What Makes Schools Work?
Why vouchers are not the answer

By Richard D. Kahlenberg and Bernard Wasow
Boston Review
November 2003

     There is a strange tenor to the current debate over school vouchers. Conservative groups, not known historically for their commitment to the downtrodden, are evincing a newfound moral passion for liberating poor kids from bad schools. They note that the wealthy can choose high quality schools by purchasing a home in an area with top-ranked public schools or by paying tuition to a good private school, and they argue that poor families should have the same right. This sentiment represents an enormous shift from the days when conservatives led the effort to halt school busing. Then, conservatives championed the "neighborhood" school. Now they point out that the neighborhood school is not such a great deal for children stuck in poor neighborhoods and argue forcefully that everyone should have the right to choose a good school for their children. There are at least three possible responses.

      One response is to join the conservatives. The new emphasis on equity has convinced some liberals that vouchers are the way to go. Joseph Califano, Andrew Young, Robert Reich, Matthew Miller, Arthur Levine, William Raspberry, Martha Minow, and the editorial page editors of the Washington Post have all come around to thinking that vouchers are worth a try. Polls find likewise that young African-Americans support vouchers by more than two to one. A failure to sign on to vouchers is seen by many of these liberal proponents as a simple matter of interest-group politics: powerful teachers' unions won't let Democrats do what is right for poor kids.

     A second response looks more critically at conservatives' motives in seeking school vouchers: are vouchers a tool to lift poor children out of failing schools, or are poor children a tool to lift schools out of the public sector? The idea that conservative promoters of school vouchers are not sincere in their advocacy for the poor is typically prelude to a defense of the broad status quo, including ongoing efforts to improve public schools. This position points out that conservatives aren't serious when they say "Throw open the doors to everyone"; certainly they don't mean that the poor have a right to an equal representation at St. Albans or Andover. Instead, this position dismisses the conservative rhetoric about liberating poor children and looks to traditional approaches to improve neighborhood public schools: reductions in class size, increases in teacher pay, and the like. The leadership of the National Education Association, for example, says Americans "want quality public education in their neighborhood school, and that's what we should be working toward."

     A third line of response sees the conservative pro-voucher argument as an enormous opportunity. However sincere conservatives may be in their new concern for fairness, school choice may indeed be a legitimate means for dealing with the outrage of poor kids trapped in failing neighborhood schools. Vouchers themselves may be wrongheaded, but school choice makes sense and is in fact already taking place within the public school system. Magnet schools, charter schools, alternative schools, and other non-geographically defined schools are growing much faster than any other segment of the public school system, and that trend should be the foundation of future school reform.

     We pursue the third option. Our core thesis is that public school choice, structured specifically to achieve socioeconomic integration, can garner the benefits of vouchers-allowing poor kids to escape bad schools, providing more variety in schooling, and shaking up the bureaucracy with competitive pressures-while avoiding the many pitfalls-increased racial and economic segregation, reduced social cohesion, and increased reliance on unaccountable institutions. Whereas many school reformers have rhetorically embraced "public school choice" as a crisp and convenient rebuttal to the argument for school vouchers, we will suggest that properly constructed public-school choice is the most significant step policymakers can take to improve poor children's school achievement and so must be at the center of public-school reform efforts.

Education and Skills in America

     Voucher advocates deserve credit for creating so much buzz over so little action. Despite all the discussion of vouchers only about 60,000 students currently participate in publicly and privately funded voucher programs nationwide-with only a quarter of these financed by taxpayers. The studies of voucher experiments, which we will review below, involve an even smaller handful of the nation's nearly 50 million primary and secondary school students. The vast majority of American children, 90 percent, continue to attend public schools. But the nature of public education is changing rapidly, with various chosen options-magnet schools, theme schools, alternative schools, and charter schools-growing much faster than traditional neighborhood schools.

