School choice advocates have long agreed on the importance of understanding what parents value when selecting a school for their children. A new study from Mathematica seeks to add to that conversation and generally finds the same results as prior research. What makes this study relatively unique, however, is that its analysis is based on parents’ rank-ordered preferences on a centralized school application rather than self-reported surveys.
To analyze preferences, researchers utilized data from Washington, D.C.’s common enrollment system, which includes traditional district schools and nearly all charters. D.C. families that want to send their children to a school other than the one they currently attend (or are zoned to attend) must submit a common application on which they rank their twelve most preferred schools. Students are then matched to available spaces using a random assignment algorithm.
The study tests for five domains of school choice factors: convenience (measured by commute distance from home to school),[1] school demographics (the percentage of students in a school who are the same race or ethnicity as the chooser), academic indicators (including a school’s proficiency rate from the previous year), school neighborhood characteristics (crime rates and measures of residents’ socioeconomic status), and other school offerings (including average class size, uniform policies, and the availability of before- and after-school care). Findings suggest that, of the five factors, convenience, school academic performance, and student body composition are the most predictive of how parents rank school alternatives. (The analysis focuses on only entry grade levels—pre-K and kindergarten for elementary schools, grades five and six for middle school, and grade nine for high school—since these are the most common levels for which families submit applications.)
In terms of subgroup breakdowns, the economic status of choosers impacted preferences for elementary and middle school applicants, but not high school applicants. For example, in elementary school, higher-income applicants preferred schools with high percentages of students of the same race and lower percentages of low-income students; low-income applicants didn’t share the same preferences. In middle school, both low- and higher-income applicants were influenced by school academic performance; however, low-income choosers focused on school proficiency rates (which were observable on the application website) while higher-income choosers were more influenced by accountability ratings (which were not immediately available on the application site). Breakdowns for the three largest race/ethnicity groups (white, Hispanic, and African American) in elementary school showed that while white choosers preferred schools with larger percentages of students from the same racial group, African American choosers “essentially showed indifference for own-group racial composition.” In middle school, however, all but the Hispanic group of applicants had a “pronounced own-group preference and a slight preference for diversity.”
To round out their analysis, the researchers use their model to predict how parents would rank schools under alternative scenarios. For example, if capacity constraints were eased so that more applicants were able to attend their most preferred schools, enrollment in high-performing schools would increase and segregation by race and income would decrease. Closing the lowest-performing schools would also increase enrollment in high-performing schools and decrease segregation.
Overall, while there are limitations to this particular study and others like it, it’s a valuable analysis of what parents look for in schools—and the importance of expanding their options.
SOURCE: Steven Glazerman and Dallas Dotter, “Market Signals: Evidence on the Determinants and Consequences of School Choice from a Citywide Lottery,” Mathematica Policy Research, (June 2016).
[1] To ensure accurate distance measurements between schools and residences, the study restricts the student sample to only lottery applicants with a valid D.C. address (approximately 98 percent of applicants).
