What the Louisiana Voucher Studies Found About Test Scores

In the spring of 2012, Louisiana's legislature passed a package of education bills pushed by Governor Bobby Jindal, and one of them took a voucher program that had operated only in New Orleans and opened it across the state. Low- and moderate-income children attending schools the state graded C, D or F could apply for a scholarship covering private tuition, up to the amount the state would have spent on them in public school. Thousands of families applied that first summer. Because more students wanted seats in some private schools than those schools had room for, the state ran lotteries. Nobody designing the program was thinking about social science when they set up those lotteries. But a lottery is the closest thing education policy has to a controlled experiment, and the researchers who later studied Louisiana's program found results that would follow the voucher debate for years.

In short: Lottery-based studies of the Louisiana Scholarship Program found large drops in math scores for students who used vouchers in the first years, the biggest negative effects in the history of voucher research. Most of those losses later faded to roughly no difference. Parents who chose the program generally reported being satisfied, and both facts belong in any honest account of it.

How the program was built

The Louisiana Scholarship Program had several features that matter for reading the research.

First, eligibility was limited. A family had to have income at or below 250 percent of the federal poverty line, and the child had to be entering kindergarten or coming from a public school with a C grade or lower under the state's rating system. This was a targeted program, closer in spirit to the original Milwaukee model than to the universal accounts some states have adopted since.

Second, participating private schools had to accept the voucher as full payment and admit voucher students by lottery if they were oversubscribed. They also had to give scholarship students the state's standardized tests, which most voucher programs in other states do not require. That testing requirement is the reason the Louisiana evidence is so strong. Researchers could compare students who won a lottery and used a voucher with students who applied, lost and stayed in public school, using the same exam.

Third, the private schools that chose to participate were not a random slice of Louisiana's private schools. That fact turns out to be important, and we will come back to it.

The program also faced early legal trouble. In 2013 the Louisiana Supreme Court ruled that the state could not pay for vouchers out of the main public school funding formula, which the state constitution reserves for public schools, so lawmakers moved the program to a separate line in the budget. The same year, the U.S. Justice Department went to federal court asking whether vouchers were undermining desegregation orders still in effect in some parishes, a dispute that ended with the state agreeing to provide data on the program's effects on school racial balance.

One more design detail shaped the evidence. The voucher's value was capped at what the state would have spent on the child in public school, which in Louisiana was often less than tuition at well-established private schools. A program that pays less than the going rate will mostly attract schools willing to accept less. That is an economic fact before it is an educational one.

The first results were not small

Two research teams studied the lottery data.

A team of economists from MIT, Duke and the University of California at Berkeley, Atila Abdulkadiroglu, Parag Pathak and Christopher Walters, analyzed the first year of the statewide program. They found that students who used a voucher scored substantially lower in math than students who lost the lottery, an effect large enough that it stunned people who had followed voucher research for years. Previous studies of vouchers in Milwaukee, Washington and elsewhere had usually found small effects in one direction or the other, often not distinguishable from zero. Losses in reading, science and social studies were also present but less dramatic.

A second team, based at the University of Arkansas and working with the Education Research Alliance for New Orleans at Tulane, followed students over several years. Jonathan Mills and Patrick Wolf, whose department has generally been sympathetic to school choice, reported large negative effects on math after the first and second years. Then the pattern shifted. By the third and fourth years, the differences between voucher users and lottery losers had shrunk, and in their four-year analysis the math and reading effects were no longer statistically significant. The researchers also looked at longer-term outcomes, such as college enrollment, and did not find the voucher users doing better.

That is the honest summary. Big early losses, unusual in their size, followed by a fade to roughly no difference on test scores. No sign that the program raised achievement, and no strong evidence that it permanently lowered it.

Why might the scores have dropped?

Researchers have offered several explanations, and they are not mutually exclusive.

The schools were different from the ones in brochures. When economists looked at which private schools joined the program, they found that participating schools tended to have had falling enrollment and lower tuition than private schools that stayed out. A school with empty seats and a shrinking budget has a strong incentive to accept a voucher worth several thousand dollars. A well-resourced school with a waiting list has less reason to accept the state's testing requirements and admission rules. In other words, the program may have connected low-income students not to the best private schools in Louisiana but to the ones that needed the money most.

Curriculum mismatch. Private schools are not required to teach to state standards. Students arriving from public schools may have encountered different material in a different sequence, while being tested on the public school curriculum. That might explain part of the first-year drop and part of the later recovery, as schools adjusted. It cannot explain all of it, because the losses were too large.

Disruption. Changing schools is hard on children, especially in the middle of the year and especially for children from families already under financial strain. Any school transfer tends to produce a short-term dip.

Louisiana's experience was not an isolated one. Studies of Indiana's statewide program and of Ohio's EdChoice program, using different methods, also found negative effects on math scores, at least in the early years. Taken together, this wave of research ended the hopeful assumption, common in the 1990s, that private schools would reliably outperform public ones once poor children could get in.

