School voucher programs, which allow eligible families to send their children to private schools with the help of public funds, have sparked controversy since the first such initiative was launched in Milwaukee in 1991. Today, 28 states and the District of Columbia (D.C.) operate 54 private-school-choice programs, which include not only government-issued vouchers but also tax-credit scholarships, education savings accounts (ESAs), and town-tuitioning programs for rural families. In spite of, or perhaps because of, the spread of such programs across the country, the debate surrounding their merits continues. Fortunately, many studies on the outcomes of private-school-choice initiatives have enabled us to begin evaluating their effectiveness. While the jury is still out on the effects of these programs on student test scores, there is significant evidence that they positively influence how far students continue in their schooling.
Private-school-choice programs disproportionately attract students from disadvantaged backgrounds. Choice participants are considerably more likely to be low-income, lower-achieving, and African American, and much less likely to be white, as compared to the average public-school student in their area. Moreover, 12 percent of the 446,000 participants in private-school-choice programs in 2016–17 were in initiatives limited to students with disabilities, which is slightly higher than the 11 percent average rate of student disability in public schools nationally.
These participation trends are not surprising, since most voucher programs are targeted to low-income urban students or students with disabilities. Even in the Indiana Choice Scholarship Program, however, which is open to low- and middle-income families statewide, the percentage of low-income students enrolled is slightly higher than their percentage of the overall K–12 population.
The private schools that participate in choice programs also are distinctive. Yujie Sude, Corey DeAngelis, and I examined the patterns of private-school participation in choice programs in D.C., Florida, and Louisiana. Private schools were more likely to participate if the gap between their tuition level and the usually lower voucher amount was smaller, if they already had experience serving disadvantaged students, and if they were Catholic schools. Stringent regulations appear to dissuade some schools from opting in: more than two thirds of private schools participate in the relatively low-regulation Florida Tax Credit Scholarship Program, while only one third participate in the relatively high-regulation Louisiana Scholarship Program. Brian Kisida, Evan Rhinesmith, and I surveyed private-school leaders who declined to participate in their state’s choice program and found that most of them feared that regulations would increase in the future. They also viewed certain regulations as restricting their independence and organizational identity, especially mandates involving curriculum and requirements to administer the state accountability test to their choice students.
Douglas N. Harris, in his contribution to this forum, states that even religious schools willing to participate in school voucher programs “often have academic and behavioral admissions requirements.” Only three voucher programs—those in D.C., Indiana, and Ohio—permit participating private schools to apply test-based admissions standards to applicants using vouchers. None of the U.S. voucher programs permit schools to deny admission to students based on their disciplinary records. Private schools can decline to participate in voucher programs, but if they agree to serve students on vouchers, in most cases they must accept all comers.
The effects of private-school-choice programs on the achievement of student participants have been extensively studied using a variety of research designs. Sixteen evaluations of eight programs in Charlotte, North Carolina; Dayton and Toledo, Ohio; D.C.; Louisiana; Milwaukee, Wisconsin; and New York City used “gold standard” experimental designs. Danish Shakeel, Kaitlin Anderson, and I conducted a meta-analysis of the 16 experimental studies, finding that the private-school-choice programs evaluated in the United States have increased student achievement by an average of .13 standard deviations in reading by the fourth year after the study started. In total, programs have had no significant effect on average math scores. The reading effect represents a gain of about four months of learning, depending on student grade level and background. The achievement effects from school-choice experiments follow a consistent pattern. They begin slightly negative, then turn positive and cumulate over time (see Figure 1).
Four recent non-experimental studies of choice programs also tended to report positive effects in reading achievement, with some qualifications. David Figlio concluded that the Florida Tax Credit Scholarship program had a statistically significant positive effect on reading outcomes for students close to the program’s income eligibility cutoff. Mark Berends and colleagues, as reported in their essay for this forum, found that students who persisted in the Indiana Choice Scholarship Program for four years experienced reading gains. I led a research team that concluded that the combination of access to school vouchers and a high-stakes-testing policy boosted the reading test scores of students in the Milwaukee Parental Choice Program. An evaluation of the Ohio EdChoice Scholarship Program, conducted by David Figlio and Krzysztof Karbownik, was an exception, reporting negative effects of that voucher program on both reading and math scores that persisted over time.
That study, while reporting negative achievement effects for participants in Ohio’s largest voucher program, also found that students remaining in public schools performed higher on tests, owing to program-induced competition. Their study is the 15th evaluation of the competitive effects of vouchers to report consistently positive results. Six other such studies reported that competitive pressure from vouchers had effects that ranged from neutral to positive. Only one study, conducted by Jay Greene and Marcus Winters and focusing on the D.C. voucher program, found that voucher competition had no effect on the test scores of non-participants, while no empirical study of acceptable rigor has found that a U.S. private-school-choice program decreased the achievement of public school students.
The effects of choice programs on educational attainment—how far an individual goes in school—are both larger and more consistent than their achievement effects. Attainment is typically measured by benchmarks such as high school graduation, college enrollment, persistence in college, and college graduation. Higher levels of educational attainment are associated with a longer, healthier life; higher lifetime earnings; and lower probabilities of divorce, welfare receipt, and incarceration.
