What the Data Show on School Choice and Segregation

By 03/22/2017

Print | NO PDF |

The Century Foundation has published a report by Halley Potter that claims private school choice will increase ethnic segregation in schools. Although the text of the report constantly invokes words like “evidence,” “studies” and “data,” its conclusions are actually defended almost entirely by appeal to a lengthy recitation of hypothetical, ideological speculation. The report’s actual engagement with empirical research is as scanty as it is misleading. A real review of the evidence shows that private school choice has never been found to increase segregation and often seems to have provided a more integrated classroom experience.

There are a number of serious methodological challenges involved in empirical research on how education policies affect ethnic segregation. I wrote about them at some length in a report for EdChoice a while back. For example, some data don’t permit causal conclusions; some methods of comparison are unfair because they compare elementary grades to secondary grades inappropriately. Reviewing all of the empirical research on school choice last year, I found that 10 studies had been conducted that examine the relationship between school choice and ethnic segregation in some respect. Some are causal, some descriptive; all shed some light on the question. Nine of those studies found that school choice provided a more integrated classroom experience, one found no visible difference, and no empirical study had ever found that a school choice program made ethnic segregation worse.

That is the empirical evidence. Nine out of the 10 studies that have been conducted report positive findings on the actual, real-world impact of school choice programs when it comes to ethnic segregation.

Unfortunately, Potter seems not to be much interested in empirical methods or in the full body of research. Her desire to transform hypothetical, ideological speculation into serious science is captured in sentences like this: “From a purely mathematical perspective, vouchers have a greater potential for increasing segregation in public schools than for increasing integration.” We need not ask what actually happens, because our theories apparently tell us what to expect, and that’s sufficient. It’s a matter of “pure math,” you see.

This pure math, pure theory approach is what justifies the report’s scaremongering. “On balance, voucher programs are more likely to increase school segregation than to promote integration or maintain the status quo,” the first bullet point of the summary solemnly intones. When I read that, I think it implies we have evidence of at least one (1) school choice program increasing segregation somewhere. Maybe that’s just me.

If you dig very, very deep into the report, you do eventually find a discussion of empirical studies. But this doesn’t mean the report gets much better, for Potter examines only two of the 10 studies that exist – and she has described them in a misleading way.

Looking at longitudinal studies in Milwaukee and Louisiana, she describes them in a way that will leave the impression that the results were negative for school choice: “In both cases, programs were used primarily by black students and generally did not exacerbate segregation in public schools; however, students using vouchers did not gain access to integrated private schools, and segregation in private schools actually increased.”

Now, even that misleading description would be enough to call into question the huge mountain of hypothetical, ideological speculation that occupies the overwhelming majority of Potter’s report. However, a more precise description of these two studies would look even worse for Potter, because it would look good for school choice.

The Milwaukee study found the voucher program made no visible difference to segregation, at least during the period under observation. It is the only such study ever to find no visible difference. Other studies in Milwaukee using different methods have found more encouraging results, though because of methodological restrictions, none of these studies can be considered a final word. The longitudinal study’s null finding is not as encouraging as a positive finding would have been, but the nightmare world of increasing school segregation promised by Potter’s lengthy speculations apparently did not come to pass in Milwaukee.

Potter draws her claim that “segregation in private schools actually increased” from the Louisiana study. But – as Potter is later forced to admit, though she quibbles about it in ways that obscure what would otherwise be a clear conclusion – the same study found that segregation in public schools decreased by a much greater amount than the increase in private schools. So this study actually found the voucher program created a significant net decrease in ethnic segregation!

After reciting some methodological quibbles, Potter summarizes the results with the statement that the program had a “mixed effect.” When you’ve been in the education research business long enough, your eyes automatically roll by reflex whenever they read the words “mixed effect.” A mixed effect is a positive effect produced by a policy that the researcher doesn’t like.

It’s interesting that there have actually been two longitudinal studies of the Louisiana voucher program, and Potter didn’t review the other one. That study found the program decreased segregation in public schools while having no effect on segregation in private schools – and in school systems under court orders for desegregation it had a positive effect in both public and private schools.

As we look at the evidence on private school choice — the actual evidence, not speculation — we should consider it in comparison with the continuing epidemic of ethnic segregation in the public school system. Jim Crow began in the public schools, and its legacy has never really left. The Government Accountability Office recently found that from 2000 to 2014, the percentage of all K-12 public schools with ethnic or socio-economic “isolation” (defined as those with over 75 percent of students from the same group) grew from 9 percent to 16 percent.

Choice may tend to produce positive results because you can’t get much more segregated than a system in which students are assigned to schools based on where their parents can afford to live. If Potter wants to fight segregation, she’s looking in the wrong place.

Within the limitations of available data and methods, the empirical evidence is very encouraging for private school choice on ethnic segregation – just as it is on academic outcomes, effects on public schools, fiscal effects and effects on civic values and practices. If there were no data, speculation would be all we had; in light of the data, speculation ought to give way to reality.

— Greg Forster

Greg Forster, Ph.D. is a Friedman fellow with EdChoice.

Sponsored Results

The Hoover Institution at Stanford University - Ideas Defining a Free Society

Harvard Kennedy School Program on Educational Policy and Governance

Thomas Fordham Institute - Advancing Educational Excellence and Education Reform