Program Evaluation on Early Head Start

I. BACKGROUND LITERATURE REVIEW PERTAINING TO THE EARLY HEAD START STUDY

HelenH.Raikes, JeanneBrooks-Gunn,and JohnM.Love

Although considerable information exists about the effects of early interventions for at-risk children, we know more about intervention effects and results of investments in programs for children of specific ages, for example, 3- to 5-year-olds, than about when to invest or for whom specific investments are most advantageous across the age span from birth to 5. Outside of a few program evaluations that span the period from birth to age 5 (Garber & Heber, 1981; Ramey & Campbell, 1984; St. Pierre, Goodson, Layzer, & Bernstein, 1994), little is known about cumulative intervention experiences during the period from birth until formal schooling begins. Moreover, variations in program models and timing of services have not been examined. Investigation of the variability in quantity and timing of services over the years from birth to age 5, the focus of this monograph, is likely to be useful to programs for fine-tuning the timing and intensity of interventions, to policy makers for optimizing early childhood investments, and to developmental scientists for better understanding trajectories of develop- ment especially of poor children in the context of environmental influences.

This monograph reports on a program evaluation of 3,001 children in 17 sites from poor families, half of whom were randomized to receive Early Head Start (EHS) services in the first 3 years of life (for some families in some sites EHS services began prenatally; ACF, 2002a; Love et al., 2005), and half of whom were not. Children’s outcomes were examined at ages 2 and 3 when concurrent impacts of the EHS programs could be examined, and at age 5, 2 years after EHS ended and when formal program experiences subsequent to EHS could be examined. Analysis of program impacts for families and children in the first 5 years of life are based on an experimental evaluation. However, looking at the services that families obtained for their children in

Corresponding author: Helen Raikes, Child, Youth and Family Studies, 257 Mabel Lee Hall, University of Nebraska-Lincoln, Lincoln, Nebraska 68588-0236, email: hraikes2@unl. edu

1

the preschool years allows for a description of how combinations of infant and toddler (for children from birth to age 3) and preschool (for 3- to 5-year- old children) services may influence school readiness in low-income children.1 Thus, the monograph addresses both longitudinal treatment impacts, as well as longitudinal variations in experience and associated outcomes.

The advantages of early intervention programs for vulnerable children prior to school entry are generally acknowledged (Barnett, 2011; Camilli et al., 2010; Duncan & Magnuson, 2006). Although programs may have multiple purposes, most ultimately aim to reduce gaps in school readiness between low- income and more advantaged children. Low-income children typically enter school from half to a full standard deviation below more advantaged children in academic-related domains such as vocabulary, cognition, and specific literacy-related skills, and often demonstrate challenges in social–emotional and regulatory functioning (Brooks-Gunn & Duncan, 1997; Duncan & Magnuson, 2005; Hart & Risley, 1995; Reardon & Galindo, 2009; Rouse, Brooks-Gunn, & McLanahan, 2005; Stipek & Ryan, 1997). Moreover, many of these gaps are observable much before school entry, as early as age 3 (Klebanov, Brooks-Gunn, McCarton, & McCormick, 1998) and possibly in the toddler years. Differences in children’s environments during the earliest years have been linked to school-age outcomes, typically operating through links with early outcomes (Duncan, Brooks-Gunn, & Klebanov, 1994; Hart & Risley, 1995). Several notable studies document that low-income children receive less cognitive and linguistic stimulation than do children in higher- income families (Bradley et al., 1989; Brooks-Gunn & Markman, 2005; Klebanov et al., 1998; Zill, 1999). For example, low-income children receive about a fifth of the language inputs from parents that more-advantaged children receive (Hart & Risley, 1995).

Children who receive high-quality intervention services, compared to children who do not receive these services, enter school with greater skills in school-success domains (Barnett, 2011; Karoly, Kilburn & Cannon, 2005; Ramey & Ramey, 2006; Schweinhart, 2006), and continue on relatively higher success trajectories at least through elementary school (Camilli, Vargas, Ryan, & Barnett, 2010; Heckman, 2006). As a consequence, many states now invest in prekindergarten programs and/or services for younger children (Barnett, et al., 2006). Policy makers agree that high-quality early childhood services for low-income children are valuable (Brooks-Gunn, 2011; Gormley, 2011). However, most evaluations have examined effects of a single program (Olds et al., 1997; Ramey & Ramey, 2006) or for a discrete portion of the preschool years (e.g., Schweinhart, 2006). Investigations of relative and cumulative contributions of combinations of program services from birth to age 5 have been limited, although it has been recommended that comprehensive and systematic assessment of early childhood experiences over time would be

2

more productive than relying on investigations of brief or one-time programs (e.g., Reynolds, Wang, & Walberg, 2003).

