Background

Modern ECCE research began in the 1960s, with the birth of Head Start in 1965. Because national spending on such research has been light—compared to national spending on medical research and national spending on ECCE itself, for example—most studies are small-scale, lacking both random assignment of children and wide representativeness, and attention has concentrated on those few studies that have at least random assignment—such as the HighScope Perry Preschool Study (Schweinhart et al. 1969). This apparent fadeout of a crucial effect led many people, notably hereditarian Jensen (1969), to conclude that all the effects of early childhood program experience fade away. Despite the theoretical predictions, permanently improved intellectual performance was not to be the vehicle of long-term effects.

The HighScope Perry Preschool Study and studies like it continued to collect data as children grew up. These studies have become exemplary in part because the interventions were highly intentional. Toward the end of elementary school, the Perry study began to find that children with early childhood program experience had better age-grade placement than their randomly assigned peers—fewer of them were assigned to special education or repeated a grade (Schweinhart and Weikart 1980). In addition, the young people with early childhood program experience had higher achievement test scores and more of them graduated from high school. These effects occurred despite, possibly even because of, the temporary effect on intellectual performance (Schweinhart 2016). These findings for age-grading and high school graduation were also found in several other studies of the Consortium for Longitudinal Studies (Lazar et al. 1982). The same combination of effects was found in Michigan’s Great Start Readiness Program longitudinal evaluation (Schweinhart et al. 2012).

Over the years, the Consortium disbanded, and a new generation of researchers replaced the older one. The Carolina Abecedarian Enhanced Child Care Project began in 1972; like Perry, it employed random assignment techniques. The Chicago Child-Parent Centers that began in 1967 began its longitudinal study in 1986 (Reynolds et al. 2011); while it did not employ random assignment, it was a relatively large-scale program, operating across the city of Chicago. The HighScope Perry Preschool Study found that the early childhood program group adults did better than the no-program-group adults by having higher school achievement, a higher high school graduation rate, a higher employment rate, higher earnings, a lower crime rate, and strong return on investment (Schweinhart et al. 2005). The Abecedarian Project (Campbell et al. 2002) found all these effects except crime and also found persistently higher intellectual performance and college attendance. The Chicago project (Reynolds et al. 2011) found the same effects as Perry. These three studies have emerged as the standard-bearers for the finding that high-quality early childhood programs for children living in poverty have long-term effects and strong return on investment. Table 1 presents the principal characteristics and outcomes of these studies. The diversity of the programs finding similar results suggests that successful programs can be designed in many different ways.

Table 1 Characteristics of three long-term preschool studies
Full size table

Other studies, however, have not found such promising short- and long-term effects for early childhood programs as these studies found. In particular, the Head Start Impact Study (Puma et al. 2012)—which was essentially the second national evaluation of Head Start—found only weak and ephemeral effects on children’s literacy, mathematics, and social skills. This was particularly disappointing because it sought to remedy the defects of the first evaluation, and Head Start had driven forward on the strength of the longitudinal studies that had found long-term effects. It sought to employ random assignment, although changing conditions made it difficult to achieve—crossovers bedeviled the design, with 60 % of those assigned to no Head Start finding a way to attend an early childhood program anyway. It also sought to draw a nationally representative sample of Head Start programs, despite the challenge of some Head Start programs serving most of the eligible children in their catchment areas. Another study that sought to apply random assignment to a large population and found disappointing results was the Tennessee Voluntary Prekindergarten Study (Lipsey et al. 2013). It found immediate but not longer-term effects on the children it served.

Yet other studies have continued to find strong short-term early childhood program effects, in Tulsa (Gormley et al. 2005), Boston (Weiland and Yoshikawa 2013), and various states (Barnett et al. 2005). Heckman and his colleagues have conducted reanalyses of the Perry and Abecedarian studies that do not rely on parametric assumptions and basically confirm the original results (Campbell et al. 2014; Heckman et al. 2010a, b). The question is no longer whether early childhood programs can have long-term effects, but rather which ones do and which ones do not. A leading hypothesis for explaining the difference that fits the data well is that high-quality early childhood programs have long-term effects while low-quality programs do not.

