Advocates of expanding preschool point to studies conducted in programs such as those in Boston, New Jersey and Tulsa, Okla. as proof that early education has a powerful positive effect on young children.
But those studies have weaknesses that may be leading to exaggerated effects, argue Grover J. “Russ” Whitehurst, a senior fellow at the Brookings Institution and the former director of the Institute of Education Sciences within U.S. Department of Education, and David J. Armor, a professor emeritus at George Mason University in Fairfax, Va.
In an interview, Whitehurst said that the essay should not be taken to mean that the federal government should not have any role in expanding preschool. Rather, “there’s a lot more that we need to know,” he said. “Supporters of pre-K investment shouldn’t be sold a bill of goods.”
Whitehurst and Armor criticize the use of “regression discontinuity” to measure effects of preschool. As the essay explains, regression discontinuity studies take advantage of age cutoffs in state preschool programs to measure the set of children who have just completed a pre-K program and are entering kindergarten (the treatment group) against the set of children who are just entering the pre-K program (the control group).
Researchers make statistical adjustments for the one-year difference in the ages of the two sets of children, to make the two groups more comparable. As Whitehurst and Armor write, studies that use regression discontinuity “assume that the only systematic difference between the two groups of children other than age, for which they control statistically, is that the older group has attended the state pre-K program for 4-year-olds and the younger group has not. Accordingly, any differences in test scores must be due to the state pre-K program.”
But the paper offers reasons why there could be differences between the control group and treatment group that are not just caused by age and enrollment in preschool. For example, they say, the younger children in the control group may be less likely to have been enrolled in a program that focuses on school readiness at the time they are tested.
Whitehurst and Armor argue in favor of research that uses randomized samples, where one set of students who attend state preschool is measured against a set who do not. One example of such a study was conducted on students in Head Start; that study’s findings was that while Head Start children started school academically ahead of their peers, by 3rd grade such effects had faded out. (The Obama administration proposal, while directing more money to Head Start, would devote most of its funds to helping grow state-run preschool.)
Not all research into preschool uses regression discontinuity in its design. But the authors argue that other studies, such as the High/Scope Perry Preschool program, which has tracked children from the time they were toddlers until age 40-plus, are so unlike today’s state preschool programs that they cannot be generalized.
Whitehurst said he believes randomized research will find positive effects for preschool in children who are most in need.
“It’s another year of school. There will be some impact,” he said. But he said he is not pleased with the notion that preschool will make a major difference in the education trajectory of most children. “I don’t think it’s the magic cure. I think the results will be modest,” he said.