Why I Was Wrong About Head Start

By Kevin D. Williamson

Friday, May 16, 2025

 

One should always be open to reevaluating long-held beliefs—and an especially good time to reevaluate them is when a guy with a Nobel Prize in the relevant subject tells you that you’ve got it wrong.

 

In at least a half a dozen articles and speeches, probably more, I have repeated something that I’ve understood to be a well-established fact for so long that I do not even remember when or where I first learned it: that Head Start does not work, that it provides no meaningful lasting results. Professor James Heckman of the University of Chicago, inconveniently enough for my longstanding belief, not only was awarded the Nobel Prize in economics (that is, the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, as Jay Nordlinger taught me) but was so honored specifically for his work on developing rigorous methods for the evaluation of social programs. I do not immediately knuckle under to appeals to authority, but I am inclined to listen to guys who have equations named after them.

 

Speaking at the Old Parkland Conference at the American Enterprise Institute in Washington this week, Heckman insisted—and not for the first time—that the mostly conservative critics of Head Start have it wrong, and that conclusions about the program’s ineffectiveness are based on bad information.

 

(The Old Parkland Conference is a recurring symposium on black socioeconomic mobility and related subjects, inspired by Thomas Sowell’s 1980 Fairmont Conference. Old Parkland, where the first meeting was held, is the Dallas office campus owned by Harlan Crow, a longtime AEI trustee and financial supporter of the conference who is, I should note in the interest of full disclosure, an investor in The Dispatch and, more important, my friend.)

 

Heckman, who does not want for confidence in his convictions, rejects the notion that randomized trials should be understood as the “gold standard” and mocks those who believe otherwise as a “cult.” But, as he tells the story, even if we were to accept the primacy of randomized trials here, we’d want them to be good randomized trials. “This all really comes from one experiment,” he says, referring to the 2005 Head Start Impact Study. “Students were randomized out of Head Start, and the ones randomized out were the control group. But what were they randomized out into?” Head Start, and pre-K education more generally, is a varied and decentralized enterprise, and many of the students randomized out of Head Start in the experiment in question ended up attending other Head Start programs or other kinds of preschool. “Some of them went to Head Start elsewhere. Some of them went to something better.” Better data from a better sample produces different results—results that point to a different outcome about Head Start’s efficacy.

 

“I don’t love Head Start,” Heckman says. “There are better ways to do it. But the notion that it just doesn’t work at all isn’t supported by the evidence.” I asked Heckman if the focus on randomized experimental data was another example of the hard-science envy among economists noted by F.A. Hayek in his Nobel Prize lecture on “the pretense of knowledge,” as he put it. “That’s it, yes,” he said.

 

Heckman’s work has focused in part on the Perry Preschool Project, which was (ahem!) a randomized study of children “treated” with preschool in the 1960s—long ago enough that we have a great deal of information not only about their life outcomes but also those of their children. Heckman’s research summary reports:

 

Children treated with early childhood education have significantly better life outcomes than the untreated children. Treatment in Perry significantly increased the participants’ employment, health, cognitive and socioemotional skills and reduced the male participants’ criminal activity, especially violent crime. Improvements in childhood home environments and parental attachment are seen as an important source of the long-term benefits of the program.

 

. . .

 

The children of participants were less likely to be suspended from school, and more likely to complete regular or any other form of high school and to be employed full-time with some college experience. While present for both male and female children of participants, the wide range of beneficial effects are particularly strong for the male children of participants, especially those of male participants.

 

Good preschool programs, in Heckman’s telling, give students some of the same things they get from good parenting: attention, examples to learn from, mental stimulation, etc. It is these things that matter, not whether the benefits are transmitted through a federally supported program.

 

I thought of an observation from Yale psychologist Paul Bloom made on a recent episode of The Remnant podcast: On average, people with psychological problems do better with therapy than without therapy—so, in that sense, therapy works. But there does not seem to be much difference in terms of patient outcomes between different therapeutic methods and techniques—rather, outcomes seem to vary most strongly by therapist. In that sense, it would be less accurate to say that this or that form of therapy works than that this or that therapist produces good results. We might expect to see something similar when it comes to preschool education.

 

And that would fit in with the evidence we have from K-12 education, which suggests that of all the in-school factors that shape educational outcomes, teacher quality matters most: “When it comes to student performance on reading and math tests, teachers are estimated to have two to three times the effect of any other school factor, including services, facilities, and even leadership,” as one RAND report put it.

 

The Trump administration has been making desultory war on Head Start even as the lunatic who runs the Department of Health and Human Services, conspiracy quack Robert F. Kennedy Jr., assures Congress that funding for the program will not be cut, and that the recent financial chaos in the program was not the work of the Trump administration (which, in reality, has suggested eliminating the program entirely) but rather that of disgruntled employees “who wanted to make the Trump administration look bad.”

 

The question of whether early-childhood education works is separate from the question of whether a federal program is the best way to go about providing it, which is again separate from the question of whether this federal program is the best way to go about providing it. My best guesses right now would be: yes, no, and no—and if I were a policymaker looking for advice on the question, I’d be more interested in James Heckman’s views than in those of Robert F. Kennedy Jr. and those in his wobbly, messy orbit.

 

And as for my earlier views on the subject of early-childhood education: I think I stand corrected—but I may be wrong about that.