     According to the National Center for Educational Statistics, between neighborhood ("assigned") public schools and chosen public schools, 91 percent of children attended public school in 1999, about a half percentage point less than in 1993. Most of these were in traditional neighborhood schools-36 million in 1999, up from 34 million in 1993. Over this period chosen public school enrollment also grew, by about two million students, from five to seven million. Over the same period all private schools, religious as well as independent, added fewer than 700,000 students, reaching a total of about 4.5 million. Voucher hype notwithstanding, we remain a nation of public schools.

     Voucher advocates also must be credited with creating an unjustified sense of crisis in public education. Children in the United States on average perform in the middle compared with children from other industrialized countries, as has been the case for some time. Contrary to what one might think, there is no general negative trend in the performance of American children relative to others internationally, at least for the past two decades. American adults also perform near the average in international comparisons of literacy, with some countries producing better average results (Sweden, for example, and the Netherlands) and others worse (Great Britain and New Zealand).

     Test scores from the United States do stand out internationally in one respect, however, and it has nothing to do with averages: American scores are exceptionally widely dispersed from top to bottom. At the top of the distribution Americans excel, while at the bottom they perform remarkably poorly. For example, Richard Freeman of Harvard and Ronald Schettkat of Utrecht University report that on a test of "quantitative literacy" non-immigrant Americans and Germans performed similarly at 15 years of education (where education is adjusted to include formal apprenticeship training in Germany). But adults in the United States with 12 years of education performed more poorly than those with nine years of education in Germany, while at the top those with 20 years of education in the United States performed substantially better than those with similar training in Germany. The performance of Americans is far less uniform than that of Germans at each level as well, especially among those with less than 12 years of education. In other studies, too, the bottom quarter of the American work force performs at a very low level of cognitive skills, far below both the average level in America and below the bottom of almost all European skill distributions.

     At its best the American education system leads the world in imparting skills. But beyond their contributions to literacy and numeracy, American public schools have helped make Americans out of wave after wave of immigrants, helping to build a common core of values in a very diverse society that is always changing. That said, the public school system fails a substantial segment of the population, and this failure aligns sharply with class and race. Reform must preserve the achievements of the system while correcting the failures.

     We accept the premise that parents must be convinced to buy into any reform agenda. That is one reason why we believe that school choice must be part of the solution. Still, insufficient choice is not the principal problem with our education system. Indeed, we must preserve the effort to build a common core of civic values. The primary goal of school reform must be to provide more equal education opportunities.

     To make the case for school choice, we will start by explaining the limits of traditional public school reform and why vouchers are not the solution to the problem of educational inequality that they claim to be.

Limits of Traditional Reform

     Scholars typically see the outcomes of schooling as depending on qualities of students and qualities of the education process. Traditional school reform has focused on the education process, including expenditures per child (which can affect the number of years and days per year of schooling, class size, the quality of material inputs, and the credentials of teachers), curriculum, and school governance. Qualities of the student apart from the school experience typically are beyond the reach of the school. School reform hardly can be expected to change innate ability, parental education and income, or conditions in a child's neighborhood or home. Therefore, education reform has focused on money and what it can buy, on curriculum, and on accountability and other aspects of governance. Indeed, evidence supports the idea that early childhood programs, reduced class size, teacher experience, and curriculum do affect outcomes.

     This traditional approach to school reform faces two main problems. First, while variables such as class size do affect outcomes, their effect is not great. There is little evidence, for example, that the expenditure of Title I funds, which provide federal budget support to high-poverty schools, have a significant effect on outcomes. Second, the quality of a school is not an independent variable that can be manipulated at will by policy makers. Of course better teachers produce better results, but we find most of the experienced, successful teachers in low-poverty schools. Of course students perform better with encouragement from their peers, but we find such an environment typically in low-poverty schools. Of course schools work better when parents are actively engaged, but we typically find such parents in low-poverty schools.

     The key problem with fixing neighborhood schools by traditional remedies is that bad neighborhoods produce bad schools. Too many good teachers burn out or ask for transfers from schools with high-risk children. High concentrations of at-risk children increase the risk to each child. Traditional school reform has made so little progress in high-poverty schools because high-poverty/low-quality schools are in a self-reinforcing trap.

