Rich or poor, black or white, a family’s decision of where to enroll their child in school is one of the most important, gut-wrenching, and revealing choices they can make. In DC, parents can choose from over 200 charter and district schools. By analyzing that data for a recent study, we were able to shed some light on what drives parents’ choices.

Images from the study.

What we analyzed

As in other cities, any DC student in kindergarten through grade 12 has a right to attend a neighborhood public school based on his or her home address. But students can also enter a lottery for an open spot at any neighborhood school or public charter school in the city.

In 2014, DC moved from separate lotteries for each school to a common system, MySchoolDC, where applicants rank up to 12 schools. A random lottery process chooses which students get the spots at any school which has more applicants than spaces.

In a recent study for Mathematica Policy Research we had the opportunity to analyze over 20,000 rank-ordered lists that parents submitted in 2014, the first year of the unified lottery. We combined these lists with data describing characteristics of the students and their families, the schools themselves, and information on household and school neighborhoods, including crime and demographics.

This allowed us to estimate the importance parents place on different school attributes, including commuting distance, transit access, test score proficiency rates, programmatic offerings, the school’s racial-ethnic composition, the percentage of disadvantaged students, school neighborhood characteristics, and a variety of other factors.

DC is a city of liberal values and unconscious biases, of racial diversity and racial tension, of rich and poor, newly arrived and long-time residents. Parents’ individual decisions add up to collective social outcomes, patterns of racial and class composition, that have lasting effects on the social fabric of DC.

Nikole Hannah-Jones wrote a moving piece in the New York Times capturing these themes as they played out in her gentrifying neighborhood in Brooklyn. The same issues come into stark relief in DC.

What parents value when choosing schools

The analysis suggests that parents do prefer schools close to home, but (not surprisingly) they are willing to tolerate a longer commute to a school with higher test scores.

The preference for academic performance was quite strong: if two schools were identical in every way except for their “tier” rating, parents would travel an average of seven miles for a school in the highest category over one with the lowest.

Academic performance was not the only factor. We also found that parents choose schools based on the race and income of students, but did not weight that as strongly.

Parents tend to rank schools higher if there are more students in the same racial or ethnic group as their own children. But the strength of this “own-group” preference differs by grade level, the applicant’s race/ethnicity, and the percentage of a school’s students in the child’s own group.

If the own-group percentage is low, parents show a strong preference against a school. But as the percentage rises, the relationship weakens and even becomes negative, suggesting a taste for diversity.

In short, parents on average seem to want their children to not be in the vast minority at their school, but as long as there are some students of similar racial or ethnic backgrounds, this stops being a priority. (More detailed results are available in the technical paper here).

The analysis found that typical middle school parents would be willing to send their children over two miles farther just to get from a school where 10% of students share the child’s race/ethnicity to one with 20%. But if choosing between schools with 40% or 50% of the same race/ethnicity, they would only be wiling to travel a half mile more to school.

Does choice affect segregation?

One fierce debate in education is about whether school choice — the policy allowing families to select a school besides the local one — worsens segregation. Some people may opt out of higher-poverty schools or those with high numbers of racial and ethnic minorities. Does this entrench segregation? Or is the segregation already there in residential living patterns?

We compared the current levels of racial and income segregation in DC to alternative scenarios where everyone got his or her first choice (which would unrealistically require some schools to be far, far larger) and where everyone went to his or her neighborhood school (though in reality, not everyone would attend that school in the absence of choice; some families would move or choose private schools).

It turns out that the existing policy results in less segregation by race for middle schools than if every student simply attended the in-boundary neighborhood school.

What if everyone could attend their most preferred school (ignoring any space limitations)? This would not increase racial segregation. Rather, the analysis showed nearly the same amount of racial segregation as the current policy or perhaps slightly lower.

We repeated this exercise, but removed the lowest-performing schools as school choice options to simulate a policy that directs more students toward high-performing schools. In this case, we found again that racial segregation would not increase under these circumstances, but fall further, to a value of 68 on a scale where 100 is the most segregated and 0 is the least

In addition to racial segregation, we looked at segregation of students by income level (low-income versus non-low-income). using the same type of metric, this time with students who are certified as low-income versus all other students. The overall segregation level by income was only 41 under the current policy, but interestingly, that level is even greater than it would have been if students had simply attended their neighborhood schools.

However, we found that if everyone could attend their most preferred schools, it would result in segregation by income roughly equivalent to a policy of no school choice (32 points, just one point lower than with neighborhood schools).

These findings for race and income segregation looked slightly different for families applying to elementary and high school, where the context is different in terms of both the applicants and the diversity of schools they are applying to. For example, white and Hispanic families’ own-race preference was stronger among applicants to elementary schools, as emphasized in a recent Slate article, than applicants to middle and high schools.

Nevertheless, choice policies that expand seats at popular schools were predicted to reduce segregation by both race and income in elementary schools. For high schools, neighborhood school assignment was predicted to lower both types of segregation compared to the current policy (school choice with a lottery for oversubscribed schools).

Again, choice with no cap on the number of accepted applicants and removal of the lowest-performing schools always results in the lowest indices of both race and income segregation (this assumes it would be possible to increase capacity at individual campuses).

What is consistent between all levels of school is that policies which let all students into their first choice (the two blue bars in the graphs above) led to the lowest segregation. For high schools, putting everyone into the neighborhood school (purple bar) also lowered segregation compared to the current policy.

Parents in DC are not race-blind, nor do they ignore the socio-economic status of their children’s potential peers. But they also are sensitive to distance and indicators of academic quality. There are also numerous unmeasured determinants of choice.

Based on the data we have available, though, we don’t see evidence that the worst fears of choice opponents are true. That is, we don’t see evidence that school choice by itself worsens the level of school segregation produced by residential patterns.

However, we also don’t see choice as a very powerful mechanism for voluntary desegregation. There remains much work to be done to understand the impacts of choice on equity and access to quality schooling for the most disadvantaged. We also need to better understand how disadvantaged families access information about school options.

The model in this study, however, provides a promising tool for leveraging data to predict the effects of policy changes on sorting of students across schools throughout the city.

Steven Glazerman is an economist who studies education policy and specializes in teacher labor markets. He has lived in the DC area off and on since 1987 and settled in the U Street neighborhood in 2001. He is a Senior Fellow at Mathematica Policy Research, but any of his views expressed here are his own and do not represent Mathematica.

Dallas Dotter is an economist whose research is primarily focused on education policy. He currently lives in Oakland and is a researcher at Mathematica Policy Research. Any views expressed here are his own and do not represent those of Mathematica.