Abstract

INTRODUCTION: Studying autism spectrum disorder (ASD) globally is not only important for estimating preva- lence rates, but also for evaluating the whole range of interplay of genetic, neurobiological, and environmental correlates leading to ASD, as well as available treatment options, while targeting the same aspects of the disorder across different settings and regions worldwide. Thus, the research of ASD globally requires sound and sustainable evaluation/diagnostic tools that would be available for use in different research and clinical settings across various high and low-to-middle- income regions. From a psychometrical point of view, the latter is only possible if evaluation methods operate in the same way and underlying constructs have the same theoretical structure across different regional/cultural groups, which implies that measurement invariance is present. This aspect is not readily tested for available ASD instruments. OBJECTIVES: To evaluate whether the Childhood Autism Rating Scale (CARS) original measures ASD the same way across different language/cultural groups (i.e., the measurement invariance is present). METHODS: The sample included children 656 with ASD for whom the CARS was completed in English (234), Span- ish (171), Turkish (150), or Hindi (101). The ordinal logistic regression was used to assess differential item functioning (DIF), as a mode of testing measurement invariance, controlling for the effects of age, sex, and country income. An item exhibiting DIF is considered non-invariant. The study was organized by the Autism Spectrum Disorder International Consortium (ASDIC). RESULTS: Overall, 14 out of 15 (93.3%) CARS items were flagged with DIF across all four language samples, while be- tween any two samples, there were 4 (26.7%) to 10 (66.8%) DIF items. Six items (40%; body use, object use, adaptation to change, taste, smell, and touch response and use, fear or nervousness, and level and consistency of intellectual response) displayed uniform DIF, indicating that children with ASD from any sample would be rated differently with these items compared to children in the other samples at the same ability level. In addition, item level and consistency of intellectual response (6.7%) had non-uniform DIF, indicating that rating intellectual functioning among children with ASD is likely dependent on one’s language/cultural group. Items object use and fear or nervous had the largest DIF. The most evident DIF items were across the Turkish and Hindi and the least across the Spanish and Hindi samples. Table 1. shows DIF summary in four language groups. CONCLUSIONS: This preliminary analysis revealed that children with ASD are likely evaluated differently by almost all CARS items across language/cultural groups, but there are items which might be more/less non-invariant. To gain a better insight into the measurement of non-invariance pertaining to the CARS for cross-cultural comparisons, further analyses will focus on children with different neurodevelopmental difficulties, typically developing children, and chil- dren with ASD across different world regions.