What's working (or not), why, and how?

An initial assessment of any study's potential contribution to policy should begin with three questions: What is being studied? Why is it expected to work? How is program performance assessed?

All choice proposals are predicated on the notion that giving public school students an exit option will in some fashion improve education. Similarities end here. What varies from program to program, theory to theory, and thus study to study, is who is allowed to exit, how they exit, and where they can go with a public subsidy. Even programs with similar conceptual underpinnings can vary greatly in specifics. The high-profile voucher programs in Milwaukee and Cleveland have similar goals, yet they differ significantly in eligibility requirements, funding mechanisms, voucher values, enrollment caps, and restrictions on participating service providers.¹⁰

These details matter because choice is a highly localized policy phenomenon, and the specifics of any given program sets boundaries on conclusions. What happens in Cleveland cannot necessarily be used to predict an outcome in Florida, and findings in Milwaukee might have little relevance for Colorado's voucher proposal. Does choice work? The question cannot be answered without knowing specifically how choice is being defined and implemented.

Choice programs can be sorted into a handful of useful categories. “Controlled choice” programs are typically implemented at the district level and allow parents the freedom to choose from among public schools. Controlled choice does not undermine the basic institutional structure of traditional public education, in which a public authority exercises centralized control over a system of taxpayer-funded schools within a designated jurisdiction. It differs from the traditional system only in that parents may choose which public school a student attends. Though the concept of a traditional neighborhood school is diminished, in practice these programs often constrain choice on the basis of space availability and the racial or demographic balance of a school. Controlled choice plans are common, not infrequently associated with magnet programs and desegregation efforts, and tend to be the least controversial choice plans.

“Option demand” programs retain the basics of the traditional public school institutional structure and the neighborhood school, but the public authority also creates a set of nontraditional educational options. Option demand allows parents to opt out of the traditional, assigned school and choose a range of alternatives created by the local school district. A well-known example of option-demand choice is District 4 in New York City's East Harlem, which created a group of “mini-schools” with varying pedagogical orientations. Charter schools operating under the umbrella of local education authorities also fit into this category.

“Universal choice” programs eliminate the concept of the neighborhood school and extend choice to private as well as public schools. Though most private schools are subject to some degree of government regulation, they are not accountable to public school authorities. They are not bound by many of the reporting, personnel, and other regulations that apply to public schools; they are not assigned or guaranteed students; and their governing bodies are not elected by public ballot. The hallmark of a universal choice program is that the public sector underwrites the purchase of educational services while exercising little control over who supplies those services. Voucher programs, such as those in Milwaukee, Cleveland, and Florida, are the best-known form of universal choice.¹¹

Also fitting into this category are the more loosely regulated charter school programs (such as Arizona's) that license a variety of groups to establish schools and compete for students and the tax dollars they represent with little or no regulatory oversight by local education authorities. In such programs, parents are directly (with vouchers) or indirectly (with tax dollars that automatically follow students) given the resources to purchase educational services, and schools must compete for those resources.¹² Universal choice plans, especially those in which religious schools and for-profits are allowed to compete for tax-funded education subsidies, are the most controversial choice proposals.

Why do some expect that providing an exit option will improve education generally? Of course, the option of school choice cannot by itself increase test scores or lower dropout rates. Does choice somehow affect teaching, learning, or the broader educational environment? What we are talking about here is “merely” theory, but theory explains connections, and for applied policy considerations there is no more practical information than understanding the relationship between implementation of a program and what happens next. Policy makers need to know how program X and outcome Y are correlated: theory lets us understand correlation as causation. Without theory, it is hard to assess the evidentiary basis for regarding a program as a success or a failure. William Howell and Paul Peterson, for example, report finding a positive achievement gain for African Americans—but not Latinos or whites—participating in a privately funded voucher program.¹³

This is described as evidence of the positive benefits of vouchers. Yet what, exactly, did the voucher schools—and more specifically the presence of vouchers—do to cause such gains? Why were the gains racially confined? Is this something specific to the voucher approach? We do not know. There is no satisfactory causal explanation for the reported correlation.¹⁴

In addition to the what and why, it is also important to ask how, that is, how is success or failure being assessed? The default dependent variable for many education studies is a test score measure (one review found roughly three-quarters of 377 studies used test scores as their dependent variable).¹⁵

While much is made of achievement scores, they represent only a narrow dimension of education. The same program that increases test scores may also increase dropout rates.¹⁶ Increasing civic engagement among students may also decrease their tolerance for dissenting views.¹⁷ Satisfied parents do not necessarily mean higher-achieving children.¹⁸ Which outcomes are analyzed and how they are measured can themselves determine the results of a policy study.¹⁹ In short, what constitutes “success” or results in a “better” school is very much in the eye of the beholder, and whether a program works or fails depends on what facets of education are considered politically important. Whatever choice does or does not affect, consumers of choice studies are well advised to put dependent variables into context before buying into sweeping policy conclusions.

