This page gives a brief overview of what we know about pre-K day care programs and their effect on later life outcomes for the children they serve. We believe that there are strong reasons to believe these programs have great potential; however, of the organizations we have reviewed in this area, none has given us strong reason to believe that it stands above the others, and none has shown the level of self-monitoring and -evaluation we would need to feel confident in it (either through directly monitoring long-term outcomes or demonstrating fidelity to a proven model).
We found a literature review by W.S. Barnett, "Does Head Start Have Lasting Cognitive Effects?", particularly useful in summarizing the conclusions of relevant research. We read several of the papers it refers to and found its assessment broadly similar to our own; in addition, Barnett lays out clear, firm criteria by which he chose which papers to examine (his criteria are based entirely on methodology, not conclusions) and provides tables with the research design and results of each paper, so that the reader need not rely on his subjective conclusions. Unfortunately, to our knowledge this essay is only available in hard copy, as Chapter 16 of The Head Start Debates (edited by Edward Zigler and Sally Styfco, published in April 2004).
We focused our research for this cause on New York City; below we detail a few relevant terms for different types of day care programs.
Model programs is a term frequently used in the academic literature (including the Barnett paper) to refer to programs that are developed, executed, and closely monitored by experts and researchers, generally with the express intent of demonstrating what child care is capable of. Such programs are often very high-intensity: for example, the Carolina Abecedarian Program was a full-day, year-round program that lasted from infancy through the age of five.
Head Start refers to a federal program - legal specifications available here - that directly reviews applications from child care programs, and assigns funding (in addition to managing ongoing assessment of programs). As outlined in this overview, Head Start is a "comprehensive child development" program, targeting children below the federal poverty line, that includes "educational, health, nutritional, social and other services to enrolled children and families." Head Start funding is generally used for part-year, part-day programs for 4-year-olds; more details, and examples, are available in the Barnett paper.
Universal Pre-K is a New York State program (legal specifications here) that, like Head Start, directly funds child care programs and carries out its own evaluations. Funding can be used in combination with other sources - for example, to provide full-day instead of part-day care. More details are available here. The term "Universal Pre-K" can also refer to other state programs, but we have focused on its New York context.
New York child care centers have a variety of other funding sources available to them (one overview is available in the early pages of this document from United Way). One such program is the state-run Universal Pre-K program, which conducts its own instructional reviews; many of our applicants have their written reviews available, providing an additional third-party opinion on the quality of a program.
Evidence of effectiveness: model programs
The Barnett paper cited above summarizes the results of 15 studies of model programs. Each of these studies tracked both a "treatment group" of children who participated in model programs and a "comparison group" of children who did not, and compared the two groups after several years had passed (each study goes at least as late as third grade). About half employed experimental design: families pre-applied to the programs, and participants were chosen by random lottery, meaning that any differences between the two groups could only be attributed to the effect of the program or to random chance.
Out of 15 such studies of such programs, 9 showed the treatment group as statistically significantly better off, as of 3rd grade or later, by at least one of the following measures: IQ (rare), special education and grade retention, achievement test scores and statistically equivalent on the others; 5 didn't show strong effects one way or the other; in one, the treatment group had lower IQ. The studies that were methodologically strongest, and most careful in their selection of at-risk children, showed the most impressive effects. These include the Perry Preschool project, which found preschool participants to have significantly higher achievement test scores and high school graduation rates at age 18, and the Carolina Abecedarian project, which found preschool participants to have significantly superior achievement test scores and college enrollment rates at age 21. These two programs are frequently cited as examples of how much preschool programs can accomplish. See pages 226-229 of the Barnett study for more details as well as full citations.
We feel that these results give strong reasons to be optimistic about what quality preschool can accomplish. However, because these programs were designed to be "exemplary" - and were often better funded and more intense than typical Head Start programs - we feel it would be a mistake to generalize their results to all preschool programs.
