Voucher supporters have suffered a major setback. At least that’s what one might think after reading the New York Times’ education section yesterday. There, Michael Winerip spent a whole column gloating over a technical disagreement between researchers who have studied the results of a voucher experiment in New York City. The total of the controversy is less than the sum of its parts.

It boils down to this: In 1997, the School Choice Scholarships Foundation offered 1,300 New York City children scholarships to private schools. The students were selected by lottery, and their progress was tracked against a parallel 1,300 students who had applied for the scholarships but were denied and remained in the public school system.

Initially, one researcher, Paul Peterson, reported modest gains in the test scores of African-American students and no gains among other groups. Partners of his in the study disagreed with Mr. Peterson’s results, and argued that the gains, which showed up only in one grade, might be a fluke. Now, two outside researchers have concluded that if you play with the definition of “African American” and add into the mix some students who entered the study in kindergarten — meaning they have no meaningful baseline test scores by which to measure improvement — one can make the test-score gains disappear altogether.

Those who have been fighting for vouchers would have preferred clearer — and, yes, more favorable — results from this study. At the same time, it is not as if the voucher movement has been hanging its hat on one study in one city. There are various studies that show improvements in test scores when students are offered vouchers. In Milwaukee, for example, where the nation’s oldest publicly funded voucher program operates, a 1997 Harvard/University of Houston study found gains in both reading and math scores among voucher students. After three years, reading scores of voucher students were 3.13 percentile points higher than their peers in the Milwaukee public schools. After four years, voucher students’ reading scores were 4.81 points higher. Math scores for voucher students were 4.98 percentile points higher in the third year, and 11.59 points higher in the fourth year. Furthermore, studies have found that public schools threatened with losing students to vouchers tend to shape up. A study in Florida in 2001 found that schools labeled failing under the Florida A-plus program — which offers vouchers to students’ whose schools are labeled failing a second year in a row — showed improvements in test scores far larger than those at other schools.

Ultimately, however, the jury is still out on how vouchers would affect test scores if more widely deployed. The samples have been small and the programs are still relatively new. But it’s worth noting that not even vouchers’ most vehement opponents suggest that they actually harm test scores; the teachers unions restrain themselves to claims such as that vouchers will drain money from public schools. The worst that critics such as Mr. Winerip are able to come up with is that the gains are too small to see, in a small sample with questionable data.

Parents in the city, though, seem to think that there is some benefit to vouchers. Even setting aside consistently high minority support for vouchers, which registers in poll after poll, look at concrete numbers. When the 1,300 scholarships were offered by the School Choice Scholarships Foundation in 1997, 20,000 New York City students applied. And, according to a spokeswoman for the Children’s Scholarship Fund, Elizabeth Toomey, 168,000 students applied for 2,000 scholarships when that organization offered them in 1998 and 1999. The fund still offers a handful of new scholarships every year, but according to Ms. Toomey, “We can’t accept any new applications, we already have such a long waiting list.”

At the end of the day the soundest conclusion is that the parents trapped in our city’s failing schools know something that the New York Times and other voucher opponents in the education establishment have yet to learn.