What's the motivation?

Academics have an “irrational taboo” against publicly confronting the possibility of individual bias.²⁰

This is understandable since most researchers want to be viewed as scientists and scholars rather than shills for special interests. The scientific method and good scholarly practice are superior to instinct, experience, political expediency, or ideological preference as the basis for constructing effective policy. Yet even the most scrupulous and conscientious researcher will fall short of the positivist ideal of applying value-neutral methods and objectively assessing the results.

Does this mean the contradictory messages of the choice research are nothing more than a political catfight in academic garb? It can seem that way. Studies by researchers affiliated with Harvard University's Program on Education Policy and Governance (PEPG) consistently attribute positive effects to choice, especially in its universal variants.²¹

The work of other academics—Jeffrey Henig, Peter Cookson, and myself, for example—tend to consistently cast doubt on the positive impact of choice, especially in its universal variants.²² Are we taking sides? Certainly.

How, then, can consumers trust our research? They should form independent judgments of the veracity of our studies by asking this basic question: Were the data treated fairly?²³

Fair means that the researcher makes known his or her preferences and offers demonstrable assurance that he or she has adhered to scholarly conventions designed to minimize the influence of those preferences. Armed with such a reference point, consumers can be reasonably confident that data are driving a study's policy conclusions rather than vice versa.²⁴

For choice studies these reference points are not always clear, and consumers may lack an insider's knowledge of what “good scholarly practice” means. These problems can be reasonably, albeit imperfectly, dealt with by following a series of simple steps:

What's the method?

Critics argue that if education policies were held to the same standards as medical research, the vast majority would be rejected for lack of evidence.³¹

Choice research is heavily based on variants of regression analysis—the standard statistical tool kit of social science research. Regression has well-known limitations, though.³² The big fear of partisans on both sides of the choice debate is that their opponents will exploit the drawbacks of regression analysis to lie with statistics; after all, a study can be methodologically flawed and still look impressively “scientific” to the uninitiated.

Most choice studies are also correlational rather than truly experimental. There are good reasons for this. The power of experimental research is based on random assignment to “treatment” and “control” groups, but randomly assigning students to different schools or choice programs presents formidable legal, ethical, financial, and logistical problems. A number of choice studies are in fact based on the “gold standard” of randomized field trials, in which students are randomly assigned into treatment (choice) and control (traditional public school) groups. This allows researchers to claim, with some legitimacy, that their results are the product of something really akin to the classic scientific laboratory experiment. Even randomized field trials, however, are several steps from the experimental ideal.³³

For example, by definition those who apply for vouchers—whether they are awarded them or not—are a self-selected rather than randomly chosen group. Of this self-selected group, a significant percentage (up to 50 percent or more) awarded a voucher will not actually use it. Receiving schools may select nonrandomly among these students offered a voucher. Of those who do use their voucher, many will use it temporarily: voucher programs tend to have high turnover among participants.³⁴ When the treatment group gets whittled down to those who are more motivated, or those with fewer academic or socioeconomic problems—a reasonable speculation—it gets harder to separate the independent and generalizable impact of the voucher from the individual characteristics of students.

Thus the “gold standard” of existing randomized field trials is no solution to the problem of extracting reliable policy information from academic studies. How groups are selected, who stays or goes, how schools select students, and how dropouts are methodologically compensated for can all help determine the results and conclusions of a study. In judging contradictory conclusions, consumers should closely examine differing methods and statistical models. If a consumer lacks the statistical skills to make these judgments, the cautions already made become paramount: peer review reduces the probability that data are being “squeezed” to produce ideologically pleasing policy inferences. Replication can also expose points of methodological controversy, even if it cannot entirely eliminate them.

What's the prescription?

