The factor structure and measurement invariance of the Autism Spectrum Quotient-28: A cross-cultural comparison between Malaysia and the Netherlands

Abstract

Despite several psychometric advantages over the 50-item Autism Spectrum Quotient, an instrument used to measure autistic traits, the abridged AQ-28 and its cross-cultural validity have not been examined as extensively. Therefore, this study aimed to examine the factor structure and measurement invariance of the AQ-28 in 818 Dutch ( Mage = 37.4, 581 females, 233 males, 4 others) and 437 Malaysian ( Mage = 23.0, 328 females, 99 males, 10 others) participants from the general population. The hierarchical structure of the AQ-28 showed fair and good fit in Malaysia and in the Netherlands, respectively. A multi-group invariance analysis supported that the AQ-28 is cross-culturally invariant. Malaysians ( M = 68.63, SD = 8.33) scored significantly higher than Dutch participants ( M = 51.48, SD = 10.30) on the AQ-28 while gender was controlled for. While the measurement invariance suggests that the AQ-28 functions similarly in Malaysia and the Netherlands in terms of structure, exploratory analyses showed 11 items with differential item functioning. Hence, while the AQ-28 possesses a stable factor structure and appears to measure the same latent traits in Malaysia and the Netherlands, some items potentially display cultural bias which, in turn, might explain the differences in AQ scores. Lay abstract The AQ-28 is a questionnaire measuring autistic traits, that is, traits that are related to Autism Spectrum Conditions, but its reliability in other cultures has not been thoroughly evaluated. We, therefore, tested whether the properties of the AQ-28 are comparable between two countries with different cultures, Malaysia and the Netherlands. A total of 437 Malaysian and 818 Dutch participants completed the AQ-28 online. We measured whether the AQ-28 measures autistic traits similarly in Malaysia and the Netherlands. The AQ-28 measures autistic traits similarly, and the reliability was acceptable and good in the general population of Malaysia and the Netherlands, respectively. However, Malaysians scored higher than Dutch participants. Moreover, 11 AQ-28 items showed cultural bias, indicating that these items are answered/interpreted differently in Malaysia and the Netherlands. Cross-cultural differences in interpreting, reporting, and/or expressing autistic traits highlighted in this study could potentially explain why some items are culturally biased and why Malaysians score higher on these items. The findings of this work imply that cutoff scores derived from one culture should not be generalised to another culture. Moreover, the findings are informative for future development of culturally neutral or appropriate screening and diagnostic tools for autism.