Rasch modeling and confirmatory factor analysis of the systemizing quotient-revised (SQ-R) scale.

Abstract:

This study assessed the dimensionality of the Systemizing Quotient-Revised (SQ-R), a measure of how strong a person's interest is in systems, using two statistical approaches: Rasch modeling and Confirmatory Factor Analysis (CFA). Participants included N = 675 with an autism spectrum condition (ASC), N = 1369 family members of people with ASC, and N = 2014 typical controls. Data were applied to the Rasch model (Rating Scale) using WINSTEPS. The data fit the Rasch model quite well lending support to the idea that systemizing could be seen as unidimensional. Reliability estimates were .99 for items and .92 for persons. A CFA parceling approach confirmed that a unidimensional model fit the data. There was, however, differential functioning by sex in some of these items. An abbreviated 44-item version of the scale, consisting of items without differential item functioning by sex was developed. This shorter scale also was tested from a Rasch perspective and confirmed through CFA. All measures showed differences on total scale scores between those participants with and without ASC (d = 0.71, p < .005), and between sexes (d = 0.53, p < .005). We conclude that the SQ-R is an appropriate measure of systemizing which can be measured along a single dimension.