Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 Effects of Child Care O ne by-product of the increase in women's employment in the United States has been a transformation in how children are cared for. This transformation has affected children of all ages--from the youngest infants to adolescents. Children in the United States typically begin full-time (i.e., 40 hours a week) nonmaternal care during their first year and this often lasts through elementary school, since the regular school day is typically shorter than parents' workdays. A critical issue for parents, educators, and policy makers is whether these care experiences are a source of enrichment that contributes positively to children's developmental out- comes or are a source of risk that undermines development. Substantial progress has been made in the past 15 years in determining the effects of child care on children's cognitive and social functioning. This progress reflects a convergence of conceptual and methodological advances and the availability of several large-scale research projects (see Box 5-1). From this research base, it is possible to specify the effects of nonmaternal care on children's development with greater confidence and precision than was possible when the National Academies published the 1990 report en- titled Who Cares for America's Children? (NRC, 1990). This chapter reviews the conceptual and methodological advances that have informed recent research. Then we evaluate the research evidence pertaining to the effects of three aspects of early child care--quality, type of care, and quantity--on a wide range of child developmental outcomes. Included in this review is consideration of experimental studies of center- based early education programs, which fulfill both child care and educa- 99
100 WORKING FAMILIES AND GROWING KIDS BOX 5-1 Multisite Child Care Studies 1. National Institute of Child Health and Human Development (NICHD) Study of Early Child Care The Study of Early Child Care (SECC) is a prospective longitudinal study of 1,364 children recruited at birth from 10 research sites: Little Rock, Arkansas; Irvine, California; Lawrence, Kansas; Boston, Massachusetts; Morganton, North Carolina; Philadelphia, Pennsylvania; Pittsburgh, Pennsylvania; Charlottesville, Virginia; Seattle, Washington; and Madison, Wisconsin. The sample includes eth- nic minority children (24 percent), mothers without a high school diploma (10 per- cent), and single mothers (14 percent). The recruited families did not differ from the eligible families on any of a substantial number of variables, except that moth- ers in the study were more likely to plan to be employed in their infant's first year. Of the 1,364 families who began the study, 1,216 continued through 36 months, 1,062 continued through 1st grade, and 1,033 continued through 3rd grade. Extensive information was collected about child care, families, and child functioning (see Annex Table A5-4). Extended observations of children's primary child care arrangements were conducted at 6, 15, 24, 36, and 54 months. Mothers reported amount and types of care during phone interviews every three to four months. In addition, extensive information about the children's families and homes was obtained during home and lab visits at 1, 6, 15, 24, 36, and 54 months. Child developmental (cognitive, social, academic, and health) outcomes were assessed using multiple methods (standardized tests, observations, questionnaires) and multiple respondents (mother, father, teacher). The study's design has made it possible to examine quality, quantity, and type of child care in the same analyses in order to estimate the unique contribu- tions of each factor. It also is possible to examine effects of timing (see Brooks- Gunn et al., 2002) and trajectories of care (see NICHD Early Child Care Research Network, 2002c). Another strength is that it is possible to include extensive con- trols for family factors, including controls for such observed factors as mother-child interaction and the home environment. The longitudinal data have permitted ex- aminations of changes in scores (NICHD Early Child Care Research Network and Duncan, 2002). Although it is a remarkably rich dataset, the NICHD study is limited in some important respects. The sample is not nationally representative. Compared with Census Bureau figures from all births in the United States in 1991, white, non- Hispanic children are somewhat overrepresented in the sample and children from ethnic minority groups are somewhat underrepresented (NICHD Early Child Care Research Network, 2001c). Mean household income and maternal education also were higher than the U.S. average. There also are indications (NICHD Early Child Care Research Network, 2000a) that poorer quality child care settings were less likely to have been observed, meaning that effects associated with quality of child care may be underestimated. 