GARS-3™

Reliability and Validity

• Internal consistency (content sampling) reliability coefficients for the subscales exceed .85 and the Autism Indexes exceed .93
• Test-retest (time sampling) reliability coefficients exceed .80 for subscales and .90 for the Autism Indexes
• Interrater reliability intraclass coefficients exceed .80 and .84 for the Autism Indexes.
• Correlations of the GARS-3 scores with those of other well-known diagnostic tests for autism are large or very large in magnitude
• All new validity studies show that the test results are valid for a wide variety of subgroups as well as for the general population.
• Binary classification studies indicate that the GARS-3 is able to accurately discriminate children with Autism Spectrum Disorder from children without autism (i.e., sensitivity = .97, specificity = .97, ROC/AUC = .93.
• Confirmatory and exploratory factor analyses demonstrate the theoretical and empirical validity of the subscales.
• Other validity evidence is provided in the manual.

Normative data (N = 1,859) were collected in 2010 and 2011. Demographic characteristics of the normative sample are keyed to those reported by U.S. Bureau of the Census, 2011.