It places the three standard-bearing programs and programs in Tulsa and Boston in the high-quality camp and the national Head Start program and Tennessee’s Voluntary Prekindergarten Program in the mixed-quality camp. It is easy to see one difference between the first programs and the federal and state programs. The scale is different, even between the city-wide Chicago program and programs that serve the nation or Tennessee. The national program, Head Start, has a complex history, involving many policy decisions made for purposes other than to maximize program effectiveness. But surely scale alone cannot explain the difference between success and failure, and the difference between the Tulsa early childhood program and the Tennessee prekindergarten program is more highly nuanced.

The dissemination

Like virtually all countries, the U.S. is governed by policymakers—heads of state, bureaucrats, legislators, and judges at federal, state, and local levels. Because the U.S. is a democratic republic, citizens vote on candidates for public office and big issues, delegating most of their authority to elected policy makers. Other countries have various forms of government, but everywhere policy makers are responsible for most policies.

The dissemination of early childhood program research findings began in academic journals and later extended to policy makers and the public through the mass media. HighScope played a special role in disseminating the results through monographs; articles; press conferences; and speeches to behavioral scientists, policy makers, educators, early childhood educators, and groups of early childhood advocates in most states.

The dissemination of program research and evaluation almost always extends between the early childhood researchers who conduct such studies and the administrators who commission them. The participating children and their parents have a special right to know the results as do others like them. But early childhood program research has extended beyond informing to advocacy, because of the nature of the results and their history.

Early childhood program research that began in the 1960s was based on the bold hypothesis that early childhood programs could affect lives. It echoed an earlier generation’s Freudian belief in the priority of early childhood (Freud 1905), but in cognitive rather than psychosexual development. This belief was confirmed by a first round of studies, seemingly disconfirmed by a second round of studies, reconfirmed by a third round of studies, leading to current studies seemingly supporting both conclusions. At each round, the hypothesis was refined, first to the idea that not all early childhood programs have long-term effect, only high-quality ones.

The first round of studies was disseminated by the usual academic journals, papers, and presentations, as was the second round challenge. Thus, when the Consortium’s finding came out, academic journals were not considered enough to get the word to relevant policy makers. The Consortium published its findings in a scientific monograph published by the Society for Research in Child Development (Lazar et al. 1982) and in a book to which each team of investigators contributed a chapter (Consortium for Longitudinal Studies 1983), but also communicated with groups of policy makers directly and immediately, by numerous speeches by Irving Lazar and other members of the Consortium.

For several reasons, HighScope Foundation played a special role in the dissemination of longitudinal findings of the effects of high-quality early childhood programs. First, the HighScope Perry Preschool Study was persistently present, from the first round of studies to the third round and beyond. Second, the effects found in this study were themselves persistent; indeed, it was one of the first studies to find many effects—the intellectual boost; age-grade effects; return on investment; school achievement at 14; high school graduation at 19; and employment, earnings, and crime reduction at 27 and 40. Third, Carnegie Corporation of New York funded a policy center at HighScope, which at first focused broadly on the relationship between early childhood research and policy, but came to focus almost exclusively on the dissemination of the findings of the Perry study and similar studies to policy makers and the public. HighScope received 10 years of funding from Carnegie Corporation to sustain this policy center with several professional staff. Fourth, the interest of policy makers in hearing the results of these studies was crucial to maintain this dissemination effort.

HighScope Press published six monographs that comprehensively reported the study and its findings at the end of the program (Weikart et al. 1970), and at ages 10 (Weikart et al. 1978), 15 (Schweinhart and Weikart 1980), 19 (Berrueta-Clement et al. 1984), 27 (Schweinhart et al. 1993), and 40 (Schweinhart et al. 2005). A follow-up study at 50 is under way.

