Skill building in early care and education programs

Abstract

Preschool policy is often motivated by the role of early childhood care and education in imparting valuable school-readiness skills to young children, which provide the foundation for lifelong skill advantages and greater success in adulthood. Our review raises questions about the degree to which a simple skill-building model—whereby preschool programs boost a set of domain-based competencies that fuel longer-run success—is a good framework for guiding preschool policy. Policy investments in preschool education boost school readiness in the short run, but in some recent studies these initial gains either quickly disappear to even turn negative. Considerable effort will be required to take advantage of these end-of-preschool gains by optimizing classroom learning experiences, integrating the learning goals of the preschool year and K-12 schooling, and continuing to push for new discoveries regarding truly fundamental early skills and capacities, such as child health, that preschools could target, that would not develop in counterfactual conditions. Supporting parental employment and access to affordable child care can boost family income and increase home investment in child enrichments, which can also benefit children’s skill development. We argue that the preschool policy portfolio should include efforts to more broadly expand child care access to support family capacities.

I. Introduction

Given the emphasis placed on maternal employment by the U.S. safety net, policies supporting the provision of child care and education for low-income mothers are essential (Currie, 2006). Both preschool and child care are important in allowing parents to work or attend school. And while only preschool is commonly viewed as fundamentally educational, nearly all forms of non-parental care for young children contribute to the development of certain cognitive and socioemotional skills. The central aim of the safety net may be to reduce poverty and its hardships, but it is clearly in the public interest for the care provided to children while parents work to be as developmentally supportive as possible. Skill development in early care and education settings is the focus of our review.

Successful societies provide educational and socialization experiences that prepare children to succeed as they take on adult roles and responsibilities. In industrialized countries, the capacities and skills needed to secure middle-class success include advanced literacy and communication skills, high-level analytic thinking and, increasingly, the social skills that allow individuals to get along with others and work in teams (Deming, 2017). Developing these skills and capacities across childhood and adolescence requires a complementary sequence of investments – by parents in the child’s home environment, by educational institutions, and by governments supporting the efforts of parents and educational institutions.

How might early care and educational environments best promote these crucial skills? We use the term “skills” broadly to encompass any skill, behavior, capacity or psychological resource that helps children attain successful outcomes. And we follow the practice common in most social and behavioral sciences of referring to collections of “cognitive” skills (measured, principally, by IQ and achievement test scores) and a catch-all “socioemotional skills” category (termed by economists “noncognitive skills,” despite the fact that most of these skills are inherently cognitive). In the “socioemotional skills” category, self-regulation and relationship skills appear to be of particular importance in early childhood (McClelland et al., 2007).

It is common to think of budding literacy, mathematics and socioemotional skills early in life as comprising the “school readiness” skills that lay the foundation for the advanced literacy, analytic thinking and teamwork skills needed for adult success. As reviewed in Section II, skill-building models proposed by cognitive psychologists and economists flesh out the conceptual pathways by which skills often build upon one another over time. These models provide a compelling case for the importance of skill-building processes and would appear to support the argument that preschools should target the kinds of basic literacy, math and socioemotional skills that are the initial building blocks for the advanced skills needed for adult success.

However, we then turn in Section III to several problems with drawing policy conclusions about early care and education design based on skill-building logic. First, and most important, even if school readiness skills are in fact important building blocks for future skill development, this does not mean that early childhood education programs targeting these capacities will necessarily produce lasting effects. In some cases, the skill-building process may indeed be at work, yet weak in the sense that skills at time t are to a much greater extent a product of concurrent school, home and other environmental influences than are those same skills at time t-1. Evidence suggests that concurrent environments are important for early achievement and socioemotional skills. Furthermore, skill-building links are weakened when preschool skills are soon mastered by children who do not attend preschool programs but are taught those skills in kindergarten classrooms or, in some cases, home environments. This weakening is likely responsible for much of the “fadeout” observed in skill-building interventions, which we view as a component of modern skill-building theory.

In light of pervasive evidence of fadeout in school readiness skills, perhaps those aren’t the right skills to be targeted by preschools. Indeed, when we attempt in Section IV to reverse-engineer skill development insights from the evaluations of two prominent model preschool programs (Perry and Abecedarian) with demonstrated impacts well into adulthood, their disparate impacts on IQ and socioemotional skills complicate efforts to identify common “active ingredients.” As explained in Section V, moreover, no clear picture of important ingredients, nor a consistent story of the skill building processes emerges from rigorous evaluations of more recent, scaled-up, preschool programs. Section VI provides a perspective on characteristics of the “right” skills, while Section VII discusses the kinds of curricula needed to build them.

Given these problems, it isn’t surprising that we fail to reach definitive answers to the question of what skills early childhood education programs should develop. We conclude, in Section VIII, with some ideas for moving forward, including placing more emphasis on rigorous implementation of proven curricula, examining possible health pathways to long-run outcomes, a more deliberate approach to integrating skill building across the preschool and K-12 years, and a greater appreciation for the potential value of early care policies that encourage children to move from home and informal care arrangements into center-based care. Reinforcing the child care safety net would support family capacities (via employment and household investments) alongside child development. Although our review suggests that boosts to the latter may fade over time, impacts on the former can improve the economic stability of families with children in the longer-term.

