Search Type Title Author/Investigator Search Advanced Search

Table of Contents | Previous | Next

VII. VARIATIONS IN PROGRAM IMPACTS AMONG FAMILIES

Beyond examining impacts overall and in key subgroups of programs, it is important to look at variations in impacts among key subgroups of families. For whom did Early Head Start make a significant difference in outcomes? How did the impacts vary among key subgroups of families? Variations in impacts might provide insights into how the programs influenced children and families and could identify demographic groups that merit special attention in future training and technical assistance.

Our analyses of variations in impacts among family subgroups show that the Early Head Start research programs had significant impacts on some outcomes in every subgroup of families we studied, although the extent and pattern of impacts varied:

• The Early Head Start programs reached all types of families with child development services. They significantly increased service receipt in all subgroups of families we examined.

• Most groups of children benefited in some way by age 2 from participating in Early Head Start. The Early Head Start programs had significant favorable impacts on at least one child outcome in most of the subgroups that we examined.

• The programs had unfavorable impacts on a few child outcomes in key subgroups of families, and these tended to be unfavorable impacts on social-emotional outcomes. The unfavorable impacts on social-emotional outcomes were usually accompanied by unfavorable impacts on negative parenting behaviors.

• While the pattern of program impacts on boys was similar to that of impacts on girls, the impacts were smaller and not statistically significant. The programs had significant positive impacts on parenting among both parents of girls and parents of boys.

• Most parents benefited from participating in Early Head Start in some way related to their roles as parents. Primary caregivers in all subgroups of families that we examined experienced significant improvements in at least one aspect of parenting and family functioning by the time their child was 2 years old. In most subgroups, Early Head Start improved parenting outcomes in more than one domain.

• Some of the less disadvantaged subgroups-families who were not initially receiving welfare, families with more educated parents, two-parent families, and low-risk families-who may have had relatively less need for program services, experienced the fewest impacts, and in some cases experienced unfavorable impacts. Control children and families in these groups tended to experience more favorable outcomes than control children and families in other subgroups.

• Among families with many risk factors, a group that program staff reported was hard to serve, the programs had few significant impacts on children and parents in the short term, and some of the impacts were unfavorable.

• Participating in Early Head Start also helped some parents work toward self-sufficiency. Primary caregivers in many subgroups of families were significantly more likely than their control group counterparts to participate in education or training activities during the first 15 months after they enrolled. However, the programs did not have a significant impact on activities oriented toward helping the primary caregiver become self-sufficient among (1) low-risk families, (2) families who enrolled before their child was born, (3) two-parent families, (4) families in which the primary caregiver lived alone with her children, and (5) families in which the primary caregiver had a high school diploma or GED. For the most part, these were less disadvantaged subgroups in which control families were relatively more likely to participate in self-sufficiency activities.

In the following sections we detail the variations in Early Head Start impacts on service receipt, self-sufficiency outcomes, and child and parenting outcomes for key subgroups of children and families. For a listing of the subgroups analyzed, their sample sizes, and the percentage of the sample in each subgroup, see Table II.10 in Chapter II. Because this chapter reports the results of a very large number of analyses, the detailed tables are included in Appendix E.VII.

Caution must be used in interpreting the variations in impacts among subgroups of families. The subgroups are defined on the basis of a single family characteristic, yet they may also differ in other characteristics. These other unaccounted-for variations in family characteristics may also influence the variations in impacts. Thus, in our analyses, we focus on patterns of impacts across outcomes and consider the potential role of other differences in characteristics that may have influenced the outcomes being examined.1 We attempted to gain insights into the effects of potential confounding by examining variations in impacts among clusters of families formed on the basis of several characteristics and by examining variations in impacts among families grouped by their number of risk factors. In the discussion that follows, we highlight potentially confounding factors that may help explain the patterns of impacts. Because of the large number of subgroups and outcomes, we focus primarily on patterns of impacts within the subgroups. The chapter concludes with a discussion of the implications of the variations in impacts across family subgroups.

A. VARIATIONS IN IMPACTS BY RACE/ETHNICITY

The impacts of Early Head Start may differ among racial/ethnic subgroups because of cultural differences affecting families' receptiveness to services and, in the case of Hispanic families, language barriers that may interfere with services, especially services and resources to which Early Head Start refers them in the community. The impacts may also differ if some cultural practices or attitudes related to parenting or child development are harder to change than others. Early Head Start programs are expected to provide services that meet families' needs and are given wide latitude for designing services that are culturally appropriate. If they do those things well, we would not expect to see large differences in impacts among racial/ethnic subgroups unless they also differed in other family characteristics or there were important differences between the programs in which they enrolled.

Cultural biases in the child and parenting outcome measures could also contribute to variations in impacts by race/ethnicity. We attempted to minimize these biases by choosing measures that had previously been shown to work well in varied racial and ethnic groups. In addition, as we examined the psychometric properties of the child and family measures, we calculated internal consistency alphas separately for each of the three major racial/ethnic subgroups. For the most part, the measures appeared to be appropriate for all groups of children and families.2

1. Hispanic Families

The pattern of impacts among Hispanic families suggests that intensity may be a very important element of effective services. Although the Early Head Start programs greatly increased receipt of any services by Hispanic families, the impacts on receipt of intensive home-based services were much more modest and often smaller among Hispanic families than among other families. Moreover, the Hispanic families in the research sample were more likely to include older mothers and later-born children, groups for whom some opportunities for intervening early had already passed.

We found much larger impacts on receipt of any services, receipt of home visits and case management, and participation in group parent-child activities among Hispanic families, which reflects the much lower levels of service receipt among Hispanic control families relative to white and African American control families (Appendix Table E.VII.1). However, Hispanic families tended to experience much smaller impacts than white families on receipt of intensive services. The impacts on use of child care were somewhat larger among Hispanic families, again because of the lower levels of use of center-based child care among Hispanic control families, and, in the case of hours of center-based child care, higher hours per week in center-based care among Hispanic families in the program group.

The Early Head Start programs had significant impacts on participation in education or job training activities among Hispanic families, but did not have any consistent significant impacts on their employment or welfare receipt (Appendix Table E.VII.2). The large impacts on participation in education or job training among Hispanic primary caregivers reflect the much lower levels of participation in these activities by Hispanic control families compared with control families in the other groups.

The Early Head Start programs had very few impacts on child outcomes when Hispanic children were 2 years old. Only one impact on the development of Hispanic children was statistically significant-Early Head Start reduced the proportion of children who scored below 100 on the Bayley Mental Development Index (MDI) (Appendix Table E.VII.3).

The Early Head Start programs increased stimulation of language and learning and improved knowledge of child development among Hispanic parents. Among these families, Early Head Start significantly increased scores on the Home Observation for Measurement of the Environment (HOME) cognitive, language, and literacy support subscale and increased reading to children at bedtime. Early Head Start also increased the primary caregiver's knowledge of child development.

The general lack of impacts on child development and parenting among Hispanic families, despite the relatively large impacts on service use, may reflect several factors:

• Hispanic families were more likely to be in Early Head Start programs that were later implementers. The early services they received may not have been sufficiently intense or focused on child development to have had extensive impacts on parenting or child development.

• Hispanic families were much more likely not to speak English as their primary language and were much less likely to have completed 12th grade or a General Educational Development (GED) credential.

• Hispanic children in the program group were more likely to have language learning environments that included speakers of two languages (for example, one of the center-based programs that served predominantly Hispanic children operated bilingual classrooms), and children may have been learning both languages more slowly than they would have learned either language in a monolingual environment.

• Hispanic families tended to include older mothers and later-born children. The Early Head Start intervention may not have been early enough in the lives of these families to have had more extensive impacts on parenting and child outcomes.

• Limitations of the data collection in the Spanish language may have affected the impact estimates.

2. African American Families

Early Head Start appears to have been most effective for African American families. Among these families, Early Head Start had a larger number of significant impacts in all key areas-parenting, child cognitive development, and child social- emotional development.

