Collaborative ECHO research led by Aaron Kaat, PhD of Northwestern University, and Kristen Lyall, ScD of the AJ Drexel Autism Institute evaluates two different versions of the full 65-question Social Responsiveness Scale (SRS) for measuring autism-related traits in preschool and school-age children. This study compared the questions used on each version and developed a comparable scoring method for both forms. Then, the researchers used computer-adaptive testing, where a computer program tailors the questionnaire in real-time to each participant, and found that a median of 14 questions was sufficient to reach a reliable score. This research, titled “Modifying the Social Responsiveness Scale for Adaptive Administration,” is published in Quality of Life Research.
Prior research has developed and examined shortened versions of the SRS questionnaire, but researchers haven’t been able to evaluate the reliability of these questionnaires for capturing broader traits in the general population or for preschool-age participants. This study included over 7,000 children ages 2.5-18 years old, most of whom did not have an autism diagnosis. The researchers conducted several analyses to assess a variety of factors that may affect how the full SRS and shortened versions of the SRS are scored. They were then able to use these data to simulate how computer-adaptive testing would score each individual after completing the minimum number of questions required to reach a comparable result.
“The results of this study suggest opportunities to reduce the amount of time participants need to spend completing questionnaires like the SRS,” said Dr. Lyall. “These results also suggest computer-adaptive questionnaires can be used to collect information in a way that is comparable to standard paper-based questionnaires, offering additional flexibility.”
The researchers are continuing to evaluate the computerized version of the SRS questionnaire developed in this study, comparing it to other versions of the questionnaire. Future research is needed to examine how these different questionnaires perform in different groups of people, and if results differ when captured according to self vs. parent report.