In this monograph, we present findings from multiple analyses from the Early Head Start Research and Evaluation Project (EHSREP) that tell a story about when and how intervention experiences throughout the period birth to age 5 may have contributed to the outcomes for poor children at school entry. Outcomes are examined for children who were randomly assigned before age 1 into EHS or a control group and followed through age 5 (shortly before they began kindergarten). Specifically, in this monograph, we (1) examine the patterns of impacts of EHS on children ages 2, 3, and 5, emphasizing age 5 impacts measured 2 years after the program ended; (2) report on children’s formal program experiences after EHS and examine the impact of EHS on the probability of children receiving follow-up program experiences; (3) report on the influences of formal preschool programs on age 5 outcomes above and beyond the effects of EHS; and (4) consider how various patterns of early experiences from birth to age 5 are associated with the observed prekindergarten outcomes. A formal preschool program is defined as a center-based program that may include child care, prekindergarten, or Head Start.2 We present each of the findings for selected subgroups of children and families, examining variations according to race/ethnicity (African Ameri- can, Hispanic, or White), and level of cumulative family demographic risk (summing across whether mother was a teenager at the time child’s birth, unemployed or not in school, unmarried, receiving TANF/AFDC, and not a high school graduate at the time the study began). We also examine three EHS program models (comparing infant and toddler experiences that were initially center-based, home-based/home-visiting, or a combination of these). The monograph addresses these questions using several design and analytic approaches including experimental and nonexperimental regressions as well as more exploratory, descriptive approaches. Although confidence in inferring conclusions varies as a result of these different approaches, the several questions, linked by an overarching conceptual framework, provide fertile ground for more comprehensive ways of thinking about children’s early development in the context of poverty and early childhood intervention programs than has been typical in the literature to date.

This literature review begins the monograph by laying out the theoretical foundation for the work, and it surveys what is known about the prevalence of formal program participation prior to school entry and the effects of intervention programs. We also provide an overview of the EHS program. Next, we provide an overview of the service context for low-income children birth to age 5; examine outcomes from programs for children targeting infants and toddlers, preschoolers, or children from birth to age 5; and examine extant research that pertains to subgroups of children of interest in this monograph. These subgroups are defined by race/ethnicity and level of

3

BACKGROUND LITERATURE REVIEW

cumulative demographic risk factors. We also examine the impacts for families enrolled in different program models.

THEORETICAL FRAMEWORK

Our work is informed by a developmental perspective that suggests interventions are likely to change the slope of developmental trajectories and that differential experiences of children during various periods may influence different aspects of child well-being (“principle of developmental timing” proposed by Ramey & Ramey, 1998). This principle leads us to expect (1) that parents will be most affected by parenting services during the infant and toddler years when they are forming their expectations of the child and for their own parenting; (2) that intervention effects for the parent-child relationship and social–emotional development will be greatest during the period of relationship and attachment formation; and (3) that intervention effects for children will be greatest during periods of most rapid learning in specific developmental areas. For example, language will be most affected during periods of rapid language learning; and school-like tasks will be most affected by periods close to school entry. Our thinking has been guided by a perspective suggesting that concurrent intervention effects are most likely to influence readiness in the first 5 years. However, we have been further informed by alternative views about timing of intervention effects. Few research teams have considered the additive or interactive effects of intervention and time. We acknowledge the relative newness and potentially post hoc nature of this way of thinking currently; however, we also note the soundness of the basic principles, the importance of questions about the most effective timing of intervention for different forms of development, and challenge the field to further develop time by intervention hypotheses to guide studies in the future.

The NICHD Early Child Care Research Network (NICHD ECCRN, 2001) has proposed four alternative explanations or hypotheses for how the timing of experiences may influence development: (1) The hypothesis of Experiences at Specific Points in Development posits that interventions will have more value when timed to coincide with sensitive periods in children’s development, for example, we might expect that providing responsive parenting during the first year of life might be most important for subsequent relationships; (2) the Early Experiences hypothesis suggests that early experiences will outweigh later ones; consistent with this hypothesis, we would expect some effects from EHS to outweigh those of later prekindergarten experiences because of the former’s primacy in setting trajectories, as might be the case for social and emotional development; (3) the Contemporaneous Experiences hypothesis posits that current experiences may be more important than earlier ones; here, we might expect prekindergarten

4

education to matter more than earlier services vis-à-vis school readiness; (4) the hypothesis of Incremental or Augmented Experiences suggests that early experiences produce effects that are maintained by later experiences. We might expect children who received EHS together with preschool education to be better prepared for school than children who did not receive it.

Certainly, developmental processes are amplified by transactional processes (Sameroff & Fiese, 2000), by which children’s development and experiences at one point in time contribute to experiences that children receive subsequently. This way of thinking is consistent with that of Heckman and Masterov (2004), who suggest that “more begets more.” Children who have more skills attract more experiences subsequently. However, little is known about the relative contributions of “early” (e.g., 0–3) and “later” (e.g., 3–5) experiences to later development.

In their conceptual approach to early childhood intervention, Ramey and Ramey (1998) postulate that intensity, directness, breadth, and flexibility of program experiences maximize opportunities for children at risk. However, precisely when to target intervention within the first 5 years of life, which is generally a time of rapid development, and how to vary intensity and breadth and for different subgroups has yet to be explicated. Moreover, as stated by the Rameys’ “principle of environmental maintenance of development” (p. 117), adequate supports must be in place to maintain children’s positive gains after programs end. The current study investigates variability in timing and type of intervention for children with differing characteristics. Particularly, our study investigates variability in supportive early childhood experiences in the two years following the 3 years of EHS intervention.

Heckman’s (2006) analyses of early childhood provide continued support for the notion that early learning influences later learning. One set of analyses (Cunha, Heckman, Lochner, & Masterov, 2006) suggests that cognitive processes not only relate to later school success but are particularly amenable to early intervention whereas social and emotional processes seem to respond to later intervention. However, these researchers did not have the opportunity to examine social and emotional processes as early in life as we did (i.e., before 3 years of age), an investigation worth making given the importance of this early period for attachment and self-regulation. The current study provides an opportunity to investigate whether specific aspects of development are more amenable to early (0–3) invention and whether other aspects of development may be more amenable to later (3–5) intervention.

