Pre-K Can Be Bad For Some Kids

As the universal child care proposals in Build Back Better take a break from the spotlight—Democrats are regrouping after failing to pass the president’s high-priority bill last year—a team of researchers has dropped a major pre-K study in the journal, Developmental Psychology. The news is not good, and it joins a mixed literature showing, at minimum, that the effectiveness of pre-K programs is far from assured.

The study is based on what was, essentially, a controlled experiment: Some schools in Tennessee’s state-funded pre-K program (TN-VPK) were “oversubscribed,” and the contested seats were handed out through a random lottery. The researchers have been following a group of kids from the cohort that started in the 2009-10 and 2010-11 school years, comparing those who won these lotteries with those who lost. The new study tracks the kids all the way through the sixth grade.

The group’s previous studies had already produced some discouraging results. Sure, like many pre-K programs, Tennessee’s had some positive short-term  results: At the end of the program, participants outscored the lottery losers on achievement tests. But the achievement effects had basically faded out by the end of kindergarten, and by the end of third grade, the participants were scoring measurably worse on some achievement measures. 

The non-academic measures weren’t much more promising: “children who attended TN-VPK had marginally significant higher rates of school rule violations and a significantly greater proportion of TN-VPK participants had special education placements.”

The new results from sixth grade continue the pattern. The non-participants scored higher on math, reading, and science,¹ and by larger margins than they had in third grade. Non-participants had slightly higher school attendance rates and substantially lower rates of disciplinary events.

This study is, of course, limited to one program, so it’s worth thinking through how it fits into the broader literature on the subject. Pre-K takes kids out of one setting—either a family member’s care, or whatever other arrangement the family prefers and can afford—and places them in another setting. It stands to reason that both parts of that equation matter, and pre-K will have better effects as it targets more deprived children and provides higher-quality care. Indeed, the most spectacular claims for these programs come from tiny, intensive efforts targeted at poor children decades ago, such as the Perry Preschool and Abecedarian projects, while studies of other, broader efforts have produced decidedly mixed results. The idea that programs will be especially helpful for kids from more disadvantaged backgrounds, or are more likely to be harmful to less disadvantaged children with better alternative care arrangements, is also fairly common in studies of child-care options for younger kids.

But beyond these generalities, it seems extremely hard to predict what long-term effect a program will have—and when, exactly, those effects will show up or fade out—before putting it into effect. Those looking to defend pre-K, for example, might point to another recent study of pre-K among Boston kids who started around the turn of the century (about a decade before the kids in the Tennessee study, allowing even longer-term follow up). Using a similar admissions-lottery strategy, the authors  found that while the Boston program didn’t boost achievement-test scores, it did promote “college attendance, as well as SAT test-taking and high school graduation,” while decreasing “several disciplinary measures including juvenile incarceration.”

One can hope that good effects of Tennessee’s program will magically pop out in another five or  10 years, similar to how early studies of the Moving to Opportunity program found disappointing results from moving poor families to better areas, but later research found positive effects. Another easy way to reconcile the difference would be to say that the Tennessee program is just bad—but as the authors of this new study explain in detail, it’s actually a decent program along numerous metrics:

A recent review of statewide programs by the NIEER group (Friedman-Krauss et al., 2019) praised the program in Tennessee for being among those in 27 states that paid its pre-K teachers at parity with elementary teachers, one of only 26 states to offer pre-K teachers retirement benefits, health care and paid time off, and one of only 25 to require its teachers to have a bachelor's degree plus certification. Among state-funded pre-K programs, the TN program is above average and arguably in the top tier on characteristics many believe mark high quality. ... Judged alternatively by its performance in producing [short-term] student gains on commonly measured cognitive outcomes, ... [a study has] found that TN-VPK ranks as a top tier program when compared to the results of similar studies in other states.

In other words, quality-wise, the TN program seems to be about what we can expect from state pre-K programs, maybe even above average, until we figure out in more detail what a “high-quality” pre-K program even is. (The study briefly discusses some possibilities in that regard.)² And in terms of focusing on disadvantaged kids, the new study includes only those eligible for free or reduced-priced lunch. Yet in this case, pre-K seems to cause some negative effects for kids who attended—at least through sixth grade.

The upshot isn’t that all pre-K is bad. Some kids certainly benefit, some programs seem to help, and many working and/or single parents have little choice but to rely on these programs, anyway (my own middle child is in preschool as I type this).

If pre-K is sometimes harmful to kids, though, perhaps the takeaway here is that the government shouldn’t spend a bunch of money on preschool-specific subsidies and on promoting universal early child care programs—and find other ways to help families instead.

Robert VerBruggen is an IFS research fellow and a fellow at the Manhattan Institute.

1.  The “treatment on the treated” effect sizes in sixth grade–which estimate the impact of actually attending the program, not just being randomized to be accepted–are measured as “the coefficient for the treatment-control difference divided by the pooled standard deviation,” with most results falling between −0.1 and −0.4 in a variety of models that vary by subsample, outcome, and weighting. See Tables 2 and 3 for full details, as well as the by-definition-smaller “intent-to-treat” results measuring the effect of being accepted without necessarily attending.

2.  One key distinction is between “constrained” or “finite” skills like learning the alphabet, versus broader “unconstrained” skills like attention and working memory. Pre-K programs already do a decent job of teaching the former, but these effects are the ones that fade out as non-attenders catch up. The latter might, in theory, be the overlooked key to lasting gains.