NYC Vouchers Lift Minority Test Scores

Published December 1, 1998

Low-income students in New York City who were able to use vouchers to transfer to private or parochial schools last fall scored higher on reading and math tests than did their peers who remained behind in the public schools, according to a new study conducted by Harvard University and Mathematica Policy Research.

Since almost all of the students in the voucher program are black or Hispanic, the sizable gains achieved by fourth- and fifth-grade students have important implications for efforts to close the minority-white achievement gap.

If students continued to score similar gains in middle school and high school, the persistent academic achievement gap between white and minority students in public schools would be eliminated, according to Harvard professor Paul E. Peterson, who conducted the study with Mathematica senior fellow David Myers and William G. Howell of the Harvard Program on Education Policy and Governance.

After one year, students in grades two through five who received vouchers from the New York School Choice Scholarship Program scored about two percentile points higher in reading and math than students who did not receive vouchers. Voucher students in grades four and five showed much larger differences–four percentile points in reading and six points in math.

School choice opponents quickly tried to discredit the results by impugning Peterson’s motives and dismissing the gains as an inevitable consequence of smaller class sizes. Although class sizes in voucher schools were smaller than those in non-voucher schools, the difference amounted to barely three students–23.6 versus 26.3, much less than 15 versus 25 class size reduction in a Tennessee study touted by choice opponents.

The New York City study is in fact the most carefully designed research project on school choice that has been assembled to date, far better controlled than the Tennessee class size study. The New York City study was designed from the beginning to eliminate many of the analytical problems that cloud the results from state-funded choice programs, such as self-selection of motivated parents and the establishment of an appropriate control group.

When the New York School Choice Scholarship Program offered three-year $1,400 scholarships to low-income elementary schoolchildren in 1997, more than 20,000 parents applied for the 1,300 scholarships. Mathematica collected benchmark data on the applicants and their families before the vouchers were awarded by lottery for use starting with the 1997-98 school year. This spring, the researchers retested 1,000 students who had received vouchers and a control group of 960 students who had applied for but not received vouchers.

“Because scholarships were awarded at random, the two groups may be assumed to be, on average, statistically equivalent, save the offer of a scholarship,” note the researchers in their report, “An Evaluation of the New York City School Choice Scholarships Program: The First Year.” Thus, they add, “any differences between the two groups can be attributed to the offer.”

As expected, this approach eliminated a criticism of earlier analyses of publicly funded school choice programs in Milwaukee and Cleveland: that academic gains result from the self-selection of motivated parents into the programs. In this study, parents in both the test group and the control group reported the same very high involvement in school and engagement in the education of their children.

Another frequent criticism of the superior performance of choice schools over public schools is that the latter must take all students, while private and parochial schools achieve a superior student body by expelling their problem students or encouraging them to leave. The New York City study, however, reported no difference between test and control groups with regard to expulsion and suspension rates, which were low for both groups. In addition, student mobility rates were the same for scholarship students and members of the control group.

“Using a scholarship reduced somewhat the racial isolation of minority students,” note the authors. They point out that while 18 percent of scholarship parents reported that less than half of their child’s fellow students were of minority background, only 11 percent of the parents in the control group gave this response. Conversely, 37 percent of the control group parents said all students in the classroom were minority, as compared to 28 percent of the scholarship parents.

George A. Clowes is managing editor of School Reform News.

For more information …

The paper by Paul E. Peterson, David Myers, and William G. Howell, “An Evaluation of the New York City School Choice Scholarships Program: The First Year,” from Mathematica Policy Research and Harvard University’s Program on Education Policy and Governance, is available from PEPG’s Web site at It is also available in six parts through PolicyBot. Point your Web browser to, click on the PolicyBot icon, and request old documents #2185406 (part 1), #2185407 (part 2), #2185408 (part 3), #2185409 (part 4), #2185410 (part 5), and #2185411 (part 6).