THE EARLY HEAD START PROGRAM

We begin by providing background on the EHS program, on the EHSREP, and on general findings when the program ended; this monograph

5

BACKGROUND LITERATURE REVIEW

emphasizes children’s experiences after EHS, during the period when children were 3–5, and child and parent outcomes just prior to children’s entry into kindergarten as well as describing earlier treatment impacts (at ages 2 and 3) to illustrate patterns of program impacts over time. The 3,001 children and families in the EHSREP enrolled in one of the 17 EHS programs chosen to participate in the study (the first 68 EHS programs were funded in fall 1995). These 17 communities were reflective of all EHS sites at the time in geographic and family diversity.

The EHS program was created through the reauthorization of the Head Start Act in 1994 and was expanded in 1998. EHS is a Head Start (HS) program and, as is true for all HS programs, must serve at least 90% families at the federal poverty level or below, must implement additional criteria so that children in families with greatest needs in communities are served, and must recruit at least 10% children with disabilities. EHS programs can begin services during pregnancy and continue to serve children and families until children reach age 3. All HS programs are two-generation, providing services for both children and parents, although many parent services may be referred to community organizations. At the time of the evaluation, up to 10% of the national HS budget could be allocated for EHS services. In 2009, the program served about 70,000 families in 65 communities. However, EHS received $1.1 billion under the American Recovery and Reinvestment Act of 2009, and in 2010 began to serve an additional 55,000 children.

EHS programs conduct community needs assessments to determine the most appropriate of four specific service delivery models for families: center- based, home-based, combination, or locally designed options. For purposes of the research, sites were classified as to whether they were offering home- based, center-based, or mixed services (a combination of home-based and center-based services). Program protocols for services to families were established by the Head Start Program Performance Standards (U.S. Department of Health and Human Services [DHHS], 1996). Requirements for each program model are specified (e.g., curriculum, educational requirements for teachers, group sizes, frequency and length of home visits, health and developmental screening) and programs are monitored every 3 years for their adherence to the Program Performance Standards. The Program Performance Standards are undergirded by the Report of the Advisory Committee on Services for Infants and Toddlers (US DHHS, 1994), the committee that designed EHS.

The authorizing Head Start Act required evaluation of EHS and, therefore, an evaluation was launched at the same time the program began. Because the program was new, an extensive implementation study was conducted as well. A national contractor was selected (Mathematica Policy Research, Princeton, NJ, together with Columbia University’s National Center on Children and Families), and local researchers in 15 universities partnered

6

with sites to collect data and to pursue specific local questions. They formed a research consortium with the contractors and ACF. Using the experimental design, analyses compared program and control groups when children were 2 years old and when they were 3 years old and their enrollment in the program had ended. EHS services end when children reach age 3. What expectations were there for children’s education experiences from the time they left EHS until they entered kindergarten? As noted, the Advisory Committee on Services to Families with Infants and Toddlers (U.S. DHHS, 1994) specified that EHS programs “transition” their 3-year-old children into community formal care and education when the EHS program services ended. The Advisory Committee did not expect impacts to be maintained if children did not have subsequent early childhood education experiences. In the next section of this literature review, we first examine the prevalence of formal programs for children 3–5 (and the likelihood that the vision of the Advisory Committee for post-EHS experiences could be achieved) and then report on child and parent outcomes from various combinations of program experiences for children from 0 to 3, 3 to 5, and 0 to 5.

PREVALENCE OF PARTICIPATION IN FORMAL CARE AND EDUCATION

Although we tracked children’s experiences after EHS into formal programs when they were 3–5 years of age, it is important to know what would have been available to them. What was the prevalence of the types of follow-up experiences the Advisory Committee sought for EHS children? Low-income children may be served in multiple community settings, with services that vary in intensity, breadth, and scope. Of particular interest were formal preschool programs as emphasized by the Advisory Committee (primarily center-based services that included a structured early childhood program). National prevalence statistics provide a context for the types of care and education that could have been available for children starting at age 3. Next we present data on the prevalence of formal programs for 3- to 5-year-olds, and for specific types of formal programs (e.g., HS, child care, prekindergarten) during the period (early 2000s) when follow-up data were collected.

Data from the Early Childhood Longitudinal Study-Kindergarten Cohort (ECLS-K), which studied a nationally representative sample of children who attended kindergarten in the fall of 1998, showed that 68% of children attended structured early childhood education programs (prekindergarten, Head Start, or center-based child care) during the year before kindergarten (Rosenthal, Rathbun, & West, 2005). Attendance figures were about 10% lower for disadvantaged children (58%; Reynolds, Magnuson, & Ou, 2006). Similarly, data from the National Household Survey of Education Programs show that 66% of 4-year-olds and 43% of 3-year-olds were in center-based care

7

BACKGROUND LITERATURE REVIEW

and educational settings in 2001 (Mulligan, Brimhall, & West, 2005), and in 2005, 55% of 3- to 5-year-olds were in a center-based program (basically the average of 3- and 4-year-olds since most 5-year-olds are in kindergarten; Iruka & Carver, 2006). These included three common forms of formal care and education—prekindergarten, Head Start, and center-based child care.

Prekindergarten programs provide increasing opportunities particularly for 4-year-olds. State funding for prekindergarten has grown in the past 10– 15 years although the rate of growth has recently slowed. In 2001–2002, about the time EHSREP sample children were in or completing preschool, 40 states served about 700,000 children, mostly 3- and 4-year-olds, including both general and special education, with state spending of about $2.4 billion. Nine years later these figures about doubled: in 2010–2011, 39 states served 1.3 million children, at a total cost of $5.5 billion. (Barnett, Robin, Hustedt, & Schulman, 2003; Barnett, Carolan, Fitzgerald, & Squires, 2011). Many state programs target low-income children, but in some states preschool services are nearly universally available, typically for 4-year-olds. Most prekindergarten programs do not serve children under age 3, although there are exceptions.

