Last week, we released our research brief on segregation by race and ethnicity in the District of Columbia. The analysis is unique insofar as it includes regular public schools, charter schools, and private schools, thus providing a comprehensive look at segregation in our nation’s capital.

Private schools serve only about 17 percent of the D.C.’s students, but almost 60 percent of its white students. This means that any analysis of segregation in D.C. that excludes private schools may be missing a pretty big part of the picture. Our brief includes estimates of segregation, using different types of measures, within the private and public sectors (including D.C.’s large charter school sector). Unsurprisingly, we find high levels of segregation in both sectors, using multiple race and ethnicity comparisons. Yet, while segregation in both sectors is extensive, it is not substantially higher in one or the other.

But one of our most interesting findings, which we’d like to discuss here, is that between 25-40 percent of total citywide segregation is actually found between the public and private sectors. This is not a particularly intuitive finding to interpret, so a quick explanation may be useful.

The segregation measure we use for this “within/between decomposition” defines perfect integration as a situation in which every single school within a given unit (in this case, across the entire city) has exactly the same race and ethnicity composition as the unit as a whole. Our estimate of between-sector segregation, put very simply, expresses the degree to which total citywide integration is impossible without students switching between sectors (see Reardon et al. 2000).

(This decomposition is more commonly carried out to look at segregation within and between districts – e.g., within a metro area.)

Most notably, as mentioned above, private schools serve only about one in seven of D.C.’s students but well over half of its white students. As a result, no matter how many students were reshuffled between schools within the public and private sectors, perfect integregation would be impossible because, in crude terms, the public sector would run out of white students and the private sector would run out of minority students. Our analysis shows that between 25-40 percent of total segregation (depending on the race and ethnicity categories included) is due to this and other differences between the two sectors in their racial composition.

This means that, even if the public school sector was somehow completely desegregated, as much as half of total citywide segregation would remain intact, since there would still be segregation within the private sector and between sectors. This finding carries important policy implications insofar as desegregation efforts have tended to focus exclusively on public schools. Even the complete success of these efforts, while absolutely worthwhile and critically important, would not come close to solving the problem on a citywide basis, at least in D.C.

The decomposition also illustrates the important idea, which is far from original but not always fully acknowledged in our public discussions about school (and residential) segregation, that segregation occurs both within and between units. So, for example, students within a given big city district may be quite integrated, with schools in the district having roughly the same mix of students by race and ethnicity, but these students may still be segregated from their peers in surrounding districts (e.g., if the big city district serves the vast majority of minority students in the area).

We also perform a similar decomposition that looks at the degree to which total segregation is due to: 1) the separation of white from minority students; and 2) the separation of minority groups (in this case, Black and Hispanic) from each other. And we find that, in the public sector, both “factors” are roughly equally important. That is, the segregation of white from Black and Hispanic students contributes as much to citywide segregation as does the segregation of Black from Hispanic students. In the private sector, in contrast, total segregation is driven mostly by the segregation of white from minority students.

The is yet more evidence that the “traditional” definition of segregation (i.e., between whites and minorities) is insufficient in an increasingly multiracial society. The segregation of minorities from each other is an important part of total segregation, and, as we show, its contribution to total segregation can vary between public and private sectors, even within the same city.

But the primary lesson we draw from the analysis as a whole is that researchers, when possible and appropriate, should think about including private schools in segregation analyses, and, more importantly, that policymakers should consider them when looking at segregation and what to do about it. There are good reasons why this is not common, such as data availability and the fact that desegregation efforts tend to focus on public schools. The bottom line, however, is that private schools, particularly in big cities, serve pretty large chunks of students, and these students tend to be different from their public school peers in terms of characteristics such as race and ethnicity or income. Excluding private schools can therefore result in an incomplete picture of segregation in a given city, area, or state.

As discussed above, this also matters for policy, since desegregation efforts that “ignore” the private sector may have what amounts to an impermeable ceiling on their efficacy. Moreover, if desegregation efforts caused some parents to flee to private schools, as has been the case in the past (e.g., Clotfelter 2004), monitoring the private sector situation would be even more important.

We will be replicating this analysis in other cities in the future, to see how the situation compares to that in D.C.

In the meantime, you can read the full report here.