The Perry Preschool Study Stands the Test of Time, but It Doesn’t Stand Alone

Perhaps because the Perry Preschool study is cited so often to demonstrate the long-term benefits of preschool, it seems the landmark study is often criticized – or at least its flaws underscored in an attempt to discredit its findings as relevant to today’s world. This week I decided to address the validity of some of the criticisms I hear most often. At the same time I add a word of caution about relying too much on this (or any) single study.

Claim: The Perry study was poorly randomized and so its results cannot be trusted.

Response: No study is perfect, and the Perry study had minor problems in random assignment. However, these do not invalidate the results. For example, a few children assigned to the treatment group crossed over to the control group because their mothers worked and getting their children to and from a half-day preschool program was difficult. As children of mothers who were employed tended to do better on tests and in school than others, this had the effect of making the control group look better, biasing estimated effects downward. A recent re-analysis of the Perry study by James Heckman and others at the University of Chicago that takes into account the limitations of the randomization finds that the basic results were not altered.

Claim: The Perry program was not preschool education in the usual sense, because it had an extensive home-visiting program to provide maternal training.

Response: The home visits were designed to work with child and mother.  Their primary value may have been to help the teacher focus on individualizing their teaching to each child and providing one-on-one tutoring.  Data collected on the families do not show any effects on the home learning environment, including such activities as reading to the child. After the Perry study, a curriculum comparison study was conducted in which the frequency of home visits was reduced in the next study to once every other week with no apparent impact on children’s learning gains.  Neither meta-analyses nor direct experimental comparisons have confirmed the hypothesis that home-visiting involving parents adds to the effectiveness of preschool programs.  The Perry study’s results point to the importance of the teacher’s interaction with the child with primary emphasis on the classroom.  Other studies of programs that did not have home visits have produced similar results.

Claim: The Perry program results are because teachers in the school knew which children attended preschool and this advantaged those children.

Response: In 1962 many people thought that preschool might actually be harmful.  It seems unlikely that there was a strong expectation that these children would perform better because they had attended a preschool program.  The fact is that the children who had attended the preschool entered kindergarten with substantially stronger cognitive skills than did the control group.

Claim:  The Perry Preschool study is no longer relevant.  Too much has changed since the 1960s.

Response: Times have changed.  Most of the Perry study children were at home if not at the preschool program.  Only about a quarter of the mothers were employed, and 80 percent of the children lived in extended family households. Whether this was a better or worse learning environment than is provided by the child care many young children in low-income families experience today is difficult to judge. Learning environments provided today by child care for disadvantaged young children who do not attend Head Start or publicly-funded pre-K on average provide little or no advantages over the home and may be slightly worse for child development. So it is unclear that quality programs today face tougher “competition” by comparison.  Still, the basic point that much is changed has merit.  The Perry Preschool study is not the most relevant study today.  Fortunately, it does not stand alone. It remains relevant as one part of a larger body of evidence that includes many more recent studies that replicate key findings.

Claim:  The Perry Preschool study’s results have not been replicated.

Response: Replication of the results of the Perry study is important both for establishing that these results were not accidental and that they generalize to broader ranges of children and families, programs, and times.  So it is replication with different programs and populations in different places at different times that is particularly valuable.  In a statistical summary of the results of studies conducted since 1960, the Perry study appears to be rather average in its results compared to others with stronger research.

Among the studies that replicate the kinds of results produced by the Perry program are:

• the randomized trial of the Abecedarian program,
• the Institute for Developmental Studies randomized trial of preschool education in New York City (though data beyond elementary school are not available because they did not keep the sample up),
• a randomized trial of preschool education for disavantaged children in Mauritius that followed children to adulthood, and
• the quasi-experimental Chicago Longitudinal Study of children.

Conclusion:  The Perry Preschool Study has stood the test of time, but it remains just one important piece of evidence. Understanding the value of intensive, high-quality preschool education depends on the full body of scientific knowledge regarding child development and early education.  This includes scholarship—theoretical as well as empirical—that is not focused on the effects of preschool education per se.  Even the more narrowly defined evidence base of studies focused on the effectiveness of preschool programs is large and growing, as shown by the meta-analysis by Greg Camilli and colleagues cited above and by more traditional reviews (see, for example, my review at: http://epicpolicy.org/publication/preschool-education)

Only by taking into account all of the literature can we make adequately informed decisions about public policy and practice.  No single study, whether it is a 40 year follow-up like the Perry Preschool study or the most recent study of Head Start, state pre-K, or child care should be relied on in isolation.  In seeking the best for our children, we need to avoid the tendency to rely too much on either classics like the Perry Preschool study or the latest hot off the presses study claiming to provide definitive answers.

Steve Barnett*

Co-Director, NIEER

* As most readers know, I contributed to the Perry Preschool age 19, 27, and 40 benefit-cost studies.