II. Skill-building models¹

Skill-building processes are most readily apparent in math and literacy, where early academic skills provide the foundations upon which later skills are built. Counting is a causal precursor of addition problem solving, and addition is used in multiplication problem solving (Baroody, 1987; Lemaire & Siegler, 1995). In the case of reading, children’s ability to match letters to sounds helps them learn to recognize written words, which in turn supports vocabulary learning and then reading comprehension (LaBerge & Samuels, 1974).

Economists’ skill-building models of human development formalize thinking about the human capital production function and emphasize how investments and child endowments interact to create a child’s accumulating stock of human capital. Cunha and Heckman (2007) posit that human capital accumulation results from two distinct processes. As described above in the context of literacy and math skills, the first is “self-productivity,” which suggests that the impacts of preschool interventions are particularly likely to persist when those interventions are designed to support incremental skill building within developmental domains such as mathematics or literacy. Indeed, in most K-12 schools around the world sequential learning goals are embedded in curricular materials and instructional guidelines (e.g., the U.S. Common Core State Standards) in an effort to drive sequential skill-building.

Suppose learning to count before the beginning of kindergarten results in a higher probability of learning to add by the end of kindergarten, which in turn results in a higher probability that the child will learn some basic multiplication skills before the end of 1^st grade. This would seem to imply that an intervention designed to boost preschool counting skills would accelerate multiplication learning. But the probability of subsequent math learning is far from perfect from one grade to the next because math learning also depends on what is taught in schools and at home. As long as these environmental influences are at work, the correlation between preschool and later-grade math skills weaken over time. Bailey et al. (2018) conduct a meta-analysis of estimates of math skill-building from end of treatment through grade 2, and find that self-productivity in skill development may not be as important as the quality of the ongoing learning environments that children encounter in school and at home.

The next processes introduced by Cunha and Heckman (2007; also see Ceci and Papierno (2005)) are dynamic complementarity and substitutability. Bailey and colleagues (2020) explain these concepts as follows:

“Consider two children, Child A and Child B, who differ in their levels of counting knowledge; Child A has higher foundational skills than Child B. If both receive the same amount of teaching time and effort from teachers or parents to learn addition and subtraction, which child will profit the most from the instruction? If Child A will profit most, we say that the skill-building model features dynamic complementarity—the teaching investment complements a child’s incoming level of foundational skills and produces a Matthew effect, where the rich get richer. On the other hand, if teaching investments are more productive for Child B, then we say that the model features dynamic substitutability—the teaching investment is compensatory by raising skills already mastered by Child A but not Child B (p. 68).”

Education planners and teachers must struggle with the dilemma of wanting sequential investments to be as complementary to one another as possible, while at the same time providing enough compensatory and individualized instruction to ensure that all children meet the learning goals of every grade.

The hypothesis of dynamic complementarity suggests that children in preschool environments that enable them to enter school with the strongest cognitive and socioemotional skills will profit the most from K-12 schooling, for example by learning the most from classroom instruction. Support for this hypothesis comes from a study by Johnson and Jackson (2019), which finds strong positive interactions between attending Head Start and K-12 school funding.

But evidence supporting the flip-side hypothesis – of dynamic substitutability – appears to be even more common in early childhood programs (Bailey, Duncan, et al., 2020). Consider a kindergarten teacher with a class comprised of some students who have enjoyed the benefits of a skill-building year attending a preschool program, while others have not. It takes considerable teaching skill to ensure that children who don’t know their letters and numbers learn them, while at the same time helping children who are ready for advanced learning to build more sophisticated literacy and math skills. Using national data, Engel et al. (2013) find that teachers spend a great deal of their teaching time on skills that almost all students have already acquired. In effect, these kindergarten classrooms are substituting for the preschool classrooms by promoting the most rapid learning among children who begin the year with the lowest level of skills.

An empirical example of complete dynamic substitutability in early childhood programs comes from Rossin-Slater and Wüst (2020), whose study looked at a Danish preschool program and a nurse home-visiting program (whereby trained nurses visited the homes of mothers with newborn infants providing health and parenting services) that were both rolled out across communities in a haphazard way during the same period. Some birth cohorts stood to benefit from the availability of both programs, while others were too old to participate in either program by the time it was introduced in their communities. Still others could benefit from only one of the two programs. The authors found that each program taken alone generated long-term impacts on educational attainment. But for children in both programs there was neither evidence of additive effects of the two programs (1+1=2) nor of dynamic complementarity (1+1>2). Children attending both programs did not better than children attending must one of them (1+1=1). In other words, these programs appeared to be dynamic substitutes for each other – a most discouraging result.

More recent attempts to discover dynamic complementarities have compared shorter-run academic outcomes for children who were randomly assigned to preschool education and then encountered either high- or low-quality instruction in the early grades of their K-12 schooling. Dynamic complementarity would be evident in positive interaction effects on school success of the combination of preschool and high-quality instruction. A recent meta-analysis by Bailey et al. (2020) of these possible interactions found that the average coefficient on the interaction measures was approximately zero.