     This is not to say that traditional reform cannot work in high-poverty schools, nor that bad neighborhoods cannot produce very good schools. But the very fact that the Heritage Foundation, despite years of trying, can list only 21 high-performance, high-poverty schools while the U.S. Department of Education lists 8,600 under-performing high-poverty schools suggests the odds against traditional remedies. Likewise in December 2001 the Education Trust published a study which purported to find some 3,592 high-poverty schools that achieve at high levels, but a reanalysis of the data by Economic Policy Institute researcher Douglas Harris found that students in high-poverty schools perform much less well than students in more affluent schools, particularly when one looks at performance over a period of years. Using the Education Trust definition of a high-performing school (scoring in the top third of the state in either reading or math), Harris noted that there are more than 21,000 low-performing, high-poverty schools. Whereas 18 percent of high-poverty schools are high-performing, 55 percent of low-poverty schools are-three times the rate of success. Because test scores fluctuate from year to year, so that individual years can represent a large number of "flukes," Harris sought to look at which schools have sustained success-for two years, in two grade levels, in two subjects. Under that definition he found that just one percent of high-poverty schools are consistently high-performing, compared to 24 percent of low-poverty schools; that is, high-poverty schools are 24 times less likely to be consistently successful than low-poverty schools.

     Traditional reform fails to consider one obvious candidate for school reform: the mix of students. Public policy cannot easily change the educational attainment of a child's parents and most other aspects of life outside school, but it can change the child's schoolmates. Since a predominantly middle-class school is more likely than a high-poverty school to have good teachers and a large number of active parents, changing a child's schoolmates is a strategy that affects not only the child's peers but the quality of teachers and parent-school engagement as well. Dozens of reports, dating back to the Coleman Report of 1966, find that low-income children have higher levels of achievement and larger achievement gains over time when they attend predominantly middle-class rather than high-poverty schools. Studies have found that the average socioeconomic background of students in a school is as important as a student's own background in determining the student's performance in school, which can translate into success in life. For example, Claude Fisher of the University of California at Berkeley found that, controlling for individual ability and home environment, attendance at a predominantly middle-class rather than a high-poverty school reduces the chances of adult poverty from 14 percent to 4 percent.

Vouchers to the Rescue?

     School voucher proponents leap from the valid premise that traditional reforms of neighborhood schools are insufficient to a radical proposal for privatization. School vouchers place all or part of the provision of education, but not its financing, in the private rather than the public sector. As in the construction of public infrastructure, private vendors would bid to provide goods and services, paid for, at least in part, by tax revenues. Like food stamps, education vouchers would leave the individual household to determine which of the permitted products they purchase.

     Many voucher advocates accept the superiority of vouchers on the basis of a general faith in markets. Since the private sector is assumed to outperform the public sector always and everywhere, voucher schools will be better than public schools. Less ideologically driven defenses of vouchers proceed from our earlier discussion: the public school system is failing at least a segment of the population; traditional remedies have not worked; so let parents decide what is the best alternative for their children. Just as households are expected to manage their purchases of food with food stamps, it is argued, they can be expected to make good choices when they spend their school vouchers. The competitive marketplace will generate new products (schools) in response to the needs of households. Competitive pressure will force every school to deliver or lose its customers. The final line of defense for those who admit doubts to these claims but nevertheless favor vouchers is the claim that "voucher schools could not be worse than public schools."

     These pro-voucher arguments are subject to three important challenges: Will the private sector provide quality schools and will households choose them? Will private schools serve the integrative and community-building functions our nation needs? Is the public really ready to give up on public schools? After considering answers to these questions, we will turn to evidence on the success of voucher programs to date.