As in the other groups of families, Early Head Start substantially increased service use among African American families (Appendix Table E.VII.1). Early Head Start also increased average hours per week that African American primary caregivers spent in school or job training and reduced their average hours per week in employment during the first 15 months after enrollment (Appendix Table E.VII.2). Early Head Start also reduced welfare receipt during this period among African American families.

Among these families, Early Head Start significantly improved child well-being in all three domains. Early Head Start improved child cognitive development among African American children, increasing children's average Bayley MDI scores and reducing the proportion of children with low scores who might need special services later on (Appendix Table E.VII.3). Early Head Start improved language development, increasing vocabulary production and sentence complexity at age 2. Early Head Start also reduced child aggressive behavior problems and improved child engagement during the parent-child structured play assessment.

The Early Head Start programs improved parenting outcomes among African American families in several areas, including emotional support from parents, stimulation of language and learning, and parent mental health. Among these families, Early Head Start increased parent supportiveness, increased reading to the child, reduced parental distress and parent-child dysfunctional interactions, and reduced depression.

The stronger impacts on child development and parenting among African American families may reflect the following:

• African American families were more likely to be in mixed-approach programs, which may have been better able to match services to families' needs.

• African American families were more likely to include teenage mothers and firstborn children. The earlier intervention with these families may have made the Early Head Start services more effective.

• African American families were more likely to be in school or training when they enrolled. They may have been in a "learning mode" and more receptive to Early Head Start home visit curricula and to Early Head Start efforts to teach them about parenting in center-based settings.

3. White, Non-Hispanic Families

The significant impacts of Early Head Start on white families were exclusively in the area of children's social-emotional development and parenting outcomes likely to influence social-emotional development. Children and parents in white control families tended to receive more favorable scores on the assessments of child development and parenting than African American and Hispanic control families and children, and it may have been more challenging for programs to improve outcomes among white families.

As in the other groups, Early Head Start increased service receipt by white families substantially (Appendix Table E.VII.1). The Early Head Start programs did not significantly influence self-sufficiency outcomes among white families (Appendix Table E.VII.2).

Early Head Start had mixed impacts on children's social-emotional development, but did not significantly affect other areas of children's development. Among white families, Early Head Start reduced child negativity toward the parent and increased child engagement in the parent-child structured play assessment, but it also reduced emotional regulation as assessed with the Bayley Behavior Rating Scale (BRS) (Appendix Table E.VII.3).

Consistent with the favorable impacts on children's social-emotional development, Early Head Start significantly reduced negative parenting behaviors, including parent intrusiveness and spanking (Appendix Table E.VII.3). Early Head Start also increased the extent to which white parents indicated that they would use only mild discipline strategies in response to conflict with their child and reduced the severity of discipline that they would use, on average. Early Head Start reduced family conflict in white families. Finally, Early Head Start also increased the extent to which white parents reported reading to their child daily.

This pattern of impacts may reflect the following:

• White families were more likely to be in home-based programs and less likely to be in center-based programs.
  
  
• White families were more likely to be in programs that were implemented early.
  
  
• White families were less likely to be in states with early work requirements.
  
  
• White families were more likely to be low-risk families (have no risk factors or only one risk factor among the five we considered: (1) being a teenage mother, (2) being a single parent, (3) receiving welfare, (4) having low educational attainment (less that 12th grade or GED), and (5) being neither employed nor in school).

B. CHILD'S AGE AT ENROLLMENT

The age of children at enrollment and the duration of Early Head Start services they could have received by age 2 are closely linked. Some children were not yet born when their families enrolled in Early Head Start, and those who were born ranged in age from 0 to 12 months. Thus, by age 2, children could have received between 12 and 24 or more months of Early Head Start services. Because younger children could have received more services and because earlier intervention with parents may be more effective, we expected children who were enrolled at earlier ages to experience larger program impacts.

1. Children Who Were Not Yet Born at Enrollment

The positive program impacts on service use were consistently largest among families who enrolled before their child was born (Appendix Table E.VII.4). However, Early Head Start did not significantly influence self-sufficiency activities or welfare receipt during the first 15 months among families who enrolled while pregnant (Appendix Table E.VII.5).

Earlier intervention and a longer period of exposure to services may be needed to influence child language outcomes at age 2. Consistent with this hypothesis, Early Head Start had significant impacts on all key language development outcomes among children in families who enrolled before their child was born, but did not significantly improve language outcomes among children in families who enrolled after their child was born (Appendix Table E.VII.6). In the case of sentence complexity, the impacts were large. Impacts on child development outcomes in the other domains were generally not significant in any of the subgroups by child's age at enrollment.

In families who enrolled before their child was born, Early Head Start also had some significant impacts in most areas of parenting. Among these families, Early Head Start (1) increased parent supportiveness, (2) increased some aspects of stimulation of language and learning (improved the support for cognitive, language, and literacy at home and increased the proportion of families who established a bedtime routine for the child), (3) reduced parent detachment, and (4) increased the extent that parents reported that they would use only mild discipline strategies (and reduced the average severity of discipline strategies they would use) in situations of conflict with their child (Appendix Table E.VII.6).

This pattern of impacts may reflect the following:

• These families had the greatest exposure to program services before the children were assessed at age 2.
  
  
• Families who enrolled while expecting a child were more likely to be in home-based and mixed-approach programs.
  
  
• Families who enrolled while expecting a child were more likely to be in programs that were implemented early.
  
  
• Families who enrolled while expecting a child were more likely to be receiving welfare when they enrolled.
  
  
• Families who enrolled while expecting a child were more likely to be high-risk families.

2. Children Who Were Born at Enrollment

In contrast to the impacts on families who enrolled before their child was born, Early Head Start had fewer significant impacts on child outcomes among the families who enrolled with infants. Early Head Start significantly improved several parenting outcomes among these families, however.

a. Families with Infants Who Were 0 to 4 Months Old at Enrollment

Early Head Start substantially increased service receipt among families with infants at the time of enrollment, as they were for other subgroups. Program impacts on the use of center-based child care were somewhat larger among families who enrolled after their child was born (Appendix Table E.VII.4). Early Head Start also increased participation in education and training programs, increased average hours per week in education or job training activities, and increased the extent of welfare receipt among families with younger infants.

Early Head Start had few significant impacts on child outcomes among families with young infants at enrollment. However, Early Head Start significantly decreased the proportion of families with young infants whose child scored below 85 on the Bayley MDI, which indicates that the children may be at lower risk for developmental delay and less likely to need special services later on (Appendix Table E.VII.6).

Early Head Start also significantly improved at least one outcome in most domains of parenting among families with young infants. The Early Head Start programs increased parents' emotional responsivity as assessed using the HOME; increased support for cognitive development, language and literacy at home (HOME); and increased the verbal/social skills of primary caregivers (HOME); but they decreased nonpunitive interactions with children (HOME) (Appendix Table E.VII.6). Early Head Start also increased knowledge of child development among parents with young infants at enrollment and reduced their parenting distress.

b. Families with Infants Who Were 5 to 12 Months Old at Enrollment

Impacts on service use were also large and positive among families with older infants at enrollment, but tended to be smaller than the impacts for families with younger infants and unborn children (Appendix Table E.VII.4). Early Head Start increased high school attendance and receipt of high school diplomas among these parents (Appendix Table E.VII.5). Among families with older infants, the programs also increased receipt of Temporary Assistance for Needy Families (TANF) cash assistance in the first two quarters after random assignment.

Early Head Start did not have a consistent pattern of impacts on child outcomes among families with older infants at enrollment. However, among such families who received services for a shorter time, Early Head Start significantly increased children's vocabulary production and reduced child aggression.

Early Head Start also had significant impacts on parenting outcomes in several domains. The programs improved stimulation of language and learning among families with older infants, and the pattern of impacts in this domain is strongest for these families (Early Head Start improved support for cognitive development, language, and literacy in the home; increased reading frequency, reading at bedtime, and reading daily; and increased parent-child activities to stimulate cognitive and language development). Early Head Start also reduced conflict in families with older infants at enrollment.

