Is there an optimal self-report measure to investigate autism-related sex differences?

There is growing research interest in autism-related sex differences. Many behavioural and cognitive sex differences have been identified, with implications for research and clinical practice. Much of this research has relied on self-report autism measures, which are assumed to measure autistic traits equally in males and females. However, robust evidence for this assumption is lacking. Previous findings have not been replicated and no study has directly compared sex differences across multiple self-report autism measures in the same sample. To address this gap in research, a large sample of adults ( N = 1000, 500 female) completed a series of self-report autism measures (AQ-50, −28, −26, −20, −10, −9, BAPQ, CATI). Following pre-registered measurement invariance analyses, only the AQ-9, AQ-28, and CATI showed good-to-acceptable invariance to sex when specifying a multi-factor structure, and all 8 measures showed non-invariance to sex when capturing a general autism construct. We discuss the implications of these findings for investigating autism-related sex differences in future research. • Few studies have tested if self-report autism measures can capture autistic traits equally across sex. • We tested eight self-report autism measures in a large, nationally representative sample. • Most measures did not capture the same autism constructs in males and females. • The consequences of these findings for research are discussed.