Three Challenges

     Some voucher skeptics have expressed doubt that the private sector ever could meet the needs of so many more students. After all, if vouchers could be spent only on nonreligious private schools, a fifty-fold increase in the capacity of that type of private school would be required if everyone were to use vouchers. If religious schools were eligible too, the capacity of private schools still would have to increase by a factor of 10. Clearly, we cannot assume that the existing population of independent and Catholic schools would simply expand to become 10 times their present size if everyone, or even a substantial segment of the student population, used vouchers.

     Surely educational institutions would respond to surging demand. The response to the GI Bill by institutions of higher learning (public more than private, it may be noted) gives some reason for optimism. But while colleges and universities responded to the GI Bill's subsidies to education by offering more and better opportunities, other examples are less encouraging. Trade schools, such as schools of cosmetology and unscrupulous business programs, have expanded their provision of useless training. Their students end up failing to repay their federal student loans, to the tune of $1.5 to $2 billion annually from trade schools alone. These operations often spend more on marketing than on running their training programs. The federal Medicaid program similarly has spawned a host of "Medicaid mills" that prey on poor and ignorant households, claiming to offer services but really only seeking the public dollar. Large-scale voucher programs might lead to the establishment of new schools similar to the (lower-priced segment of) private schools we have today, or they could lead to something quite different. We might get innovative, energetic new educational leaders. Or, we might get a rash of hucksters, hustlers, and bigots, attracted by the smell of public funds. In all likelihood, we will get both, in an unpredictable mix.

     What about civic integration? As we have discussed, Americans rightly expect more from publicly funded schools than a literate and numerate citizenry. Civic engagement, participation in community activities, and responsible parenthood all can be influenced by experiences in school. If our schools produced graduates who read well, but who read only the holy books of their own religion, the social cohesion of the United States might be jeopardized, even as test scores rise. In short, we expect our schools to build good citizens. As Justice Felix Frankfurter put it in 1948, public education is "the most powerful agency for promoting cohesion among heterogeneous democratic people . . . at once the symbol of our democracy and the most pervasive means for promoting our common destiny."

     Do we really want to subsidize schools which advertise that they teach Christian Values or Wahhabi Islamic Values? Are we prepared to let Creationism replace science, or to allow families to choose whether their children will learn about the germ theory of disease? Opponents of vouchers are frightened by the prospect of taxpayer support of schools that teach ideas regarded as nonsense by most scientists, or curricula that promote religious intolerance or ethnic chauvinism.

     When the Supreme Court ruled segregated schools unconstitutional in Brown v. The Board of Education (1954), many communities responded by establishing a system of private segregated schools. Should we look forward to a future in which such Balkanization is subsidized by taxpayers?

     Evidence that these fears are grounded comes from the extensive experience with vouchers overseas. Martin Carnoy of Stanford reports that Chile's voucher program (introduced under military rule in 1980 and continued when democracy returned) increased private-school enrollment to 43 percent of all students by 1996, more than twice the proportion in 1980. Most of the new schools were for-profit and catered to better-off families. Segregation of students by socioeconomic class increased following the introduction of vouchers. In New Zealand, similar increases in segregation by class and race appear to have followed the introduction of vouchers.

     Finally, opponents of vouchers take issue with the idea that Americans are fed up with the public school system and that vouchers look good to them. True, 63 percent of Americans give the public education system a grade of "C" or below. But when they are asked to rate the public school their own children attend, that figure drops to 29 percent. Seven in 10 parents with a child in public school are happy with the quality of the education there. Other evidence corroborates the idea that most people think education in high- and middle-income neighborhoods is pretty good, but that in low-income neighborhoods it is not. Consistent with this nuanced and basically favorable view of public education, 69 percent of those questioned in a recent Gallup poll favored reform of the existing public school system, while only 27 percent favored finding an alternative system. In fact, the consistently unfavorable public response to suggestions that we replace public schools with "school vouchers" and "privatization" has led advocates to use the language of "school choice" instead. This response by the American public reinforces our conviction that public school choice is the way to go.