III. Fadeout in skill-building processes

In the absence of strong dynamic complementarity between preschool and subsequent educational investments, the skill advantages wrought by a preschool investment will disappear – a phenomenon known as “fadeout.” Take the example of the high-quality evaluation of a pre-K mathematics curriculum, Building Blocks, embedded with professional development, in the TRIAD study (Clements et al., 2011; Clements et al., 2013). Building Blocks adds 15–20 minutes of daily math activities to an existing preschool classroom curriculum and, in contrast to “drill and kill” classroom methods, embeds learning in playful activities. The TRIAD study involved random assignment of the curriculum at the school level to 834 students in 30 elementary schools serving low-income neighborhoods in either Buffalo, New York, or Boston, Massachusetts.² Students’ mathematics achievement was measured at the beginning and end of the preschool year and at the end of kindergarten, first grade and fifth grade, as well as during the fourth-grade year. Drawn from Clements et al. (2013) and Bailey et al. (2017), Figure 1 shows that treatment/control differences at the end of the pre-K year amounted to .63 SD – a large impact. A more detailed look at the nature of the end-of-treatment impacts showed that the curriculum boosted all four subdomains that it targeted – counting, patterning, geometry, and measurement – all of which have been shown to predict later mathematics knowledge (Nguyen et al., 2016).

Figure 1:

Experimental Impacts on Math Achievement from the TRIAD/Building Blocks Program

Note: Vertical lines depict 195% confidence intervals. Based on Bailey et al., 2017

But Figure 1 also shows that this large initial treatment effect at the end of preschool quickly faded – to about 40% of its initial value by the end of first grade and almost completely by fourth grade. Figure 2 recasts Building Blocks data through age 7 by showing vertically-scaled math scores separately for treatment and control groups. The more rapid growth in math skills across the pre-K year for the treatment relative to the control group produces the large end-of-treatment effect shown in Figure 1. Between the end of treatment and the end of first grade, math scores grow rapidly (about 1 SD per year) for both groups, but they grow faster, on average, for control- than for treatment-group children. Thus, the fadeout observed in Figure 1 appears to be largely the result of control-group catch-up.

Figure 2:

Math Achievement During and After the Pre-K TRAID/Building Blocks Program

Note: Rasch score standard deviation = 1. Based on Bailey et al. 2017

Meta-analyses of interventions targeting specific skills typically find a Building Blocks-type pattern of impacts, with end-of-treatment impacts that begin to fade out quickly and, in some cases, disappear completely. Two examples:

• Bus & van IJzendoorn (1999) reviewed 36 evaluations assessing impacts on phonological awareness targeted by phonological awareness training programs. Only 7 of the 36 assessed “long-term” effects, and those assessments, on average, occurred only 8 months after the end of the programs. Effect sizes fell by about half of their end-of-treatment values. For reading outcomes, the measurement period was longer (averaging 18 months), but the amount of fadeout was larger as well (nearly 80%).

• Takacs and Kassai (2019) summarize the literature on executive function interventions. As with the other literature reviews, only a small fraction of studies (15 of 90) included follow-up assessments, which were conducted, on average, 22 weeks after the end of treatment. In contrast to end-of treatment effect sizes, which were substantial, the authors did not find convincing evidence that impacts persisted at follow-up.

These studies suggest that some degree of fadeout is the norm. Although it is difficult to generalize to the universe of skill-based interventions, the Bailey et al. (2020) review of this literature could not identify any meta-analyses of interventions targeting specific skills in children – which varied substantially in terms of follow-up period – in which full persistence prevailed.

IV. Historical evidence of skill building in the Perry and Abecedarian programs

Much of the empirical work undergirding economists’ and developmentalists’ models of skill development is derived from two legacy studies in early childhood intervention: the Perry Preschool Program and Abecedarian Project. The famous Perry Preschool program provided one or two years of part-day educational services and home visits delivered by highly educated, and well compensated, teachers to a sample of low-income, low-IQ African American children aged three and four in Ypsilanti, Michigan during the 1960s, at an average per-pupil cost in 2021 dollars of $23,800 for one to two years of the intervention. Control-group children experienced low-quality counterfactual conditions in their home environments, and most of the children in both groups attended relatively low-quality public schools. Benefit/cost ratios for Perry ranged between 6:1 and nearly 9:1, with benefits driven in roughly equal measure by increases in earnings and reduced crime (Heckman et al., 2010).

What drove these long-run impacts? Perry’s impressive initial impacts on IQ scores famously faded out by third grade, but the program generated a steady set of impacts on achievement and various socioemotional skills, like reducing aggressive and rule-breaking behaviors, and improved motivation, that persisted through childhood, adolescence and early adulthood (Heckman et al., 2010; Heckman et al., 2013; Schweinhart, 2005).

The Abecedarian program began in 1972, enrolled infants of low-income, mostly African American women from Chapel Hill, North Carolina, and provided year-round, full-time individualized educational activities for five years, at a cost of approximately $20,000 per child, per year in 2021 dollars (~$105,000 total intervention cost per child). As with Perry, counterfactual conditions in the home were of relatively low quality, although K-12 schools in Chapel Hill were integrated and of relatively high quality. Benefit/cost ratios for Abecedarian exceeded 6:1, with more than two-thirds of the benefits driven by crime reductions and the remainder reflecting differences in adult health and the labor income of participants and their parents (the five years of full-time childcare enabled parents to establish and maintain more continuous and higher-paying careers; García et al., 2020).

