Discriminant analysis of the autism spectrum disorder observation for children

Daniene Neal, Louisiana State University, Baton Rouge, 70803, USA.
Johnny L. Matson
Brian C. Belva


PURPOSE: To run a discriminant analysis on the individual items and the total scale of the ASD-OC to determine if they significantly discriminated between ASD and atypical groups. METHOD: The measure was administered to 78 children as part of an outpatient evaluation. RESULTS: The DA revealed that all of the items, excluding five, were significant predictors by themselves. Additionally, Wilks' lambda was significant, λ = 0.16, χ(2)= 115.91, p < 0.001 for the function, indicating that all of the items together significantly discriminated between groups. The DA was run again excluding those items mentioned above and the variability accounted for by all of the items and prediction of group membership decreased. Therefore, all 45 items were retained for inclusion in the final version of the ASD-OC. CONCLUSION: The ASD-OC is able to discriminate between ASD and atypical groups.