Boston Review
table of contents
new democracy forum
new fiction forum
rave reviews
writers’ guidelines
bookstore locator
literary links
RSS feed

Search the Web


Devil in the Details

Richard J. Murnane

8 Kahlenberg and Wasow argue that controlled public-school choice offers much more promise than private-school vouchers for achieving the goal of better education for children from low-income families. While I agree with the extraordinary importance of this goal, my view is that their article does not shed light on how to achieve it. My concerns are of two related kinds. First, their basic setup—contrasting private-school vouchers with controlled public-school choice—diverts attention from the specifics of voucher systems and controlled-choice systems, but those specifics are of essential importance. Second, by suggesting that controlled choice is the answer they fail to address the difficult trade-offs in the design of any choice system. I focus my comments on these points.

Too-simple Categories

The authors point out that the operation of large-scale voucher systems in Chile and New Zealand led to segregation of students by socioeconomic status (SES). Kahlenberg and Wasow are right that this is a troubling outcome, because peer groups have a substantial impact on educational achievement. However, they do not point out that in the Chile and New Zealand systems all children received vouchers of equal value. Several analysts, including Christopher Jencks and Caroline Hoxby, have advocated voucher systems in which children from poor families and/or children who are especially expensive to educate would receive higher valued vouchers than other children. The logic is to make it attractive to schools to recruit such students rather than to shun them. It is not clear how politically viable a system with differentially valued vouchers would be or how large the differentials would need to be to avoid the segregation by SES that the Chile and New Zealand systems produced. The key point is that the impact of a voucher system on educational opportunities for children from low-income families would depend critically on details of the system’s design.

The same point holds for public-school controlled-choice systems. Critical design considerations include whether money follows students from school to school, whether charter schools may participate (and, if so, what types of organizations are eligible to start charter schools), what happens to educators in schools that do not appeal to many families, the circumstances under which families have the right to move children from one school to another, and the obligations of schools to accept students at different points in time, such as the middle of the school year. Without knowing a great deal about such design details, it is not possible to judge how any choice system, whether labeled a voucher system or a public-school controlled-choice system, would impact the education of children from low-income families.


There is no one best design for a choice system. Instead, there are difficult trade-offs, and this is just as true of public-school controlled-choice systems as it is of voucher systems. Consider trade-offs between rapid supply responses and quality control. It is important in choice systems to rapidly expand the supply of seats in the types of educational programs parents and children find attractive. Making it easy for groups to start new schools—for example, charter schools—contributes to this objective. However, the easier it is to start a new school, the more likely it is that the quality of some new programs will be low and some children will suffer. On the other hand, requiring an elaborate review process may help with quality control but it is likely to discourage supply expansion.

Trade-offs also exist between facilitating the success of individual schools and protecting educational options for children from especially troubled families. The more difficult it is for schools to expel a student whose behavior impacts negatively on the achievement of peers, the more difficult it is for schools to succeed. On the other hand, the easier it is for schools to expel students, the greater is the problem of what to do with troubled students. Another example concerns the obligations of schools to accept students in the middle of the school year—a major issue in many urban districts serving substantial numbers of transient families. The more control individual schools have over midyear admissions, the easier it is for them to succeed in delivering coherent instructional programs. But if schools can restrict midyear admissions, what happens to students from mobile families?

These are only a few illustrations of the many trade-offs that characterize the design of any choice system. Focusing on the issue of vouchers versus controlled choice is less helpful than the authors suggest.

Cambridge as a Case Study

Cambridge, Massachusetts, the authors point out, has had a public-school controlled-choice system for many years. To my knowledge the system has not been subjected to a rigorous evaluation. However, it seems reasonable to conclude that it has contributed positively to the integration of schools by race and socioeconomic status. It probably also contributed to integrated housing patterns, because families can choose where to live without worrying about the quality of the nearest public school. For these reasons Cambridge’s version of controlled choice may have contributed to cohesion in an extremely diverse community.

At the same time, the controlled choice plan has not led to excellence in student achievement. Forty-one percent of Cambridge eighth-graders scored at Level 1 (warning) on the 2002 MCAS math exam, compared to 33 percent of students across the state. The comparable figures on the seventh-grade English-language exam are 17 percent and nine percent. One could argue that these comparisons are not fair because Cambridge serves above-average percentages of immigrant and low-income families. On the other hand, Cambridge spends more money per student ($13,296 in 2000) than any public school district in the state, with the exception of Provincetown.1 Thus, it does not seem that Cambridge’s version of controlled choice has been successful in translating very high expenditures into consistently high student achievement. A controlled choice system with a different design might be more effective in improving students’ achievement.

Standards-based Reforms

The authors state that “the nature of public education is changing rapidly,” and they mention the increase in the number of charter schools and magnet schools as examples. I found it odd that they do not mention standards-based educational accountability, the most powerful force affecting public education in most states, including Massachusetts. These state initiatives, especially as impacted by the provisions of the 2001 federal No Child Left Behind (NCLB) legislation, are putting enormous pressure on public schools serving concentrations of low-income children to improve students’ performances on state-mandated tests. NCLB also includes provisions mandating that students in underperforming schools have the right to choose another school. It is not clear how the provisions of NCLB will play out in local districts or how the law will be revised over time. However, it seems likely that NCLB will have a marked effect on student-choice systems in the years ahead.

*  *  *

Kahlenberg and Wasow make a contribution to debates about how to improve education for low-income children by drawing attention to the importance of peer groups. It is important to create systems of schools that are attractive to both children from middle-class families and children from poor families. Such schools are more likely to succeed in providing a high quality education to all children than are schools that serve only poor children. However, I do not see contrasting generic voucher systems with generic controlled-choice systems as contributing to clarity about how to achieve this objective. The design details are what matter. <

Richard J. Murnane, Thompson Professor of Education and Society at the Harvard Graduate School of Education, is author with Frank Levy of Teaching the New Basic Skills and The New Division of Labor (forthcoming).


1. Source: Provincetown has 288 students and a per student operating cost of $14,199. There are two very small “districts” with fewer than 40 students that also spend more per student than Cambridge. Operating cost per student in Boston was $9,428, $3,868 less than in Cambridge.

Click here to return to the New Democracy Forum, “What Makes Schools Work?”

Originally published in the October/November 2003 issue of Boston Review

Copyright Boston Review, 1993–2006. All rights reserved. Please do not reproduce without permission.

 | home | new democracy forum | fiction, film, poetry | archives | masthead | subscribe |