We’re more than half a century past Brown v. Board, but segregation remains a prominent feature of American public schools. New York City, the nation’s largest school district and a prime example of the persistence of segregation, has recently embarked on an initiative to desegregate some of its schools, and early results suggest this plan is working and could better integrate more schools.

The way the plan works (in theory)

Although New York City is nominally one district, it is divided into 32 community school districts. Three are “choice districts,” where students have priority for any school in that district. In addition, there are “nonzoned” schools that are open to students from anywhere in the district or borough. School choice within and across districts has become popular, with 40 percent of students attending schools other than their zoned public school.

In the 2016–17 school year, six choice and nonzoned schools and one zoned school were part of the Diversity in Admissions pilot program designed to increase the representation of underserved demographics. These schools created “set-asides” that reserved a minimum number of seats for certain groups of students: for example, 40 percent of seats for students eligible for free or reduced-price lunch (FRPL), the most common criterion. (For legal reasons, school districts avoid using race or ethnicity as a criterion.)

Under the set-aside program, schools use a two-step lottery. In the first stage, only students who meet the set-aside criteria are eligible, and the lottery proceeds until the set-aside is reached. Any qualified student who does not receive a seat in the first stage is entered into the second-stage lottery for the remainder of the seats.

This means that at least 40 percent of the seats will be filled by FRPL-eligible students. These set-asides encourage more students from underserved demographics to apply in the first place.

The chart below illustrates how a set-aside would work for 10, 30, and 60 percent set-asides. A 30 percent set-aside, for example, guarantees that at least 30 percent of the class is FRPL eligible unless there are not enough such students in the applicant pool to fill the set-aside seats.

The way the plan works (in practice)

The actual set-aside plans in the New York City pilot schools, described in the table below, are more complicated than the theoretical model. Many use a mix of FRPL eligibility, English language learner (ELL) status, and other factors.

These schools had low percentages of FRPL-eligible students in the 2015–16 academic year, before the set-aside program, compared with their districts. In fact, part of the program’s impetus was that applicant pools were becoming wealthier than the neighborhood.

View data for the table above via Chalkbeat and the New York City Department of Education.

To see whether the set-aside plan was effective, I compare the FRPL percentage among kindergarteners in the year before implementation (2015–16) to the first year of implementation (2016–17). I focus on FRPL eligibility, as it is the most common criterion, and limit the analysis to kindergarten students, as the plan essentially only applies to them.

Five of the seven schools had an increase in the percentage of FRPL-eligible kindergartners, with three exceeding the floor they determined with the set-aside. Although only one of the differences was statistically significant, in part because of small sample sizes, the model pooling all seven schools is statistically significant, a sign that the program as a whole had a positive impact on desegregation.

View data for the chart above via the New York City Department of Education.

For the two schools that did not see an increase, the drops were small. Moreover, the Brooklyn Arts and Science school used a different eligibility criterion, enrollment in the child welfare system, which is similar to but not the same as FRPL eligibility. This school is also the only zoned school among the seven pilot schools, meaning that priority for zoned students could mitigate the set-aside’s effect. Preliminarily, the program seems to be effective, but with only seven schools and small kindergarten populations, this is far from conclusive.

In the 2017–18 school year, the program was rolled out to 21 schools, and other plans are in the works. Revisiting this analysis when those enrollment data are available will be important. Examining additional schools will help indicate whether the program integrates these schools by FRPL eligibility and ELL status.

An additional year of data can also be added to the initial seven schools, allowing us to study whether the results persist and how much sibling preferences can magnify the set-aside’s effect.

Finally, an anticipated program benefit is to promote racial and ethnic diversity. Our current data don’t allow us to do this, but as more data are released, we hope to revisit this question to study whether, and how much, FRPL, ELL, and other criteria can be used to improve not only economic diversity but racial and ethnic diversity as well.