2. Cost, Quality, and Outcome Study The Cost, Quality, and Outcomes Study (CQO; Peisner-Feinberg et al., 1999) was conducted in four states (California, Colorado, Connecticut, and North Carolina) that varied in the stringency of their child care regulations. The initial sample was recruited in 1993 and consisted of 579 children (30 percent ethnic minority) who were enrolled in 183 preschool classrooms. At the start of the study, the children were in their next-to-last year of preschool before entering school. Classrooms were observed and rated for quality of the classroom environment, teacher sensitivity, and teaching style. These quality indicators were combined
EFFECTS OF CHILD CARE 101 into a single process quality composite. Children were followed through two years of child care and the first three years of school (kindergarten through 2nd grade) and completed tests of receptive language ability, reading ability, and math skills. Child care staff and schoolteachers rated the children's cognitive and attention skills, sociability, and problem behaviors yearly. 3. The Three-State Study The Three-State Study (Scarr et al., 1994) was conducted in Georgia, Mas- sachusetts, and Virginia, three states that varied in child care regulations. The sample consisted of 120 centers that included randomly selected programs from national chains, with nearby nonprofit programs, local for-profit centers, and church sponsored centers situated. Each program was observed during a single full-day visit that included observations of an infant classroom, a toddler classroom, and a preschool classroom. A total of 718 children (176 infants, 291 toddlers, and 251 preschoolers) were observed at the centers. Mothers and fathers completed ques- tionnaires about the study child's problem behaviors, as well as information about family income, education, parenting attitudes, and parenting stress. 4. The National Day Care Study The National Day Care Study (Ruopp et al., 1979) included a quasi-experi- ment that was conducted at 49 publicly funded centers in three cities (Atlanta, Geor- gia; Detroit, Michigan; and Seattle, Washington) and a random assignment experi- ment that was conducted in eight centers (29 classrooms). In the quasi-experiment, ratios were improved in some centers, high ratios were maintained in some centers, and low ratios were maintained in other centers. In the experiment, classrooms were assigned to one of three levels of staff education (master's degree, completed 2-year training program, had not completed 2-year training) and one of two ratios (5:1 versus 7:1). Outcomes included observed teacher and child behavior at the centers and child performance on standardized cognitive assessments. 5. Family and Relative Care Study This study was conducted in three communities (San Fernando Valley, Cali- fornia; Charlotte, North Carolina; and Dallas, Texas) that differed in the stringency of their child care home licensing regulations (Kontos et al., 1995). Participants were identified from random digit phone calls, birth records, and referrals for child care providers. A total of 820 families and 226 child care homes and providers of relative care were contacted. The final sample of children consisted of 145 cases (35 per- cent ethnic minority, 54.7 percent low or very low income) in which both mother and provider agreed to participate. Three-hour observations were conducted at each home by trained field staff, and quality of care was assessed using the Family Day Care Rating Scale (FDCRS). 6. The National Child Care Staffing Study (NCCSS) The National Child Care Staffing Study (NCCSS) was conducted in 1988 in 227 centers in five metropolitan areas in the United States (Atlanta, Georgia; Bos- ton, Massachusetts; Detroit, Michigan; Phoenix, Arizona; and Seattle, Washing- ton). Approximately 45 centers were randomly selected from the licensed full-day programs in each city. In each center, an infant, toddler, and preschool classroom was randomly selected, and two teachers in these classrooms (six per center; total number of teachers = 1,309) were interviewed about their training, education, wages, experience, and personal background. The selected classrooms also were observed by the research staff, who rated process quality using the Early Child- hood Environment Rating Scale (ECERS), the Infant/Toddler Environment Rating Scale (ITERS), and the Arnett Scale of teacher sensitivity.