Fewer choice studies have examined attainment than achievement because doing so requires tracking students for many years. The five studies undertaken so far all report positive effects of private-school-choice on attainment for all participants or key subgroups, and these effects are both statistically significant and substantively large. An experimental study I led for the U.S. Department of Education of the D.C. Opportunity Scholarship Program found that using a voucher increased the likelihood of high-school graduation by 21 percentage points, representing a 30 percent boost. In a similarly rigorous experimental evaluation, Matthew Chingos and Paul Peterson reported that participating in the New York City private-school scholarship program increased college enrollment rates for African American and Hispanic students by 6 percentage points, which represented a 10 percent hike. The program also increased those students’ college-graduation rates by 3.5 percentage points, an increment of 35 percent. In a non-experimental analysis, Chingos and Daniel Kuehn found that participation in the Florida Tax Credit Scholarship Program increased the student rate of college enrollment by 15–43 percent, depending on how many years the individual used a scholarship. Two non-experimental studies of the Milwaukee Parental Choice Program reported that it increased high-school graduation rates, but by smaller amounts than the programs in D.C., New York, and Florida. One of those studies, by my research team, also followed students into college, finding that voucher students enrolled and persisted in four-year colleges at higher rates than their matched public-school peers. The fact that Milwaukee voucher students advanced through their college years at better rates than the comparison group indicates that their higher high-school graduation rate was not driven by possibly-lower diploma standards in the private-school sector.
Why Do Effects Vary?
Private-school choice begins with a school switch for all participants except rising kindergartners who did not attend a preschool. Eric Hanushek and Steven Rivkin have established that student achievement tends to drop the year after a school switch, as students adjust to their new schools. Such a decline is likely larger for voucher students who move to a private school immediately after a choice program is created, because the schools also have to adjust—to an influx of new, disadvantaged students. The achievement effects of choice programs after just one or two years may well turn out to be misleading indicators of the longer-term effects on test scores and attainment. Parental choice is a commitment to a journey that takes time to deliver clear learning benefits to students.
It should not surprise us that private schooling boosts student attainment more than it does test scores. Most private schools focus on educating the whole child: intellectually, behaviorally, and spiritually. For example, the Alliance for Catholic Education program based at the University of Notre Dame, which trains college graduates to teach in Catholic schools, speaks of “preparing students for college and heaven.” Although research on the question remains at a formative stage, private schools tend to focus on molding student character by fostering grit, conscientiousness, and tolerance of others. Such character traits are more predictive of educational attainment than of future educational achievement. The attainment effects of choice programs may be outstripping their achievement effects because private schools prioritize character over test scores. That prioritization of long-term over short-term outcomes likely pleases their customers: parents.
The evidence also suggests, though by no means conclusively, that voucher programs targeted to low-income urban students have larger and more consistent positive effects on participants than do statewide programs that are less narrowly targeted. Because of the entrenched practice of assigning students to public schools based on their neighborhood of residence, urban public schools tend to concentrate highly disadvantaged students in schools characterized by low levels of safety and achievement. Prior research by William Howell and Paul Peterson suggested that the reason low-income inner-city African Americans benefit most from private-school choice is that moving to the new school represents a more dramatic improvement in the school environment for them than for less-disadvantaged white and Hispanic students.
Statewide choice programs are too new to generate a clear comparison with the more established urban voucher programs. Statewide programs in Florida, Louisiana, and Ohio, however, already have demonstrated clear positive effects on the achievement of students who remain in public schools, confirming Caroline Hoxby’s claim (see “Rising Tide,” features, Winter 2001) that competition from choice generates “a rising tide that lifts all boats.”
We know precious little about what makes some private-school-choice programs more successful than others—and success itself can be defined in various ways. A choice program can be called a success if it serves a large number of students, attracts a high percentage of private-school providers, improves racial integration in the community’s schools, increases test scores of participants, improves the test scores of non-participants, boosts student attainment, or enhances student civic values. Different program designs are likely to favor some of these desirable outcomes at the expense of others. There is great risk in thinking, with so little evidence, that we know exactly how to design voucher programs to optimize student outcomes.
Still, I will offer a few humble suggestions. Voucher programs narrowly targeted to income-disadvantaged urban students reach a particular student population that appears to benefit most from access to private schooling. Meanwhile, programs broadly available to both low- and middle-income students statewide attract a diverse and likely higher-quality set of participating schools. Combining these two features in creative ways, such as by providing higher-value vouchers to lower-income students, might be the best way to match disadvantaged students with a wide array of private schools to serve their educational needs. Education savings accounts could prove to be the most effective mechanism for delivering private-school choice: they provide parents with more flexibility to customize the educational experience of their child, potentially drawing from multiple schooling providers. The ESA model also encourages parents to obtain the best value for their child’s education dollars, as unspent money in one year can be rolled over to the next and even be spent on college costs.
A specific debate rages over what forms of government accountability to impose on private schools participating in choice programs, which already are accountable to parents, who can vote for or against them with their feet. There is merit to the arguments on both sides of this dispute. My main concern is that, in trying to perfect private-school choice, we could accidentally destroy it. Policies requiring private schools to administer the official state tests, which are aligned with the public-school curriculum, appear to discourage distinctive private schools from participating. Such policies also create incentives for schools that do accept voucher students to change their educational programs to match what the state tests. The fundamental purpose of vouchers is to permit parents to choose from among a diverse array of educational models for forming their children into successful adults. If policymakers impose regulations that limit the range of choice—either by disqualifying certain types of schools or by encouraging uniformity in curriculum and school identity—they will ultimately narrow family options rather than expanding them. That would be a regrettable choice.
This is part of a forum on private school choice. For alternate takes, see “Still Waiting for Convincing Evidence,” by Douglas N. Harris, or “Lessons Learned from Indiana,” by Mark Berends, R. Joseph Waddington, and Megan Austin.