Methods: We analyzed sub-samples participating in two nationally representative datasets to estimate the risk of being identified as having disabilities attributable to race or ethnicity. In the first set of analyses, we used student-level data (e.g., student N = 44,630) to estimate the risk for identification while attending schools in 11 U.S. states with histories of de jure and de facto racial segregation, prior to and following statistical control for strong confounds. We then replicated and extended this first set of analyses by analyzing data from a federal dataset of U.S. school districts (e.g., district N of 2,571) in which we again control for potential confounds (e.g., exposure to poverty, achievement gaps).
Results: Logistic regression analyses of student-level data from 11 U.S. Southern States (i.e., Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Texas, and Virginia) yielded no evidence of minority over-representation in special education attributable to systemic bias. Instead, we repeatedly observed that White students were more likely to be identified than similarly situated Black or Hispanic students (e.g., covariate adjusted odds ratio range for Blacks relative to White students of .26 to .59). Replication analyses of district-level data across the U.S. again indicated that White students were more likely to be identified than Black or Hispanic students.
Conclusions: Minority students are less likely to be identified as having disabilities in the U.S. than White students even when displaying similar clinical needs. These inequities may partially explain racial achievement gaps in the U.S., as well as disparities in the school-to-prison pipeline (e.g., Ramey, 2015), and may require systematic efforts to address.