Test Scores

     Voucher advocates give short shrift to the idea that education is essential to socialization into American culture and focus instead on the urgent need to provide every student with marketable skills. Literacy and numeracy certainly are the first goal of education, so if vouchers were a way to achieve them more efficiently, some people would be willing to risk the social consequences. But there is little evidence that vouchers improve test scores of recipients. In fact, as the evidence starts to accumulate, it does not look good for voucher advocates.

     Some advocates defend vouchers simply by comparing the results of public- versus private-school education. School vouchers may be a new idea, but private schooling surely is not, and if private-school graduates do better in life than public school graduates, isn't this a prima facie case for the superiority of private education? The problem with such comparisons is that public- and private-school students are two fundamentally different populations. Private-school students have different socioeconomic characteristics from those in public schools; they have motivated parents who seek out private schools; perhaps most important, they are admitted and retained at the discretion of the private schools. Public schools, in contrast, must teach those who would not think to apply to private schools in the first place, as well as those who are rejected and expelled by such schools.

     Other studies compare the performance of voucher recipients to public-school students who are otherwise similar. The best way to make this comparison takes advantage of the fact that voucher programs often are oversubscribed. As a result, the voucher recipients are selected by lottery. Everyone in the lottery meets the income and other selection criteria, and all are concerned enough about education to have sought out vouchers in the first place. Lottery winners are selected from this population by blind luck: they form a random sample of the eligible. They become the "treatment group"; the losers in the lottery become the "control group."

     In spite of these sound methodological plans, however, the actual studies that use such an approach still face a large problem. Education takes time, and over time students change schools, move away, or simply fail to continue providing information. While the difference between lottery winners and lottery losers is random, the difference between those who remain in the study population and those who do not almost certainly is not. Since the populations are now different, we cannot know how much of any variation in educational success is due to vouchers and how much stems from the unobservable influences that keep students in the study population (such as family motivation or unreported resources-such as aid from grandparents-that help pay for private schooling). Moreover, this problem of "self selection" increases over time. But we are most interested precisely in the long-run effects of vouchers. In fact, according to the General Accounting Office (GAO), private voucher programs (most of which only cover part of the expenses of attending a private school) typically have attrition rates of 20 percent per year.

     The best data we have on a publicly funded program are from Milwaukee, Wisconsin, which has provided vouchers to some low-income students since 1990, initially to attend non-sectarian private schools. Seven private schools participated in the first year, rising to 20 in 1996 and to nearly 90 in 1998, when religious schools were added. If a school had too many voucher applicants (the maximum allowed was 49 percent of enrollment), a lottery was used to choose who would be admitted. Princeton professor Cecilia Rouse has compared the standardized-test results of students admitted through such lotteries and those not selected from the same lottery pool. Earlier research on the Milwaukee experiment was inconsistent and contentious. In using the losers from the lottery as the control population, while also correcting for deviations from randomness in the lottery process, Rouse avoided the issue of comparability that bedeviled earlier studies. She concluded that those selected by private-school lotteries gained 1.0 to 1.5 percentage points per year on their mathematics test scores, but performed the same as the control group on reading tests.

     Rouse undertook a follow-up study in which she compared Milwaukee's voucher programs not only to ordinary public schools but also to magnet schools and schools participating in a program for particularly low-performing public schools ("P-5" schools). That study suggested that magnet schools performed like ordinary public schools (for the population of voucher seekers), but that P-5 schools matched voucher schools in generating slight improvements in mathematics test scores and did better than voucher schools in generating small increases in reading scores as well. Rouse notes that the relatively good results in the P-5 schools, which received extra funding in return for meeting norms for staffing, may be the result of small class size. Other experiments have demonstrated that African-American students in smaller classes have slightly better test scores than their peers in larger classes. She conjectures that the modest advantages of voucher schools, too, may lie in nothing more complicated than smaller classes.

