Abstract

The purpose of the present study was to examine the reliability and validity of the Korean Version of the Childhood Autism Rating Scale(K-CARS), and to determine it's cut-off scores in diagnosing autism. Subjects consisted of the following four clinical groups: 10 autistic children, 32 PDD NOS children, 31 communication disordered children, and 24 MR children within the age range of 24 to 80 months. K-CARS and SMS were administered to mothers of all clinical groups by clinical psychologists. Analysis of data revealed that K-CARS has inter-rater reliability of .94, internal consistency of .87, and test-retest reliability of .91, indicating that K-CARS is a highly reliable test. K-CARS total scores correlated negatively with SQ of the Social Maturity Scale and DQ of the EWHA-PEP. Three factors were revealed through factor analysis of K- CARS, which were labelled "socialization-communication", "repetitive-stereotyped behavior", and "unusual sensational response" factors. There were significant differences between the autistic group(autistic disorder and PDD NOS) and the non-autistic group(communication disorder and MR) on K-CARS total scores. These results strongly support the construct and discriminant validity of K-CARS. For further evaluation of the discriminant validity of K-CARS, the discriminant power of each item was calculated, and 11 of the 15 items discriminated significantly the autistic from the non-autistic group. The cut-off score for diagnosing autism was found to be twenty-eight, which classified 73.8% of the autistic group and 85.5% of the non-autistic group correctly. The limitations of the present study and further issues for the future study were discussed.