John Chubb and Terry Moe issued probably the best known, and certainly the most sweeping, prescription based on a scholarly study of school choice: “We think reformers would do well to entertain the notion that choice is a panacea … Choice is a self-contained reform with its own rationale and justification. It has the capacity all by itself to bring about the kind of transformation that, for years, reformers have been seeking to engineer in myriad other ways.”³⁵

Their advice to engage in a wholesale, market-based transformation of education is remarkable since it springs from a study that did not examine any sort of choice program. Chubb and Moe inferred the effects of universal choice from a study of public and private school differences. Their prescriptive leap was to assume that universal choice would result in more schools maximizing the benefits they associated with private schools while minimizing the problems they associated with public schools. Perhaps choice will do this, but their actual study contains no direct analysis or empirical evidence of any such impact.

Does choice significantly improve student achievement?

No single choice program has been consistently applied in different settings with uniformly positive results on achievement. The case for choice increasing academic achievement rests primarily on studies examining small-scale programs whose results are methodologically debatable³⁸

or are inferred from studies examining public-private school differences.³⁹ Even the studies that do find a positive effect of choice often lack clear causal explanations or unassailable results.⁴⁰ Many of the best-known studies used to support achievement claims have one or more of these characteristics: association with policy centers with a consistent prochoice orientation; sponsorship by organizations with a vested ideological stake in vouchers and choice; publication in non-peer-reviewed outlets; and susceptibility to methodological criticism.⁴¹

The case for choice negatively affecting achievement is similarly weak. For example, my own study with Kenneth J. Meier concludes that the competition among service providers at the heart of a universal choice program ends up suppressing achievement at some schools. Yet, like Chubb and Moe, we do not study an actual choice program; rather, the conclusion is inferred from a statistical study of the correlation between the district-level market share of private schools and public school performance in Florida.⁴²

Replications of this study generally confirm our results,⁴³ though they report more mixed relationships if the sample is split into high and low income schools.⁴⁴ Replicating the Smith and Meier model using Texas data, others find private school market share to have no impact on public school performance.⁴⁵ Critics of Smith and Meier argue that our original model is improperly specified,⁴⁶ which lends more weight to the inference of no impact. That inference summarizes the conclusion that comes through most clearly from the scholarly record: choice is unlikely to have anything but a trivial effect, negative or positive, on student achievement.

A similar conclusion emerges from the controversy surrounding evaluations of the Milwaukee voucher program. In this case there is a single voucher program, so the policy specifics are more or less constant. John Witte, Troy Stern, and Christopher Thorn found no consistent evidence that achievement was different for voucher students and public school students.⁴⁷

Their state-funded research design was based on comparing voucher students with a random sample of public school students. A study funded by the Olin, Annie E. Casey, and Smith Richardson foundations challenged this conclusion, arguing it was based on a flawed methodology: the appropriate comparison group was nonselected voucher applicants.⁴⁸ Reanalyzing Witte's data on this basis, Jay Greene, Paul Peterson, and Jiangtao Du concluded that vouchers did have a significant positive impact. Witte in turn replicated the challengers' study and found, among other methodological problems, that attrition rates were so high in their randomized block research design that comparison groups were quickly reduced to a handful of students. In some cases, the control group N dropped all the way to zero.⁴⁹ Cecilia Rouse reanalyzed the data one more time with an eye toward the thorny statistical issues raised by the previous studies. Her results ended up splitting the difference: she found a small, positive impact in math achievement, but no impact in reading achievement.⁵⁰ Rouse's methodology is top notch (and peer reviewed). Yet the results raise rather than answer a question about the efficacy of vouchers. Why would voucher schools teach math better than public schools, but not reading?

Research on Cleveland's voucher program follows a similar pattern. The contract researcher (Kim Metcalf), funded by the Ohio Department of Education, like Witte, found no firm evidence pro or con. PEPG-affiliated researchers challenged these results; they found positive impacts in studies equally vulnerable to methodological criticisms.⁵¹

Thus in the two most high-profile voucher experiments in the country, research undertaken by academics raises reasonable questions about why the program should (or should not) work, as well as additional concerns about motivation, methods, and paths to publication.

Such concerns are little attenuated by broader research reviews, which largely come to the same cautious conclusions reached by Witte and Metcalf. Brian Gill and his colleagues conclude that few studies show anything other than very modest gains for choice students compared to performance expected in traditional public schools.⁵²

Clive Belfield and Henry Levin report that most studies examining the impact of competition (not necessarily attributable to a specific choice program) on academic outcomes result in statistically insignificant findings.⁵³ A recent meta-analysis considered 18 studies and found no significant effect of vouchers on academic performance, showing them capable of explaining less than one-half of one percent of the variation in test scores.⁵⁴ Given the highly limited findings, any broad or sweeping policy conclusions about the impact of vouchers and choice on student achievement should be treated with due skepticism.