     The GAO has surveyed research on results of privately funded voucher experiments. While there are many such experiments currently underway, most are small and lack information good enough to use in an evaluation. Even the research cited in the GAO study (of voucher schemes in New York, Dayton, and Washington, D.C.) is hampered by data problems. None of those programs had been in existence for more than three years when studied, none had more than 1,500 voucher recipients, and all suffered high rates of attrition and inconsistency in test-taking. The results, too, hardly invite confidence:

     Recently, Alan Krueger of Princeton has looked more closely at the results for New York, the only ones that show some significant effect of private vouchers on test scores, and called them into question. Krueger reports that when he uses all observations in statistically correct specifications (including observations excluded by William Howell of the University of Wisconsin and Paul Peterson of Harvard from the original study cited by the GAO), the estimated effect of vouchers on African-American scores is greatly diminished. Similarly, when he experiments with the definition of race and ethnicity Krueger finds that the results are not robust. Howell and Peterson define race by characteristics of the mother: only the children of black, non-Hispanic mothers are classified as black, regardless of the race and ethnicity of the father. Krueger found that the results are sensitive to inclusion of the father's race and to reclassification of black Hispanics as black.

     Krueger's critique has been accepted by two of Howell and Peterson's coauthors, David Myers and Daniel Mayer, both of Mathematica Policy Research. But Peterson and Howell, in a June 2003 letter to the Wall Street Journal, continue to claim that "[p]rivate schooling had a positive impact on student performance" and sidestep questions about who it is that benefits and the issue of statistical significance.

     So where does all this leave us? The case for any effect of private vouchers on test scores is exceedingly weak. The only results that may be positive apply to children of non-Hispanic African-American mothers in New York City. If any qualifier-"non-Hispanic," "African-American mother," "New York City"-were dropped from this statement then there would be no evidence at all. And Dayton and Washington, D.C., where nearly all of the voucher recipients were African-American, show no similar positive effects.

School Choice and Competition

     Advocates of voucher programs argue that vouchers will improve the education of all children, even those who do not use them. Competition for students will force poorly performing schools to do better or face crisis; families with vouchers will "vote with their feet" to seek education elsewhere. The entire school system will be forced to meet the challenge of competition.

     The evidence in support of this proposition, however, is fairly weak. Caroline Hoxby of Harvard finds positive effects from competition in Milwaukee, Arizona, and Michigan, but her method tautologically assumes, without testing, that effective school reform is implemented only because of competitive threats. In fact, important public-school reform in Milwaukee (the P-5 schools) began more than a decade before vouchers were introduced. Moreover, whatever competitive benefits do arise from private-school choice should apply to public-school choice as well.

Controlled Public-School Choice

     The polarization of the school reform debate between vouchers and traditional school reform-neither of which offers much promise for solving the pressing problem of educational inequality-obscures the best reform option we have: controlled school choice within the public school system.

     The controlled choice strategy was formulated by Charles Willie of Harvard and Michael Alves of Brown University. Rather than assigning students to neighborhood schools, which tend to reflect stratified residential patterns, school districts allow parents and students to choose the public school they would like to attend within a given geographical region. Districts then honor these choices in a way that promotes integration. Most of the existing plans have stressed integration by race, though a few such as Cambridge, Massachusetts, San Francisco, and St. Lucie County, Florida now integrate based on socioeconomic status (measured primarily by student eligibility for free and reduced-price lunch, that is, family income at or below 185 percent of the poverty line). Many other districts-from La Crosse, Wisconsin to Wake County (Raleigh), North Carolina-integrate by socioeconomic status through a mixture of choice and more traditional redistricting schemes.

     Under most controlled-choice plans, families provide a first, second, and third choice of schools at the levels of kindergarten, sixth, and ninth grades. Information and outreach programs attempt to ensure that parents are well informed. Placement decisions are subject to fairness guidelines, to make certain that all schools fall within a range of the district's demographic average. For instance, in a district in which 30 percent of students are eligible for subsidized lunches overall, each school might be permitted to have a mix of students with no fewer than 15 percent and no more than 45 percent eligible for subsidized lunches. Preferences are normally given to applicants whose siblings already attend the school in question and to those within close walking distance. And the actual assignment is carried out by computer, so that individual school principals cannot pick those promising students they believe will be easiest to teach.

