The Predictive Validity and Differential Item Functioning of Autism Spectrum Rating Scale (ASRS)

Abstract

The study aimed to reveal the predictive capability of the Autism Spectrum Rating Scale (ASRS) and to assess the Differential Item Functioning (DIF) of the scale items. Data from a parent of 198 children aged between two and five years were recruited. Half of them were diagnosed with autism spectrum disorder at the Growth and Behavior Disorders Center at Al-Waladah Children's Hospital in Mecca, Saudi Arabia. Children were selected using a simple random sampling method. To assess the predictive ability of the scale, data was analyzed using Artificial Neural Networks (ANN), and the results showed a strong predictive capability of the scale, making it a suitable tool for classifying children with autism spectrum disorder aged two to five years. To assess the Differential Item Functioning (DIF), two subscales, one for social communication and one for unusual behaviors, were analyzed separately using the Iterative Hybrid Ordinal Logistic Regression/Item Response Theory (OLR/IRT) method. The results indicated that the scale's items exhibited no regular, irregular, and uniform differential performance. The study recommended using the scale for diagnosing children with autism spectrum as an effective tool for identifying children with autism spectrum disorder. The study also proposed conducting a study to calibrate the scale based on the Multidimensional Graded Response Theory model, which is one of the models of Item Response Theory (IRT).