Since its release, the Government Accountability Office’s (GAO) recent report on discipline disparities has generated substantial heat, but no new light. Based on an analysis of the most recent discipline data collected by the Office of Civil Rights, it concludes that “Black students, boys, and students with disabilities were disproportionately disciplined…in K–12 public schools.” But if that sentence contains any new information, it is well hidden. And as the report acknowledges, by themselves these disparities “do not establish whether unlawful discrimination has occurred.”
Using a generalized linear regression model—basically a more flexible version of ordinary linear regression—the authors of the report investigate the relationships between various school-level characteristics and discipline outcomes. However, as they acknowledge, their methodology has at least two important limitations.
First, because they don’t have student-level data, the authors can’t actually control for poverty and other factors at the student level. Thus, although the study finds that schools with more black students have higher suspensions rates—even after controlling for the number of poor kids—it doesn’t show that poor black students are more likely to be suspended than poor white students.
Second, as the authors once again acknowledge, “some variables that may be related to student behavior...