Supplementary Materials

In order to estimate the potential of interdistrict choice to offer students better schooling options, we needed to make several assumptions about students' choice of schools, driving distances, and the capacity of higher-performing schools to accept students. Each of these assumptions impacts the final results.

Driving Time: We chose a 20-minute driving distance to represent the time most students spend commuting to school—according to data from the 2001 National Household Travel Survey, the average commute to school is 18 minutes. While there are examples of programs in which students are bused long distances from city to suburban schools, often riding the bus for an hour each way, using such a long driving distance could overstate the potential impact of choice. Since we're estimating the potential of choice to operate on a statewide basis and for more than a select number of students, we chose a commuting distance that would likely be considered reasonable to most parents. In addition, because the driving-time estimates do not take into account additional drive times due to rush-hour traffic or indirect bus routes, the 20-minute limit underestimates actual driving times.

Moreover, expanding travel time beyond 20 minutes does not necessarily expand choice substantially (see sidebar, More Miles to Go, Page 17). While increasing the maximum drive time does increase the number of potential higher-performing schools for any given student, it also increases the number of other students who have access to those same schools. Because of this “competition effect,” more travel time does not necessarily equal more choice. Indeed, our analysis suggests that increasing the maximum drive time assumption beyond 20 minutes has a negligible effect on the percentage of students with additional options. Beyond that point, the benefit of additional accessible schools is, for the most part, cancelled out by increased competition for limited spots from other students.

Capacity: Higher-performing schools cannot infinitely expand to accommodate students transferring from lower-performing schools. Therefore, we needed to include some measure of school capacity that would neither artificially limit nor overstate the impact of interdistrict choice. We chose a 10-percent increase in enrollments, an amount we estimate schools could reasonably accommodate. In their recommendations for NCLB's school transfer provision, which allows students in low-performing schools to attend higher-performing schools, the Aspen Commission on No Child Left Behind supports this assumption, proposing that higher-performing schools be required to make at least 10 percent of their seats available to transferring students. We found no research suggesting that any assumption other than 10 percent is more empirically justified as the best estimate of maximum increased capacity. An analysis that did not assume some limitation on capacity—one that only counted the number of higher-performing schools within range of a given school, for example—would in many cases overstate the true potential of interdistrict choice. For an example illustrating this, see Map 4, The Piedmont Bubble (Oakland, Calif.), Page 8.

To estimate how our results might differ if this capacity assumption were changed only requires some basic math. If 12 percent of California students enrolled in lower-performing grade three schools could transfer under interdistrict choice with a 10-percent capacity assumption, 24 percent could transfer if we increased our capacity assumption to 20 percent.

Choice of Schools: We also made an assumption about who would be offered the choice to transfer schools and which schools they would transfer into. Rather than assume that all students would have the option to transfer to any school, no matter how much better or worse that school was performing, we limited choice only to students in the bottom 40 percent of schools and limited their choices to schools that were substantially better performing—at least two quintiles above in student performance rankings. We ranked schools from 1, the lowest quintile of performance, to 5, the highest quintile. In our analysis, a higher-performing school, ranking a 3, 4, or 5, was only considered a viable transfer option if it was at least two quintiles above a lower-performing school, ranking a 1 or 2. A school ranked a 3, for example, is only considered a viable option for students in a school ranked a 1. These limitations follow with good interdistrict choice policy design—policies that target choice to students who attend the lowest-performing schools and ensure transferring students move into substantially higher-performing schools.

If we remove this restriction and allow students in schools ranking 2 to attend schools ranking a 3, for instance, the number of students who could benefit from increased choice will increase by a few percentage points. Combined interdistrict and intradistrict choice among students in California grade three schools, for example, would increase from 11.9 percent to 12.1 percent under this changed assumption. Similarly, if students in schools ranking a 1 are allowed to attend schools ranking a 2, choice would expand further. Alternately, if students in higher-ranking schools were also allowed to choose, the competition for space in higher-performing schools would increase, thereby decreasing the number of spots available to each school and decreasing the percent of students in low-performing schools with available choice.

Impact of Assumptions: Our assumptions—limiting choice to a 20 minute driving radius, assuming higher-performing schools can expand their capacity by 10 percent, and limiting choice to schools with at least two quintiles difference in performance—necessarily limit the percent of students who can transfer, in addition to any limitations due to the geographic distribution of schools. To isolate the impact of geography, we calculated the percent of students in each state who could transfer, without using the GIS analysis, by simply taking 10 percent of the enrollments in higher-performing schools and calculating those slots as a percent of total enrollment in lower-performing schools. This allowed us to determine the impact on choice of driving distance and other geographic limitations that were included in the GIS analysis.

For California schools, geography only slightly lowers the percent of students with the option to transfer among grade three and grade seven schools. Among grade 10 schools, geography has a greater impact, lowering available choice by several percentage points. In both Texas and Florida, geography lowers the percent of students who could transfer by an average of 4.5 percent.