Evidence of effectiveness: Head Start and other large-scale programs
Head Start comes with its own assessment program, and participants have been observed to meet certain quality guidelines relatively consistently (see page 222 of the Barnett paper for an overview of the research on implementation, along with citations). Therefore, a broad enough survey of outcomes may be somewhat generalizable. A large-scale impact study is currently underway, but no data on later life outcomes is yet available.
Barnett reviews 24 studies of Head Start or similar programs ("12 studied the effects of Head Start programs, 4 studied a mix of Head Start and state or local programs, and 8 studied state or local programs"). Like the studies of model programs cited above, these studies track both a "treatment group" of children who participated in model programs and a "comparison group" of children who did not, and compared the two groups after several years had passed (each study goes at least as late as third grade). However, none of these studies employed experimental design; most simply compared children who had attended preschool to those who had not, raising the concern that any differences could reflect pre-existing differences in families - related to the fact that some families sent their children to preschool, and others didn't - rather than actual effects of the program. The only two studies that made any attempt to address this issue suffered from high attrition, i.e., many of their subjects dropped out of the study before it was complete.
15 of these 24 studies (including 1 of the 2 that seems to have attempted to control for selection bias) showed Head Start children with a statistically significant advantage as of grade 3 or later (and in the last year of the study) on at least one of: high school graduation, achievement tests, special education, grade retention. 9 showed mixed or non-statistically significant effects. See pages 233-236 of the Barnett paper for details.
Barnett also observes that differences between Head Start and non-Head Start children may be understated in some cases, because groups with weaker academic performance also lost more students to grade retention and special education, causing their average test scores (as measured) to rise artificially.
While we are concerned about potential sources of bias, we believe these studies give some reason to be cautiously optimistic overall, and we look forward to learning more from the large-scale impact study currently in progress; but knowing what we know, we are not confident in the lasting effects of non-model preschool programs.
Our applicants
We had a total of 9 applicants running preschool programs for disadvantaged children, but none provided us with either:
A compelling case that their program resembles an already proven "model program."
Independent evidence for their program's effect on later life outcomes.
We were in some cases sent reports on children's cognitive development, but these reports either did not show consistent progress for enrolled children, or did not use measures that allowed us to compare the progress to what might be expected in the absence of a preschool program. In addition, none of these tracked children after their enrollment in kindergarten. This is a particular concern since there are many studies of preschool programs showing short-term improvement that fades out entirely - or weakens considerably - over time (the studies mentioned above, summarized in the Barnett paper, provide many examples of this phenomenon). All of the reports in this category were sent in hard copy only.
Below, we list all of the applications we received from day care programs, and mark which ones receive funding from Head Start and Universal Pre-K.
You can find more Head Start programs using this tool from the Administration for Children and Families.
A note on costs
We estimate that the "per-participant" costs for the types of programs discussed here are roughly in the same range as those of Nurse-Familiy Partnership (NFP).
Two of our applicants provided program specific budgets as well as figures for the number of children they serve for 2007:
The Bloomingdale Family Center Inc. served 194 children at a total cost of $2,954,226 in 2007 (see the Round 1 application, Pg 2, and Budget Attachment). This implies a program cost of about $15,228 per child served.
The Brooklyn Kindergarten Society served about 324 children at a total cost of $4,196,052 in 2007 (see Round 1 Application, Pg 1 & 3). This implies $12,950 per child served.
While the majorities of these expenses are Federally funded (The US Department of Health and Human Services reports that New York State Head Start Centers received on average $8,800 per child enrolled per year in 2007) our applicants appear to rely partially on non-federal funding to finance their programs.
Conclusion
Although we see great potential in preschool programs, we have more confidence in NFP - in terms of its long-term impact on children's lives - than in the preschool programs run by any of our applicants. These programs are roughly similar on a cost-per-child-served basis. Therefore, we believe that NFP is a better bet for a donor seeking to make a difference through early childhood care.