Does choice increase parental satisfaction?

Higher satisfaction levels are reported across different types of choice programs, as well as among private school parents. This finding is consistent regardless of methods, indicators of motivation, path to publication, and the survey instrument used to assess satisfaction. Between 36 and 56 percent of voucher parents award their schools an A, whereas only between 25 and 30 percent of comparable public school parents award an A.⁵⁵

A minor caveat for these results comes in the form of questions about the size of the satisfaction levels and their substantive meanings. Both issues center on the why of satisfaction levels. Does higher satisfaction reflect access to a higher quality education, or something else? For voucher parents a psychological effect independent of school quality that may drive reported satisfaction levels. We are, for example, reasonably certain that those who seek a voucher are more dissatisfied with public schools that those who do not.⁵⁶

In a group of voucher-seeking parents, differences in satisfaction levels are to be expected between recipients and nonrecipients—one group got what they wanted, and the other did not. And even these differences narrow over time; studies show a decrease in satisfaction after the first year.⁵⁷

Does choice increase equality of educational opportunity?

Choice programs can be designed to target underprivileged groups. In Cleveland's program, for example, priority is given to voucher applicants who come from families with a family income lower than 200 percent of the federal poverty line. In Milwaukee, family income below 175 percent of the federal poverty line is a voucher eligibility requirement.⁵⁸

Such targeted plans help to equalize opportunity by underwriting a realistic set of alternative educational options for disadvantaged students. Regardless of researchers' methods, motivation, and path to publication, studies find Milwaukee and Cleveland voucher families to be overwhelmingly minority, more likely to be single-parent households, and more likely to have lower incomes than public school families.⁵⁹

The fact of an exit option for such groups, however, sets a low benchmark for judging policy success. The academic policy literature generally sets the bar much higher, defining equality of educational opportunity in terms of the average outcomes for different population groups.⁶⁰

Thus equality of educational opportunity between, say, blacks and whites is judged to exist when the average educational outcomes (test scores, graduation rates, and so forth) between these groups are roughly the same. The record of choice programs on academic achievement provides little to suggest they are capable of such a leveling effect.⁶¹

Advocates of choice tend to favor the lower benchmark, for there is powerful political incentive to do so. Casting choice as a civil rights argument has a strong appeal, especially in central cities, where public schools often offer a lower quality education compared to wealthier suburban public schools. Indeed, equity arguments are increasingly overshadowing free-market advocacy as the primary justification for choice.⁶²

Advocating choice on this basis carries a moral rather than an ideological imperative; choice is a means to reduce the “savage inequalities” that constrain the social and economic opportunities of disadvantaged and minority students.⁶³ This policy goal, however, does not require universal choice (that is, vouchers), the option most ideologically pleasing to free-market advocates. It can be accomplished by a controlled or option-demand choice program.⁶⁴

The role of universal choice (and, to a lesser extent, option-demand) programs in redistributing opportunity is difficult to determine in part due to the variety in participating students and parents. Voucher students in Milwaukee and Cleveland, though mostly from disadvantaged families, tend to have better-educated parents. Students with disabilities are underrepresented in choice programs.⁶⁵

Furthermore, when the exit option is facilitated not by a targeted voucher, but by a more general subsidy such as a tax advantage, universal choice programs are disproportionately used by middle- and upper-income families.⁶⁶ This makes sense since tax credits are useless to the poor, and it highlights the fact that how vouchers are funded can make a big difference in the nature of the program.

Does choice improve the overall quality of public schools?

Some studies have concluded that choice programs provide real competition to public schools, and that traditional public schools improve dramatically in response to the competition. The best known of such studies, however, are sponsored or funded by organizations with clear policy preferences and have been introduced into the policy debate without peer review. They are vulnerable to criticisms that they do not control for the full range of possible determinants of school improvement.

An example is Jay Greene's analysis of Florida's “A-Plus” statewide voucher program (officially called the Opportunity Scholarship Program), which was contracted through Florida State University as part of an evaluation effort funded by the state.⁶⁷

The report itself was published by the Manhattan Institute and supported by Harvard's PEPG. As a contract report, it was not peer reviewed (such reports are typically reviewed by the sponsoring or funding agencies, or by the principal investigators themselves and their colleagues). In the A-Plus Program the state regularly grades schools on an A–F scale and allows students to attend an alternate public or private school at taxpayer expense if their public school receives two consecutive failing grades. All the schools that received failing grades in the initial stages of this program quickly showed performance improvements. Greene attributed this improvement to the direct threat of losing students to the voucher program. His study attracted considerable national media attention and was praised as evidence of competition's salutary impact on education.

