What we know about the effects of school choice

A pretty African American university woman reading at the parkWhile there has been a tremendous expansion of school-choice programs, Dan D. Goldhaber, The Urban Institute, and Eric R. Eide, RAND Corporation, contend that the empirical evidence on the academic effects of choice programs is mixed and incomplete.

In their study of existing research, they use the term choice to refer to any policy that enables students to attend schools other than their neighborhood school. Within the public-school arena this refers to magnet and focus schools, open enrollment, charter schools and inter-district choice of schools.

There are also three publicly funded voucher programs in Milwaukee, Cleveland, and the state of Florida, and a number of privately funded voucher programs that subsidize the enrollment of students in private schools. Goldhaber and Eide point out that charter schools are public schools using public funds, but unlike magnet schools they are generally subject to less regulatory control than other public schools and so have more freedom to innovate.

The most compelling argument in favor of school choice is that it provides an alternative to traditional public schools that are failing to educate students adequately and puts pressure on public schools to improve. Although results are mixed, these researchers report that there is evidence of positive effects for minorities, particularly African-Americans, living in urban areas.

Studies reveal that both urban and suburban families that choose schools are more likely to say they trust teachers and more often are involved in parent-teacher associations and volunteer programs at the school.

Evidence of benefits to minority students

Choice among public schools has been around longer than voucher programs, and there is evidence that, at least in high-poverty districts, minority students sometimes demonstrate dramatic improvements in achievement when they are able to choose their schools. But there is still controversy about these achievement increases in choice schools.

For example, some researchers point out that the successful choice program in New York City’s District 4 imported higher-achieving students into the district, and that leadership, resources and school size also played roles in improving students’ scores.

A study of open enrollment in Chicago revealed that being able to choose a particular school led to higher graduation rates for some groups of students but did not have much impact on grades.

Promotes competition among schools

Arizona, with the most charter schools of any state, found that greater competition appears to create some changes in school principals’ behavior and district practices. Following the development of charter schools, districts began to advertise their strengths to their communities, and some instituted new curricula.

This was particularly true for lower-performing districts that served large numbers of minority students, but the study did not assess the impact of the charter schools on academic achievement.

In Michigan, the state’s standardized testing program has not shown improvement in charter-school students’ scores, and the availability of charter schools has had almost no effect on the test scores of neighboring public schools. Goldhaber and Eide point out that charter-school laws differ significantly from state to state, and no state’s charter schools have shown
consistently significant positive effects on students’ achievement.

Selection bias in studies

One of the major problems with the research data on choice programs is selection bias. Few studies compare randomly selected groups of students, despite the fact that such randomization is essential for accurate, unbiased results. Because families choose whether to participate in non-neighborhood public schools, it is difficult to randomly select students to compare.

Families (and therefore their children) who choose options other than their neighborhood school may be significantly different from families who do not.

For example, non-experimental (non-randomized) early studies found that Catholic schools were more effective than public schools at educating students. They appeared to be better at equalizing educational opportunities for students of differing racial, ethnic and socioeconomic backgrounds. However, there has been a lot of criticism of the statistical techniques and the conclusions of this research, mainly because it did not control for selection bias.

Follow-up studies found that when the previous academic achievement of students was taken into account (in an attempt to control for selection bias), the outcomes in favor of Catholic schools largely disappeared. When populations were sorted by ethnicity, however, positive effects continued to exist for urban minority students.

In particular, African-American students in large cities fare much better in terms of test scores at private schools than at public schools. And when controlling for socioeconomic status, researchers found modest gains for Catholic schooling for low-income, urban white students but not for suburban white students.

Need to rule out individual characteristics of students

It is difficult to distinguish the effects of competition from other factors influencing students’ educational achievement. The main issue for researchers is controlling for individual characteristics that might influence achievement and therefore distort the effects of choice alone.

