US autism identifications skew toward White boys

Who is and is not being identified as autistic?

In Autism for 20 April 2026, Paul Morgan and Eric Hu reported their investigation of sociodemographic disparities in children identified as having autism in US schools. Working with data collected from 19 years of the National Assessment of Education Progress at fourth grade, they examined differences in identification by race, ethnicity, biological sex, family income, and language use among 100s of thousands of children. They found that “students of color, females, students from low-income families, and multilingual learners (MLs) are less likely to be identified with autism.”

Here is the abstract for Professors Morgan’s and Hu’s article:

Whether and to what extent sociodemographic disparities in school-based autism identification have been occurring in U.S. elementary schools is currently unclear. We investigated for disparities attributable to race, ethnicity, biological sex, family income, and language use by analyzing repeated cross-sectional data collected on very large samples of U.S. fourth graders participating in the National Assessment of Educational Progress from 2003 to 2022 (ns = 103,150–205,860). Multivariable logistic regression models accounting for potential confounds including student-level academic achievement and school-level resources repeatedly indicated that students of color, females, students from low-income families, and multilingual learners (MLs) are less likely to be identified with autism while attending U.S. elementary schools. These disparities have been largely stable over time, particularly for Black students, females, and MLs. Health and educational policies that ensure equal access to autism supports and services in U.S. elementary schools including by students from historically marginalized communities are warranted.

These findings provide important insights about US schools’ identification of children with autism. The report by Professors Morgan and Hu used a large and representative sample so that the sociodemographic disparities if showed clarify what had been poorly understood aspect of autism identification. It also showed that these sociodemographic factors are implicated in identification of children with autism, but that these factors have been in play for nearly two decades; they were not only influencing identifications only at some historical time nor just recently, but they have been affecting identifications across time.

The findings are important, too. Because schools are the main point of access for services for children with autism, if children with certain characteristics are not identified, they miss access to those services. Even if the services provided in school could be improved, only providing them to some students and not to others is, well, unfair. Regardless of whether a student is green-skinned and speaks Martian, if she needs help, we special educators should provide it.

As of this writing, the full article is available on the Web site of Autism. Just follow the digital object identifier in the reference.

Reference

Morgan, P. L., & Hengyu Hu, E. (2026). Over-time estimates of sociodemographic disparities in autism identification in U.S. elementary schools. Autism. Published online 20 April 2026. https://doi.org/10.1177/13623613261434432