In contrast to Perry, Abecedarian generated persistent impacts on IQ. Like Perry, it produced consistently substantial impacts on verbal and quantitative achievement test scores across childhood and adolescence (Campbell et al. 2001). Consistent with their emphasis on the importance of language development mentioned above, Burchinal et al. (2022) argue that rich language interactions were a key ingredient of the Abecedarian study, acquired through extended scaffolded interactions between the teachers and either small groups of children or individual children. Abecedarian did not measure socioemotional skills as persistently as Perry, and there were no consistent impacts for the skills it did measure. In fact, Haskins (1985) documents that the program generated an unexpected increase in teacher reports of children’s aggressiveness in the early school years, although these effects appeared to fade with time.

It appears that achievement-related skill advantages may well have been an active ingredient in the longer-run successes of both Perry and Abecedarian. The evidence is mixed regarding socioemotional skills. Perry persistently improved those skills across childhood, but Abecedarian did not. The fact that both programs showed achievement-related impacts might suggest that preschool programs should concentrate on those impacts. However, Section V shows that evaluations of contemporary models of preschool programs targeting specific skills often fail to show even medium-term impacts on achievement.

V. Recent evidence of skill-building in modern, scaled-up, preschool interventions

The historical evidence provided a proof of concept for early childhood interventions. Yet the scaled-up public preschool interventions available to today’s four-year-olds differ quite substantially from the conditions in these legacy studies. Adult outcomes are not yet available for cohorts born during the past 25 years, but shorter-run outcomes have been tracked in several recent RCT studies of preschool programs. We review three of them: the National Head Start Impact Study, an evaluation of the Tennessee Voluntary Pre-K program, and a lottery-based evaluation of Boston’s pre-K program. Collectively, they raise more skill-building puzzles than they solve.

Puma et al. (2010) track elementary school outcomes of the $10,000-per-slot Head Start program in the National Head Start Impact Study (NHSIS). The study was able to draw a large probability sample of Head Start centers, and then worked with center directors to generate waitlists and randomly select children (n=4,667 in all) to be offered enrollment. The study included both three- and four-year-olds applying for entry into sampled Head Start centers and conducted follow-up data collections through third grade.

Apart from rigorous random assignment, what distinguishes the NHSIS is the quality and breadth of its measurement of academic and socioemotional skills, both prior to random assignment and in all follow-up waves. School achievement skills were measured with vertically scaled assessments. Both parents and teachers reported on a range of socioemotional capacities, including problem behavior, prosocial skills and, beginning in kindergarten, school engagement. Many of these skills were measured using several different validated instruments. It is hard to imagine a broader assessment of cognitive and socioemotional skills in the early grades.

Impact results from the study are summarized in Table 1 with simple counts of the number of coefficients that exceeded a low threshold (p<.10) in the absence of adjustments for multiple testing both at the end of the Head Start year and at the end of 1st grade (3rd grade results are similar to 1st grade results). Because the evaluation included quite a number of assessments within the broad domains of literacy, math, and problem behavior, the table also shows the total number of assessments. Given our p<.10 filter, we are interested in domains in which considerably more than 10% of impact coefficients pass the filter.

Table 1:

Ratio of statistically significant (p < .10) treatment impacts to outcomes examined in the Puma et al. (2012) Head Start Impact Study

	End of Head Start year	End of 1st grade
Language and Literacy	6 of 8	1 of 11
Math	0 of 2	0 of 4
Socioemotional skills	0 of 9	1 of 9

Results in Table 1 are striking: They show that end-of-treatment impacts are significant only for measures in the literacy domain and for virtually none of the assessments at the end of 1^st (or 3^rd) grade. With effect sizes at the end of the program for literacy measures averaging about .20 SD, it appears that children whose parents were offered Head Start slots enjoyed modestly higher growth in literacy skills. No impacts on math or behavioral measures were detected. At the end of 1st and 3rd grade, there was virtually no evidence of above-chance gains in any of the three domains.

Table 1 suggests that academic and behavioral gains are unlikely to persist even a few years beyond the Head Start year. Of course, the 24 outcome measures gathered in the 1st-grade follow-up might not capture the “dark matter” skills that undergird most long-term impacts (Gibbs et al., 2011). But with the evaluation’s intensive focus on measuring the broad range of early-grade skills that the research community had identified as important, all of which are incorporated into Head Start’s “whole child” intervention approach, the dark-matter channel to longer-run benefits from the current implementation of Head Start seems far from certain.

It is worth noting that two key components of the Head Start model – child physical and oral health – were not assessed. Quasi-experimental evidence from the first years of the Head Start program (albeit with substantially different counterfactual health care than in the NHSIS; i.e., Medicaid expansion, desegregated hospitals) shows Head Start-related reductions in child mortality causes between ages 5–9 (e.g., measels, anemias, pneumonia; Ludwig & Miller, 2007).

The Tennessee Voluntary Pre-K program (TN VPK) is a large, state-funded pre-K program that annually enrolls about one-quarter of the state’s four-year-olds at a cost of $4,460 per slot. Although technically available to all four-year-olds living in the state, TN VPK grants priority to children whose families qualify for the means-tested free or reduced-price lunch program. By conventional U.S. standards, the quality of the program is fairly high: It is a full-day program (5.5 hours) with relatively small class sizes (maximum of 20 students), and it requires both a licensed teacher and a teacher aide in each classroom (this keeps the child-to-adult ratio at or below 10 to 1). The curriculum must be chosen from a state-approved list. Researchers at Vanderbilt University randomly assigned children to TN VPK through a lottery system implemented at over-subscribed programs (Lipsey et al., 2018).

