Test-Retest Reliability and Validity of the Autism Symptoms Questionnaire

Abstract

Although many instruments can measure symptoms in Autism Spectrum Disorder (ASD), only a few are directly based on the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision (DSM-IV-TR) criterion. One of these tools called the Autism Symptoms Questionnaire (ASQ) is directly based on the DSM-IV-TR criteria for ASD and has been used in several clinical studies. However, its basic psychometric characteristics such as reliability and validity have not been studied. In this study we analyze the data from 165 individuals that were seen in an autism multispecialty clinic whose caretakers completed the ASQ at least twice. In addition, concurrent to completing the ASQ, most caretakers also completed two other commonly used validated questionnaires used in autism research: the Social Responsiveness Scale (SRS) and the Aberrant Behavior Checklist (ABC). The data was analyzed using a mixed-model to control for within-subject variation and time between assessments. First, we found excellent test-retest reliability for the ASQ with statistically significant r values for the mixed-model. Reliabilities of 0.93, 0.94, 0.91 and 0.99 were found for the social, communication and stereotyped behavior subscales and total score, respectively. The Minimal Clinically Important Difference (MCID) was calculated as 0.11, 0.09, 0.12 and 0.04 points for the social, communication and stereotyped behavior subscales and total score, respectively. The r values for the mixed-model associating the ASQ with the SRS and ABC varied between subscales but were moderate to good in magnitude and statistically significant, demonstrating that the ASQ measures a valid psychometric construct. Overall, this study suggests that the ASQ has acceptable reliability. The ASQ’s close correspondence to the DSM criteria is a strength of this instrument. With further research, it may be possible to develop a similar tool based on the DSM-V criteria. It appears that a Minimal Clinically Important Difference is approximately 0.1 point on any subscale, providing validation that changes on the ASQ can be considered important.