Journal of Autism and Developmental Disorders, Vol. 24, No.

Z 1994

Domains of the Childhood Autism Rating Scale:

Relevance for Diagnosis and Treatment I
David L. DiLalla 2
Southern Illinois University

Sally ,1. Rogers

University of Colorado Health Sciences Center

The Childhood Autism Rating Scale (CARS) was factor analyzed. Three fac-
tors emerged: Social Impairment (SI), Negative Emotionality (NE), and Dis-
torted Sensory Response (DSR). Unit-weight factor scales showed moderate-
to-good internal consistency. Cross-sectional analyses demonstrated that autis-
tic (A UT) subjects were distinguished from subjects with pervasive develop-
mental disorders (PDD) and nonpervasive developmental disorders (NPDD)
by higher scores on SI. An SI cutoff score of 26 classified individuals as autistic
vs. nonautistic with 78% accuracy. Longitudinal analyses showed that DSR
was stable over 6 months of treatment, with little indication of symptom re-
duction. SI decreased over time across the diagnostic groups, but still showed
significant continuity over the period_ N E was most malleable and apparently
sensitive to the effects of treatment.

INTRODUCTION

The Childhood Aiatism Rating Scale (CARS; Schopler, Reichler, De-

veillis, and Daly, 1980; Schopler, Reichler, and Renner, 1988) is widely used
in diagnosis, treatment planning, and treatment evaluation of children with
1The treatment program described in this article was supported by Grant No. G00811002~7
to Sally Rogers from the Office of Special Education and Rehabilitation, U.S. Department
of Education. The research was also supported in part by an award to Sally Rogers from
the Developmental Psychobiology Research Group Endowment Fund provided by the Grant
Foundation.
2Address all correspondence to David L. DiLalla, Department of Psychology, Southern Illinois
University at Carbondale, Carbondale, Illinois 62901.

115
0162-3257/94/04fl0-0115507.00/0D 1994PlenumPublishingCorporation
116 DiLalla and Rogers

autistic disorders. The CARS is an objective, behaviorally based rating sys-

tem with demonstrated reliability and validity (Garfin, McCallon, & Cox
1988; Morgan, 1988). The CARS comprises 15 items on which children
can receive scores ranging from 1 to 4 with a score of 1 indicating age-ap-
propriate behavior and a score of 4 representing severely abnormal behav-
ior. Each CARS item taps behavior related to one of the following areas:
relationships with others; imitation, emotional expression, body use, pecu-
liarities in object use; resistance to change; visual, auditory, and tactile re-
sponsiveness; anxiety; verbal and nonverbal communication; activity level;
and intellectual ability. The 15th item represents the rater's global evalu-
ation of level of autistic symptoms. Traditionally, the sums of scores for
the 15 items is used to make broad diagnostic classifications. Children with
scores lower than 30 are characterized as nonautistic. Individuals who score
above 36 on the CARS and have scores of 3 or higher on a 5 or more of
the CARS items are described as "severely autistic." Children scoring
higher than 30 but not meeting criteria for severe autism are characterized
as mildly to moderately autistic (Schopler et al., 1980).
Total CARS score has been shown to effectively discriminate autistic
from mentally retarded individuals without Pervasive Developmental Dis-
orders (PDD) during childhood (Teal & Wiebe, 1986), adolescence (Garfin
et al., 1988), and in non-Western cultures (Kurita, Miyake, & Katsuno,
1989). The CARS is sensitive to developmental changes in autistic symp-
tomatology (Mesibov, Schopler, Schaffer, & Michal, 1989) as well as treat-
ment effects on autistic symptoms (Rogers & Lewis, 1989).
Whereas the overall score for the CARS has substantial diagnostic and
clinical utility, inspection of the content of CARS items, as well as correla-
tions among items, suggests that the CARS may not be unidimensional. To
the extent that robust subdomains of the CARS exist, scales derived from
such domains could provide additional information relevant to diagnosis and
treatment planning. In the present study, we present an exploratory factor
analysis of the CARS and apply factor-based scales to diagnostic and treat-
ment outcome data from a day treatment program for young children with
Autism, Pervasive Developmental Disorders, and other psychiatric disorders.

METHOD

Subjects
Participants in the study were 69 children between the ages of 24 and
73 months (M = 46.5; SD = 10.7) who were enrolled between 1981 and
1990 in an educational and psyehotherapeutic day treatment program at