At the end of the TN VPK program year, children attending the program outscored their comparison-group counterparts by about one-quarter of a standard deviation on an achievement test composite based on Woodcock-Johnson assessments– a result consistent with many of the regression discontinuity estimates of end-of-pre-K impacts. But these gains had disappeared completely by the end of kindergarten and became negative and statistically significant by second grade. A follow-up in sixth grade provides even stronger evidence of perverse achievement and behavioral impacts (Durkin et al., 2022). As with the NHSIS, the pattern of fading achievement impacts and largely null to negative behavioral impacts makes it seem unlikely that longer-run impacts will emerge in a later TN VPK follow-up.

The NHSIS and TN VPK studies are deeply discouraging for policymakers who hope that current preschool programs will generate the kind of positive long-run returns seen in Perry and Abecedarian. But neither the NHSIS nor the TN VPK study has tracked outcomes beyond sixth grade, much less into adulthood. The third program we review was able to track children into late adolescence. Its results are more promising, but raise other puzzles.

Gray-Lobe and colleagues (2021) conducted a lottery-based evaluation of cohorts entering Boston’s public pre-K program between 1997 and 2003, using administrative data available for virtually all children who won or lost the lottery. Little is known about counterfactual conditions and socioemotional skill development. Program costs were around $13,000 per pupil for a full-day class and half that much for half-day classes.

As in the case of the NHSIS and TN VPK studies, there is a striking absence of impacts for the 24 measures of K-8 achievement scores and early school progress (e.g., avoiding grade failure and special education placement (Table 2).³ But positive impacts on achievement measures (e.g., state and SAT test-taking and scores) and favorable behavioral impacts (e.g., suspensions, juvenile incarceration) emerge in high school. Moreover, the researchers find favorable impacts on 7 of 11 measures of high school graduation and college attendance.

Table 2:

Ratio of Statistically significant (p<.10) treatment impacts to outcomes examined in the Gray-Lobe et al. (2021) Boston Pre-K study

	K-12	Post-secondary
K-8 test scores	0 of 18	-
K-8 on-track progress	0 of 6	-
High school behavior problems	3 of 7	-
High school tests	7 of 21	-
Completed schooling	2 of 2	5 of 9

The Boston pattern of no detectable short- or medium-run achievement and behavioral impacts is very different from patterns seen in Perry and Abecedarian; with those two programs, favorable impacts on achievement were apparent throughout the elementary and middle-school years. At the same time, the generally null elementary-school achievement impacts in the Boston study are similar to patterns seen in the NHSIS and Tennessee pre-K evaluations. But, as shown in Table 2, the Gray-Lobe et al. (2021) Boston study showed longer-run impacts in high school and beyond despite the absence of impacts on outcomes measured earlier. It is difficult to imagine that the late-adolescence Boston impacts were not fueled by some kind of short-term improvements, although it is quite possible that the “secret sauce” isn’t captured by the administrative data available to evaluators. Overall, no clear picture of important ingredients, nor a consistent story of the skill building processes emerges from contemporary, scaled-up, preschool programs.

VI. Target more fundamental skills?

If contemporary preschool programs targeting narrow conceptions of school readiness skills fail to accelerate mastery of achievement skills, perhaps broader conceptions of fundamental skills should be considered. Bailey et al. (2020) examine general intelligence and conscientiousness as two potentially promising examples. IQ can be considered a fundamental capacity because it supports performance across a wide variety of important tasks and is the single strongest predictor of many important life outcomes (Cawley et al., 1997; Gottfredson, 1997). Unfortunately, attempts to augment general intelligence experimentally through education interventions have rarely proved successful (Jensen, 1998; but also see Nisbett et al. 2012 for a more optimistic review). Protzko’s (2015) meta-analysis of 23 different interventions targeting IQ show end-of-treatment effect sizes of .37 SD, but these effects declined by about .10 SD per year, caused mainly by declines in the IQs of treatment-group children across the follow-up period. Indeed, only Abecedarian program generated persistent effects on children’s general intelligence. Perhaps this success was due to the intense nature of Abecedarian, combined with the conditions of relative deprivation facing control group children and their families.

Among socioemotional skills, conscientiousness – one of the “Big Five” personality traits – also appears to be fundamental. It is a powerful correlate of both schooling and later life outcomes (Almlund et al., 2011; Judge et al., 1999). But while ample evidence suggests that many socioemotional measures can be altered with interventions (Almlund et al., 2011) and may be important mediators of the impacts of early childhood education programs on adult outcomes (e.g., Heckman et al., 2013), no study has established that interventions in early and middle childhood can generate persistent improvements in conscientiousness (or Duckworth et al.’s (2007) related concept of “grit”).

Developmentalists emphasize the importance of the related skills of self-regulation and executive function. Self-regulation generally refers to control of one’s behaviors and emotions in a social context, and captures skills like paying attention to, remembering, and following rules and directions, planning and completing multi-step tasks, and problem solving (McClelland et al., 2007). Relatedly, executive function is a broad term that captures cognitive skills such as inhibitory control and working memory as well as the ability to flexibly shift one’s attention (Blair, 2016). It is intuitive that such skills would support children’s success in school-based settings but most unclear whether such skills are amenable to intervention (Nesbitt & Farran, 2021).

