Investigating the Validity of the Social Responsiveness Scale In a Clinical Sample of Preschool Children with Autism Spectrum Disorder

Background: The Social Responsiveness Scale (SRS) is a widely used instrument for distinguishing autism spectrum disorder (ASD) from other developmental disorders and for quantifying the severity of autistic social impairment. The SRS has been validated in children 4 to 18 years of age from clinical and referred samples. However, current understanding of the psychometric properties of the SRS in younger children with ASD is limited.

Objectives: This study examined the validity and psychometric properties of the SRS in a clinical sample of newly-diagnosed preschool children with ASD. 

Methods: The data came from a Canadian longitudinal study investigating the development of children with ASD. The study sample consisted of 339 children with an ASD diagnosis (mean age: 40.8 months (SD 9.3 months); 284 males (84.4%)). First, mean scores for the total scale and five SRS subscales were compared between the current sample and the norms. Second, confirmatory factor analysis (CFA) was used to assess the best fitting model for the structure of the SRS in this clinical sample. Third, Rasch analyses were used to evaluate the performance of the overall scale as well as each of the 65 items of the SRS.  

Results: Study results indicate that: (a) Children in the current clinical sample score significantly higher than the published scale norms on the total scale and five SRS subscales (all effect sizes > 1.0 SD). Differences were largest for the Social Communication and Autistic Mannerisms subscales. (b) CFA results showed that the 1-, 2-, and 5-factor solutions for the structure of the SRS, previously proposed in the literature, provided an inadequate fit for the data (CFI and TLI varied from 0.46 to 0.50; RMSEA varied from 0.074 to 0.075); the best fit for the data was provided by the Autistic Mannerisms subscale (CFI = 0.85, TLI = 0.82, RMSEA = 0.08). (c) Results from the Rasch analyses confirmed those derived in the CFA; they also indicated that 17 out of the 65 SRS items did not perform as expected.    

Conclusions: Study findings show that compared to the published scale norms, preschool children with a diagnosis of ASD have elevated scores on social impairment as measured by the SRS. Furthermore, the data suggest that the structure of the SRS cannot be described as uni-dimensional and that a substantial number of SRS items function poorly and do not discriminate well between children with ASD in this clinical sample and population norms. Overall, the Autistic Mannerisms subscale performed the best out of the five SRS subscales and provided an adequate fit for the data. Although the SRS has proved to be a useful assessment screening tool , caution is warranted for its use and interpretation in clinical samples of preschool children with ASD.