HS provides services primarily to 3- and 4-year-old children from the lowest- income families, those living at or below the federal poverty line, particularly. With a federal budget of $7.6 billion in 2011, HS served 942,354 children in 2011 and 912,345 in 2001–2002 (Schmit, 2012b), 90% of whom were at the federal poverty level or below. Of these, the vast majority, 90%, were age 3 and older. EHS, serving children under age 3, drew 17.6% of the federal HS appropriation and served an additional 165,522 children and pregnant women nationally (Schmit, 2012a). (EHS services were recently expanded to these numbers by a one-time funding increase through the American Recovery and Reinvestment Act of 2009.) HS is believed to serve between 40% and 50% of the eligible children living in poverty in the United States; however, state-sponsored prekindergarten programs (in some states HS and state prekindergarten programs may be combined) also serve children eligible for HS. The smaller EHS program, even with expansion, serves only a small proportion of eligible infants and toddlers, approximately 3%.

Child care programs (including center-based and home-based programs) serve large numbers of children beginning when children are infants. In 2002, data from the nationally representative National Survey of America’s Families (NSAF) showed that more than 68% of children younger than 5 with low- income, employed mothers were in some form of nonparental care (Capizzano & Adams, 2003). The authors reported that 25% of these children were in center-based child care, an option more commonly used for 3- and 4-year-olds than for younger children. Other prevalent forms of out-of- home care were relative care (30%), family child care homes (11%), and nonrelative in-home care (4%). Federal contributions to child care in 2005, mostly for low-income children, were approximately $5.3 billion (Child Care

8

Bureau, 2006a); states added at least $2.2 billion in maintenance of effort funds (Child Care Bureau, 2006b) and TANF transfers and other funds further contributed. About 1.8 million children were served by federal/state subsidy child care programs (Child Care Bureau, 2006a).

The foregoing are the common forms of formal programs. In this monograph, formal program settings (i.e., preschool education) for children aged 3–5 years were the focus.3 Altogether, several million children living in low-income families are served in various combinations of formal programs when children are 3–5 years of age (HS, prekindergarten, and child care). Yet, as is clear from this brief review, more formal program services tend to be available for 4-year-olds than for 3-year-olds. EHS programs were charged with placing children in quality formal care and education through their remaining preschool years. It is important to determine if services were available and whether it was possible to meet that goal.

Moreover, early childhood educational services represent billions of dollars in state and federal investments and time spent in these programs constitutes substantial proportions of children’s early years. As we will demonstrate later, many of these programs intend to influence children’s development and, in the case of those serving poor children, to help reduce the gaps in school readiness skills.

Children’s access to preschool education programs could vary by community context (e.g., whether communities are in rural or urban areas and whether there are good quality formal programs available in a community for low-income children aged 3–5), family race/ethnicity, family risk factors, and EHS program model. Anecdotally, it is well known that variability exits in access to quality early childhood programs across communities. The striking differences in access to prekindergarten programs for 4-year-olds versus for 3-year-olds (Barnett, Hustedt, Hawkinson, & Robin, 2006), and the greater number of 4-year-olds than 3-year olds served in HS—in 2002–2003, 53% of HS children were 4 years old and 34% were 3; the rest were 5 or under 3 years of age (Early Childhood Learning and Knowledge Center, 2004)—suggest that some communities might not have services to carry out the EHS vision of quality preschool follow-up services. For example, studies also show that children are more likely to have been in center-based care during the year prior to kindergarten in urban areas than in rural areas. Among African American children entering kindergarten (in most U.S. states when children enter kindergarten they must be age 5, although some may be as old as 6), those in urban areas were more likely to have been in center-based care the year prior to kindergarten (37%) than those in rural areas (14%); 55% of urban White children had been in center care the year before kindergarten as compared to 35.3% of rural White children (Grace et al., 2006). On the other hand, the same study showed that children were more likely to have been in HS if they were rural (17% of all rural children in the kindergarten sample)

9

BACKGROUND LITERATURE REVIEW

versus urban (8.7%). Additionally, 37% of rural African American children attended HS the year before kindergarten versus 20% of urban African American children.

Not only are disparities seen within racial groups according to urbanicity; they are seen between racial/ethnic groups. Low-income African American children tend to be more likely to use center-based care than are other low- income children, and Hispanic children tend to be least likely to use this form of care (Capizzano, Adams, & Ost, 2007; Magnusson & Waldfogel, 2005). Though studies examining formal program use by cumulating family demographic risk factors do not exist, prekindergarten programs serve children across a wider socio-economic range than does Head Start. However, one of the variables that comprises the risk index used in the current monograph—mother’s education—is known to be associated with program participation. In one study, children were more likely to enroll in preschool if their mothers had a graduate or professional degree (66% in at least one weekly care arrangement) than if they had a high school diploma or GED (55%) or had not completed high school (35%; Iruka & Carver, 2006).4 We found no studies examining availability of programs for children ages 3–5 related to types of program services children had received during the preceding three years, a matter we will pursue in the current study. In general, the variability in differential supply and access is an issue for the field of early childhood education (Burchinal, Nelson, Carlson, & Brooks-Gunn, 2008).

