Head Start is one of the most important, and confusing, anti-poverty programs in existence. It serves nearly a million kids a year, at a cost of about $8.5 billion dollars this year to the federal government. It's been in place for a half century. And we have no solid idea if it really works.
Which is not to say that there isn't research on Head Start. There's a pile of research on Head Start. But the findings of various studies are contradictory. And the biggest, most-widely cited study of Head Start's effectiveness is routinely misinterpreted.
One of the problems with assessing Head Start is that Head Start isn't one thing. It's run as a grant program. That means the government sends a check to a local group to operate its own Head Start classrooms. There are rules for how those classrooms should be run – lots of rules – but each Head Start grantee does have flexibility in choosing a curriculum, offering certain services, and in hiring its own teachers (actually, Head Start parents get a big voice in choosing teachers, but that's another thing altogether).
That creates a challenge for researchers, because there can be wide variation between different Head Start centers around the country. And some Head Start centers seem to provide much bigger benefits to kids than others.
A new working paper by Berkeley economist Christopher Walters tries to pinpoint the factors that separate the good from the not-so-good in Head Start programs across the country.
Walters uses data from the large-scale Head Start Impact Study mentioned above. He applies a statistical analysis to a number of variables in both the child's home environment and the Head Start center characteristics to try to find out which variables matter, and which don't.
We've reported here in the past that teacher training is one of the core-characteristics of a high-quality preschool program. But Walters' analysis says neither teacher certifications, nor the number of teachers with bachelor's degrees is a significant factor in the effectiveness of a Head Start center.
In this analysis, there were two variables that stood out: full-day classes and home-visitation programs. Head Start programs that offered both of these services showed bigger gains in student growth than those that did not.
But the biggest surprise for me about this study is what didn't turn out to be significant: teacher education level. We've reported here in the past that teacher training is one of the core characteristics of a high-quality preschool program. But Walters' analysis says neither teacher certifications, nor the number of teachers with bachelor's degrees is a significant factor in the effectiveness of a Head Start center.
This finding is significant, because the federal government has been phasing in new requirements for Head Start teachers to make sure that each classroom has a qualified teacher. If Walters' analysis is correct, this may turn out to be the wrong policy prescription for improving Head Start's effectiveness.
To get a quick take on Walters' research, I turned to Tim Bartik, of the Upjohn Institute. Bartik has written two books evaluating the research on all kinds of preschool programs (his latest, From Preschool to Prosperity, is available free online here).
First, Bartik told me we need to take into account the study's method. Because the study used a statistical analysis rather than a randomized experimental model, there's less certainty about its result. In fact, Walters says as much in his paper.
Bartik writes in "From Preschool to Prosperity" that teacher interactions with kids definitely make a difference in learning outcomes. The question is how to improve those interactions. It turns out, the overall message from previous research is pretty mixed when it comes to requiring teacher certifications, or of even minimizing class sizes. Still, Bartik argues both factors could still be important. We just don't really know for sure yet:
Despite this mixed research, it seems prudent for policymakers to assume that smaller class sizes and stronger teacher credentials will facilitate greater program effectiveness. One rationale for this policy advice is that it seems plausible: Better teacher-child interactions are probably easier to attain if the class size is more modest and the teacher is better trained. Second, some of the best pre-K programs with favorable evaluations used certified teachers and had moderate class sizes ... Lowering quality standards too much from these models increases the risk of lower effectiveness.
Bartik says another important factor to consider is teacher pay. Walters didn't include teacher pay as a variable in his latest paper. But Bartik says low pay can lead to high turnover, especially if teachers are required to be highly trained. Those with training will find it easier, and more enticing to seek a job with better pay.
So, this new paper on Head Start characteristics adds to the discussion, but it doesn't close the case. That would be too easy, and too straightforward.