102 WORKING FAMILIES AND GROWING KIDS tional functions. We then examine the effects of different types of child care during middle childhood. CONCEPTUAL AND METHODOLOGICAL ADVANCES Various aspects of child care have informed the advances of researchers working to understand the effects of child care on the cognitive and social functioning of children. Relationship Among Contextual Factors Bronfenbrenner's ecological systems theory (1979, 1989; Brofenbrenner and Morris, 1998) has guided much of the research by developmental psychologists who study child care effects (see NICHD Early Child Care Research Network, 1994; Vandell and Posner, 1999). A key element of the theory is a framework of nested relations among contextual factors, which are conceptualized as microsystems, mesosystems, and exosystems. A microsystem is described as "a pattern of activities, roles, and interpersonal relations experienced by the developing person in a given setting with par- ticular physical and material characteristics" (Bronfenbrenner, 1979:22). Consistent with the formulation, child care researchers have developed detailed descriptions of children's activities, roles, and interpersonal rela- tions at centers and day care homes, and with nannies and grandparents (see Clarke-Stewart, Gruber, and Fitzgerald, 1994; Howes, 1983; NICHD Early Child Care Research Network, 1996, 2000a). These descriptions proved instrumental in the development of measures that distinguish high- quality and low-quality care. Bronfenbrenner conceptualizes the mesosystem as "the interrelations among two or more settings in which the developing person actively partici- pates, such as, for a child, the relations among home, school, and neighbor- hood peer group" (1979:25), and this also has influenced research in this area. One child care and family linkage that has been extensively investi- gated is families' selection of care arrangements. In some cases, selection reflects active decision making, which occurs when parents visit several providers and then select one. In other cases, parents may use an arrange- ment because it is the only one that they can afford, even if they have concerns about it. The critical point is that family preferences and circum- stances influence the particular care that children receive. Child care and the family also are interconnected because child care may affect family functioning. An example of such effects was reported in the Wisconsin Family and Work Project. Early and extensive child care was related to maternal and paternal emotional well-being (Vandell et al., 1997). Increases in maternal and paternal depression, anger, and anxiety were
EFFECTS OF CHILD CARE 103 found in parents whose infants were in full-time child care during the first four months, but not in parents whose children were not in early and extensive child care. These associations between child care and the family highlight a chal- lenge for child care research, namely, to distinguish between aspects of the family that influence placement into care and aspects of the family that change in response to child care. Only the former reflect selection differ- ences, whereas the latter may be indicative of changes in family functioning that mediate child care effects. Longitudinal studies have begun to track the interplay between child care and the family over time. Bronfenbrenner conceptualizes the exosystem as "one or more set- tings that do not involve the developing person as an active participant, but in which events occur that affect, or are affected by, what happens in the setting containing the developing person" (1979:25), and this also has guided research about child care. For example, as discussed earlier, the mother's work environment (her schedule, working conditions, etc.) has implications for the child's development even if the child is not typically cared for at the mother's work site (Hoffman and Youngblade, 1999). Such effects are consistent with the conceptualization of the exosystem. Multidimensional Aspects of Child Care A second advance that has occurred in child care research since the mid-1980s is the move from simple comparisons of day care versus no day care to studies that focused on quality, quantity, and type of child care (see Lamb, 1998, for a comprehensive review). Studies of child care quality have asked whether structural and caregiver characteristics as well as more process-oriented indicators of caregiving are related to child developmental outcomes. Studies of child care quantity have asked if cumulative hours in child care as well as when care begins are related to child outcomes. Type- of-care studies have primarily focused on the effects of center care, al- though some research has considered the effects of child care homes, nan- nies, and relatives on child outcomes. A limitation of much of the research in this area is that these three aspects of care (quality, type, and quantity) have been studied in isolation, that is, without consideration of the other aspects. Thus, quantity of care has been investigated without consideration of the quality of care, and quality of care was studied without consideration of quantity. Findings in these studies are sometimes difficult to interpret because it is not possible to rule out alternative explanations for purported effects (Phillips et al., 1987a; Vandell and Corasaniti, 1990). Several recent projects, described below, have sought to disentangle effects associated with quality, amount, and