IMPACTS OF INTERVENTION PROGRAMS OFFERING SERVICES PRENATAL TO AGE 3

Since the 1960s, a number of programs have provided intervention services for children from birth (or prior to birth), during infancy, and in some cases up to age 3. The programs we consider first do not provide follow- up services after age 3. During the infant and toddler years, the programs frequently offer two-generation services to both children and parents. Here, we review outcomes for children and parents from the Infant Health and Development Program (Brooks-Gunn, Klebanov, Liaw, & Spiker, 1993; Infant Health and Development Program [IHDP], 1990; McCarton et al., 1997; McCormick et al., 2006); the Parent Child Development Centers (PCDCs; Johnson & Blumenthal, 1985); the Yale Child Welfare program (Seitz, Rosenbaum, & Apfel, 1985); Healthy Families America (HFA; Daro & Harding, 1999); Nurse Family Partnership (NFP; Olds et al., 1997; Olds, Henderson, Kitzman, & Cole, 1995), UCLA Home Visiting/Mother Infant Group Intervention (Heinicke, Fineman, Rodning, Recchia, & Guthrie, 2001), and Parents as Teachers (PAT: Wagner & Clayton, 1999). Services these programs offered during the infant and toddler years were home-based,

10

center-based, or a combination (Howard & Brooks-Gunn, 2009; Sweet & Appelbaum, 2004). For example, IHDP offered home visits early after infants’ discharge from the hospital; at age 1, children began center-based care as well.

Intervention effects have been reported at age 2 or 3 (when the programs ended) from participation in intensive infant/toddler programs. Favorable effects of intervention were found (1) for children’s cognitive development in the IHDP (IHDP, 1990; Brooks-Gunn et al., 1993) and the Houston PCDC (Johnson & Blumenthal, 1985); (2) on language development in the IHDP (Brooks-Gunn et al., 1993) and Yale Child Welfare participants (Seitz & Provence, 1990); and (3) on social and emotional development among treatment groups in the IHDP (Brooks-Gunn et al., 1993; IHDP, 1990). Overall effect sizes in these studies ranged from small to large. Few home visiting programs studied using rigorous experimental design have demon- strated positive impacts on child outcomes at or before age 3 across their entire sample (Howard & Brooks-Gunn, 2009; Sweet & Appelbaum, 2004).

Many infant and toddler programs offer parenting support; most of these are home visiting programs and a number of them have reported positive effects for parents (see Howard & Brooks-Gunn, 2009, for a recent review). Home visiting programs have reported improvements in maternal mental health and reduced subsequent pregnancies (Kitzman, Cole, Yoos, & Olds, 1997), increased parental reading to children (Johnson, Howell, & Molloy, 1993), greater reliance on nonviolent discipline (Heinicke et al., 2001), increased sensitivity in interactions (Olds et al., 2002), reduced depressive symptoms (Gelfand, Teti, Seiner, & Jameson, 1996), and less child maltreatment (Daro & Harding, 1999; Olds et al., 1997; Olds et al., 1995; Wagner & Clayton, 1999). Programs that affect parenting have been criticized because these changes in parents have not been consistently linked to changes in children’s development (Duncan & Magnuson, 2006, Howard & Brooks-Gunn, 2009). However, Olds and colleagues have demonstrated longer term favorable child outcomes following early gains in parent-related behaviors (Olds, 2006) using impact analyses at multiple points in time.

Of particular relevance is that some of these evaluations have followed their children through the elementary school years and, in some cases, even longer. Positive effects attributable to services received during the infant and toddler years were found on vocabulary test scores in the Yale Child Welfare Study (Seitz & Provence, 1990), on standardized Iowa Basic Skills test scores among boys in the Houston PCDC (Johnson & Blumenthal, 1985), and on IQ, reading, and math achievement in heavier low-birth-weight infants in IHDP (McCarton et al., 1997) at age 8. Achievement scores were higher at age 18 in the IHDP-treated group (McCormick et al., 2006). Long-term effects of the Yale Child Welfare study also included reduced behavior problems for boys at age 8 (Seitz et al., 1985) and reduced need for remedial and support services (Zigler, Taussig, & Black, 1992). The Nurse Family Partnership study found

11

BACKGROUND LITERATURE REVIEW

fewer arrests, convictions, and probation violations among intervention participants during adolescence (Olds, 2006). Effect sizes from these evaluations range from small to moderate, although Hill, Brooks-Gunn, and Waldfogel (2003) report a dosage effect with large effect sizes on cognitive development at age 8 among children who experienced more than 350 days of center-based care in the IHDP study over 2 years.

IMPACTS OF PROGRAMS SERVING PREKINDERGARTEN CHILDREN, AGES 3–5

Programs focused specifically on the prekindergarten period (ages 3 and 4, but most typically for 4-year-olds), have an extensive history. These programs typically include center-based experiences for young children, and some include comprehensive services for parents, although the parent component tends to be less prevalent than in programs serving children under age 3. Fewer programs exclusively target parents of children 3–5 years of age but rather most tend to focus more directly on children.

Some notable evaluations of preschool programs include those of the Perry Preschool Program (Schweinhart, 2006); Chicago Child-Parent Program (Reynolds & Temple, 2006); Tulsa Prekindergarten Program (Gormley, Gayer, Phillips, & Dawson, 2005); five state-sponsored prekinder- garten programs in Michigan, New Jersey, Oklahoma, South Carolina, and West Virginia (Barnett, Lamy, & Jung, 2005), and Head Start (US DHHS, 2005).

