scales compared to children with low ADOS-G scores. The social and nonsocial autistic-like trait scales showed significant positive correlations with the Quantitative-Checklist for Autism in Toddlers (Q-CHAT; Allison et al., 2008) which supported their validity (r = .37, p = .001, and r = .43, p < 0.001, respectively). Low-to-modest, mostly nonsignificant, correlations were found with general temperament scales (the Early Childhood Behavior Questionnaire; Putnam et al., 2006), supporting the specificity of these social and non-social scales to autistic-like traits rather than temperament or behaviour problems more generally.

Conclusions: A new, brief parent report measure of social autistic-like traits (12 items) and nonsocial autistic-like traits (9 items) was developed for research on young children.  In part this was developed because the majority of existing dimensional assessments of autistic-like traits are for children over the age of 36 months. The principal components analysis and correlations supported the division of social and non-social autistic-like traits into separate scales.  The practical uses of these scales are as brief measures of social and non-social autistic-like traits in samples of typically developing young children, in epidemiological research, and in at risk infant sibling studies. The scales may also be useful in molecular genetic research of ASDs, which is moving towards incorporating dimensional assessments of degree of autistic-like traits in addition to the conventional case-control design (e.g., Ronald et al., 2010).