To rule out the effects of individual characteristics, experiments are needed in which students are randomly assigned to either a choice school or the neighborhood school. If random samples are used, there is no need to account for differences in individual characteristics. Research using experimental data can more easily attribute differences between groups to program effects.

Recent interest in voucher programs to improve education has resulted in a number of both publicly and privately funded projects. One publicly funded program, the Milwaukee Parental Choice Program, provides some unique data that can be used to evaluate the effect of private-school attendance on student achievement.

This program focuses primarily on low-income, urban minority students. Beginning in the fall of 1990, private nonsectarian vouchers were available to families whose income did not exceed 1.75 times the national poverty limit. Schools were not allowed to discriminate on the basis of race, ethnicity or prior academic performance, but could exclude students with an identified disability if schools felt they did not have adequate services to meet their needs.

Schools could limit their choice students to 49 percent of their enrollment (65 percent after 1994). This data is still only quasi-experimental, because while students were randomly selected by schools, families still had to make the decision to apply. Results are mixed and yield no conclusive effects.

One study (Greene, Peterson, and Du, 1998) compared students who participated in the choice program and attended private schools with those who applied but were rejected because of full subscription through a process of random selection. This comparison avoids the problems associated with selection bias.

Researchers assume that there are no inherent differences between the group that attended and those who applied but for whom there was no room, because the selection was random. The authors of this study found little evidence of difference between private and public-school performance in the first two years of the program, but they found a large private-school
advantage starting in the third year.

Controlling for family background, researchers estimate a statistically significant private-school advantage on standardized tests of seven percentile points in math and six percentile points in reading in year three of the program.

Do private schools benefit students?

Evidence on privately financed experimental voucher programs in New York City; Dayton, Ohio; Washington, D.C.; and Charlotte, North Carolina, suggests that private schooling leads to dramatic improvements in achievement for at least some students. Although these programs are relatively new and have slightly different eligibility requirements, all of them are targeted toward low-income students receiving free or reduced-price lunch, and all offer modest subsidies.

The most generous offers $1,700 per year. Each program established a lottery system to randomly select students for available spaces, and this provided an opportunity to evaluate the programs on an experimental basis. Surprisingly, private schools tended to benefit some types of students more than others. Positive effects have been found more consistently
for African-American students who participated in the program for at least two years.

Therefore, while private schools have not demonstrated much greater overall efficiency than public schools, they can be very beneficial to certain students. The question is why private schools benefit African-American students more than other minorities. Researchers speculate that African-American students’ public-school experiences may differ in important ways from other students. It is also possible that language barriers impede the ability of other minorities to take advantage of private-school settings.

Opposition to choice programs

Those who oppose choice believe that these programs will lead to skimming the best students and teachers from public schools, leaving behind those most in need of help and ultimately resulting in greater disparities in educational achievement.

Since only a portion of tuition is covered by any voucher system, opponents say the poorest students will always be excluded. In addition, parents have widely differing access to and ability to use information about choice programs. Opponents also suggest that the effects of peer influences may be important to results and they speculate that as private schools take in more and more diverse students from public schools, they will become more like public schools and the differences in achievement will decrease.

Goldhaber and Eide report that research on choice programs would benefit from having more fine-grained distinctions of race, ethnicity, achievement and income to ascertain different effects of choice programs. In addition, all the data available currently is based on averages and shows little evidence of generalized positive or negative effects.

These researchers believe that comparing different types of students (for example, students who fall along different points of the achievement spectrum) would help to determine how choice programs affect them. Research demonstrates that the forms of school choice that have long existed within public schools system can lead to some improvements in students’

The implications of the newer forms of choice — charter schools and vouchers — are less clear, but these researchers conclude that school choice holds promise for improving educational outcomes for urban, minority students in particular.


“What Do We Know (and Need to Know) About the Impact of School Choice Reforms on Disadvantaged Students?” Harvard Educational Review Volume 72, Number 2, Summer 2002 Pp. 157-176.

Published in ERN October 2002 Volume 15 Number 7

Leave a Reply

  • (will not be published)