A recent meta-analysis identifies 123 experimental or quasi-experimental evaluations (Camilli et al., 2010). Early childhood programs have short-term effects on children’s cognitive development and on specific school readiness measures such as reading. Effect sizes have often been notable, ranging from 0.2 (US DHHS, 2005) to 0.6 (ACF, 2006; Camilli et al., 2010; Reynolds & Temple, 2006). Short-term school-related achievement gains are possible, with some evidence pointing to the importance of direct instruction (Barnett, 2011).

Those evaluations that have followed children into later years suggest that children receiving early childhood education are less likely to be held back in school or to receive special education services (Camilli et al., 2010). Notable too are findings showing that treatment children are more likely to finish high school, less likely to engage in crime, and more likely to be productively engaged throughout adulthood (Reynolds & Temple, 2006; Schweinhart, 2006). Larger, publically funded programs also have shown evidence of success, most notably the evaluations of prekindergarten programs (Barnett et al., 2005; Gormley et al., 2005).

With regard to HS, the HS Impact Study found few lasting effects through 1st grade, although effects were maintained for some subgroups, most notably

12

African American children (ACF, 2010). Other nonexperimental research designs have been used to examine effects of HS. Using a sibling design, Garces, Thomas, and Currie (2002) found that White children who had attended HS showed a significantly greater likelihood of completing high school and attending college, as well as some evidence of higher earnings in early adulthood. African Americans who were former HS participants were significantly less likely to have been charged or convicted of a crime. Ludwig and Miller (2006), using a regression discontinuity design, also found evidence of increased high school graduation rates and postsecondary participation, irrespective of race/ethnicity, in poor counties with enhanced HS participation. And, using propensity score matching procedures, preschool children who attended HS have been compared to those receiving prekindergarten, other center-based care, noncenter-based care, and parental care, with school readiness scores being higher for children attending HS compared to parental or noncenter-based care, but being similar or lower for children attending prekindergarten programs (Zhai, Brooks-Gunn, & Waldfogel, 2011; Zhai, Brooks-Gunn, & Waldfogel, 2011). The largest effects for IHDP were found when comparing children who received the treatment with those in parental or noncenter-based care (Hill, Waldfogel, & Brooks-Gunn, 2002). Using nonexperimental data from FACES, 2 years of HS was shown to confer more benefits than 1 year (Wen, Leow, Hahs-Vaughn, Korfmacher, & Mancus, 2011).

IMPACTS OF PROGRAMS BIRTH TO SCHOOL ENTRY

Studies that focus on services from birth to age 5 are more limited. These programs are likely to be quite expensive to implement and so it is perhaps not surprising that their prevalence is not widespread and that research comes largely from single-site demonstrations. Studies focused on children’s early care and education experiences from birth to age 5 include the Abecedarian Project (Ramey & Ramey, 2006); the Milwaukee Project (Garber & Heber, 1981), the Brookline Early Education Project (BEEP; Bronson, Pierson, & Tivnan, 1984; Pierson, Bronson, Dromey, Swartz, Tivnan, & Walker, 1983; Pierson, Walker, & Tivnan, 1984), and the federal Comprehensive Child Development Program (CCDP; St. Pierre et al., 1994). Most of the studies employed a treatment versus control group experimental design. BEEP used a within-treatment experimental design, randomly assigning participants to varying levels of program services.

The Abecedarian Project (Ramey & Ramey, 2006) provided continuous services for children in center-based settings from 4 months to the start of kindergarten. Of the 111 children, half were randomly assigned to receive the center-based program and half to a control group that received no program

13

BACKGROUND LITERATURE REVIEW

(both groups received health and social support services); children’s mothers were primarily primiparous, African American, and single. Similarly, the Milwaukee Project provided intensive, continuous, center-based and family support services to a special population: an extremely small sample of 20 African American mothers with very low IQs (75 or below; Garber & Heber, 1981). The BEEP study randomly assigned parents and children to three levels of intensity— (1) monthly home visits, meetings, and center-based child care; (2) similar but less frequent services; or (3) information and support at the center and by phone from birth to kindergarten. The CCDP provided intensive family support services in 21 locations. CCDP emphasized case management with family services and to a much lesser extent child development services (St. Pierre et al., 1994).

Two of these evaluations reported very large effects on children’s cognitive development (effect sizes of 1.0) at the time the program ended and/or when children entered school (Garber & Heber, 1981; Ramey & Ramey, 2006). The BEEP study observed social and emotional benefits among treatment children (Pierson et al., 1983). The CCDP program did not demonstrate cross-site positive outcomes, leading researchers to conclude that family support programs not providing child development services were not sufficient to affect child development (St. Pierre et al., 1994).

Birth to 5 programs have also been found to have positive impacts beyond the intervention period. Most notably, the Abecedarian Project reported that treatment children were less likely to be retained in grade or require special education and were more likely to graduate from high school and attend college (Ramey & Ramey, 2006). The treatment group had higher IQ, math, and reading scores from age 8 to 21 (McLaughlin, Campbell, Pungello, & Skinner, 2007; Ramey & Ramey, 2006). Milwaukee Project children at age 10 demonstrated an average IQ that was 20 points higher than that for the control group (105 vs. 85; Garber & Heber, 1981). BEEP children were observed in the spring of second grade, 3 years after the program ended, with significant differences favoring the most intensive BEEP intervention children found for reading and teacher-reported “learning skills” (Pierson et al., 1984).