After each monograph was published, we held a news conference, the first in 1980 at Carnegie Corporation headquarters in New York City, and subsequent ones in Washington, DC. These news conferences led to enormous news coverage of the findings throughout the U.S. and around the world. Reporters and feature writers frequently called (and, later, sometimes emailed) to get information and quotes for their articles. Michigan newspapers, The New York Times, and The Washington Post were especially interested over the years.

We engaged in an extensive effort to write articles reporting the study and articles of larger scope that included the study, 87 in all: 6 up to when Perry study participants were 10, 11 when they were 10–19, 42 when they were 19–27, 21 when they were 27–40, and 7 since they were 40. We were invited to write these articles, so they are a good gauge of interest in the study. They appeared mostly in journals for researchers and educators. A few were directed specifically at policy makers and the public; we collaborated on policy papers with the Council of the Great City Schools and the National Governors Association. These articles, HighScope’s website, and a few other websites, became the hub of dissemination as the Internet grew from novelty to institution. A current Google search yields about 1,260,000 hits for “Perry Preschool.” “High/Scope,” the name of the organization that houses the Perry study and the curriculum first developed in Perry, yields about 7,652,000 results.

We made many presentations on the long-term benefits and return on investment of high-quality early childhood programs, throughout the U.S and around the world. We met with a dozen national associations of behavioral scientists, including the American Society of Criminology (with special recognition by the Academy of Experimental Criminology) and the American Psychological Association. We made eight presentations at meetings of the American Educational Research Association and six presentations at meetings of the Society for Research in Child Development. We met with staff of national policy-making groups including staff in the White House and Congress and the Committee for Economic Development; the Education Writers of America, the National Education Goals Panel in 1991, the National Governors Association, the National Conference of State Legislatures, the National Association of Counties, and the National League of Cities. We met with dozens of national associations of educators, including the Council of Chief State School Officers, the National Association of State Boards of Education, the American Association of School Administrators, the Council of the Great City Schools, the National Association of Elementary School Principals, the National Education Association, and the American Federation of Teachers. We met with another several dozen national associations of early childhood educators, including the National Head Start Association, the Administration for Children, Youth and Families, and the Gesell Institute. We were deeply involved with the National Association for the Education of Young Children, presenting at conferences annually and serving on the Michigan board from 1987 to 1994 and the national board from 1993 to 1997. We also played a leadership role in the first years of Michigan’s Early Childhood Investment Corporation from 2005 to 2008. We held an invited national conference of early childhood advocates annually from 1980 to 1990 and in 1998.

We met with state associations of early childhood advocates in most states, except the less populous states, with multiple meetings in California, Florida, Idaho, Illinois, Indiana, Kentucky, Minnesota, New York, North Carolina, Ohio, Oklahoma, Pennsylvania, South Carolina, Tennessee, and Vermont. Of course, we made many presentations in Michigan. We established Voices for Children projects in Michigan, Ohio, North Carolina, and South Carolina, meeting with state policy makers ourselves and training groups of early childhood advocates to give speeches to influential local groups. We were similarly active in California. In Michigan, we met with key state legislators prior to the large expansion of the state’s Great Start Readiness Program in 2013.

We also made presentations in Canada, Mexico, England, Ireland, the Netherlands, Sweden, and West Germany.

These papers and presentations contributed to the worldwide introduction and expansion of ECCE programs of higher quality intended to contribute to children’s development and reduce inequality. In the U.S., examples are expanded funding of Head Start and state prekindergarten programs.

Conclusions

We draw three conclusions from this experience. The first is that the findings of longitudinal studies of high-quality early childhood programs profit from broad dissemination to early childhood advocates at national and state conferences. The second is that there are few early childhood longitudinal studies. The third is that early childhood researchers and policy makers should work more closely together.

