Jump to:Page Content
You are seeing this message because your web browser does not support basic web standards. Find out more about why this message is appearing and what you can do to make your experience on this site better.
Evaluating Education Reform: Early Childhood Education
Anne Mitchell, Heather Weiss, Tom Schultz
Over the past three decades an enormous body of research literature has been amassed on early childhood education, and on parent education and family support programs. The literature on collaboration and partnerships relevant to young children and their families, while not nearly so vast, also continues to grow. A search on early education limited to the ERIC database reveals literally hundreds or citations. In addition to completed works, a number of critical studies relevant to policymakers are currently underway across these three areas. This review attempts to summarize these three areas of research as well as to report on relevant research in progress.
Over the past 20 years dozens of literature reviews on early care and education have been written, culminating in the efforts of the National Academy of Sciences. Their publication, Who Cares for America’s Children? Child Care Policy for the 1990s (Hayes, Palmer, & Zaslow, 1990), incorporates a thorough review of all child care research to date. On the broader topic of child and family-focused programs, the National Commission on Children has produced an excellent review of research literature woven throughout its final report, Beyond Rhetoric: A New American Agenda for Children and Families (National Commission on Children, 1991). United We Stand (Kagan, 1991) offers a review of collaboration literature from the education and human service fields as well as a concise history of collaboration in early education and child care.
From a policy perspective, the term ‘early education’ is a convenient label for a large category of services encompassing child care, prekindergarten, preschool, and early intervention programs.
Review of research on program effects and quality in early intervention, prekindergarten, and child care programs
There are basically two bodies of literature that provide information about the quality of early childhood programs, where by quality we mean, the characteristics of a program that contribute to better program outcomes. One is research on program efficacy; the other is research in which studies specifically examine the effects of variation in program characteristics on program outcomes.
Research on the efficacy of early childhood programs can be conveniently divided into short-term and long-term studies, where long-term is defined as having data for more than one or two years beyond the end of the program. There are literally hundreds of studies in the short-term category and relatively few of these have data beyond the end of the program period. While there are only a handful of long-term studies, these tend to be the strongest methodologically and have provided many of the most interesting results. Indeed, for many years the central issue in early childhood was whether positive outcomes are persistent. Now the focus is shifting to research on how lasting positive outcomes are produced and the best approaches to producing such outcomes (Barnett, 1988; Barnett et al., 1988).
Short-term studies. By 1984, as part of a federally funded project, the Early Intervention Research Institute (EIRI; Casto, White, & Barnett, 1985 and 1986) had collected hundreds of reports of research studies on the effects of early childhood programs for disadvantaged children. Of these, most involved classroom-based programs for three- or four-year-olds that measured short-term outcomes. Additionally, Hubbell (1983) has reviewed over 1,500 studies of the Head Start Project since 1970. With such a large number of relevant studies, a conventional comprehensive literature review is impossible. However, quantitative syntheses or meta-analyses are possible and have been conducted (Hubbell, 1983; White & Casto, 1985; McKey et al., 1985). Although many of the 1,500 Head Start studies annotated by Hubbell followed the children for up to three years after Head Start, the majority measured only immediate effects. The overall findings of the meta-analyses were that children who attended Head Start showed immediate improvement in cognitive ability and school readiness. Given the nature of the data the authors of the Head Start meta-analysis were understandably cautious in drawing conclusions about program quality, but they did tentatively conclude that, at least in the short-run, classroom curriculum did affect cognitive development (more structured programs were better) and achievement motivation (Piagetian-based programs were better), class size did not have any effects, emphasis on language led to improved achievement motivation, and children of more involved parents appeared to gain more.
The EIRI meta-analysis (Casto & Mastropieri, 1986) dealt with studies of programs serving populations designated as disadvantaged and handicapped. These two groups of studies were analyzed separately, but interestingly, there were few differences in findings between the two groups of studies. Overall, there were educationally meaningful effects of approximately the same magnitudes for disadvantaged and handicapped samples, which help up across handicapping conditions and domains of development. As with the Head Start study only very gross measures of program quality were examined. The most interesting findings were that very intensive parent involvement does not appear to be critical to strong child outcomes, and intensity (the number of hours) of services seems to be positively related to success for children identified as handicapped.
The meta-analyses have been strongly criticized for a variety of problems. The quantitative reviews of Head Start were criticized for including many studies with questionable internal and external validity (Schweinhart & Weikart, 1986; Woodhead, 1988). It was suggested that Head Start attendees would have shown even greater benefits had the reviewers been more selective in including studies. Fortunately, the analyses by White and Casto (1985) and Casto and Mastropieri (1986) reported results according to the quality of the studies. They found that weaker studies consistently produced higher estimates of program effects. This strongly suggests that criticism of the Head Start meta-analysis based on inclusiveness is off-target. However, the more serious criticism is that the measures of program characteristics were poorly constructed. For example, parent involvement is a nebulous term that was essentially defined as training parents as intervenors. Another is that the attention devoted to understanding each study was so small that important nuances are missed and the idiosyncrasies of each study are missed. Perhaps most importantly there are apt to be correlations among program characteristics and between program characteristics and the characteristics of the children and families served. The statistical analyses that were performed did not take these into account and so run a very high risk of confounding the effects of program and sample client characteristics.
Long-term studies. Since a major goal of most public early childhood programs is to contribute to the long-term development of children, short-term studies are not entirely satisfactory. They become more satisfactory to the extent that their findings correspond to those of long-term studies. If the immediate effects of short-term studies are found to persist or to be linked to other long-term outcomes of importance, then the short-term studies can be used to guide public policy. To the extent that long-term findings diverge, policymakers and practitioners have to depend more heavily on the results of more expensive and less frequent long-term studies.