The case the parents made

Here is where the story requires care. Studies of the Louisiana program also surveyed families, and parents who used vouchers generally reported high satisfaction with their new schools. Many valued things a state test does not measure: a religious environment, smaller classes, a sense of safety, a school culture they trusted.

It would be condescending to dismiss that. Low-income parents are not confused about what their children need. They may weigh discipline, faith or a fresh start differently from how an economist would. Some may reasonably prefer a school where their child is known and safe, even if math scores are no better.

There are two honest responses, and both deserve space.

One is that satisfaction and achievement measure different things, and public money should be judged on both. A state that spends tax dollars on a program in the name of rescuing children from failing schools should not be indifferent to whether those children learn more math. If the policy's promise was better academic outcomes, the evidence matters.

The other is that the parents' choice points to something public schools in Louisiana and elsewhere have often failed to provide: an environment that low-income families trust. When a sizeable number of parents prefer a struggling private school to their assigned public school even with no academic gain, the public system should treat that as information about itself, not just about the parents.

A short guide to reading a voucher study

Most readers will never open the academic papers, and they shouldn't have to. But a few questions will help anyone judge the claims made about them, in Louisiana or anywhere else.

Who is being compared with whom? The strongest studies compare lottery winners with lottery losers, because chance decided who got the voucher. Weaker studies compare voucher users with students who never applied, and those comparisons are vulnerable to the fact that families who apply may differ in ways that matter. When a report says "voucher students did better," ask: better than whom?

Which test, and when? An effect measured after one year may not hold after four. Effects in math and reading often differ. A study that reports only the subject or year that looks good is telling you less than it seems.

What happened to the students who left? Many voucher users return to public school within a few years. A study that follows only students who stayed in the program is studying a self-selected group. The Louisiana researchers dealt with this by analyzing students according to whether they won the lottery, regardless of where they ended up, which gives a cleaner but more conservative answer.

Is the result statistically distinguishable from zero? A finding that is "not significant" does not mean the program had no effect. It means the study couldn't tell an effect apart from chance. That is a common reason the same study gets described as "no harm" by one side and "no benefit" by the other.

Who paid for it, and does it matter? The Louisiana work was carried out by researchers with differing priors, including some at a department known for favoring school choice, and they reached broadly compatible conclusions. That convergence is one reason the findings carry weight.

Keep those questions handy. Every legislative session in a state debating vouchers brings a new round of studies summarized in a sentence.

How both sides use these studies, and what they skip

Voucher opponents cite the first-year math results constantly, sometimes as if they were the final word. They were not. The fade toward zero is part of the record, and leaving it out is no more honest than voucher boosters leaving out the early losses.

Voucher supporters have their own selective reading. Some emphasize the later years and the parent surveys and treat the initial drop as a transition cost. Others argue that test scores are a poor measure of what matters in schools, which is an argument worth having, but it sits awkwardly beside the decades in which school choice advocates used test scores to condemn public schools.

The more useful lesson from Louisiana concerns program design. A voucher only works as well as the schools that accept it. If the best private schools stay out and the struggling ones sign up, the program becomes a pipeline into the weakest part of the private sector. Louisiana actually had stronger accountability rules than most later programs: testing requirements, enrollment lotteries and the possibility of being barred from taking new voucher students. Many of the universal account programs adopted since 2022 have less, not more. That should worry anyone who believes public money ought to come with evidence that it works.

In 2024 Louisiana itself moved toward a broader education savings account program, under a new governor, Jeff Landry. The evidence from the earlier, more tightly regulated program is an important baseline for judging what comes next.

Another lesson concerns time. The first-year results produced headlines; the four-year results, which told a more complicated story, received far less attention. Policy debates move faster than evidence, and a single striking number often outlives the fuller picture that replaces it.

What the evidence asks of progressives

If you believe, as this site does, that public dollars should go first to the schools most children attend, the Louisiana research supports your position, but carefully. It shows that vouchers did not deliver the gains their sponsors promised. It does not show that every family who used one made a mistake, and it should not be wielded against parents.

The constructive response is to build public schools that low-income families want to choose: schools with strong teachers, safe buildings, real counseling, serious academic programs and the respect families feel is often missing. That takes money and patience. It also takes listening to the parents who signed up for a voucher, because the reasons they gave are a report card of their own.

What you can do

Whether your state has a targeted voucher like Louisiana's original program or a newer universal account, the details of eligibility and accountability are decided by lawmakers who rarely hear from voters about them.

• Ask your state legislators what results your state's voucher or account program has to report and who checks them. Find them, with phone numbers and their recent votes, at the representatives lookup.
• Read new research and reporting on choice programs as it appears on the education topic page.
• If you want to support public schools in low-income communities directly, look for local education foundations, tutoring programs and parent organizing groups in the organizations directory.
• If your district has a public comment period on its budget or school improvement plan, go and speak. Bring the specific question parents in your area are asking, and ask the district what it will do about it.