Greene's study and its conclusions were eventually subject to replication and peer review. A version of the report appeared in Education Next—a peer-reviewed publication, albeit run and supported by, among others, the Hoover Institution, PEPG, and the Manhattan Institute. Critical replication and peer review from other scholars came after the report's findings had been widely reported in the media. Researchers challenged Greene's conclusion on both methodological and substantive grounds. A key methodological objection was that Greene failed to adequately control for the statistical phenomenon known as regression to the mean—a tendency for movement in any distribution to be toward the middle rather than the tails. Schools that initially got Fs, for example, are at the bottom of the scale and cannot go any lower. If the school changes at all, it has to go up. When replications of Greene's study addressed these sorts of methodological issues, the impact of the voucher program was greatly attenuated or disappeared altogether.⁶⁸

Scholars also raised objections to Greene's conclusions about “why” performance improved. While acknowledging other plausible explanations for the improvements, Greene concluded that “one should follow the general advice to expect horses when one hears hoof beats, not zebras”—a reasonable point. However, an alternate and equally valid explanation for the observed improvements is the stigma associated with publicly receiving an F—the so-called “red letter” effect. High-stakes testing and accountability systems in other states have resulted in similar dramatic improvements without the voucher threat.⁶⁹

The red-letter impact may simply require a public system of evaluation.

Broader reviews of the academic research on the potential impact of competition show a mixed bag. Belfield and Levin report weak positive relationships between competition and educational efficiency, teacher wages, teacher quality, future earnings, and housing prices (used as a proxy for educational quality because of the strong relationship between property values and school funding). Reviewing numerous studies, they conclude that “[t]here are benefits from higher competition, but the substantive effects … appear to be modest.”⁷⁰

The benefits are so modest, they suggest, that it may not be politically feasible to raise competition to the levels required to have any substantive impact. Finally, whatever competition's beneficial effects, they do not seem to grow appreciably from controlled to option-demand to universal choice programs.⁷¹

Some claim, legitimately, that a stronger impact for competition has not been detected because there is so little real competition in public education. The choice programs that currently exist are too small and too limited to generate the sort of systematic change envisioned by many advocates. Thus it is unknown whether a more comprehensively marketized system of education will produce anticipated benefits. The research on the small-scale programs that do exist provides, at best, very mixed support for such a proposition.

Does choice promote segregation?

Even prochoice theorists concede that programs allowing complete freedom can produce racially, religiously, or socioeconomically segregated schools.⁷²

The private education market is currently dominated by religiously affiliated institutions, as well as including a significant number of high-quality prep schools catering mainly to the wealthy. Critics of choice programs argue that market forces will replicate these race, class, and religious divisions in public education. Proponents, in response, argue that choice programs may play a positive role by promoting integration in a public system that is already segregated because of racial housing patterns.

Academic studies support both points of view. To understand why, we must take into account which type of program is being studied. Controlled choice programs are sometimes adopted with the explicit goal of promoting racial balance in schools, and even targeted option-demand or universal programs may produce some increases in racial diversity in areas of high racial stratification (PEPG-affiliated researchers have studied the latter programs).⁷³

Inner-city minorities who take advantage of a voucher, for example, may leave a racially homogenous public school for a more diverse inner-city parochial school. Even if there is no increase in diversity, furthermore, many small-scale choice programs seem unlikely to decrease diversity in the short term; initial analyses of most option-demand choice programs (such as charter schools) show schools with racial and ethnic distributions roughly equivalent to comparable public schools.⁷⁴

Whatever the type of choice program, in general, the larger and less regulated it becomes, the greater the probability it will exacerbate stratification by race, religion, or class.⁷⁵

