School choice — what the data actually says

Public charter schools are tuition-free schools operated by independent entities under a charter from a public authorizer, funded by public dollars, subject to most state academic accountability requirements, and required to admit by lottery when oversubscribed. Vouchers, tax credit scholarships, and education savings accounts use public dollars to fund tuition at private schools, including religious schools, often with reduced or no academic accountability to state systems. Open enrollment policies allow students to attend public schools other than the one assigned by residence. Conflating these three under a single label obscures that they answer different questions, have different evidence bases, and produce different effects.

CREDO's three national charter studies have used a "virtual twin" methodology, matching each charter student to a statistical composite of similar district-school peers. The 2009 study found charters performing on average worse than district peers in reading and math. The 2013 study showed improvement, with charters now performing slightly better in reading and equivalent in math. The 2023 study showed continued improvement, with charters outperforming district peers in both subjects on average, driven by gains for Black and Latino students in urban networks. The methodology is contested — selection on unobservables remains a concern — but the pattern is consistent with the lottery-based studies, which are methodologically stronger.

Studies that exploit oversubscribed charter lotteries, in which students are randomly assigned, are the gold standard. Joshua Angrist, Parag Pathak, and Christopher Walters's work on Boston charters; Caroline Hoxby's work on New York City charters; and the multiple evaluations of KIPP networks have consistently shown large positive effects for students who win charter lotteries in urban networks. KIPP middle school students gain roughly 0.20 standard deviations per year in math and 0.15 in reading relative to lottery losers. Boston charter middle schools produce comparable gains. The lottery design rules out selection on unobservables. The effects are real for the children who win the lotteries. The effects do not generalize automatically to charters operating outside these intensive networks.

The bottom third of the charter sector — the schools that underperform their district counterparts — tend to share features. They are often small (fewer than 200 students), under-resourced, run by founders without instructional expertise, and inadequately monitored by their authorizers. Virtual charters, which operate online, perform particularly badly in nearly every rigorous study; the 2015 CREDO study of online charters found students losing 72 days of learning per year in reading and 180 in math compared to district peers. Authorizer quality matters enormously: states with strong authorizing standards and willingness to close failing schools produce stronger sector averages.

Milwaukee launched the first modern voucher program in 1990. Two decades of evaluations by John Witte and others produced findings that have been read in opposite ways by partisans: voucher students did not outperform public school peers on most academic measures, but parental satisfaction was high, graduation rates appeared modestly elevated, and the program survived politically. The Milwaukee experience set the template: vouchers as a policy that delivers parental satisfaction and graduation gains without producing test score improvement, and that funds, alongside legitimate religious and Montessori schools, a tail of low-quality operations that would not have survived in a regulated market.

The Louisiana Scholarship Program, which expanded statewide in 2012, was evaluated by Jonathan Mills, Patrick Wolf, and others using a lottery-based design. Students who used vouchers to leave public schools for private schools performed substantially worse in math after one year — losses of roughly 0.4 standard deviations — and continued to lag after several years. The participating private schools were disproportionately small, religious, and experiencing enrollment declines. The finding shocked voucher advocates, including Wolf himself, who had previously published more favorable findings on D.C. and Milwaukee. The Louisiana data are not an outlier; they are consistent with Indiana and Ohio findings.

Indiana's Choice Scholarship Program, the largest in the country by enrollment, was evaluated by Mark Berends and R. Joseph Waddington using matched comparisons. Students who used vouchers showed math losses of roughly 0.1 standard deviations per year, persistent over four years. Ohio's EdChoice program, evaluated by David Figlio and Krzysztof Karbownik, showed comparable losses. The pattern across the three largest statewide programs is consistent enough that the previously dominant claim — that vouchers improve outcomes — cannot be defended on the current evidence. Advocates have moved to arguments about parental satisfaction, religious liberty, and long-run outcomes that the test-score data do not capture, which are legitimate arguments but different ones.

In the mid-2000s, Susan Aud, then at Education Sector, and Erin Dillon produced summaries of the voucher evidence that already acknowledged the weakness of academic gain claims and shifted advocacy toward parental satisfaction, graduation rates, and civic outcomes. The shift was honest and prescient. The subsequent statewide data have confirmed that the academic case was weaker than mid-2000s advocates hoped. Proponents who maintain the academic case today are doing so in the face of evidence that does not support it.