This pattern of impacts may reflect the following:

• These families had less exposure to program services before their children were assessed at age 2.
  
  
• Families with older infants at enrollment were more likely to be in center-based programs.
  
  
• Families with older infants at enrollment were more likely to be low- or moderate-risk families.

C. AGE OF MOTHER AT CHILD'S BIRTH

Early Head Start may have had differential effects based on the mother's age when the focus child was born. Not only are teenage mothers likely to be less emotionally mature than older mothers, they also may be struggling with their own developmental needs and less receptive to some services directed toward their child's development. Program staff also regarded teenage mothers as harder to serve. If the Early Head Start programs had significant positive impacts on teenage parents and their children, they may provide a useful model to others trying to serve this at-risk group in other places.

1. Teenage Mothers

The Early Head Start impacts on families headed by teenage mothers were concentrated in the social-emotional domain and in parenting factors likely to affect children's social-emotional development. The greater immaturity of teenage mothers, on average, may provide more opportunities for Early Head Start programs to make differences in their parenting behavior and interactions with their children-differences large enough to improve their children's social-emotional development.

As in other subgroups, Early Head Start substantially increased service use by teenage parents. However, the impacts on service use by teenage parents were smaller than those for older mothers on nearly all measures (Appendix Table E.VII.7). Early Head Start increased participation in education or job training (high school attendance, specifically) and average hours per week in education (Appendix Table E.VII.8). Early Head Start also reduced employment and increased receipt of TANF cash assistance during the first part of the follow-up period. Local ethnographic research reported in Box VII.1 provides insights into factors that influence how teenage mothers engage in various program services.

Among teenage mothers, Early Head Start significantly improved child outcomes in two of the three domains (Appendix Table E.VII.9). The programs had significant impacts on language development (increased vocabulary production at age 2) and social-emotional development (reduced child aggressive behavior problem and increased child engagement in the parent-child structured play assessment).

Early Head Start also had some significant impacts in all the areas of parenting that we examined. Among teenage parents, Early Head Start increased parent supportiveness, increased reading at bedtime, reduced parent detachment, increased knowledge of child development, and reduced parenting distress and parent-child dysfunctional interactions. One significant impact was unfavorable: Early Head Start decreased the primary caregiver's verbal and social skills as assessed using the HOME (it is possible that teenage mothers in the program group were more self-conscious or felt like they were being tested on what they learned in the program and talked less during the assessments).

 

BOX VII.1 FAMILY GOALS AND ENGAGEMENT WITH THE PROGRAM: PERSPECTIVES OF TWO TEENAGE MOTHERS Rebecca Ryan and Barbara Alexander Pan Harvard Graduate School of Education For three years, researchers from the Harvard Graduate School of Education have been following two teenage mothers, Rachel and Kristen,3 as part of an ethnographic study of Early Head Start research families in Brattleboro, Vermont. The purpose of the study is to examine how factors such as parent-child dynamics, child care, work, welfare and other assistance interact over time in families' lives and how they influence engagement with the program. Understanding what young parents want for themselves and their children, and why, is crucial for understanding program efficacy, because participants' goals and beliefs determine what services they find useful. Rachel and Kristen differ strikingly in their present lives and in their plans for the future. These differences explain, in part, how they value and engage differently in the home-visiting, child care, and adult services the program provides. Two months after her 16th birthday, Rachel gave birth to her daughter Daisy. She and Daisy currently live in an apartment paid for in part by the local Land Trust. Since her daughter's birth, public assistance has been Rachel's main source of income. She works 20 hours a week as part of Vermont's welfare-to-work program. She also takes a full course load at a local community college. While Rachel is at school and work, her daughter Daisy attends full-time child care that is fully subsidized by Early Head Start. Rachel hopes to earn a college degree and secure a good job. She sees attending school full-time, working part-time, placing Daisy in full-time child care, and temporarily remaining on welfare as necessary steps toward self-sufficiency. When asked what she values most about her involvement in Early Head Start, Rachel explains how crucial good-quality, subsidized child care is to her plan. Thus, Rachel uses the Early Head Start services primarily to invest in her skills and training to achieve professional and financial goals. She describes the program as helping her chart a realistic path toward those goals and supporting her emotionally. Kristen had her baby Emily at age 17. Kristen, her husband Jack, and Emily live in a trailer home in Brattleboro. Jack works full-time as a mechanic, and Kristen stays home part-time to care for Emily, who is in child care two days a week provided by Early Head Start. Kristin wants to have time to care for her daughter, both now and in the future. Unlike Rachel, Kristen is not investing time and resources in her own skills now to work toward a future goal; rather, her priorities are how best to meet Emily's immediate needs. When asked what she gains from participating in Early Head Start, Kristen mentions information about child development and healthy ways to care for children. She values the child care because she believes it benefits Emily's development immediately and directly. For Kristen, Early Head Start is valuable because it helps her care for Emily and supports her daughter's development during these first three years. Low-income parents choose both whether to apply for Early Head Start and when and how to use Early Head Start services. These choices are rooted in how they understand their present and future lives and in turn influence the impact that the program can have. Mothers like Kristen and Rachel can help researchers and policymakers understand the perspectives of young mothers in similar situations.

 

The pattern of impacts on teenage mothers and their children may reflect the following:

• Teenage mothers were more likely to be African American and less likely to be Hispanic.
  
  
• Teenage mothers were much more likely to enroll with a firstborn child-Early Head Start intervened earlier in the lives of these mothers.
  
  
• Teenage mothers were much less likely to have completed 12th grade and much more likely to be in school or training when they enrolled.
  
  
• Teenage mothers were more likely to be in high-risk families (have four or five risk factors among the five we considered: (1) being a teenage mother, (2) being a single parent, (3) receiving welfare, (4) having low educational attainment (less than 12th grade or GED), and (5) being neither employed nor in school).

2. Older Mothers

As in other subgroups, program impacts on service use by older mothers were positive and large, and tended to be larger than for teenage mothers (Appendix Table E.VII.7). Early Head Start also increased older mothers' participation in school or job training (vocational training, specifically) and their average hours/week in education or job training during the first 15 months after enrollment.

Early Head Start had a clear pattern of significant impacts on the development of children of older mothers. Among these children, Early Head Start significantly increased average Bayley MDI scores and reduced the proportion of children with scores below 100. The programs also significantly improved several language outcomes among children of older mothers and increased children's sustained attention with objects during the parent-child structured play assessment. The greater maturity of older mothers may make them more receptive to services that teach them about their children's development and more likely to respond in ways that promote their children's cognitive and language development.

The Early Head Start impacts on parenting among older mothers were concentrated in the areas of emotional support and stimulation of language and learning. Early Head Start improved parents' emotional responsivity as assessed using the HOME and increased parent supportiveness as assessed in the parent-child structured play assessment. Early Head Start also increased support for cognitive development, language, and literacy at home (HOME); increased reading at bedtime and establishing bedtime routines; and improved parents' verbal/social skills (HOME). In addition, Early Head Start significantly increased older mothers' knowledge of child development.

This pattern of impacts may reflect the following:

• At enrollment, older mothers were more likely to have children who were not firstborn children.
  
  
• Older mothers were less likely to be in school or training when they enrolled and were more likely to have completed 12th grade or higher.
  
  
• Older mothers were more likely than teenage mothers to live with a spouse or to live alone.
  
  
• Older mothers were more likely to be low- or moderate-risk families.

D. CHILD'S BIRTH ORDER

Early Head Start may have differential effects depending on the birth order of the child. Intervening earlier in parents' experiences as parents may be more effective than intervening after they have already had several children and established patterns that may be difficult to change.

1. Families Who Enrolled with Their First Child

Early Head Start had a strong pattern of positive impacts on firstborn children and their parents, which suggests that early intervention, when parents first become parents, may be especially effective.