This can be seen in the evolution of the Milwaukee and Cleveland programs, where racial enrollment patterns have shifted over time with the programs' expansion. Initially, these programs were dominated by minorities (particularly blacks), with whites proportionately underrepresented compared to public school enrollments. Voucher students went to relatively more white schools, thus increasing levels of desegregation. White participation, however, has increased over time. In Milwaukee, white participation increased significantly after the voucher program incorporated religious schools in 1995. Whites represented 6 percent of voucher students in Milwaukee in the early 1990s and almost 19 percent in the late 1990s. In Cleveland, white students are overrepresented in the voucher population compared to public school enrollments. The data are not conclusive, but are consistent with the expectation that whites will disproportionately participate in voucher programs as they mature, and that students already in private schools will disproportionately benefit from vouchers.⁷⁶

The evidence is also fairly unequivocal that, if allowed to participate, religiously affiliated schools will play a large role in voucher programs. In Cleveland, for example, approximately 97 percent of voucher students attend religious schools.⁷⁷

Such enrollment patterns are controversial because they mean the bulk of voucher program expenditures transfer resources from traditional public to private religious schools. Whether providing greater resources to religious schools promotes religious or racial stratification is largely unknown, though. No studies of programs like those in Cleveland and Milwaukee have examined the possibility of a systematic relationship between individual religious affiliation and voucher program participation. Anecdotal evidence, and systematic evidence from programs outside of the United States, however, suggest that there may be a connection.⁷⁸

I have already addressed the possibility that choice programs promote segregation by class. Even in small-scale, targeted programs it is the “most advantaged of the disadvantaged” that participate.⁷⁹

Greene suggests that this is not a problem because other programs targeted to the disadvantaged (such as food stamps) also have difficulties in reaching the most “disadvantaged of the disadvantaged.”⁸⁰ However, there is a crucial difference between education and other social welfare programs: If I am better off for taking food stamps, you are not left worse off for not doing so. But, if I, an educated and involved parent, remove my child from a school for the sake of a better educational opportunity, all else being equal, the school we leave loses the civic capital I represent. This loss may indeed leave those behind worse off.

Does choice affect the level and focus of civic engagement?

A central justification for creating systems of public education in the United States was to perpetuate the dominant political values and beliefs of society and to promote a civic culture.⁸¹

It is unsurprising, then, that social scientists have taken considerable interest in assessing the actual and potential civic impact of market-based education reform.

Much of this research examines the potential effects of choice programs on social capital, the “features of social life—networks, norms and trust that enable participants to act together more effectively to pursue shared objectives.”⁸²

Robert Putnam distinguishes between bonding social capital and bridging social capital.⁸³ The former refers to the connections between individuals within homogenous groups, such as communities characterized by race, religion, or class; the latter refers to the connections between groups—the social networks that “bridge” diverse social groups and allow them to find common ground. The differences between bonding and bridging capital are important because choice programs seem to boost the former while eroding the latter.⁸⁴

For example, in a series of well-regarded (and peer reviewed) studies, Mark Schneider and various colleagues found that parents in option-demand choice programs were more likely to join the PTA, trust their child's teachers, and talk to other parents. All this suggests that carefully crafted choice programs can boost levels of bonding social capital. Yet they also note that better-educated, wealthier, and white parents tend to accrue the most new social capital in such programs.⁸⁵

Amy Wells and Alejandra Lopez also found that California charter schools promote social capital of the type that strengthens group boundaries. They conclude that choice programs atomize public education into smaller school communities that are separate, inwardly focused, and self-interested.⁸⁶

Values such as political tolerance and civic confidence (an individual's perception of their ability to make a difference through involvement in the public arena), as well as the political knowledge required for productive civic engagement, are the cornerstones of a healthy democratic civic culture. Private school students score high marks on these types of civic sensibility. Advocates of choice argue that this provides evidence that market-based educational institutions are not antithetical to values that can bridge differences across social groups. Opponents agree that private schools are sometimes better at transmitting value systems to students, but find that these values are more sectarian and ideological than egalitarian and democratic.⁸⁷

Again, the evidence is mixed. David Campbell found that the positive civic socialization impact of private schools is primarily a Catholic school phenomenon; secular private schools are not noticeably different from public schools. Non-Catholic religious schools, moreover, “demonstrate a lower level of political tolerance than any other group of students. They also display the highest level of civic confidence—an explosive combination.”⁸⁸

The studies just cited do not examine how any form of choice program will change the inculcation of civic values. In their broad ranging survey of the choice research literature, Gill and his colleagues conclude that there is simply too little research to draw even tentative conclusions about the effect of choice on the transmission of democratic values to individual students.⁸⁹

Reviews by Patrick Wolf and Jay Greene are more hopeful, though still tentative, and they are largely limited to examinations of PEPG-affiliated research.⁹⁰