METCO in Boston, Hartford's Open Choice, and St. Louis's voluntary transfer program have all produced gains for participating low-income, minority students. Susan Eaton's work on METCO documented graduation rate improvements and college attendance gains that persisted into adulthood. The mechanism is consistent with the Coleman finding: the move is from a high-poverty school to a lower-poverty school, and the composition of peers is doing most of the work. The programs remain small not because they fail but because suburban receiving districts ration seats.

Diane Ravitch, once a charter advocate as a Bush-era assistant secretary, became one of the most prominent critics of choice after reviewing the evidence in the 2000s. Her The Death and Life of the Great American School System (2010) and subsequent work argue that the choice movement has functioned as a Trojan horse for privatization, that it has destabilized district systems without producing reliable gains, and that the energy spent on choice has displaced energy that should have gone to teacher development, curriculum, and school funding. The argument is contested but worth engaging on its own terms: the political economy of choice — the foundations, the advocacy networks, the legislative push — is a real phenomenon, and the effects on district systems are real, regardless of one's views on whether the substitution is worth it.

The newest wave of school choice policy, education savings accounts now operating in Arizona, Florida, Iowa, and several other states with near-universal eligibility, deposits per-pupil state funding into accounts families can use for private tuition, tutoring, homeschool curricula, and other education expenses. The evidence base on ESAs is thin because the programs are new and the spending is heterogeneous. Early data from Arizona, the longest-running near-universal program, suggest that most users were already enrolled in private schools or homeschooling, meaning the program is operating substantially as a transfer to families who were not using public schools rather than as an exit ramp for public school students. Whether this is a feature or a bug depends on one's policy goals.

The Think Law requires distinguishing what the data show from what one wishes they showed. The honest summary: high-quality urban charter networks produce real academic gains; the average charter is roughly comparable to the average district school in its area; the average voucher recipient in recent statewide programs performs worse than matched public school peers; open enrollment programs that move students into lower-poverty schools produce gains. Choice is not magic, and the supply of high-quality school seats is the binding constraint that no choice mechanism addresses. A serious choice policy is paired with serious supply-side investment. A choice policy that is not so paired is rearranging the same seats among the same families with marginal improvement.

1. CREDO (Center for Research on Education Outcomes). As a Matter of Fact: The National Charter School Study III. Stanford, CA: Stanford University, 2023.

2. CREDO. National Charter School Study. Stanford, CA: Stanford University, 2013.

3. Angrist, Joshua D., Parag A. Pathak, and Christopher R. Walters. "Explaining Charter School Effectiveness." American Economic Journal: Applied Economics 5, no. 4 (2013): 1–27.

4. Abdulkadiroğlu, Atila, Parag A. Pathak, Jonathan Schellenberg, and Christopher R. Walters. "Do Parents Value School Effectiveness?" American Economic Review 110, no. 5 (2020): 1502–1539.

5. Mills, Jonathan N., and Patrick J. Wolf. "Vouchers in the Bayou: The Effects of the Louisiana Scholarship Program on Student Achievement After Two Years." Educational Evaluation and Policy Analysis 39, no. 3 (2017): 464–484.

6. Waddington, R. Joseph, and Mark Berends. "Impact of the Indiana Choice Scholarship Program: Achievement Effects for Students in Upper Elementary and Middle School." Journal of Policy Analysis and Management 37, no. 4 (2018): 783–808.

7. Figlio, David, and Krzysztof Karbownik. Evaluation of Ohio's EdChoice Scholarship Program: Selection, Competition, and Performance Effects. Columbus: Thomas B. Fordham Institute, 2016.

8. Witte, John F. The Market Approach to Education: An Analysis of America's First Voucher Program. Princeton: Princeton University Press, 2000.

9. Ravitch, Diane. The Death and Life of the Great American School System: How Testing and Choice Are Undermining Education. New York: Basic Books, 2010.

10. Aud, Susan, and Leon Michos. Spreading Freedom and Saving Money: The Fiscal Impact of the D.C. Voucher Program. Washington, DC: Cato Institute, 2006.

11. Dillon, Erin. Plotting School Choice: The Challenges of Crossing District Lines. Washington, DC: Education Sector, 2008.

12. Hoxby, Caroline M., and Sonali Murarka. "Charter Schools in New York City: Who Enrolls and How They Affect Their Students' Achievement." NBER Working Paper No. 14852. Cambridge, MA: National Bureau of Economic Research, 2009.