Factor Analysis of the Autism Spectrum Rating Scales Parent Report 6-18 in a Complex Community Sample.

Abstract

There is a need for research on autism questionnaire psychometrics outside of the standardization sample. This study investigated the factor structure of the Autism Spectrum Rating Scales-Parent report for ages 6-18 (ASRS_p6) in a well-characterized community sample of 696 children (autism [AUT] n = 231; non-autism [NOT] n = 465; X̄age = 10.03) prospectively evaluated with the Autism Diagnostic Observation Schedule-2, a gold-standard autism diagnostic measure. The original author-identified structure demonstrated poor model fit. Exploratory Factor Analysis with a randomly selected half of the sample (n = 346) suggested retaining 17-items on three factors (Rigid/Sensory, Social, and Executive Function [EF]), explaining 55% of the variance. Confirmatory Factor Analysis with the remaining participants (n = 350) indicated good model fit. Partial measurement invariance was observed for diagnostic (AUT/NOT) and gender (male/female) groups. Correlations with the DSM-5_p6 (an ASRS_p6 scale) were high. Mean differences were observed between AUT/NOT groups for Social; when controlling for age, marginal differences (p = 0.02 - 0.03) were observed for all factors but EF. Social also demonstrated significant AUC regardless of control variables. However, AUC values for Social and other factors, while significant, were in the poor range; correlations with ADOS-2 scores were weak or non-significant. Despite improving the factor structure, the Total-17 does not appear to measure ASD-specific traits. However, these findings provide a basis for further research on ASD questionnaires.