Universal Pre-K: Why 2 Landmark Studies Disagree

Almost everything written about universal pre k argues from one study. Advocates cite Boston or Tulsa. Skeptics cite Tennessee. Both sides quote real research, which makes "does pre-K work" the wrong question. The useful one is which programs work, and what a state has to spend to run that kind.

Start with where the country is.

Who has universal pre-K?

Far fewer families than the phrase implies.

NIEER's 2025 yearbook, covering the 2024-2025 school year, recorded nearly 1.8 million children in state-funded preschool: 1.42 million 4-year-olds and 315,293 3-year-olds. That works out to 37% of 4-year-olds and 9% of 3-year-olds.

Six states fund no program meeting NIEER's definition at all: Idaho, Indiana, Montana, New Hampshire, South Dakota, and Wyoming. At the other end, the District of Columbia enrolls 94% of its 4-year-olds and Vermont 72%. Count state pre-K, preschool special education, and Head Start together and the national figure reaches 46% of 4-year-olds. More than half of American four-year-olds attend no publicly funded program of any kind.

Growth has also stalled. Enrollment rose by 44,000 children last year against 111,000 the year before, twenty states enrolled fewer preschoolers than the prior year, and federal COVID relief money that propped up the earlier expansion is gone.

What does the evidence show?

Boston is the strongest case in the yes column, because its design removes the usual objection.

Boston's public preschool was oversubscribed and used admissions lotteries, so Guthrie Gray-Lobe, Parag Pathak, and Christopher Walters could compare lottery winners against lottery losers instead of comparing families who chose preschool against families who did not. Published in the Quarterly Journal of Economics in 2023, the study found attendance raised on-time college enrollment by 8.3 percentage points, an 18% increase. It also raised SAT-taking and high school graduation and cut disciplinary outcomes including juvenile incarceration. Effects ran larger for boys.

One detail matters more than the headline: the study found no detectable effect on state achievement test scores. The kids who went to college did not test better at ten. Whatever preschool moved, standardized tests failed to measure it.

Tulsa's long-running evaluation of Oklahoma's universal program found a similar shape, with a reported 12 percentage point increase in college enrollment.

Why did Tennessee go the other way?

Because a different program produced a different result, and the design was strong enough that nobody can wave it off.

Vanderbilt researchers ran a randomized controlled trial of Tennessee's Voluntary Pre-K program, following children who entered in 2009-10 and 2010-11. Initial gains faded by the end of kindergarten. By third grade the effects had turned slightly negative. By sixth grade the pre-K group trailed its peers on achievement, drew more disciplinary infractions, and landed in special education more often.

That is the most rigorous pre-K trial run in this country and it came back negative. Anyone selling universal pre-K as settled has to answer it.

Sources: Quarterly Journal of Economics (2023); Vanderbilt Peabody Research Institute evaluation of Tennessee's Voluntary Pre-K program.

Read that cohort row before treating the two as a coin flip. These studies examined different programs, in different decades, with different populations and different alternatives available to the control group. They are not two answers to one question. They are evidence that the question has to be asked program by program.

What separates a program that works from one that harms?

Money per child, and what happens in the room.

NIEER rates state programs against 10 quality benchmarks covering teacher credentials, class size, ratios, curriculum support, and continuous improvement. In 2024-2025 a record six states met all ten: Alabama, Georgia, Hawaii, Michigan, Mississippi, and Rhode Island. Twenty programs met five or fewer, including three of the four largest. Those weaker programs enrolled 46% of all children in state pre-K, against 21% in the highest-rated ones.

Spending tracks that split. NIEER puts average state funding at $8,124 per enrolled child, or $9,988 counting local and federal dollars. The District of Columbia, New Jersey, and Oregon each topped $15,000 in state funding per child. Eight states spent under $4,000, and Florida came in at $2,963 from all reported sources. You cannot pay a credentialed teacher a public-school salary on $3,000 a child, so those programs do not, and teacher pay parity keeps showing up in the programs with good long-run results. Georgia's jump to all ten benchmarks this year came with $97.6 million in new money, part of it spent on K-12 salary parity.

One complication should bother both camps. Tennessee met 9 of NIEER's 10 benchmarks on paper. NIEER says so, and NIEER also warns that its benchmarks measure written policy rather than implementation, pointing to states that legislate a teacher degree requirement and then issue waivers because they cannot hire. A statute is not a classroom. Tennessee is the clearest evidence anyone has that the two can diverge, and only the classroom reaches a four-year-old.

What does the country spend on this?

States put nearly $14.4 billion into preschool in 2024-2025, and $17.7 billion counting local and federal dollars. Both are records. Spending rose 3% after inflation, against 17% the year before.

Set that against the private market it is meant to relieve. Child Care Aware of America puts the national average cost of child care near $13,000 a year per child, and care exceeds in-state college tuition in most states. At 37% coverage of a single age cohort, state pre-K is not displacing that market. It trims one year off the top of it for roughly a third of families.

Why is pre-K an affordability question?

Because for most parents the education debate is downstream of the bill.

A family enrolling a four-year-old in public pre-K is not primarily buying a test score. They are buying back a child care bill that averages near $13,000 a year and the ability to hold a job while they do it. That is why enrollment runs highest where programs offer a full school day and lowest where they run three hours, which solves nothing for a working parent.

The same logic runs through every other piece of this system. Head Start is funded for roughly one seat per four poor children under five. Child care subsidies reach about one eligible child in six. A single parent needs close to two incomes to clear a living wage with one child. Each program patches a hole, each is capped by an appropriation, and the holes keep widening because wages stopped tracking the cost of raising a child decades ago.

Universal pre-K is worth building, and the evidence says a cheap version can do damage. Fund a program at $3,000 a child, staff it with underpaid adults on credential waivers, and you have bought a building rather than an education. Fund it at $18,000 with teacher pay parity and the long-run numbers get very good. The country keeps choosing the first option and then debating whether preschool works. The data on what a family now has to earn explains why parents keep showing up anyway.