104 WORKING FAMILIES AND GROWING KIDS type of care by including measures of all three aspects of care in their design and analyses. Multiplicity of Child Care Arrangements A third advance is the recognition that children in the United States often have multiple child care arrangements, both simultaneously and se- quentially (see Chapter 3). This multiplicity of arrangements means that researchers have needed to collect information about child care over time and include secondary and tertiary arrangements as well as the primary arrangement. Otherwise, key aspects of these arrangements may not be measured adequately. Surveys such as the National Child Care Survey and the National Household Education Survey have asked parents to report both primary and secondary arrangements, and the Survey of Income and Program Par- ticipation (SIPP) asks parents to report all child care arrangements used by each child. The NICHD Study of Early Child Care obtained reports from mothers every three to four months to document the types and amount of care that were used. Correlational Versus Experimental Designs Although research examining the effects of high-quality center-based interventions on children from low-income families (e.g., Ramey et al., in press) has relied on experimental and quasi-experimental designs, most of the research examining the effects of child care quality and quantity has used correlational designs. As is the case with any correlational study, there are important concerns that unmeasured factors may account for reported effects (Blau, 2001). In some cases, child care researchers have sought to address possible selection bias by including multiple controls for family and child characteristics (examples of such studies appear in the annex at the end of this chapter, see Tables A5-1, A5-2, A5-3). In other cases, investiga- tors (Blau, 2000; NICHD Early Child Care Research Network and Duncan, 2003, discussed below) have considered the robustness of findings using other statistical methods to control for biases introduced by unobserved factors. Characteristics of the Child A final advance reflected in much of the recent research is the recogni- tion that child characteristics also may influence placement in child care. A long line of scholarship has shown that children (and parents) actively seek out environments that are consistent with children's maturity, interests, and
EFFECTS OF CHILD CARE 105 skills (Scarr and McCartney, 1983). For child care researchers, the issue is whether children with particular characteristics, dispositions, or skills are more likely to be placed in some settings and not others, and if these selection differences account for the observed findings. In some instances, it is relatively easy to identify child characteristics that affect placement. Placement based on age, for example, is easy to determine; infants are more likely to be placed in relative care, whereas preschoolers are more likely to be placed in centers. Identifying child dispositions and compe- tencies that influence placement in early child care is more difficult, because of the absence of reliable and robust measures of social and cognitive func- tioning for young infants that can be obtained before they begin child care. In several reports, the NICHD investigators used maternal reports of child temperament collected at 6 months of age as an indicator of child disposition, but by 6 months, 75 percent of the children had already been in care for 3 months (or half their lifetime). Measures of child functioning obtained after care begins may reflect the effects of that care and thus may not be measures of child selection. Measuring child adjustment and functioning for older children is more feasible because there are numerous psychometrically strong measures of social and cognitive functioning that are appropriate for preschoolers and young school-age children. QUALITY OF EARLY CHILD CARE In this section, we consider the research evidence pertaining to the effects of child care quality on developmental outcomes. First, we describe how quality is measured, focusing on measures of process quality, struc- tural characteristics, and caregiver characteristics. Next, a model is pre- sented that describes the interactions among various dimensions of child care experiences. This model has guided much research on the effects of child care quality on children's developmental outcomes. Three sets of research findings related to this model are explored. These include: (1) relations between structural characteristics and process quality, (2) rela- tions between process quality and child outcomes, and (3) relations be- tween structural measures and child outcomes. Both concurrent and longer term associations are presented. Measuring Child Care Quality Process quality refers to the kinds of experiences that children have with caregivers and other children, opportunities for cognitive, linguistic, and social stimulation, and opportunities to use interesting and varied ma- terials. Process quality is typically assessed by trained personnel who observe the arrangement for an extended period of time. Particular expe-