In summary, these evaluations provide evidence that early childhood services, especially those that are educational, can influence child develop- ment. However, these evaluations tell us little about how timing and duration affect children’s development and parenting either directly or through mediation effects. The infant and toddler programs may affect emotional as well as cognitive development (especially if the program includes a center- based component) and may influence parenting (especially if the program includes a home-based or parent focus). Programs for 3- to 5-year olds (that tend to be center-based) seem to have their strongest impacts on school- related cognitive skills. Few evaluations of programs starting in the first year of

14

life and continuing until kindergarten have been conducted. One—the Abecedarian Project with a strong center-based focus—had strong, lasting child development impacts whereas another—CCDP with no center-based focus—had few impacts, perhaps because its focus was on family support rather than child development. We speculate that an optimal package of services might include parent and child services in order to enhance child social-emotional, language and cognitive outcomes and parenting in the infant and toddler years, followed by formal care and education that is child focused in the preschool years. However, except for some preliminary findings pertaining to school readiness skills (e.g., preliteracy or math skills), we know little about how broader domains (e.g., language, broad cognitive development, or social-emotional development) might be differentially affected by program timing, duration, or two-generation versus child-only services, and we know little about how these effects might differ across subgroups determined by family or program characteristics.

EFFECTS OF EARLY CHILDHOOD PROGRAMS ON CHILD OUTCOMES WITHIN SUBGROUPS

It is somewhat challenging to determine whether programs are differentially effective for subgroups, which is a question we would like to address ultimately. In this monograph, three types of subgroups are examined, based on program model of EHS service delivery, family race/ ethnicity, and family level of demographic risk.

Because infant and toddler early childhood programs are sometimes home-based, sometimes center-based, and sometimes a combination of the two, we cluster sites by the type of model employed. Effects may to be larger and broader among programs offering a combination of home visiting and center-based services (ACF, 2002a; Gomby, 2005), although direct tests of this premise do not exist as programs were not randomly assigned to program approach. For primarily home-based services, positive child outcomes have been detected in areas related to health and safety (Johnson et al., 1993; Kitzman et al., 1997) and, to a lesser extent children’s emotional functioning (ACF, 2002a; Jacobson & Frye, 1991), whereas center-based programs tend more frequently to report cognitive outcomes for children (e.g., Field, Widmayer, Stringer, & Ignatoff, 1980; Ramey, Bryant, Sparling, & Wasik, 1985). Home-based programs often report positive effects on parents (Howard & Brooks-Gunn, 2009).

We are interested in whether early childhood program opportunities and effects vary for low-income children depending on race/ethnicity. For example, many early childhood intervention studies have focused on at-risk African American children (Olds et al., 2004; Ramey & Ramey, 2006; Reynolds

15

BACKGROUND LITERATURE REVIEW

& Temple, 2006; Schweinhart, 2006). With some exceptions (Gormley et al., 2005), fewer studies have demonstrated effects for Hispanic children. The IHDP (Brooks-Gunn et al., 1993) reported that children of African American mothers were more influenced by the intervention than children of White mothers at 2 years, possibly, as the authors note, because the former were less educated and more poor.

Programs may differentially affect children and families with different levels of risk. Previous evaluations have examined this premise. For example, IHDP had the greatest impact on children’s cognitive development in families where mothers had a high school education or less (Brooks-Gunn et al., 1992; Liaw & Brooks-Gunn, 1993). Using a cumulative risk index rather than just maternal education, effect sizes were largest for children whose families had a moderate number of risks (Liaw & Brooks-Gunn, 1994). The NFP often reports the most positive benefits for mothers with psychological and emotional risk factors (Olds et al., 1994). Most intervention programs are targeted at children at risk but there has been little standardization of the variables used to define risk factors. In addition, intervention programs serve different populations, both in terms of central tendency and distribution. Moreover, who is at greatest risk in one cohort may differ from who is at greatest risk in another cohort, even under the same definition. Thus, it is difficult to compare whether the “highest risk” in one study is comparable to “highest risk” in other studies. To distinguish EHS families with different levels of risk, we counted up to five demographic risk factors that families had when they enrolled: (1) being a single parent; (2) receiving public assistance; (3) being neither employed nor in school or job training; (4) being a teenage parent; and (5) lacking a high school diploma or GED. To form groups of reasonable size, families were divided into three subgroups based on the number of risk factors they had when they enrolled: (1) lower risk families who had zero, one, or two risk factors; (2) moderate-risk families who had three risk factors; and (3) highest risk families who had four or five risk factors. Because the current study quantifies levels of risk, the findings reported here should be helpful in beginning to clarify who benefits from what combinations of services.

RESEARCH QUESTIONS AND HYPOTHESES

Despite over 40 years of research on effects of early childhood programs on children’s development, surprisingly little is known about the timing of intervention mechanisms by which programs affect development. We offered theoretical perspectives earlier on how early and later experiences might influence children’s development. These perspectives lead to the research questions this monograph addresses in Chapters III–VI.

16

1. What were the impacts of EHS on children and parents when the children were 2 and 3 years of age (age 3 being the end of the program) and 2 years after the end of the program when the children were age 5 (Chapter III)?

