The conventional wisdom in education circles is that U.S. schools are “resegregating” (see here and here for examples). The basis for these claims is usually some form of the following empirical statement: An increasing proportion of schools serve predominantly minority student populations (e.g., GAO 2016). In other words, there are more “segregated schools.”

Underlying the characterization of this finding as “resegregation” is one of the longstanding methodological debates in education and other fields today: How to measure segregation (Massey and Denton 1988). And, as is often the case with these debates, it’s not just about methodology, but also about larger conceptual issues. We might very casually address these important issues by posing a framing question: Is a school that serves 90-95 percent minority students necessarily a “segregated school?”

Most people would answer yes. And, putting aside the semantic distinction that it is students rather than schools that are segregated, they would be correct. But there is a lot of nuance here that is actually important.

For one thing, a given school cannot really be considered segregated without some reference to other schools (or districts, metro areas, etc.), any more than a neighborhood can be considered segregated without looking at other neighborhoods. Of course, we all know the U.S. is a multiracial society, and so we know that along with largely minority schools there are also largely white schools – i.e., the comparison group is implicit. The fact remains, however, that segregation necessarily entails multiple units (schools, districts, etc.).

This is not entirely semantics. That a school or a given number of schools serve mostly minority student populations carries different implications in a district (or area) where 95 percent of the students are minorities than it does in a district where 30 percent are minorities. In the former district, it might very well be mathematically impossible for any school not to be mostly minority, whereas in the latter district a large or increasing proportion of mostly minority schools would indicate more sorting of students by race and ethnicity.

(Most of this discussion applies to income and other forms of segregation as well.)

In the former scenario (overwhelmingly minority area), the number of schools that serve mostly minority students would mostly reflect between district (or area, state, etc.) segregation. That is, there are so many mostly minority schools in this district primarily because there are so many minority students, which in turn is a result of segregation of students between districts.

In the latter scenario, in which a mostly white district contains a high number of mostly minority schools, there is more evidence of segregation within the district. Minority students are clustered in certain schools within the district.

Now, to be sure, in both of these hypothetical districts, the schools serving large minority student populations would still offer these students less potential for interaction with white students (and vice-versa) than they would have in more diverse schools. In other words, the experiences of students in these schools aren’t necessarily affected by the comparison group. They might also have fewer resources and other disadvantages.

Yet there is a difference between the racial and ethnic composition of a group of schools and the degree to which their students are segregated per se. The proportion of schools that serve large minority student populations might increase over time due solely to an increase in the minority student population overall. Put crudely, if there are more minority students, there will be more schools serving large proportions of minority students, even if those minority students are spread out perfectly randomly across schools. And, indeed, this is why, as mentioned above, an increasing percentage of U.S. schools serve large minority student populations. A big part of it is demographic change.

(Alternative approaches to segregation, such as a group of measures often called “evenness” measures, “control” for demographics by focusing exclusively on how spread out students are across schools. You can find more discussion and citations here.)

This distinction also carries policy implications. For instance, desegregation efforts usually focus within districts. Policymakers trying to increase white|minority integration have greater constraints in a district that serves an overwhelmingly minority student population than they do in a district with a more diverse student population (though minority groups are often segregated from each other, and integregation of these groups is also important).

Given the huge contribution of between district segregation to total segregation, efforts to desegregate schools within districts are limited in their potential impact, particularly in homogeneous districts. And this is in no small part a result of how school districts (and attendance zones) are delineated. It will be far more difficult to achieve integration without addressing between district segregation. A fuller discussion of this complex policy situation is beyond the scope of this post, but the point here is that it might go completely unnoticed if the measurement issues are ignored.

We now return to our initial question of whether a school serving 90-95 percent minority students is a “segregated school.” For all intents and purposes, the answer is yes. If, hypothetically, U.S. schools were integrated, there would be none serving such a large minority population. Conversely, every school serving such a high proportion of minority students is indeed evidence of segregation. But it bears remembering that schools are segregated both within and between districts (or areas), and the latter actually a more important reason why there are so many schools serving overwhelmingly minority populations.

Moreover, claims that schools are “resegregating” most often rest upon a trend – i.e., the increasing number of overwhelmingly minority schools – that is more a result of demographic change (i.e., larger minority student population) than of the sorting of students into schools by race and ethnicity. This does not change the situation “on the ground,” but it is important to understand why it’s happening if we are to understand the problem and what to do about it.