     Controlled choice shares ground with voucher proposals, which seek to expand parental choice, as compared to neighborhood school assignment or traditional busing schemes, which give little or no voice to parents. The key difference with voucher proposals (or neighborhood assignment) is that controlled choice aims to address the single most important variable within the control of policymakers: the socioeconomic diversity of the student body.

     Ample evidence suggests that economic status of students in a school drives most of the key factors that educators care about-including the level of expectations, parental involvement, peer influences, and teacher quality-and that low-income students will do better academically, and middle class children will not be hurt, so long as a majority of students in a school are middle-class. And students of every background stand to gain from increased diversity.

     In theory, vouchers or other unregulated public school choice schemes might result in greater economic integration because they permit families to seek educational opportunities outside their neighborhoods. But both of these approaches lack a key feature: mandatory choice. A crucial element of controlled-choice plans is that they require every family to choose a school, even if its choice is the neighborhood school. There is substantial evidence that when parents are under no obligation to choose, a choice plan can actually exacerbate rather than alleviate concentrations of race and class. The key problem, according to studies assembled in Who Chooses? Who Loses?, edited by Bruce Fuller of the University of California at Berkeley and Richard Elmore of Harvard, is that the least educated parents are least likely to avail themselves of choice, and the most aggressive parents, predominately middle-class and highly educated, dominate the system. This is particularly true when parents can gain an edge by investing their own resources, as when they camp out all night to be first in line. Elmore and Fuller conclude that "a large part of the stratification problem seems to result from parents and students who simply do not choose, rather than from differing preferences among those who do choose. That is, once parents and students make the decision to choose and actively exploit the opportunities that decision presents, they seem to have preferences that are remarkably similar across race and social class."

     Controlled choice protects against that stratification between choosers and non-choosers by requiring a choice. Neighborhood schools do not become repositories of the "leftovers" of the choice process; no family is guaranteed a place in any school, so everyone has incentive to choose. Business groups recognize that for the competitive aspects of choice to be fully realized, all parents must be required to choose, and the National Alliance of Business has endorsed "mandatory choice." Well-designed controlled-choice plans also provide for mail-in registration, so that there is no advantage to being first in line.

      The use of fairness guidelines avoids coordination problems as well. Middle-class parents in a given community may want to choose the new computer-based magnet school for their children, but if they have no guarantee that the school will be economically mixed, individual sets of parents may avoid selecting that school for fear that other middle-class families might avoid it as well. Under controlled choice, that factor is eliminated from the decision-making process and families can choose based on pedagogy rather than fears about the school's demographic makeup.

     The fairness guidelines for socioeconomic integration often raise opposition from those who worry about "constraints" on choice, which remind them of busing. But there are important steps that can be taken to reconcile choice and integration. Before plans are implemented families can be surveyed to see what kinds of choices they would like for their children. If the survey finds that 40 percent of parents want a highly disciplined environment with uniforms, and only 10 percent want a French immersion school, then the makeup of the options should reflect that general preference. For parents who say they believe it is too early for their children to specialize, options for "regular" schools should be made available. Once a system of school choice is established, it should remain flexible and responsive to parental demand. Schools that are undersubscribed year after year should be closed down or reconstituted. Schools that are continually oversubscribed and deemed successful should be replicated or franchised.

     Likewise, to the extent that pre-implementation parent surveys reveal any socioeconomic leanings toward certain programs, middle-class and working-class parents can be encouraged to divide their choices among schools. For example, if wealthier parents trend toward progressive or alternative schooling, the district can place those progressive programs in formerly blue-collar neighborhood schools, so that the tendency of some to prefer neighborhood schools counteracts the pedagogical preference.

