Interdistrict Choice, 3 years later

The Century Foundation published a report on Monday that seeks to answer a question I posed in two reports a few years ago—does interdistrict school choice significantly increase options for students in low-performing schools? And, related, does it make sense to expand NCLB’s choice provision to allow students to choose higher performing schools in other districts?

A great thing about having multiple organizations conducting research on the same issue is that you get people trying different research strategies and  each iteration of research provides a new insight to the problem. The Century Foundation’s analysis uses different methods than I did to answer these questions* and comes to slightly different conclusions about the potential of interdistrict choice, but ends with similar recommendations for how policy can leverage interdistrict choice to expand options for students in low-performing schools.

In my analysis, I used geographic information systems mapping to model whether interdistrict choice would expand options for students in low-performing schools as compared with intradistrict-only choice. To do this, I made a few assumptions in my model—1) I limited travel distance to a 20-minute drive; 2) I assumed higher performing schools could increase capacity by 10 percent to accept transferring students; and 3) I divided schools into quintiles based on performance and assumed that students would only choose schools performing at least two quintiles above their current school. These assumptions impacted the outcomes of my analysis, which is why I described in the paper exactly how adjusting each assumption would affect the outcomes.

The Century Foundation had some problems with these assumptions, so conducted their own study using different research techniques. Instead of setting a drive-time cut-off, they discounted the ‘desirability’ of higher performing schools based on their distance from a lower-performing school, so a school with 80% proficiency 3 miles away would be more desirable than a school with 80% proficiency 5 miles away. Makes sense and is a great way to distinguish between schools based on a continuum of travel time rather than a hard cut-off.

They also use a calculation that compares each school’s student-teacher ratio with the state average as a proxy for capacity. Assessing school capacity is a challenge in modeling interdistrict choice because there are no good, consistent data sets available to do this on a large scale. This is also problematic for designing policy—if you’re a state considering throwing incentives at higher performing schools to take more transfer students, it would be good to know objectively whether those schools are already bursting at the seams. If The Century Foundation’s student-teacher ratio assessment proves reliable, this could be a useful technique for policymakers looking to identify schools with excess capacity.

Finally, they also sought to make sure each sending school—receiving school pair represented a meaningful difference in school quality. I did this by limiting choice to schools that were at least two quintiles apart in terms of performance. The Century Foundation research calculated differences in the performance of each sending-receiving pair to assess the relative ‘attractiveness’ of each receiving school—the bigger the difference, the more attractive the school was.

Based on my analysis, I was less enthusiastic in my report than The Century Foundation folks about the potential of interdistrict choice. I found that between 10 – 20 percent of students would benefit from expanding choice from intradistrict to interdistrict, which I characterized as valuable, but unlikely to make dramatic changes in students’ options overall. Therefore, I argued, interdistrict choice needs to be one part of (but not eliminated from) current strategies to improve educational options for students in low-performing schools. It needs to be paired with policies that support creating new schools, improving existing schools, and offering students additional supports outside of school.

The authors of The Century Foundation report are more enthusiastic, concluding that interdistrict choice has the potential to “meaningfully expand students’ access to higher-performing schools beyond the existing intradistrict choice policy…” They base this on their finding that the “accessibility value” of an average sending school increases from 3.97 to 17.96—a large increase, but difficult to interpret in terms of the number of students affected, and therefore difficult to compare with my results.

Even using different research strategies and coming to slightly different conclusions, the final recommendations of both reports are pretty similar: if interdistrict choice is going to work, we need to do a lot more than just open up district boundaries. Choice needs to be targeted to students who are most in need of better school options; transportation needs to be funded; and receiving schools need incentives (usually financial) to take students. I call this avoiding “cheap choice”– a type of choice policy that is unlikely to yield better outcomes for students.

There are other challenges with focusing on interdistrict choice as a primary policy solution. These include the high cost of transportation, the reluctance of higher performing schools to accept students, and the fact that merely sitting in a better school doesn’t guarantee better performance—many of the transferring students will still need additional supports if they’re going to catch up.

I’m also concerned about relying on a policy that requires students to travel potentially long distances to get a good education—while it’s important to provide that option, states and districts also have an obligation to provide all students with a good school down the street. Expanding inter- or intra-district choice doesn’t relieve them of that obligation.

* I want to clarify two misconceptions about my research from The Century Foundation report. First, is that the methods I used in my August 2008 report were intended to model interdistrict choice under NCLB. The August report actually sought to address the issue of interdistrict choice broadly and was not focused on the NCLB choice provision. The second report, published in November 2008, examined that issue and in that report I used different methods that were consistent with NCLB’s criteria for identifying sending and receiving schools. Second, while my final conclusions were focused on choice across all schools, I did break down my analysis and findings by school geography and reported on the different effects of interdistrict choice based on whether schools were in urban, suburban or rural areas. My overall conclusions were informed by this breakdown and were not downwardly biased by the inclusion of rural schools.