EnglishDevelopmental & Educational Psychology

Autism in girls is still underdiagnosed – can assessing preschool skills help identify it earlier?

By Penny Roy and Shula Chiat

Autism Spectrum Disorder (ASD) in the general population is about three times more prevalent in boys than girls, and girls with the same level of ASD symptoms as boys are less likely to be diagnosed. If you’re an older female or have an intellectual disability, the gender discrepancy is lower, but the disparity remains.

Why might this be? Most agree that at least some of this gender difference is due to under-recognition and under-diagnosis in females, putting girls at greater risk of getting a “late” diagnosis, if at all. “Late” diagnosis for children well into their school years and beyond carries costs for their mental health and wellbeing. These children and their families miss opportunities to benefit from early intervention, support and understanding. So, the case for earlier identification in the preschool years is clear – but is it achievable, particularly for girls whose earlier manifestations of social and communication difficulties may be subtle?

Evidence suggests that careful assessment of sociocognitive skills may help. The Early Sociocognitive Battery (ESB) was developed to identify early indicators of social communication difficulties (SCD) and ASD in children with language delay, on the grounds that children with a “late” diagnosis are often reported to have had language delay in their early years. So, what is the ESB and what is the evidence that it can identify early signs of autism in girls?

The ESB assesses early developing social engagement skills that are known to be involved in early language development and that are often impaired in children with autism. Designed to be fun and engaging for children, the ESB provides a systematic and objective measure of children’s responses to a series of “triggers.” These “triggers” seek to elicit behaviors that demonstrate sociocognitive skills and may not occur spontaneously in a clinic session.a

The ESB requires no verbal responses from the child, minimal verbal understanding, and can be used with children from diverse language and cultural backgrounds. It is psychometrically robust, standardized for UK children aged 2;0-4;11 and currently finishing a census-matched data collection in the U.S. The ESB provides both categorical “Low” cut-off scores and dimensional continuous scores for more detailed profiling of children’s strengths and difficulties. The assessment was used in a follow-up study of young children with language delay but no preexisting diagnosis of autism, and a key finding was that “Low” ESB scores at preschool age were predictive of social communication difficulties and ASD reported by parents some 7-8 years later, at the end of elementary school. Sensitivity (the proportion of diagnosed children at outcome who had “Low” ESB scores when assessed as preschoolers) was 89%. However, the sample was comprised mainly of boys, so the question arises: how predictive was the ESB for the 21 girls who made up somewhat less than a quarter of the sample?

While small sample size means results need to be treated with caution, recent findings on gender in this follow-up study are promising. At preschool age, there was no evidence of gender bias in the risk of later social communication difficulties and ASD according to performance on the ESB.  A third of the girls had “Low” ESB scores indicative of sociocognitive difficulties, a very similar proportion to the boys. However, 7-8 years later, in line with wider evidence, boys were twice as likely to have received a diagnosis of ASD/SCD. So, why was there a gender bias in reported diagnosis at the end of elementary school, that was not evident in their preschool sociocognitive skills?  A closer look at the girls’ trajectories is informative.

As expected, girls with an ASD/SCD diagnosis had “Low” ESB scores as preschoolers. But what about the girls who had “Low” ESB scores, yet did not have a diagnosis? Were these false positives? Evidence from parents’ ratings on the Social Responsiveness Scale (SRS), a measure of social communication and autism symptom severity, suggests not.

Figure 1

First, in contrast to the gender bias in proportion of children with a reported diagnosis in the full sample, there was no gender difference in the proportion of children with clinically significant scores. Second, according to SRS parental ratings, all but one of the undiagnosed girls with a “Low” ESB score had clinically significant social communication difficulties despite no diagnosis. To illustrate the point, Figure 1 shows the mean scores on the SRS of undiagnosed boys and girls according to their ESB scores (Low/Not Low). As you can see, the mean score of undiagnosed but Low-ESB-scoring girls was high, indicative of severe autism. This suggests that these were cases of under-diagnosis rather than false positives on the ESB.b

However, “Low” ESB scores are only one side of the predictive coin. We also need to consider how well ESB scores above the cut-off predict positive outcomes. Referring again to Figure 1, you can see that the mean score of preschoolers on the SRS who scored above the ESB cut-off is low. The majority of the girls’ scores fell in the normal range, with just two that were very close to the cut-off and well below clinically significant scores. The long-term risk of autism for these girls with adequate preschool sociocognitive skills was small. Much more research is needed, but initial findings are promising, and suggest that if anything the ESB may be a more sensitive measure of social communication for girls at risk of autism than boys.


aClick here for a White Paper providing more information about the ESB.

bNB: It is possible these girls received a diagnosis in subsequent years, and if so were offered, albeit belatedly, appropriate support and understanding to address their needs.


American Psychiatric Association (2013). Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, DSM-5. Arlington, VA: American Psychiatric Association.

Chiat, S. and Roy, P. (2013). Early predictors of language and social communication impairments at ages 9-11 years: A follow-up study of early-referred children. Journal of Speech, Language, and Hearing Research, 56. 1824–1836.

Constantino, J.N. and Gruber, C.P. (2012). Social Responsiveness Scale. Torrance, California: Western Psychological Services.

Gould, J. (2017). Towards understanding the under-recognition of girls and women on the autism spectrum. Autism, 21, 703–705.

Hendrickx, S. (2015). Women and girls with Autism Spectrum Disorder Jessica Kingsley Publishers

Leedham, A., Thompson, A.R., Smith, R., and Freeth, M. (2020). ‘I was exhausted trying to figure it out’: The experiences of females receiving an autism diagnosis in middle to late adulthood. Autism 2020, 24, 135–146

Lord, C., et al (2022). The Lancet Commission on the future of care and clinical research in autism. Lancet, 399, 291-334.

Mandy, W., et al (2022). Mental health and social difficulties of late-diagnosed autistic children, across childhood and adolescence. Journal of Child Psychology and Psychiatry 63, 1405–1

Roy, P., Chiat, S., and Warwick, J. (2022). Early identification of girls at risk of ‘late’ ASD diagnosis. Archives of Disease in Childhood, 107, (Suppl2), A88-A89.

Roy, P., Chiat, S., and Warwick, J. (2019). The Early Sociocognitive Battery. Oxford, UK: Hogrefe Ltd.