We expected that impacts would be seen across a range of outcomes when the children were 2 and 3 years of age. Sustained impacts were expected at age 5, although effect sizes would be smaller than at ages 2 and 3, given that effects in previous programs diminished several years after the intervention. Such findings would be consistent with early experiences influencing development, since the intervention experience, occurring in the first years of life, would influence development later. We expect such effects, if found, to be most pronounced for social and emotional outcomes. Consistent with the hypotheses that pertain to intervention being particularly effective if offered during specific periods of development, and that programs focused on improving mother–child relationships, we expected that parental impacts would be seen at age 2 on parenting measures that would mediate impacts of social and emotional functioning in the children later on. We also expected some impacts on parents to be sustained to age 5 as well (Howard & Brooks- Gunn, 2009). Moreover, we expected that age 3 EHS impacts on children’s language, cognitive, social, and emotional development would mediate age 5 impacts in similar domains. Furthermore, based on the expectation that EHS would provide transition experiences for children (US DHHS, 1994), we hypothesized that EHS would increase the probability that children would participate in formal programs at ages 3 and 4. We further hypothesized that EHS would increase the probability that children would enter HS.

2. What were the impacts of EHS on children and parents within prespecified subgroups at ages 2, 3, and 5 (that is during, at the end of, and following the intervention) (Chapter IV)?

It is important to know whether effects of EHS vary depending on family and child characteristics. Differential treatment effects are examined for the three racial/ethnic groups and for three groups of families defined by number of risk factors. Some (but not all) evidence suggests that African American children, and perhaps Hispanic children, show larger benefits of early intervention than White children, although these links may be due, in part, to the fact that even within low-income families, the former two groups are less educated and poorer than the later group (Bassok, 2010; Brooks- Gunn, et al., 1993).

3. What were the impacts of EHS across ages 2, 3, and 5 in the three clusters of programs—those that offered home-based, center-based, or a combination of the two? (Chapter V).

We expected that children and families in mixed-approach programs would experience more positive age 3 outcomes than their counterparts, given that they would have had the advantages of participating in programs with the capacity to provide either or both home-based and center-based services to

17

BACKGROUND LITERATURE REVIEW

individual families in a flexible way designed to meet their specific goals and needs. We expected more child impacts for center-based programs and more parent-related impacts among participants in home-based programs.

4. What were the effects of out-of-home, formal preschool program participation during the age 3–5 period and how did effects of children’s experiences across the age 0–5 period accumulate (Chapter VI)?

This question asks whether some conditions post-EHS enhance or detract from outcomes at age 5. Controlling for EHS use during the children’s years from birth to age 3, we hypothesized that effects for preschool services would be similar to what has been found in other studies in which school achievement-related outcomes tend to be associated with formal care and education experiences, if the care settings are of good quality (Gormley et al., 2005; Magnuson & Waldfogel, 2005; NICHD ECCRN, 2005a). We also expected that emotional outcomes would be related to program experience in a negative way, consistent with the previous nonexperimental analyses (Magnussun, Ruhm, & Waldfogel, 2007; NICHD ECCERN, 2005a).

We hypothesized that children who had both EHS (infant and toddler) and preschool formal programs (ages 3–5) would fare the best at age 5, consistent with the original theory of change for EHS that emphasized early gains would be sustained and augmented by assisting EHS families find preschool programs (US DHHS, 1994). As an overarching hypothesis we hypothesized that experiences are incremental and augmented (NICHD ECCRN, 2001), whereby later experiences during the preschool years would build on early EHS impacts. At the same time, EHS impacts might act as a buffer such that if preschool program attendance was associated with aggressive behavior as others have found, EHS might offset this association (NICHD ECCRN, 2003a; Magnusson et al., 2007). For cognitive, school achievement-related outcomes, we predicted, as has been found in other studies (Gormley et al., 2005; Magnusson et al., 2007; NICHD ECCRN, 2000), that contemporaneous experiences from formal preschool education would be linked to age 5 outcomes as would EHS intervention. As for school- related outcomes, we predicted that EHS language impacts would be strengthened among children who had attended formal programs. We also hypothesized that parents in EHS would engage in more cognitive stimulation and support for children’s development and that preschool attendance would not be an influence, given that preschool education programs, in general, spend less time working with parents than infant and toddler programs do.

NOTES

1. It has been the practice of Head Start to refer to the eligible population of “low-income” families as those with annual incomes below the federal poverty level. In many research

18

contexts, however, “low income” refers to families whose incomes are below 200% of the poverty threshold. However, to be consistent with usage within most Head Start and EHS research, in this monograph we refer to the enrolled families as low income.

2. We have more than one reason for limiting our focus during the years 3–5 to formal early care and education programs. First, drawing on an extensive literature pertaining to the importance of center-based care for children during the prekindergarten years, the Advisory Committee on Services to Families with Infants and Toddlers, the committee appointed by then Secretary Donna Shalala to design what would become EHS, recommended that children receive formal program services following EHS (U.S. Department of Health and Human Services, 1994). Arguably, family support and home visiting services could be recommended for children in families following EHS but this was not the specific recommendation of the committee for all children and it is not the question addressed in this monograph. Second, few children/families in the EHS sample, in either the program or control group, were enrolled in exclusively home visiting services during the years 3–5, although families with children in HS would have received at least two home visits a year, whereas formal care and education experience for children was relatively common as will be shown.

3. Home visiting programs reach many children, most but not all of whom are served during the years 0–3 and many but not all are considered low income or are otherwise at risk. As of the early 2000s, as many as 400,000 children are served annually in home visiting programs, at a cost of approximately $750 million to $1 billion and these numbers will expand under current administration proposals (Gomby, 2005).

4. Children are less likely to enroll in HS or EHS, however, the higher their parents’ level of education attainment, as would be expected given the income requirements and the relationship between income and educational attainment (Iruka & Carver, 2006).

19

BACKGROUND LITERATURE REVIEW

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.

Users should refer to the original published version of the material.