Do vouchers reduce crime?

The potential benefits of education policies extend far beyond the usual metric of test scores. A policy might be considered beneficial, for example, if it leads to reduced criminal behavior. That is the claim made in a recent report from University of Arkansas researchers, who studied the relationship between crime and a school voucher program. A review of the report finds that the study, while suggestive of a possible correlation, has several weaknesses and certainly does not support the implication that the voucher programs caused a reduction in crime.

Using data from the Milwaukee Parental Choice Program (MPCP), the researchers compared crimes processed through the Wisconsin courts for program participants and a matched sample of public school students. Although most of these comparisons showed no association, the report finds that some subgroups of MPCP students were less likely to commit crimes as adults.

Clive Belfield, an Associate Professor of Economics at Queens College, City University of New York, reviewed the report, The School Choice Voucher: A “Get Out of Jail” Card? for the Think Twice Think Tank Review Project at the National Education Policy Center, housed at the University of Colorado Boulder’s School of Education.

Professor Belfield writes that education and crime are often found to be negatively correlated, so if a program has an educational benefit, it is reasonable to expect that criminal behavior might be reduced. Moreover, the past MPCP research, from this same group of researchers, has announced some benefits, albeit modest and mixed. But the findings in this new study do not warrant any strong claims.

One concern is that the paper employs a matching method that omits some important factors that explain school choice and crime. So differences between the voucher group and the matched (control) group cannot easily be attributed to the voucher policy.

Also, the study’s results are highly variable, with most of the comparisons between MPCP participation and measures of adult crime showing statistically insignificant results. Professor Belfield explains that a valid interpretation of the report is that vouchers and crime are, in fact, not correlated. Conversely, for subgroups and estimation approaches that do yield statistically significant associations, the MPCP effects appear to be too extreme. Finally, even assuming that vouchers do reduce adult crime, it remains unclear by what mechanisms vouchers might do so.

Professor Belfield also cautions that the report is wrong to assert that the methods used justify a causal inference. The study’s title should not imply that voucher programs are a “get-out-of-jail” card, and the evidence in the study is associational, not causal.

Belfield concludes that, though the report contributes to policy and practice in looking at how educational processes influence behavioral outcomes, “it is unclear how useful the actual findings of this paper are for policymakers.”