As for other subgroups, Early Head Start had large positive impacts on service use by families who enrolled with their first child. Among families who enrolled with their first child, the program impact on use of center-based child care was somewhat larger, while impacts on other service use outcomes tended to be somewhat smaller (Appendix Table E.VII.10). Early Head Start consistently increased participation in school or job training (specifically high school attendance) and reduced average hours per week in employment (Appendix Table E.VII.11). The programs also increased receipt of TANF cash assistance in the early part of the follow-up period for these families.

Early Head Start enhanced child cognitive outcomes and improved several other outcomes among firstborn children. Early Head Start increased average Bayley MDI scores and reduced the proportion of firstborn children scoring below 85 (which indicates that they may be at less risk for developmental delay and less likely to need special services later on) (Appendix Table E.VII.12). In addition, Early Head Start significantly increased firstborn children's sentence complexity and reduced their aggressive behavior problems at age 2.

Early Head Start significantly improved some outcomes in several areas of parenting that we examined. The programs improved stimulation of language and learning (increased support for language and literacy at home and increased establishment of regular bedtimes and reading at bedtime) among parents of firstborn children. They also reduced spanking and family conflict. Finally, Early Head Start increased knowledge of child development and increased the extent to which parents who enrolled with their first children reported that they would use only mild discipline strategies and reduced the severity of discipline that they reported they would use in situations of conflict with their children.

This pattern of findings may reflect the following:

• Firstborn children were somewhat more likely to be in center-based programs.
  
  
• Firstborn children were less likely to be in programs that were implemented early.
  
  
• Families enrolling with their first child were much more likely to be teenage mothers.
  
  
• Families enrolling with their first child were more likely to be high-risk families.

2. Families Who Enrolled with Later-Born Children

As for other subgroups, Early Head Start significantly increased service use by a large amount among families who enrolled with later-born children. Program impacts on service use tended to be larger among these families, although the impact on use of center-based care was somewhat smaller among these families (Appendix Table E.VII.10). Among families who enrolled with a later-born child, Early Head Start increased employment and overall participation in activities designed to promote self-sufficiency (education/job training or employment) (Appendix Table E.VII.11).

The pattern of program impacts on child outcomes among later-born children differed from that of firstborn children. Early Head Start improved cognitive outcomes among later-born children but did not have significant impacts on any language or social-emotional outcomes among these children. Early Head Start increased average Bayley MDI scores and reduced the proportion of later-born children who scored below 85 (Appendix Table E.VII.12). Importantly, the reduction in the proportion of children with low scores on the Bayley MDI was larger among later-born than firstborn children.

Early Head Start significantly improved parenting in several areas among parents who enrolled with later-born children. Early Head Start increased their emotional responsivity (HOME), increased their reading at bedtime, and improved their verbal/social skills (HOME). Early Head Start also decreased detachment among parents of later-born children, but increased their negative regard. Finally, Early Head Start improved knowledge of child development.

This pattern of impacts on later-born children and their parents may reflect the following:

• Later-born children were more likely to have older mothers.
  
  
• Families who enrolled with later-born children were more likely to be Hispanic or white.
  
  
• Families who enrolled with later-born children were more likely to be low- or moderate-risk families.

E. CHILD GENDER

We did not expect program impacts to vary according to the child's gender. However, it is possible that differences in the developmental trajectories of girls and boys might lead to different responses to Early Head Start services.

1. Families with Boys

As in other subgroups of families, Early Head Start consistently had large positive impacts on families' service receipt. The program impacts on service use by families with boys were similar in magnitude to those for families with girls (Appendix Table E.VII.13). Early Head Start also increased participation in education and job training activities significantly among primary caregivers with boys (Appendix Table E.VII.14).

The Early Head Start impacts on the cognitive, language, and social-emotional development of boys were favorable, but they were small and did not reach statistical significance. The programs improved parenting outcomes in several areas among parents of boys (Appendix Table E.VII.15). Early Head Start increased reading to boys, increased knowledge of child development among parents of boys, and promoted less spanking. Early Head Start also reduced parent-child dysfunctional interactions in families with boys.

2. Families with Girls

Even though the characteristics of girls' and boys' families were similar, the programs they attended were similar, and the impacts on their service receipt were similar (see Appendix Table E.VII.13), Early Head Start had a clearer pattern of significant positive impacts on the development of girls. Early Head Start increased average Bayley MDI scores among girls and reduced the proportion of girls who received low scores. The programs also improved most of the language outcomes and two key social-emotional outcomes among girls (Appendix Table E.VII.15).

Among families with girls, Early Head Start improved parenting in the areas of emotional support and stimulation of language and learning. Early Head Start increased parent supportiveness and increased support for language and literacy learning in the home (HOME), increased reading to the child, and increased parent-child activities to stimulate cognitive and language development. Early Head Start also increased knowledge of child development among parents of girls.

As was the case among parents of boys, Early Head Start increased parents' participation in education or job training, although the pattern of impacts is weaker (Appendix Table E.VII.14). However, the programs consistently increased the extent to which the primary caregivers who enrolled with girls participated in some kind of activity designed to promote self-sufficiency (education/job training or employment) and the average hours per week they spent in these activities during the first 15 months after enrollment. The pattern of impacts among primary caregivers of boys is similar, but the impacts were smaller and not significant. The stronger pattern of impacts on girls and their parents does not appear to be due to differences in other measured characteristics of families of boys and girls or differences in the programs in which they enrolled. The distributions of children by gender were very similar in the key subgroups of programs and in the key subgroups of families. Instead, it appears that the boys were less responsive to the improvements in parenting that the programs stimulated in their parents.

F. WELFARE STATUS AT ENROLLMENT

The impacts of Early Head Start on families who were receiving TANF cash assistance when they enrolled may be different from those on families who were not receiving TANF cash assistance. In many cases, families receiving cash assistance faced welfare reform work requirements sooner after they enrolled, while other families may have faced the threat of work requirements if they were to obtain TANF cash assistance but were not immediately subject to them. Families who were not receiving cash assistance may have been working or in school and had other sources of financial support and different needs.

1. Families Receiving TANF Cash Assistance When They Enrolled

The pattern of significant program impacts on families who were receiving TANF cash assistance when they enrolled (welfare families) indicates that the Early Head Start programs may have provided a safety net for children in these families. Early Head Start appears to have helped welfare families maintain a focus on parenting and their child's development while they coped with the new requirements of welfare reform.

As with all subgroups, Early Head Start had large positive impacts on service use among welfare families. However, the Early Head Start impacts on service use tended to be somewhat smaller among welfare families than among those who were not initially receiving cash assistance (nonwelfare families) (Appendix Table E.VII.16). The programs did not significantly increase the extent to which families used child care or center-based child care, although they did significantly increase the average hours per week that children in welfare families were in center-based care.

Early Head Start appears to have delayed employment and welfare exits among some welfare families (Appendix Table E.VII.17). This pattern of impacts is consistent with program staff reports that they advised many families against taking the first job they could find and encouraged them to get the education or training they needed to get a job that would pay higher wages.

Early Head Start impacts on children in welfare families were concentrated in the area of language development. In addition, the programs increased children's sustained attention with objects during the parent-child structured play assessment (Appendix Table E.VII.18). Early Head Start did not have a significant impact on these children's cognitive development.

Early Head Start had several significant impacts on parenting in welfare families. The programs increased parent supportiveness and improved stimulation of language and learning (increased support for language and literacy in the home, increased use of responsive praise, increased reading to children, and increased parent-child activities to stimulate cognitive and language development) (Appendix Table E.VII.18). Early Head Start also improved the safety practices of welfare families.

This pattern of impacts may reflect the following:

• Families who initially received TANF cash assistance were less likely to be in center-based programs and more likely to be in home-based programs.
  
  
• Families initially receiving TANF cash assistance were more likely to be in programs that were incomplete implementers overall, but were more likely to be early implementers of family development services (which suggests that programs that served welfare families gave priority to meeting the family development needs of this population).
  