     Although no student is guaranteed a place in the most favored school, in practice the number of students in controlled-choice districts who are assigned to schools not of their choice is very small. (Cambridge boasts placement in first-, second-, or third-choice schools at around the 90 percent level-a rate replicated in other jurisdictions using controlled choice.) Moreover, most of those who do not receive their first choice are turned down because of overall space limitations having nothing to do with student diversity. Charles Glenn of Boston University notes that in 1990 only 1.7 percent of students assigned to Boston public schools under its controlled-choice plan (238 of 14,041 first-, sixth-, and ninth-graders) "were either denied a place or assigned involuntarily to a place that another student was denied in order to meet the requirements of desegregation." In a 1995 Bain and Company survey, 80 percent of parents said they were satisfied with controlled choice, and 72 percent said they preferred having a choice to assignment based on neighborhood schools. (Boston has since dropped its controlled-choice plan because of legal challenges to the use of race. There is no similar legal impediment to using socioeconomic status in student assignment.)

     Although some critics fixate on the 1.7 percent whose choice is constrained by integration goals, the movement from neighborhood schools to controlled choice represents an enormous expansion of options, particularly for the poor. Today, private-school choice remains largely the province of the well-to-do; by contrast, those with little education and low income are twice as likely as the wealthy to use public-school choice.

     Research suggests that roughly 36 percent of all elementary and secondary schoolchildren attend neighborhood schools chosen indirectly by their parents in deciding where to reside-and that wealthier families are much more likely to have made such residential choices. For the poor, assignment to unpopular neighborhood schools is a fact of life.

     It is ironic, Glenn notes, that in moving from assigned schools to controlled choice, critics focus on the small element of control rather than the enormous flowering of freedom. "An inevitable cost of freedom is to experience remaining constraint as galling," he writes. "So long as children are simply assigned to school involuntarily on the basis of where they live, of course, the issue of disappointment does not arise." Moving to controlled choice means that 90 percent get one of their top choices-as opposed to the 36 percent who today choose a neighborhood for its schools.

     There are no carefully controlled studies of the consequences for test scores of socioeconomic integration through public school choice, in part because there isn't massive financial support to set up experiments and evaluate them, as there is on behalf of private school vouchers. But there is promising evidence of success in La Crosse, Wisconsin, where test scores have risen, and in Wake County, North Carolina, where 90 percent of students achieve at or above grade level. And there is forty years of social science research to back up the commonsense proposition that all poor and middle-class children achieve more in economically integrated, middle-class schools than in poverty-concentrated schools.

Conclusion

     Liberating poor children from bad schools is a moral imperative. We agree with many of the critics of traditional efforts to improve neighborhood schools in high-poverty neighborhoods: traditional remedies have not done enough. The fundamental barrier to progress is that outstanding high-poverty schools are immensely more difficult to create than outstanding middle-class schools. But it would be a grave mistake to simply give up on a public education system that is fundamentally sound and that is a bulwark of our diverse democracy.

     More choice, controlled to guarantee substantial socioeconomic diversity in every school, is the way to go. Choice in the form of public magnet schools, charter schools, alternative schools, and the like is already expanding rapidly; the public is demanding it. That beginning is the foundation on which future school reform should be built. The issue isn't choice versus no choice, but what kind of choice parents should have. Should we give up on the civic goals of public education and pour public funds instead into private and religious education enterprises that can use them to promote intolerant religious beliefs, pseudo-science, and exclusionary ideology? All in the unsubstantiated hope that test scores might improve? Or should we make public schools-the 90 percent of schools that take all comers, help bind together people of diverse races and religions, and are accountable to the public through their elected leaders-work better? Between privatization through vouchers and controlled public-school choice, we believe the choice is clear.

     Richard D. Kahlenberg is a senior fellow at The Century Foundation and the author of All Together Now. He is the editor of Public School Choice vs. Private School Vouchers.

Bernard Wasow is a senior fellow at The Century Foundation and a contributor to Public School Choice vs. Private School Vouchers.