Doctor of Philosophy

dissertationEarly identification and intervention of an Autism Spectrum Disorder (ASD) can have beneficial effects that extend into later life. However, currently used instruments have difficulties detecting children who may have an ASD. The current study investigated the utility of a newly published measure, Autism Spectrum Rating Scales (ASRS). Participants included 67 children ages 2 to 5 years old, referred for possible special education services. Participants were divided into two groups: those with an ASD (n = 37) and others suspected of having a general developmental disability (DD) (n = 30). Participants were assessed using the ASRS to examine the instrument's ability to classify them as having an ASD or a general DD. Additional testing examined the effects various levels of intellectual, adaptive, and language skills have on the ability of the ASRS to classify children. Classification ability and error rates of the ASRS were also examined with regard to base rates and error acceptability by context. Results indicate that with a recommended cut score of 70, the Parent ASRS had an overall hit rate of 64%. The Parent ASRS had a Type I error rate (i.e., false positive) of 16% and a Type II error rate (i.e., false negative) of 19%. For the Teacher ASRS, the hit rate was 62%. The Teacher ASRS had a Type I error rate of 15% and a Type II error rate of 23%. Sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio and negative likelihood ratio of the ASRS were also examined to gain insight into the measure's utility. ROC Curve analysis determined the area under the curve (AUC) for the ASRS, the most optimal point for sensitivity and specificity. It was concluded that across all ASRS forms (e.g., Parent, Teacher), the general ability of the ASRS to classify and discriminate between children with potential ASDs or general DDs referred for possible special education services were similar