  
• Families initially receiving TANF cash assistance were less likely to be in states with early work requirements.
  
  
• Families initially receiving TANF cash assistance were more likely to be African American and less likely to be white or Hispanic.
  
  
• Families initially receiving TANF cash assistance were much less likely to include a primary caregiver who was employed at the time of enrollment.
  
  
• Mothers initially receiving TANF cash assistance were more likely to live alone with their children.
  
  
• Families initially receiving TANF cash assistance were much more likely to be high-risk families.

2. Families Not Receiving TANF Cash Assistance When They Enrolled

Early Head Start had somewhat larger positive impacts on service use (especially use of center-based child care) among nonwelfare families (Appendix Table E.VII.16). Consistent with the somewhat larger impacts on use of child care and center-based care, Early Head Start also increased participation in education or job training (high school attendance, specifically) and increased average hours per week in education or training during the first 15 months after enrollment among nonwelfare families (Appendix Table E.VII.17). These impacts are reflected in several significant increases in participation in activities designed to promote self-sufficiency (employment or education/training).

The pattern of Early Head Start impacts on children in nonwelfare families differed from the pattern in welfare families. Early Head Start increased average Bayley MDI scores and reduced the proportion of children in nonwelfare families who received low scores (Appendix Table E.VII.18). Early Head Start had few significant impacts on social-emotional outcomes in this subgroup, but did decrease Bayley emotional regulation scores among children in nonwelfare families. The programs did not have any significant impacts on language outcomes among children in nonwelfare families.

Early Head Start had mixed impacts on parenting outcomes in nonwelfare families. The programs increased reading to children at bedtime and increased knowledge of child development among parents in these families. However, Early Head Start also increased punitive interactions (HOME) and increased negative regard among parents in nonwelfare families.

It appears that although Early Head Start significantly improved the cognitive development of children of mothers in nonwelfare families and improved some aspects of parenting, the programs may also have created greater pressures that increased parents' punitive interactions and negative regard relative to the nonwelfare mothers in the control group. The programs' efforts to help families work toward self-sufficiency may have created pressures that many control families, who did not face immediate work requirements and were less likely to be receiving services that would create similar pressures, did not experience. The unfavorable impacts on parenting may in turn have had a negative effect on their children's social-emotional development.

The pattern of impacts on nonwelfare parents and their children may reflect the following:

• Families initially not receiving TANF cash assistance were more likely to be in center-based programs.
  
  
• Families initially not receiving TANF cash assistance were more likely to be in programs that were fully implemented early.
  
  
• Families initially not receiving TANF cash assistance were more likely to be in states with early work requirements.
  
  
• Families initially not receiving TANF cash assistance were more likely to be white or Hispanic and less likely to be African American.
  
  
• Families initially not receiving TANF cash assistance were much more likely to include a primary caregiver who was employed at enrollment.
  
  
• Mothers initially not receiving TANF cash assistance were more likely to live with a spouse or with other adults.
  
  
• Families initially not receiving TANF cash assistance were much less likely to be high-risk families.

G. PRIMARY OCCUPATION

Early Head Start impacts may vary according to the initial activities of primary caregivers. Employed parents are likely to have different needs and different availability for program services and activities than parents who are in school or training or parents who are neither employed nor in school or training. Employed parents are more likely to need child care and help in balancing the demands of parenting and working. Parents who are in school or training are also likely to need child care, but not necessarily full-time, and may need help balancing the demands of school or training with parenting. Parents who are neither in school or training nor employed may need much more intensive help working toward self-sufficiency goals or, if they are married or living with a partner, may need mostly parenting support. These families may have had more time for program services and activities. Early Head Start programs may have been more effective in meeting some demands and needs than others.

1. Families in Which the Primary Caregiver Was Initially Employed

Early Head Start increased service receipt substantially among families in which the primary caregiver was initially employed (employed families) (Appendix Table E.VII.19). Among those initially employed, Early Head Start also increased employment in the first two quarters after enrollment, which suggests that Early Head Start helped these families retain their jobs longer or find new jobs quickly (Appendix Table E.VII.20). In addition, the programs increased high school attendance later in the follow-up period and increased average hours per week in education or training activities among employed families.

Early Head Start improved outcomes in two areas among children of employed families. The programs significantly increased the average Bayley MDI and improved two social-emotional outcomes (reduced child negativity toward the parent and increased child engagement in the parent-child structured play assessment) (Appendix Table E.VII.21).

Among primary caregivers who were initially working, Early Head Start also significantly improved parenting in several areas. The programs improved stimulation of language and learning (improved support for language and literacy in the home, increased daily reading to children, increased fathers' reading to children, and increased reading frequency) and reduced negative parenting behaviors (reduced parent detachment, parent intrusiveness, and spanking) (Appendix Table E.VII.21).

This pattern of impacts on employed families may reflect the following:

• Families in which the primary caregiver was initially employed were more likely to be in center-based programs.
  
  
• Families in which the primary caregiver was initially employed were much less likely to be receiving TANF cash assistance.
  
  
• Families in which the primary caregiver was initially employed were more likely to be white or Hispanic.
  
  
• Families in which the primary caregiver was initially employed were more likely to be low-risk families.

2. Families in Which the Primary Caregiver Was Initially in School or Training

As with all groups, Early Head Start had large positive impacts on service use among families in which the primary caregiver was initially in school or training. The impact on use of center-based care was largest in this group of families (Appendix Table E.VII.19). In addition, Early Head Start helped these parents stay in school longer, increased their average hours per week in education or training during the follow-up period, and reduced their receipt of public assistance (Appendix Table E.VII.20).

Early Head Start did not significantly affect the cognitive or language development of children of mothers who were initially in school or training, but it significantly reduced aggressive behavior problems among the children (Appendix Table E.VII.21). The reduction in child aggression was relatively large in terms of the effect size.

Among families in which the primary caregiver was in school or training, Early Head Start had significant impacts in two areas of parenting. Early Head Start influenced discipline strategies (increased the extent to which parents reported that they would use only mild discipline strategies and would respond less severely to conflict with their children), and Early Head Start reduced parent-child dysfunctional interactions.

This pattern of impacts may reflect the following:

• Families in which the primary caregiver was initially in school or training were slightly more likely to be in center-based programs.
  
  
• Families in which the primary caregiver was initially in school or training were more likely to be in programs that were incomplete implementers.
  
  
• Primary caregivers who were initially in school or training were more likely to be teenage mothers.
  
  
• Families in which the primary caregiver was initially in school or training were more likely to be moderate- or high-risk families.

3. Families in Which the Primary Caregiver Was Initially Neither Working Nor in School or Training

As with all groups, Early Head Start significantly increased service use by a substantial amount among families in which the primary caregiver was initially neither employed nor in school or job training. The program impacts on use of any services and use of child care tended to be somewhat larger among families in this subgroup, in part because levels of service use were low among control group families in this subgroup (Appendix Table E.VII.19). Early Head Start also had larger impacts on transportation and housing assistance in this subgroup. Early Head Start increased participation in school or job training late in the 15-month follow-up period in this group of families (Appendix Table E.VII.20). The programs also increased receipt of public assistance among families that were not initially employed or in school or training.

Few program impacts on the children of parents who were neither employed nor attending school or training when they enrolled were significant. However, Early Head Start significantly increased sentence complexity at age 2, and fewer children received low scores on the sentence complexity measure (Appendix Table E.VII.21).

Among primary caregivers who were neither employed nor in school or training when they enrolled, Early Head Start improved parenting in a number of areas. The programs increased parent emotional support (increased emotional responsivity and parenting supportiveness), improved stimulation of language and learning (improved support of language and literacy at home, increased reading to children at bedtime, and increased primary caregivers' verbal/social skills), and increased knowledge of child development.

This pattern of impacts may reflect the following:

• Families in which the primary caregiver was initially neither employed nor in school or training were slightly more likely to be in home-based or mixed-approach programs.
  
