By Kenneth Shores, Ha Eun Kim, and Mela Still for Brookings

The Civil Rights Data Collection (CRDC) reports large, persistent gaps between Black and white students on educational outcomes such as school suspensions, uptake of AP classes, assignment to special education and gifted and talented classes, and grade-level retention. How might we explain disparities in these outcomes? As we argue in a forthcoming paper in the American Educational Research Journal (prepublication version here), schools are responsible for these disparities—and they have the capacity to do away with them.

Conventional approaches, which tend to analyze these disparities using frameworks developed to analyze differences in test scores (often referred to as “achievement gaps”), use regression analysis to determine empirically whether Black and white differences in educational outcomes, like suspension rates, are explained by racial differences from in- or out-of-school variables. Typically, in-school variables include teacher characteristics, such as experience and race, and school demographic variables. Out-of-school variables generally include proxy measures of the student’s household income, such as whether they qualify for free lunch and their parents’ educational attainment. Our review of the literature provides an exhaustive account of how the research community has tended to document and model racial disparities in disciplinary policy, access to advanced courses, assignment to gifted and talented and special education, and grade-level retention in this way.

As with gaps in test scores, racial differences in socio-economic status tend to account for most of the variation in racial disparities in suspension rates, classification into specialized classes, and advanced course-taking. A common interpretation of these results is that the race-based sorting of students into different educational experiences is attributable to students’ characteristics or home environments. In other words, the data are interpreted to mean that Black students are suspended at higher rates (or less likely to take advanced math) because they confront steeper out-of-school challenges—not because of anything that schools do.

Our view differs from these conventional approaches. We argue that schools are the principal source of these disparities, and, to the extent out-of-school factors are emphasized over educator discretion, researchers will have failed to hold schools properly accountable. To understand our argument, first note that the educational outcomes listed above are systems of classification (e.g., gifted, suspended) that can either benefit or harm students to whom they are assigned. When these systems of classification track students’ race and ethnicity, it is sometimes described as “racialized tracking,” and following recent sociological theory, we refer to this process generically as “categorical inequality.”

The important thing to note is that schools create these socially relevant categories, and teachers and school leaders sort students into them. For this reason, categorical inequalities are distinctly within the purview of schooling.

Our empirical analysis documents that assignment to these categories is systematically demarcated by students’ race. We compile publicly available data from the CRDC for years 2011-12, 2013-14, and 2015-16 (the most recent year data are available) and calculate Black-white gaps in each of these educational outcomes for U.S. public school districts. Our goal with these data is to document the scope and clustering of Black-white categorical inequality among school districts in the U.S.

Black-white categorical inequalities are large in magnitude (as shown in Figure 1 below) and persist after controlling for racial differences in socio-economic status and neighborhood contexts (see Table 4 in our paper).

We also show that in districts where Black students are worse off academically or socio-economically, school districts tend to increase the educational disadvantages that Black students face. As evidence, we show that districts with greater inequality, segregation, and lower overall socio-economic status also have larger achievement and disciplinary gaps. Further, these categorical inequalities are commonly found clustered together in districts. That is, districts with large categorical inequalities (defined as having a categorical inequality in the top 20% of our sample) in one dimension (e.g., suspension rates) tend to have large categorical inequalities in other dimensions (e.g., AP uptake). Finally, the probability of having multiple large categorical inequalities steeply increases as test score and racial socioeconomic inequality increases. In the spirit of the CRDC’s accountability mission, we also identify specific school districts in which these disparities are consistently the largest in our sample. (See Tables 6, 7, and 8 of our paper.)

In many ways, our quantitative analysis mirrors the conventional approaches we faulted above. The key difference is in the interpretation. For instance, if, after controlling for racial differences in socio-economic status and neighborhood contexts, racial differences in categorical inequalities were eliminated, the conventional interpretation would be that schools are not contributing to these inequalities. For us, the fact that Black and white students have different socio-economic environments does not give schools license to create racially stratified categories. And, because categorical inequalities persist after controlling for racial differences in out-of-school, we conclude that schools tend to recapitulate structural inequality.

As evidence that schools need not recapitulate structural inequality, we identify practices and policies that can diminish them. At the nexus of many of these categorical inequalities is a teacher who uses a student’s race as a classifier instead of an objective standard. We cite evidence on the effect of using objective standards (such as test scores and grade point averages) in place of teacher discretion to reduce disproportionality in special education classification, gifted and talented classification and AP course-taking. Teachers can also be intervened upon: Interventions that addressed teachers’ cultural deficit thinking reduced disparities in special education classification, and a relatively brief, empathy-based intervention caused teachers to cut suspension rates in half.

Thus, schools can both perpetuate and undo categorical inequality. Indeed, our analysis of the data finds many districts where discipline, retention, classification, and AP gaps are substantially smaller than otherwise similar districts, making it possible to identify them and learn from their experience.

And because we take as our starting point that racial inequality is ipso facto bad, we believe that the “burden of justification” falls on schools to provide an account of why these disparities are permissible. For instance, a common argument for inequalities in access to AP classes can be expressed along these lines:

“Advanced Placement courses are necessary because they give students with strong academic potential the opportunity to develop their talents and prepare for further education. If everyone could take these classes, those with lower academic ability would not benefit, and those with stronger academic ability would not learn as quickly. The fact that racial differences in Advanced Placement uptake are pervasive and systematically track racial differences in socioeconomic status is a function of structural inequality in the U.S. but not an indicator of unequal treatment of racial groups by schools per se.”

The framework of categorical inequality allows us to interrogate this argument in multiple ways. The most basic objection is that the process was not fair from the beginning, as Black students have not had the same opportunities as white students (the “opportunity gap”), and Black students with similar test scores as white students are still less likely to take these courses for reasons having to do with the selection process itself.

Nevertheless, it is only a necessary first condition for schools to use a fair selection process. Supposing a fair process was in place, and the magnitudes of categorical inequalities among Black and white students declined, the mere existence of socially relevant categories will create constant pressure to leverage those categories for stratification. Therefore, any purported benefit of category creation would need to be evaluated against the corresponding racial inequalities that are unleashed.

Weighting these competing values is a difficult but not impossible task. For instance, a common moral principle states that inequalities are acceptable if they are necessary to improve the lives of society’s least advantaged members. Thus, if differentiated instruction was thought to ultimately benefit the least advantaged members of society—perhaps by allowing gifted students the opportunity to develop their skills, producing economic benefits that improve the lives of those who didn’t have access to advanced classes—then it would be justified under this standard.

Often, the existence of categorical inequalities in schools is the default. In this article, and in our published paper, we hope to disrupt this complacency. If schools take the burden of justification seriously, then they will need to explain why, by appealing to external reasons such as those described above, classification systems that tend to stratify students along racial lines are necessary.