Association for Public Policy Analysis & Management

School vouchers are a market-based reform allowing parents to choose any school for their children. Both government and privately sponsored voucher programs exist. The effectiveness of voucher programs is fiercely disputed in both academic and policy circles. Most reviews of the school voucher literature have been selective, not systematic. Prior research by Rouse & Barrow (2009); Anderson, Guzman, & Ringquist (2013); and Epple, Romano & Urquiola (2015), either did not systematically search for all the empirical evaluations of school voucher participant effects or relied heavily on non-experimental findings even when a large number of more rigorous studies were available. A thorough meta-analysis informed by a true systematic review of all available randomized controlled trial (RCT or “experimental”) studies would provide the foundation for a greater scholarly consensus regarding the impacts of school vouchers on student outcomes.

The objective of this meta-analysis is to estimate the average academic impacts that the offer (or use) of a voucher has on students. This study will be the first to systematically review all Randomized Control Trials (RCTs) internationally. Our meta-analytic results will focus on the RCTs because these are the “gold standard” of program evaluation in terms of assessing causal relationships.

Our initial search was guided by the following research question: What is the impact of private school vouchers globally on the student achievement of those students offered the vouchers?  We will present an overall meta-analytic estimate of the test score impacts of school vouchers globally as well as subgroup estimates: (1) just in the U.S.; (2) just outside the U.S.; (3) of publicly-funded programs and (4) of privately-funded programs.  For all of these topical estimates we will differentiate math and reading impacts, for a total of 10 distinct impact estimates (2 overall and 8 subgroup). In addition, impacts on traditionally disadvantaged subgroups will be reported when possible.

            To date our systematic computerized and networked search has identified 18 separate RCTs of the impact of school vouchers on test scores.  Fourteen of the studies are situated in the U.S. and four are of programs outside the U.S., specifically in Colombia (2) and India (2).  For our meta-analytic methodology we will combine the impact estimates and variances from the studies, generating Hedges g as our primary summary statistic and Cohen’s d as a robustness test.

This meta-analysis will contribute to the field of market-based education reforms by combining and systematically evaluating rigorous evidence from RCT studies of the effects of voucher funded K-12 education. This review provides a broader overview of all the rigorous experimental findings and will have important policy implications about the effectiveness of voucher programs generally. As the first meta-analysis of the participant effects of vouchers internationally, it will help establish the baseline for future studies.