  
• Families in which the primary caregiver was initially neither employed nor in school or training were slightly more likely to be in programs that never fully implemented child development services but fully implemented family development services early.
  
  
• Primary caregivers who were initially neither employed nor in school or training were more likely to be white or Hispanic.
  
  
• Families in which the primary caregiver was initially neither employed nor in school or training were more likely to be high-risk families.

H. HIGHEST GRADE COMPLETED

The initial educational attainment of primary caregivers may also lead to different needs and varying impacts of Early Head Start. Those who had not completed high school or a GED may have been younger and less mature, may have had learning disabilities, or may have had other problems that interfered with their education and could have interfered with participation in Early Head Start. Those who had a high school diploma or GED had completed an important step toward self- sufficiency, and they may have been slightly older and better able to participate in and benefit from Early Head Start services. Those with more education may have been more mature and more motivated to improve their lives and those of their children, and they may have been most able to benefit from Early Head Start services.

1. Families in Which the Primary Caregiver Initially Had Not Completed 12th Grade or a GED

As with more educated primary caregivers, Early Head Start had large positive impacts on service use among families in which the primary caregiver had not completed 12th grade (Appendix Table E.VII.22). The program impact on use of center-based child care was especially large in this subgroup. Early Head Start also increased participation in school or training (high school attendance, specifically) in this group of primary caregivers. Among those who initially had not completed 12th grade, Early Head Start also reduced average hours per week in employment (Appendix Table E.VII.23).

Early Head Start did not significantly improve cognitive or language development among children of parents who had not completed 12th grade or the equivalent (Appendix Table E.VII.24). However, Early Head Start significantly reduced aggression among children in these families.

Among primary caregivers who had initially not completed high school or a GED, most Early Head Start impacts on parenting were concentrated in two areas: emotional support (the programs increased emotional responsivity and parent supportiveness) and parent physical and mental health (Early Head Start decreased parental distress, decreased parent-child dysfunctional interactions, and reduced family conflict). Early Head Start also increased reading to children at bedtime and the extent to which parents reported that they would use only mild discipline in conflict situations with their child.

This pattern of impacts among less educated primary caregivers may reflect the following:

• Families in which the primary caregiver had not initially completed high school or a GED were more likely to be in programs that were implemented later.
  
  
• Families in which the primary caregiver had not initially completed high school or a GED were less likely to be in states with early work requirements.
  
  
• Primary caregivers who had not initially completed high school or a GED were much more likely to be teenage parents.
  
  
• Primary caregivers who had initially not completed high school or a GED were less likely to be white and more likely to be Hispanic.
  
  
• Families in which the primary caregiver had not initially completed high school or a GED were more likely to be high-risk families.

2. Families in Which the Primary Caregiver Had Initially Completed 12th grade or a GED

As with all groups, Early Head Start substantially increased service use among families in which the primary caregiver had completed high school or obtained a GED (Appendix Table VII.22). Early Head Start did not significantly affect the activities designed to promote self-sufficiency in primary caregivers in this subgroup (Appendix Table E.VII.23).

Among families in which the primary caregiver had completed a high school diploma or obtained a GED, Early Head Start significantly improved several child outcomes. The programs had a clear pattern of favorable impacts on the social-emotional development of children in this group at age 2 (Appendix Table E.VII.24). Early Head Start increased children's sustained attention with objects, reduced child negativity toward parents, and increased child engagement during the parent-child structured play assessment. Early Head Start also significantly increased the children's vocabulary production when they were 2. Among families in which the primary caregiver had completed high school or a GED, Early Head Start improved parenting mainly in the area of stimulation of language and learning. The programs increased support for language and literacy in the home, increased the extent to which fathers read to their children, and increased reading frequency (Appendix Table E.VII.24). Early Head Start also significantly reduced one of the negative parenting behaviors that were examined (parent intrusiveness).

This pattern of impacts may reflect the following:

• Families in which the primary caregiver had initially completed high school or a GED were slightly more likely to be in programs that were implemented early.
  
  
• Families in which the primary caregiver had initially completed high school or a GED were more likely to be moderate-risk families.

3. Families in Which the Primary Caregiver Had Initially Completed More than 12th Grade

As with all groups, Early Head Start substantially increased service use among families in which the primary caregiver had initially completed more than 12th grade (Appendix Table E.VII.22). Program impacts on child care use and out-of-pocket child care costs tended to be largest in this subgroup. Early Head Start also increased participation in vocational training in this group and increased the average number of hours per week that parents spent in education or training activities during the 15-month follow-up period (Appendix Table E.VII.23).

Early Head Start did not significantly affect any of the child outcomes that were measured at age 2 among families with more educated primary caregivers (Appendix Table E.VII.24). In addition, Early Head Start had few significant impacts on parenting among families with more educated primary caregivers. Early Head Start did, however, significantly improve support for language and literacy at home and substantially increased reading to children at bedtime in these families.

This pattern of impacts may reflect the following:

• Families in which the primary caregiver had more education were slightly more likely to be in programs that were fully implemented early.
  
  
• Families in which the primary caregiver had more education were more likely to be low-risk families.
  
  
• Control families in which the primary caregiver initially had education beyond high school generally had better outcomes than control families in the other groups, and it may have been more difficult for Early Head Start programs to make a difference for this group.

I. LIVING ARRANGEMENTS

The needs of families in different living situations are likely to vary, and Early Head Start effectiveness may also vary. Families in which the primary caregiver lives alone with her children are likely to have greater needs for social and possibly economic support. Families in which the primary caregiver lives with a spouse may need help with relationships and parenting in the context of family relationships. Families in which the primary caregiver lives with other adults are less likely to need social support but may need help with parenting in the context of varying family relationships.

1. Families in Which the Primary Caregiver Initially Lived with a Spouse

As with all groups, Early Head Start substantially increased service use among families in which the primary caregiver lived with a spouse. Early Head Start had small impacts on child care use and use of center-based child care but somewhat larger impacts on home visits, case management, and education and employment-related services in this subgroup (Appendix Table E.VII.25). Among primary caregivers living with a spouse, Early Head Start had few impacts on activities designed to promote self-sufficiency (Appendix Table E.VII.26).

Early Head Start did not affect children's cognitive or social-emotional outcomes in families that included a spouse. However, Early Head Start appears to have had a significant negative impact on two aspects of language development (Early Head Start increased the proportion of children with low vocabulary production scores and reduced combining words). Families that included a spouse were relatively more likely to be Hispanic, and program children in these families may have been in bilingual learning environments and learning both languages more slowly than they would have learned either language in a monolingual environment.

Among primary caregivers who initially lived with a spouse, Early Head Start significantly improved several aspects of parenting. Early Head Start increased emotional support (increased emotional responsivity and parent supportiveness), increased caregivers' verbal and social skills, and reduced spanking.

This pattern of impacts among families that included a spouse may reflect the following:

• Families in which the primary caregiver initially lived with a spouse were slightly more likely to be in home-based programs.
  
  
• Primary caregivers who initially lived with a spouse were more likely to be white or Hispanic.
  
  
• Families in which the primary caregiver initially lived with a spouse were more likely to be low-risk families.

2. Families in Which the Primary Caregiver Initially Lived With Other Adults

As with all groups, Early Head Start substantially increased service use among families in which the primary caregiver lived with other adults. In most cases, the impacts on service use were somewhat smaller in this subgroup (Appendix Table E.VII.25). However, the impacts on use of child care and use of center-based child care were much larger in this group. Early Head Start had a clear pattern of significant impacts on participation in education or job training among primary caregivers who lived with other adults (Appendix Table E.VII.26). The programs increased participation and average hours per week in these activities throughout most of the 15-month follow-up period. Early Head Start increased high school attendance in this subgroup.

Early Head Start had significant impacts on both child cognitive and child social-emotional outcomes in families in which the primary caregiver lived with other adults. Early Head Start significantly increased average Bayley MDI scores, increased child engagement, and reduced child aggression (Appendix Table E.VII.27).