The first conclusion from this dissemination effort is that the findings of the longitudinal studies of high-quality early childhood programs profit from broad dissemination at conferences, especially to relevant policy makers. This effort got the message out in a way that many professionals could understand and care about. But this dissemination involved two-way communication. In addition to our communication of the message, many audiences were affected by the way we talked about our topic. We learned how to say things that made sense to the audiences. The presentations were all alike in the gist of their message, and each was different from the others in responding to the needs and interests of each audience. For example, we were presenting differences between the mean scores of the program and no-program groups until we recognized that some in our audiences did not understand mean scores; after that, we used group percentages except with groups of behavioral scientists. We presented the economic return on investment to society, a term that includes participants and taxpayers, until a state policy maker expressed interest only in the economic return to taxpayers. For a time, we tried to combine the findings of all the studies, presenting meta-analytic statistics, but questions indicated that audiences were primarily interested in hearing from us about the results of the HighScope Perry Preschool study.

The second conclusion comes from comparing the domain of early childhood longitudinal studies to other domains, particularly in the relatively well-funded field of medical research: there is a paucity of early childhood longitudinal studies of adequate quality. Only two small-scale studies, Perry and Abecedarian, have used random assignment techniques to find long-term results and economic return on investment, and they cannot claim widespread representativeness of early childhood programs or populations of children. The Chicago study has city-wide representativeness of a type of early childhood program and the population of children it serves, but at the cost of random assignment. The Head Start Impact Study (Puma et al. 2012) has national representativeness and random assignment, but largely sacrificed a selective focus on high quality to represent Head Start nationally. The Head Start Impact Study sought to represent all Head Start programs and succeeded thereby in showing that programs that followed Head Start regulations at the time were not on the average of sufficient quality to generate long-term results.

The third conclusion is that early childhood policy makers and researchers should work more closely together. Researchers should learn what questions policy makers are interested in and how research can best be designed to answer the questions they ask. Policy makers should integrate research and research findings into their policy formulations and the questions they ask. The need for more early childhood longitudinal research findings suggests the need for this integration. We do not agree with Farran (2016) that this lack of research is a reason not to expand early childhood programs further until enough research is done. Even with its few studies, ECCE has more longitudinal research than other levels of K-12 education. We do know enough to continue expanding early childhood programs, but should set aside some of this funding—say, 5 % of new funding—for evaluative research on early childhood programs. Early childhood program evaluations are a step in the right direction, but evaluations tend to be project-specific, without systematic relation to each other. Early childhood program research needs to be larger scale and more systematic than most program evaluations tend to be so that questions and answers can be cumulative and build on each other. What are the essential elements of program quality? How can these elements be validly and reliably measured so that it is clear when they are present or absent? Here we agree with Farran (2016) that the current state of measurement of preschool program quality is not adequate. Validity means not only that a measure of preschool program quality is what it claims to be, but also that it predicts program effectiveness.

So early childhood policy making should include a 5 % set-aside of new funding for experimentation with what policies lead to effective early childhood programs. This research should be highly coordinated so that it can be cumulative. Early childhood policy makers should follow this research and inform their policy formulations with it. Early childhood researchers should listen carefully to policy makers about what questions need to be answered. This model of collaboration applies especially to Head Start, where policy develops almost independently of research. An example is Head Start’s Program Review Instrument for Systems Monitoring, which was almost completed when a couple of researchers (Martha Abbott-Shimm and myself) were added to the committee guiding it its development—a step in the right direction, but too little too late.

Early childhood researchers should fashion their reporting so that policy makers can act on it. Too often it is laden with research jargon that is unintelligible to policy makers and practitioners. Granted, it is challenging to master the complexities of research methodologies that permit intelligible framing of results. But it is in this way that early childhood researchers can communicate valid findings to policy makers and practitioners and everyone can work together to realize the great potential of high-quality early childhood programs.