African-American students who received vouchers to attend private schools in New York City derived no academic benefit from them, a Princeton University researcher has concluded.
The report, “Another Look at the New York City School Voucher Experiment,” is available from the Education Research Section at Princeton University. (Requires Adobe’s Acrobat Reader.)
Paul E. Peterson’s study “School Choice in New York City After Three Years: An Evaluation of the School Choice Scholarship Program” is available from the Mathematica Research Policy Inc. (Requires Adobe’s Acrobat Reader.)
The findings, released last week, emerged from an analysis of data that yielded a different conclusion a little more than a year ago.
In February 2002, Paul E. Peterson, a professor of government at Harvard University, unveiled the results of a study he had conducted with Mathematica Policy Research Inc. The study showed that African-American students who had won privately financed tuition vouchers in a 1997 lottery scored 5.5 national percentile points higher on standardized tests three years later than did black peers who had sought but did not receive the vouchers. The researchers called the difference a “statistically significant positive impact.” (“Voucher Plans’ Test Data Yield Puzzling Trends,” Feb. 27, 2002.)
Intrigued by the Peterson study’s finding that black students appeared to benefit from the vouchers while Hispanic students did not, Alan B. Krueger, a professor of economics and public policy at Princeton, sought to examine the data. Mathematica, a private, Princeton, N.J.-based group, made it available.
At an April 1 news conference here, Mr. Krueger argued that design features of the original study had produced misleading conclusions.
“This research has really been blown out of proportion,” said Mr. Krueger, accompanied by his Princeton co-researcher, Pei Zhou, and two of the Mathematica researchers who had worked with Mr. Peterson.
“For the most representative sample of black elementary school students,” he said, “offering a voucher had no statistically discernible impact on achievement scores in the New York City experiment.”
Mr. Peterson, who was invited to the news conference but did not attend, said he is working on a paper to respond to the technical questions raised by Mr. Krueger’s analysis. He said he stands by his original findings.
“Our original estimates, at least as far as we know at this point, are the best available estimates of the impact of the voucher program,” Mr. Peterson said.
Mr. Krueger, in detailing the aspects of the earlier study that he contends produced skewed results, pointed to the exclusion of large numbers of children from the sample. While 2,666 needy K-4 pupils were placed either in a group that received vouchers or a control group that applied for but did not get them, more than 500 were excluded from the study because they did not take the baseline standardized test.
Bringing those children back into the study sample increased its size by 44 percent, Mr. Krueger said.
The definition of race proved pivotal as well, he said. The original study inferred a child’s race or ethnicity by asking the female parent or guardian to choose one from a list of racial and ethnic groups to describe herself. That method excluded many children, including those of mixed race and those whose fathers might belong to a racial or ethnic minority.
Adding those children back in increased the sample size by another 10 percent, he said. With those additions, and more revisions made to control for other demographic factors, Mr. Krueger and his colleagues found that black children who had received vouchers scored only 1.44 percentile points higher on tests three years later than did their black peers who did not receive vouchers.
Mathematica researcher David E. Myers said the re-evaluation had been “collegial,” and a model of how social scientists should share their findings and allow the data to be subjected to multiple examinations. But he lamented what he sees as the way some academics and reporters magnified the differential between black voucher recipients and nonrecipients. They overlooked what he considers the most significant finding, he said, which is unchanged by Mr. Krueger’s analysis: In the overall group of students studied, obtaining vouchers provided no academic benefit.