Early Head Start improved several aspects of parenting in this subgroup. The programs improved emotional support (increased parent supportiveness), improved stimulation of language and learning (increased reading at bedtime, daily reading, and reading frequency, as well as parent-child activities to stimulate cognitive and language learning), and reduced negative parenting behavior (reduced parent intrusiveness).

This pattern of impacts may reflect the following:

• Families in which the primary caregiver initially lived with other adults were slightly more likely to be in center-based and mixed-approach programs.
  
  
• Families in which the primary caregiver initially lived with other adults were slightly more likely to be in programs that were implemented early.
  
  
• Families in which the primary caregiver initially lived with other adults were more likely to be teenage mothers.
  
  
• Families in which the primary caregiver initially lived with other adults were more likely to be moderate- or high-risk families.

3. Primary Caregivers Who Initially Lived Alone with Their Children

As with all groups, Early Head Start significantly increased service use among families in which the primary caregiver initially lived alone with her children (Appendix Table E.VII.25). Early Head Start did not have any significant effects on self-sufficiency-oriented outcomes in this subgroup (Appendix Table E.VII.26).

Early Head Start consistently improved child outcomes among families in which the primary caregiver initially lived alone with her children. The programs significantly increased average Bayley MDI scores and reduced the proportion of children who received low scores (Appendix Table E.VII.27). Early Head Start also had somewhat larger impacts on language outcomes among children of mothers living alone. Finally, Early Head Start improved one aspect of child social-emotional development (child sustained attention with objects).

Among the mothers who initially lived alone, Early Head Start improved parenting in several areas. Early Head Start increased stimulation of language and learning (increased establishment of bedtime routines and reading at bedtime; improved support for literacy and language learning at home) (Appendix Table E.VII.27). Early Head Start also increased knowledge of child development and increased the extent to which primary caregivers reported that they would use only mild discipline in conflicts with their child.

This pattern of impacts may reflect the following:

• Families in which the primary caregiver initially lived alone with her children were slightly more likely to be in home-based and less likely to be in mixed-approach programs.
  
  
• Families in which the primary caregiver initially lived alone with her children were more likely to be in programs that were incomplete implementers.
  
  
• Families in which the primary caregiver initially lived alone with her children were more likely to be receiving TANF cash assistance when they enrolled.
  
  
• Families in which the primary caregiver initially lived alone with her children were more likely to be African American.
  
  
• Families in which the primary caregiver initially lived alone with her children were more likely to be moderate- or high-risk families.

J. NUMBER OF RISK FACTORS

In addition to examining subgroups of families defined by one characteristic at a time, we investigated two types of subgroups based on multiple family characteristics-subgroups of families clustered statistically based on multiple characteristics and subgroups based on the number of family risk factors. Variations in impacts among these subgroups provide some insights into the potential confounding factors identified in the discussion of each set of subgroups above.

To minimize the potential confounding of family characteristics, we explored using cluster analysis to form subgroups based on multiple baseline characteristics simultaneously.4 Five clusters were identified: (1) nonwelfare mothers living alone, (2) working parents living with a spouse, (3) non-English- speaking families, (4) teenage parents, and (5) welfare families. While these clusters take into account multiple characteristics simultaneously, they are smaller than the other subgroups and less power is available for detecting significant impacts in them. The patterns of impacts among these subgroups tend to mirror patterns in the associated subgroups based on a single characteristic, so are not reported here.

In addition to investigating impacts among clusters of families, we examined impacts among families with difference levels of risk. In order to distinguish low, moderate, and high-risk families, we counted the number of risk factors that families had when they enrolled. Some of the risk factors tended to occur together, and when they did, families were considered higher risk. We counted up to five: (1) being a single parent, (2) receiving TANF cash assistance, (3) being neither employed nor in school or training, (4) being a teenage parent, and (5) lacking a high school diploma or GED. Families who had zero or one risk factor were classified as low-risk families. Families who had two or three risk factors were classified as moderate-risk families. And families who had four or five risk factors were classified as high-risk families.

1. Low-Risk Families

As with all groups, Early Head Start significantly increased service use among low-risk families. The magnitude of the impacts on receipt of home visits and participation in group activities was somewhat larger in this subgroup (Appendix Table E.VII.28). In this subgroup, Early Head Start had no significant impacts on activities designed to promote self-sufficiency in primary caregivers (Appendix Table E.VII.29).

Early Head Start did not significantly affect child cognitive or child language outcomes in this subgroup (Appendix Table E.VII.30). However, Early Head Start appears to have had a significant unfavorable impact on several child social- emotional outcomes (less child engagement during the parent-child structured play assessment and lower Bayley BRS emotional regulation scores). This pattern of impacts reflects in part the fact that control children in this subgroup were doing somewhat better than control children in the other subgroups at age 2 on all of the child outcomes we measured (as we would expect if risks are correlated with outcomes).

Among low-risk families, Early Head Start significantly influenced parenting in several areas. Early Head Start increased stimulation of language and learning (improved support for language and literacy learning at home and increased reading to children at bedtime), reduced the severity of discipline that parents reported they would use in response to conflict with their child, and reduced family conflict. Early Head Start also appears to have had unfavorable impacts on several parenting outcomes, which may have contributed to the unfavorable impacts on children's social-emotional development. Early Head Start decreased nonpunitive interactions (as assessed using the HOME) and increased negative parent regard during semi-structured parent-child interactions.

This pattern of impacts on low-risk families may reflect the following:

• Low-risk families were more likely to live in states with early work requirements.
  
  
• Low-risk families were more likely to be white.
  
  
• Low-risk families were more likely to enroll with older children.
  
  
• By definition, low-risk families were more likely to have primary caregivers who were employed, more likely to have primary caregivers who had completed 12th grade or a GED, and more likely to have primary caregivers who lived with a spouse.

2. Moderate-Risk Families

As with all groups, Early Head Start substantially increased service use among moderate-risk families. The impacts on child care use and average hours per week of center-based care were somewhat larger in this subgroup (Appendix Table E.VII.28). Early Head Start increased participation in school or training and increased average hours per week that primary caregivers spent in these activities. Early Head Start increased participation in vocational training in this subgroup (Appendix Table E.VII.29).

Early Head Start had a clear pattern of impacts on children in moderate- risk families. The programs significantly improved child cognitive, language, and social-emotional outcomes among these families. Early Head Start increased average Bayley MDI scores and reduced the proportion of children who scored below 85 at age 2, increased vocabulary production and sentence complexity at age 2, and increased child engagement during the parent-child structured play assessment (Appendix Table E.VII.30).

Early Head Start also improved several aspects of parenting in this subgroup. Early Head Start increased emotional support (emotional responsivity and parent supportiveness) and increased stimulation of language and learning (increased reading frequency, increased support for literacy and language learning at home, and increased parent-child activities to stimulate cognitive and language learning) (Appendix Table E.VII.30). Early Head Start also reduced parent detachment during the parent-child structured play assessment and increased knowledge of child development.

This pattern of impacts among moderate-risk families may reflect the following:

• Moderate-risk families were more likely to be headed by teenage parents.
  
  
• Moderate-risk families were more likely to have completed 12th grade and less likely to have attained less education.
  
  
• Moderate-risk families were more likely to be Hispanic.

3. High-Risk Families

As with all groups, Early Head Start significantly increased service use among high-risk families. High-risk families in the control group were more likely than control families in the other risk groups to receive services in the community, so the Early Head Start impacts on service use were generally not as large for this subgroup (Appendix Table E.VII.28). Early Head Start increased receipt of TANF cash assistance and food stamps by high-risk families, and reduced their average hours of employment (Appendix Table E.VII.29).

Early Head Start had mixed impacts on child outcomes in high-risk families. The programs significantly improved one child language outcome (reduced the proportion with low vocabulary production scores) but had consistently negative impacts on several child social-emotional outcomes (increased child negativity toward the parent and reduced Bayley emotional regulation and orientation/engagement scores) (Appendix Table E.VII.30). As noted in the joint effort of two local research teams (Box VII.2), adult attachment may be a significant factor in at-risk parents' ability to be effective caregivers.