Some have suggested that language, as opposed to literacy, skills are fundamental and could be emphasized more in preschool. Burchinal and colleagues (2022) argue that in attempting to prepare children for kindergarten, today’s preschool programs lose the forest of providing opportunities for child development for the trees of school readiness achievement. Specifically, in their focus on literacy and math skills, preschools fail to emphasize rich language interactions between teachers and children.

Bailey et al. (2017) posit that a number of possible skills meet their “trifecta” combination of being fundamental, malleable and unlikely to develop in counterfactual conditions. These include advanced mathematics (e.g., fractions, algebra) plus analytic skills that some students never master in many K-12 public schools, plus advanced literacy and communication skills and concrete vocational skills. Most of these skills align with the skills that are increasingly demanded by employers offering higher-paying jobs, but all are the end products of years of skill-building rather than skills that might be taught in preschool classrooms.

All in all, the research literature has yet to identify a set of fundamental skills that are amenable to intervention during the preschool years. This is obviously an urgent research priority.

VII. Can curricula build skills?

Even if we knew precisely what fundamental skills we wanted preschools to promote, we rely on preschool teachers and administrators to structure their classrooms and interactions with students in a way that promotes those skills. Traditionally, policy attempts to influence preschool classroom experiences involve licensing regulations, structural approaches such as ensuring minimal health and safety standards, teacher educational or credentialing requirements, and child-to-staff ratios. Yet it is well established that regulating structural features may assure a “floor” of educational quality, but those features do not correlate with the development of school readiness skills (Early et al., 2007).

Several features of preschool classroom processes do correlate with children’s skills: the teacher’s use of complex language and responsive conversations with children; the teacher’s ability to create interesting activities that engage children’s attention, while also managing classroom organization; and the positive nature and emotional supportiveness of the classroom. These processes involve a balance of both child- and teacher-led methods, as well as a combination of focused and small-group instruction, while active attention is paid to the needs of individual students (Farran, 2017). In theory, curricula guide can such processes, and thus they have the potential for scaling up preschool skill building.

In terms of policy, designing and regulating classroom curricula are remarkably direct levers for intervening in educational experiences in classrooms of scaled-up preschool programs. The supported implementation of domain-oriented curricula has been a central tenet of several key early childhood interventions. Perry used highly qualified and trained teachers to implement what has become the HighScope curriculum, and to conduct weekly meetings with parents in their homes. Abecedarian also relied on highly qualified and trained educators who implemented the Learning Games curriculum, which emphasized frequent language-rich interactions between teachers and students. The small scale of both of these programs ensured that the curricula would be faithfully implemented and, if necessary, adapted to solve emerging implementation problems.

How to move ahead with effective curricula? Preschool programs typically choose their own curricula, but their choices are often limited to a pre-approved list provided by state agencies and accrediting bodies. Many of these lists, including those issued by Head Start programs, consist of “whole-child” curricula, which emphasize child-centered active learning, facilitated by strategically arranging the classroom environment. It takes a great deal of skill on the part of teachers to implement a whole-child curriculum effectively. The teacher’s task is to scaffold learning with just the right amount of input – not so little that the child fails to learn, but not so heavy-handedly that the task becomes teacher-driven. In the case of Head Start, there is wide variation in learning activities and classroom quality across settings that use the same whole-child curricula (Jenkins et al., 2019).

In contrast, more targeted, skill-specific curricula are often delivered as supplements to activities in whole-child curricula or to activities teachers develop on their own. Proponents of skill-specific curricula argue that preschool children benefit most from sequenced, explicit instruction that focuses on specific academic (e.g., literacy or math) or socioemotional (e.g., emotion regulation) skills and is provided in the context of play and exploration. Building Blocks and, to a large extent, Abecedarian’s Learning Games are examples of skill-specific approaches.

Findings from meta-analytic studies (Chambers et al., 2016; Wang et al., 2016), experimental studies (Jenkins et al., 2018; Preschool Curriculum Evaluation Research Consortium, 2008), and reviews of curricula (National Center on Quality Teaching and Learning, 2014), show a consistent pattern of short-run impacts of skill-specific curricula in the targeted skill domain (i.e., math curricula boost math skills relative to either a whole-child or a locally developed curriculum). Virtually no evaluation has demonstrated longer-run impacts.

In contrast, children exposed to whole-child curricula do not outperform children in classrooms that use locally developed curricula. These results are remarkable: Despite the widespread use—and average per-classroom price tag of $2,000—of whole-child curricula, they are no more effective at boosting school readiness than the assortment of activities that early childhood education centers develop on their own (Jenkins et al., 2018). To be clear, these results do not establish that whole-child approaches are without value. After all, Perry’s whole-child HighScope curriculum generated impressive long-term impacts. But whole-child curricula as currently designed and implemented do not appear to be good policy bets when subjected to rigorous evaluation against locally developed curricula.

