Charting Pre-K's Value for All
When President Barack Obama announced his support for universal preschool in his State of the Union address this year, he rekindled a fierce debate. Supporters praised universal preschool as an excellent "investment" in the nation's future workforce. Critics lambasted it as yet another example of wasteful federal spending.
We have grown accustomed to cacophonous debates, so this is not unexpected. Yet, if you look at the research on pre-K learning in the United States, there is a surprising degree of consensus among researchers in early-childhood education. From this, bipartisan support for some type of federally supported universal preschool program could grow over time—an important development given that the National Institute for Early Education Research reported just last week that state pre-K funding has declined.
Let's begin with the common ground. Most early-childhood-education researchers would agree with the following three propositions:
First, a high-quality pre-K program can boost school readiness substantially, for disadvantaged children. This has been amply demonstrated by the Perry Preschool Project, the Abecedarian Project, the Chicago Child-Parent Centers Project, and many others. High-quality universal preschool programs in Georgia and Oklahoma have also demonstrated big gains in school readiness for poor children.
Second, the cognitive benefits from pre-K often decline from kindergarten to 3rd grade. Sometimes, the "fade out" is modest, as in the case of the Chicago Child-Parent Centers. Sometimes, it is dramatic, as with a 2010 Head Start evaluation. Note, though, that the Head Start evaluation included high-quality, low-quality, and medium-quality programs.
Third, even if some initial fade-out occurs, the long-term benefits of a high-quality pre-K program can be substantial. These include higher high school graduation rates, lower rates of juvenile delinquency, less substance abuse, and higher adult earnings.
Thus, many studies show that high-quality pre-K programs can improve outcomes for disadvantaged children in the short run and generate favorable returns for taxpayers in the long run.
So, what's the fuss about? Despite some common ground, experts disagree, to some extent, on three other questions: First, does a high-quality pre-K program benefit middle-class children? Second, how substantial is fade-out for large-scale programs? Third, do the benefits of large-scale universal pre-K programs exceed the costs?
Until recently, it has been difficult to study the effects of government-funded pre-K on middle-class children because most pre-K programs have been aimed at poor children. However, we are beginning to accumulate evidence from the largest and most mature of the universal preschool programs—namely, Georgia's and Oklahoma's. In our study at Georgetown University of the school-based pre-K program in Tulsa, Okla., we found that middle-class children who participated in pre-K were seven months ahead of their peers in reading, four months ahead in writing, and four months ahead in math. Disadvantaged children benefited even more, but middle-class children benefited substantially. A recent evaluation of Boston's universal preschool program also detected substantial effects on middle-class students. Studies of Georgia pre-K program participants have thus far not directly addressed the question of middle-class effects, though they have concluded that the presence of middle-class children in pre-K classrooms benefits disadvantaged children.
Fade-out has sparked considerable controversy, especially over Head Start. Unfortunately, these debates do not tell us what we need to know about a high-quality universal preschool program because a different mix of students is involved and because Head Start programs vary sharply in quality. We have followed Tulsa pre-K students and a control group of students over time, for two cohorts of students.
For the early cohort, which enrolled shortly after the universal pre-K program began, we found no persistent pre-K effects through grade 3. For the later cohort, which enrolled a few years later, we did find persistent pre-K effects through grade 3. Specifically, boys who participated in pre-K had higher math scores than boys who did not. We also found significant differences between pre-K program participants and nonparticipants for disadvantaged students, and marginally significant differences for middle-class students, with pre-K participants scoring higher. We believe that it took a few years for K-3 teachers to change their expectations and their pedagogy in response to the new influx of pre-K alumni. Once this happened, teachers were able to capitalize on these better-prepared students.
In a Georgia study using data from the National Assessment of Educational Progress for children as a whole (not distinguishing between pre-K participants and nonparticipants), Cornell University researcher Maria Fitzpatrick also found some persistent pre-K effects over time, though they seemed stronger for rural children than for urban children or those on the urban "fringe"—where an urban area and its suburbs give way to country.
A lively discussion has also ensued over benefit-cost analysis—how it should be conducted, and what inferences can be drawn from prior work for the current universal preschool debate. Studies of the Perry Preschool program and the Chicago Child-Parent Centers program have yielded benefit-cost ratios of 7-to-1 or higher. But are these estimates applicable to universal pre-K programs in place today, such as the Georgia and Oklahoma programs? In her study of the Georgia pre-K program, Fitzpatrick concluded that the costs actually exceed the benefits. That conclusion has been challenged on many grounds, including the fact that Fitzpatrick apparently looked only at benefits to the government, as opposed to benefits to society as a whole. Another key limitation of the study is that it relied on strong inferences from an interstate NAEP comparison that did not directly compare pre-K participants and nonparticipants. Such comparisons assume, among other things, that state demographics change similarly across states over time and that differences in education spending levels are irrelevant.
Several researchers have tried to project future benefits from Georgia and Oklahoma, splicing together data from these states with data from other states linking kindergarten test scores to adult earnings and/or other benefits. Using this technique, economist Robert Lynch estimated that the benefits of the Georgia pre-K program are likely to exceed the costs by about 5-to-1. Recently, Timothy Bartik of the W.E. Upjohn Institute for Employment Research estimated that the adult-earnings benefits of the Tulsa pre-K program alone are likely to exceed the costs by 3- or 4-to-1. If criminal-justice effects were also factored in, the benefit-cost ratio could be higher still.
What, then, can we conclude from the pre-K evidence that is in dispute? First, poor children need more help, and they benefit more from a high-quality pre-K program than middle-class children do. But middle-class children benefit, too.
Second, some fade-out of short-term cognitive gains seems likely over time, even if the pre-K program is high in quality. But fade-out can be reduced by strategic decisions that reshape K-3 curricula with higher expectations. Also, stronger "soft skills" (such as attentiveness) may lead to other benefits down the road.
Third, the benefits of a high-quality universal pre-K program are likely to exceed the costs, and by a substantial margin. In our Tulsa projections, we found this to be true for both disadvantaged and middle-class children.
Given the weight of the evidence, one solution would be federal support for state initiatives to establish, expand, or improve pre-K programs.
Most states would probably choose to invest most of their money in disadvantaged children first. But some states would probably support middle-class children as well. As this occurred, we would accumulate more valuable evidence on just how effective a universal preschool program could be.
Vol. 32, Issue 30, Pages 26-27, 29