 

BOX VII.2 ADULT ATTACHMENT IN EARLY HEAD START PARENTS Susan Spieker and Claire Hamilton University of Washington and University of Georgia Two of the Early Head Start research sites (in Kent, Washington, and Venice, California) conducted the Adult Attachment Interview (AAI) with all parents at the beginning of the project. The AAI is a structured, hour-long, semi-clinical interview during which the parent is queried about early experiences with caregivers. Transcripts are classified according to a four-category system describing an adult's current "state of mind with respect to attachment." The four states of mind are secure-freely autonomous, insecure-dismissing, insecure-preoccupied, and unresolved (reflecting mental disorganization associated with traumatic events). Parents whose AAI transcripts are classified as secure-autonomous are more sensitive caregivers of their children. The majority of parents in low-risk samples are classified as secure-autonomous, whereas the majority of parents in low-income samples have insecure classifications, in particular insecure-dismissing and unresolved. Among parents eligible for the program at the first Early Head Start research site, which involved predominantly white, non-Hispanic mothers, only 27 percent were classified as secure-autonomous, 32 percent were classified as insecure-dismissing, 7 percent as insecure-preoccupied, and 33 percent as unresolved. At the second site, consisting primarily of Latino immigrant families, the distribution was somewhat different: 38 percent of the mothers were classified as secure-autonomous, 25 percent as insecure-dismissing, 31 percent as insecure-preoccupied, and 6 percent as unresolved. The security rate at both sites was typical of other low-income samples, suggesting that Early Head Start parents are at risk for insensitive and unresponsive caregiving. Cultural differences may be involved in the relatively different distributions of preoccupied and unresolved classifications at the two sites.

 

Early Head Start had few significant impacts on parenting outcomes among high-risk families by the time children were 2. Early Head Start reduced parental distress, but also reduced the likelihood that parents set regular bedtimes for their children.

This pattern of impacts among high-risk families may reflect the following:

• High-risk families were slightly less likely to be in center-based programs.
  
  
• High-risk families were less likely to be in programs that were implemented early.
  
  
• High-risk families were less likely to be in states with early work requirements.
  
  
• By definition, high-risk families were more likely to include a teenage mother, more likely to be receiving TANF cash assistance when they enrolled, less likely to be living with a spouse, and more likely to include a primary caregiver who had not completed 12th grade or a GED.

We found this pattern of mixed impacts despite the following:

• High-risk families were more likely to enroll with firstborn children.
  
  
• High-risk families were more likely to enroll when they were pregnant.
  
  
• High-risk families were more likely to be African American.

K. CONCLUSIONS AND IMPLICATIONS

The analyses of impacts among family subgroups suggest several implications:

• The Early Head Start programs appear to have been most effective with families who enrolled with firstborn children. Programs had significant impacts in all areas of child development measured among firstborn children and in many areas related to parenting. However, the programs also had impacts on later-born children in the area of cognitive development and improved parenting of later-born children in several areas.
  
  
• The Early Head Start programs also appear to have been somewhat more effective in improving child outcomes in families who enrolled before their child was born. In particular, the significant impacts on child language outcomes are concentrated in children whose mothers enrolled while expecting a child. The programs also had a few positive impacts on children who were already born when they enrolled, and improved a number of parenting outcomes in both groups of families.
  
  
• The Early Head Start programs had a strong pattern of impacts on girls. Although the pattern of impacts on boys was similar, most of the impacts on boys were not statistically significant. The program impacts on parenting were more similar among parents of boys and girls. It appears that boys' development, which tended to be slightly behind that of girls, was less responsive to the improvements in parenting.
  
  
• The programs had more consistent, significant impacts among families in which the primary caregiver had no adult support at home and probably needed the social and other support provided by Early Head Start the most.
  
  
• The programs had the most consistent significant impacts on child well-being among families at moderate risk. These families probably needed and benefited from the support Early Head Start programs provided, but did not have so many risks that they had difficulty participating in services and acting on what they were learning.
  
  
• The impact analyses results suggest that the Early Head Start programs may need to find new strategies for serving the highest-risk families. The programs had relatively few significant impacts on these families, and a few of the significant impacts were unfavorable. This pattern of impacts is consistent with program staff reports during site visits that the highest-risk families were hardest to serve and suggests that the programs were not able, in the short run, to improve outcomes significantly, compared with the outcomes high-risk control families experienced with the help of other community services that they obtained on their own. Because high- risk families were relatively more likely to be in programs that were not fully implemented early, one important focus in efforts to serve high-risk families effectively may be full implementation of the Head Start Program Performance Standards.
  
  
• The Early Head Start programs produced some significant impacts on child well-being in the social-emotional and language domains among children of teenage parents. Moreover, Early Head Start improved parenting outcomes among teenage parents in most of the domains examined. Given the special challenges associated with serving teenage parents and the difficulties other programs have had in improving outcomes for them, this is an important accomplishment. The impacts on the children of teenage parents were not as pervasive as those for children of older mothers. Therefore, continued focus on the special needs of teenage parents and their children might strengthen Early Head Start programs.
  
  
• The Early Head Start programs appear to have provided a safety net for child development and parenting among families receiving TANF cash assistance when they enrolled, a population that was most immediately affected by the new requirements of welfare reform. The Early Head Start programs significantly improved the language and social-emotional outcomes of children whose mothers were receiving cash assistance when they enrolled. The Early Head Start programs increased the emotional support that mothers who were initially receiving cash assistance provided their children, in addition to improving the stimulation of language and learning that they provided.
  
  
• The Early Head Start programs also appear to have helped employed parents balance the demands of work and family. The programs had a strong pattern of positive impacts on child and parenting outcomes among these families.

In addition to the variations in impacts among different types of programs and contexts, the variations in impacts among subgroups of families suggest that early intervention (intervention early in children's lives and early in families' parenting experiences) may be important for maximizing program benefits. However, the programs also had significant impacts on parents and children who were enrolled during infancy or were later-born, which suggests that the program may be effective in different ways for these different groups. The subgroup analyses also show that program efforts to improve knowledge of child development and increase stimulation of language and learning at home yielded significant improvements in almost all subgroups of families.

For whom did Early Head Start achieve significant impacts? It is notable that the impacts of the Early Head Start research programs were fairly broad-based. The programs had some significant impacts in most of the subgroups we examined. The patterns of significant impacts differed among subgroups, however. The strongest patterns of impacts were found among:

• Families with moderate risks. Low-risk families tended to do better on their own than other higher-risk families. High-risk families, on the other hand, had more problems and needed more help than the programs provided in the short run.
  
  
• Families without other adult support at home. In the short run, Early Head Start did not increase families' financial resources but did increase their social resources. Families without social support from other adults at home appear to have benefited most from Early Head Start services.

The Early Head Start research programs also demonstrated patterns of significant impacts in several policy relevant subgroups-welfare families, working families, and families headed by teenage mothers. For these families, Early Head Start appears to have provided a safety net for parenting and child development while families coped with new work requirements and time limits on TANF cash assistance, balanced the demands of work and family, or attended to their own developmental needs.

---

1Appendix Tables E.VII.31 and E.VII.32 show the configuration of family characteristics across the research sites.(back)

2Because of sample-size constraints, these analyses are limited to three subgroups; Hispanic, African American, and White families.(back)

3The names of the participants have been changed.(back)

4This analysis used hierarchical clustering statistical techniques to allocate families with "similar" characteristics to the same clusters. The following family measures were used to form the clusters: (1) whether the mother had a high school credential; (2) whether the mother lived with a husband; (3) whether the mother lived alone; (4) whether the mother was a teenager; (5) whether the mother was employed; (6) whether the mother was in school or a training program; (7) whether the mother was receiving TANF benefits; and (8) whether English was the mother's primary language.(back)

 

Table of Contents | Previous | Next