Several recent experimental evaluations of supplemental curricula and teacher-training modules designed to improve children’s socioemotional skills have demonstrated greater effectiveness when compared with usual classroom practice. One of the most successful, Preschool PATHS, has shown impacts on children’s emotion knowledge, problem-solving skills, behavior, and self-regulation (Bierman et al., 2008; Morris et al., 2014). But a host of other socioemotional curricular interventions, including the widely known “Tools of the Mind” curriculum targeting children’s executive function skills, are not effective at cultivating any of the outcomes in this broadly-defined developmental domain (e.g., working memory, delay of gratification) (Morris et al., 2014; Nesbitt & Farran, 2021). And one of the longest-run evaluations of a socioemotional intervention, the Chicago School Readiness Project, has followed participants through college entry and shows overwhelmingly null effects across a broad range of outcomes (Watts et al., 2021). Perhaps a skill-specific curricula that successfully targets a truly fundamental skill (e.g., complex language) may persist in the long run, and we encourage work on this frontier. As programs continue to grow, improving teacher training and coaching is a promising approach for making curricula more effective. Kraft et al.’s (2018) meta-analysis found impressive (~.50 SD) impacts of coaching interventions on pre-K instruction, with noteworthy but more modest carry-over impacts on student achievement.

VIII. Next steps

How to move ahead, given these puzzling patterns of results? Perry and Abecedarian clearly demonstrate that high-quality (and high-cost) programs designed and run by researchers, and in the context of low-quality counterfactual conditions, can transform the lives of many enrolled children. But RCT evaluations of more recent programs show a much more pessimistic pattern of null short- and medium-term impacts, followed, in only one case, by more promising longer-run outcomes. Our takeaway from the curriculum intervention literature is that curricula are an effective way to use policy to boost specific skills during the preschool year, but many of these skills also fade out quickly. Furthermore, professional development supports are essential if these approaches are to be implemented with fidelity. Longer-run follow-ups for the NHSIS and TN VPK evaluations, as well as for the successful targeted curricular interventions reviewed above, are a clear research priority, but results are years away. In the interim, we suggest three promising areas for future work:

• Aligning learning goals in preschool with those in K-12 schooling

• Extending our understanding of preschool impacts on child health

• Expanding the idea of skill-building to include parent capacity-building via employment supports

Aligning preschool and elementary school learning

Testing for sustaining environments, several studies have examined children’s experiences in the early grades, and kindergarten in particular. The findings here suggest that kindergarten does not offer enough meaningful, challenging, new material for children who have attended preschool, and much of the catch-up from children who have not—which leads to fadeout—occurs in this first year of elementary school (Engel et al., 2012; Weiland et al., 2021). How can we align preschool learning objectives with those of K-12? We’ve created a public education system that builds skills from one grade to the next, and long-standing evidence demonstrates that children gain between half and a full standard deviation on literacy and math achievement test score in the early grades of schooling (Hill et al., 2008). It seems reasonable to expect this might be done for public preschool as well.

Curriculum is one essential component of alignment, but curricula are ideally driven by learning standards, which should be integrated across P-12. Yet, general K-12 standards for reading, mathematics, and science education—namely, the Common Core State Standards and Next Generation Science Standards—remain distinct from each state’s home-grown “early learning standards.” Although some states aim to align their early learning standards to the Common Core, there is no consensus regarding early learning standards analogous to the Common Core for state adoption. Perhaps telling of the early childhood development/K-12 educational approach differences are the typical domains for which standards for the former are produced: motor skills, socioemotional development, approaches to learning, language and literacy, mathematics, and science (the latter two may be bundled into “cognitive development”). These learning objectives are much more varied than the traditional domains of English language arts, mathematics, and science. It is clear that the way developmentalists conceive of preschool learning is quite different from the way education administrators think about K-12, and even elementary school learning.

Another complication here is that pre-K is not simply an additional grade before kindergarten. Many public pre-K programs are offered through mixed delivery; private and community-based child care providers contract with state and local government to offer pre-K in their facilities, and providers may include public schools (e.g., North Carolina, Florida, New York City). Other programs provide pre-K services solely in public schools (e.g., Tulsa, Tennessee, California). The complex governance structures of pre-K programs, the varied settings in which children are served, and the generally siloed nature of early learning and K-12 organizational arrangements and funding sources pose substantial challenges to states and communities as they seek to offer cohesive experiences (Stein & Coburn, 2021). Johnson and Jackson (2019) demonstrate that investments in high quality preschool, via Head Start, when coupled with subsequent investments in K-12 learning, can produce dynamic complementarities. But few other studies testing for complementarities in skill-building investments yield positive findings.

All of this underscores the challenge of creating developmentally appropriate curricula that are effective at cultivating hard-to-reach fundamental skills, implemented well with supports, and aligned across the continuum of preschool through elementary education. Young children learn through play and exploration; over time, children develop the ability to learn in more structured, didactic settings. The task for researchers, educators, and policymakers is to develop program models that keep play integrated during the early elementary years, and to integrate into the preschool context fundamental domain-based skill development that facilitates longer-run subject-based learning.

Understanding preschool impacts on child health

An emerging body of literature examines how children’s early physical and mental health and behavioral skills play an important role in determining later school success and long-term well-being (Currie, 2009; Figlio et al., 2014). Research has documented child health disparities across income groups, finding causal evidence that poor health in childhood, and prior to school entry, has adverse effects on future success (Case et al., 2005; Currie et al., 2010). Thus, children who begin their public-school careers with both economic and health disadvantages are particularly at risk for poor educational and later life outcomes.