The most consistent finding in the long-term studies is that early childhood programs produce immediate gains in cognitive development as measured by IQ tests and in academic ability. There is considerable consistency across studies in this regard. The most significant long-term studies (based on quality of design and length of follow up) are:
The measures of long-term effects on cognitive development used in these studies are mainly: (1) standardized intelligence tests, and (2) school achievement and placement. Socio-emotional outcomes are not consistently measured in this body of literature, and when they are, they are not comparable to each other nor predictive of behavior (Datta, 1983). Outcomes in other domains were very rarely even considered.
Standardized intelligence tests. Standardized tests of intelligence showed the result of an immediate boost in IQ for children who experienced preschool intervention. This boost ranged from one-third of a standard deviation for the Comparative Curriculum Study (Karnes et al., 1983) to more than two standard deviations for the Milwaukee Project (Garber, 1988). This range is interesting in that less intensive intervention of half-day preschool for one year resulted in the smaller IQ gains, while the more intensive intervention of full-day intervention almost from birth resulted in the higher IQ gains. Ramey, Bryant, and Suarez (1985) concluded from a review of the experimental long-term studies (which includes these two) that length and intensity are positively related to IQ gains.
All of the studies with data to age eight found a drop in IQ. Caldwell (1987) has suggested that this neglected period of development—age eight to ten—may need special intervention similar to preschool. In the Milwaukee Project, the experimental children did retain an IQ one standard deviation higher than the control group through age ten, while in the other studies, the two groups’ IQs became equivalent.
School Achievement and Placement. The “fade out” of IQ gains was a great disappointment to researchers and to the field in general, yet the experimental groups, in most cases, continued to outperform controls in elementary school on school achievement tests, grade-point average, nonretention in grade, and reduced placement in special education. With these results, the researchers were encouraged to continue looking for differences between the two groups in the area of school success. The data in the area of school achievement through high school are less complete than we would like, as attrition was heavy (more than 50%) for all but three of the studies: the Perry Preschool Study, the Early Training Project, and the Family Development Research Program. The Perry Study found abiding and significant differences in school achievement test scores through age 19 in favor of the experimental group. The results for the Early Training Project and the Family Development Research Program are less clear. Females generally outperformed males on tests of achievement, and females in the experimental groups outperformed all other groups.
More complete data are available for retention in grade and special education placement which can be more easily determined from school records. Interestingly, in most studies, if one of these factors is significant, the other is not, suggesting that school districts tend to deal with most students who are having difficulties by either retaining students or labeling them as handicapped, but not both. The Rome Head Start Study, the Early Training Project, the Perry Preschool Project, the Cincinnati Study, the Philadelphia Evaluation, the New York Prekindergarten Program, and the Abecedarian Project found reduced placement in special education for the experimentals, while the Karnes and Washington, DC studies found greater retention in grade for the controls. A related measure of number of students who did not drop out of high school also favored children with preschool experience in the Rome Head Start, Early Training, and Perry Preschool studies. Many of these longitudinal studies also investigated students’ educational aspirations and expectations (Gray, Ramsey, & Klaus, 1984; Miller & Bizzell, 1984; Karnes et al., 1983; Berrueta-Clement et al., 1984; Lally, Mangione, & Honig, 1987), and results again favored the children who had attended preschool. Taken together, these measures of scholastic behavior show a consistent beneficial effect on school achievement of participation in preschool programs.
Summarizing the major findings from early educational intervention research, Ramey and Ramey (1992) identify six principles that characterize programs with the strongest effects.
Studies focusing on quality. As has been shown in the preceding reviews of efficacy studies, a fair amount of confidence can be placed in the effects of early childhood programs in a “laboratory setting,” and some corroborating evidence exists for more large-scale settings. However, one major limitation in most of these studies is the implicit assumption that program is homogeneous, well-defined, and easily replicated. No attempt is made to measure or even describe the program provided in terms of what actually happens in the classroom. Another body of research exists that considers the effect of classroom-level characteristics and the quality of the child’s experiences on outcomes.
Initial interest in assessing the effect of various classroom experiences closely followed the original efficacy studies and basically took two paths. The first was a theory-driven “horse race” to prove that a curriculum derived from Theory A was more effective than those derived from Theories B, C, and D. This resulted in a number of curriculum comparison studies in which children were randomly assigned to one of several classrooms using the various curricula (Karnes, Schweidel, & Williams, 1983; Miller & Bizzell, 1983, 1984; Weikart, Epstein, Schweinhart, & Bond, 1978). Other possibly important variables were held constant such as teacher/child ratio, class size, teacher training, and child characteristics. No explainable differences were found among the various curricula’s outcomes. Miller found that boys who attended Montessori preschool had a higher IQ. However, there were some methodological problems with this study, including smaller sample size in the Montessori classroom, and differential attrition across gender and curriculum. Also, this finding was not replicated by Karnes. In addition, Schweinhart, Weikart, and Larner (1986) have asserted based on their long-term comparison of school and out-of-school outcomes that direct instruction models in preschool may fail to reduce antisocial behavior and promote prosocial behavior as compared to child-centered approaches. This conclusion has been questioned on methodological grounds, including a relatively high attrition rate which might have affected the results (Gersten, 1986; Bereiter, 1986), but it remains provocative. Moreover, the Schweinhart et al. result suggests that long-term outcomes may not always correspond to short-term outcomes (which favored the direct instruction group).
Free. Available online only.