Because education is a leading social determinant of health, it is not surprising that public health scholars view preschool programs as an important policy strategy (Woolf et al., 2007). A key question here is whether educational interventions—especially those like Head Start and Abecedarian that have explicit nutritional and health objectives—improve child health directly, or whether preschool interventions improve health outcomes by virtue of improving educational achievement, promoting economic prosperity, and bringing about shifts in health behaviors (Muennig, 2015). The Abecedarian study included objective measures of health only during follow-up, but showed impressive long-run impacts in reducing metabolic and cardiovascular diseases during adulthood (Campbell et al., 2014).

Moving forward, it would be beneficial for studies to include objective measures of health outcomes before and immediately after preschool intervention, an approach that would shed light on the role played by preschool programs in altering education, behavior, and health, and how impacts in the post-treatment period mediate those in the long run.

Expand the focus from child skill building to include family capacity-building supports

Our review points to a few possible avenues for identifying the key skill development targets for preschool programs, including rich oral language and health and health behaviors. However, we need to learn more about the policy levers and classroom processes through which programs may move the needle on these domains (e.g., curricular selection). Alignment of early learning and K-12 environments may support these aims, but we lack convincing evidence on whether and how alignment fosters skill building.

Where does this leave us in our review of policies to support preschool skill development? We worry that the focus on the direct benefits of preschools for child human capital, via skills building, has distracted us from the capacities that child care, in and of itself, allows families and parents to cultivate. By supporting parents’ human capital investments, employment, and ultimately their family income and home investments in child enrichments (e.g., nutritious meals), child care supports can also indirectly benefit children’s skill development (Burchinal, Whitaker, et al., 2022).

For many advocates, the aim of safety net programs is to provide a comprehensive system of supports for low-income families. The existing support structure provides near universal access for eligible families to most domains of basic needs, such as nutrition and medical care, as the other papers in this volume show (e.g., Jackson et al.), but this is not true for child care. In fact, only 15% of eligible families receive child care subsidies through the Child Care and Development Block Grant (CCDBG), a welfare program originally developed in the 90s; Head Start, Early Head Start, and state pre-K programs serve between 10–35% of eligible children (Borowsky et al., 2022). However, shoring up the supply of and public funding for child care is important for promoting parents’ education (Herbst & Tekin, 2011), employment (Morrissey, 2017), and well-being, irrespective of children’s skill development.

Furthermore, the largest skill-building benefits of preschool programs are enjoyed by children who would otherwise have been cared for in their homes (Feller et al., 2016; Kline & Walters, 2016). Indeed, experimental evidence from variations in welfare policy packages from the 1996 reform era show that policy bundles which increased both parents’ income and use of center-based child care produced the largest gains in children’s achievement (Duncan et al., 2009). By expanding the child care safety net, and reducing barriers to accessing such programs (e.g., Bassok et al., this volume) policymakers could create more such opportunities for families and support children who would benefit most.

How might we expand this safety net support for working families and provide more care for children? Early Head Start and Head Start are very comprehensive programs that cannot easily be scaled up to serve all income-eligible children. State pre-K also has a limited reach because it targets only 3- and 4-year-old children, and is often provided as part-day sessions. In contrast, child care subsidies through the CCDBG are designed for scaling. Expansions of the CCDBG were part of the Build Back Better legislation and included in separate legislation proposed by Senators Murray and Kaine (Borowsky et al., 2022), which increase maternal employment (Blau & Tekin, 2007). Although child care stability and quality have been perennial issues in the CCDBG program, major policy changes, including the 2014 CCDBG reauthorization, American Reinvestment and Recovery Act child care stabilization grants, in addition to Murray and Kaine-like proposals, have and will continue to change substantially the care system. We urge research on these combined reforms, especially in their effects on family capacities.

VIII. Conclusions

Policy discussions of preschool programs typically assume that they “work” by imparting valuable school-readiness skills to their students that provide the foundation for lifelong skill advantages and greater success in adulthood. Attention then turns to program affordability and, in some cases, considering whether responsibility for early care best resides with the government (via subsidies or direct provision) or with families.

Our review raises questions about the degree to which results from early program models such as Perry and Abecedarian are generalizable to today’s preschool programs, and whether simple skill-building models are a good framework for moving forward. We believe that policy investments in preschool education can be profitable for society, but that considerable effort will be required to maximize the student learning experience in preschool classrooms, integrate the learning goals of the preschool year and K-12 schooling, and continue to push for new discoveries regarding truly fundamental early skills (including health) that preschools should target. It is possible that researchers uncover the “right” early skills that underlie long-run success, and a reliable way to cultivate such skills in scaled-up preschool programs. In the meantime, all of this should be complemented by policy efforts aimed at supporting the capacities of families to resource (i.e., via employment) and stabilize (i.e., via child care subsidies, tax credits) their investments in child development.

Biographies

Jade Marcus Jenkins is an Associate Professor in the School of Education at the University of California, Irvine, studying early childhood policy. Her work is multidisciplinary, focusing on issues that are amenable to educational and social policy intervention, using diverse research methods to evaluate programs and understand the mechanisms that promote child and family wellbeing.

Greg J. Duncan is Distinguished Professor in the School of Education at the University of California, Irvine. Duncan’s recent work has focused on the impacts of income supplements on the cognitive and socioemotional development of infants. Duncan was elected to the National Academy of Sciences in 2010.