106 WORKING FAMILIES AND GROWING KIDS riences are evaluated because developmental theory and research have iden- tified them as important for children's healthy development. There are several robust measures of process quality that have been used by research- ers. Each of these measures is designed to serve somewhat different pur- poses. All of these measures are strong measures that have particular uses and strengths. One of the most commonly used measures of process quality is the Early Childhood Environment Rating Scale (ECERS) (Harms and Clifford, 1989), an instrument used to assess center-based care for preschool-age children. It consists of 37 items and evaluates 7 areas: personal care rou- tines, furnishings, language reasoning experiences, motor activities, creative activities, social development, and staff needs. Detailed descriptors are provided for each item, which is rated on a scale from 1 to 7 in which 1 = inadequate, 3 = minimal, 5 = good, and 7 = excellent. Ratings are com- pleted after at least two hours of observation in a classroom. The Infant/Toddler Environment Rating Scale (ITERS) (Harms et al., 1990) is a related measure for use in classrooms serving children under the age of 21/2 years. The Family Day Care Rating Scale (FDCRS) rates process quality in child care homes (Harms and Clifford, 1989), and the School- Aged Care Environment Rating Scale (SACERS) is appropriate for before- school and after-school programs serving school-age children. All of these measures have good internal consistency, and field staff can be trained to use them fairly easily The Observational Record of the Caregiving Environment (ORCE) was developed by the NICHD Early Child Care Research Network (1996, 2000a) to assess all types of child care settings. Age-appropriate versions are available for children ages 6 months to 5 years.1 Observers complete both time-sample behavioral counts and qualitative ratings during a series of 44-minute observation cycles collected over a 2-day period. The ob- server records the frequency or amount of specific caregiver behaviors, such as responds to child vocalization and asks questions, and makes qualitative four-point ratings of caregiver sensitivity to the child's needs, cognitive stimulation, positive regard for the child, emotional detachment, and nega- tive regard. The positive caregiving composite score is the mean of the qualitative scales, after reflecting the ratings of detachment and negative regard. Annex Table A5-4 provides the distribution of child care quality (cat- egorized as poor, fair, good, excellent) that was derived from positive caregiving composite scores of the ORCE in the NICHD Study of Early Child Care. In these analyses, poor quality care was defined as a composite 1 Coding manuals, including detailed descriptions of scales at each age, can be found at http://secc.rti.org.
EFFECTS OF CHILD CARE 107 score of less than 2, fair quality care as a score of 2 or more, but less than 3, good quality care was a score between 3 and 3.5, and excellent quality care was a score of 3.5 or higher. Because there is no nationally represen- tative study that has assessed process quality in the United States, the observations from the NICHD study and other multisite studies (described in Annex Table A5-2) are the only available estimates of process quality in the United States. In the NICHD study, 7 percent of the settings observed were of poor quality and 12 percent were of excellent quality. Most of the settings provided care that was only of fair quality. In Chapter 3 we estimated the process quality of child care from these studies. As shown in Annex Table A5-4, children in low-income families were more likely than children in high-income families to receive poor-quality child care: 11 percent of the low-income children compared with 4 percent of the high-income children. Children in low-income families also were less likely than children in high-income families to receive excellent quality child care: 8 percent of low-income children compared with 15 percent of high-income children. Disparities in quality associated with family income were more evident in child care homes and informal care arrangements than in centers, perhaps because children in low-income families have ac- cess to publicly supported programs such as Head Start. It is likely that these observations overestimate the amount of high- quality care and underestimate the amount of poor-quality care. Informal care settings and settings that serve children from low-income families were more likely to refuse to participate in the observations, and outcomes are lower for children whose care was not observed even after controlling for an extensive array of family covariates (NICHD Early Child Care Research Network and Duncan, 2003). The sample in the NICHD Study also did not include some groups of children (e.g., children of adolescent mothers, mothers who do not speak English, and mothers who were known sub- stance abusers). Other measures of process quality are the Caregiver Interaction Scale (Arnett, 1989), which focuses on teachers' sensitivity during interactions with children, and the CC-HOME scale, which assesses overall quality of child care homes (Clarke-Stewart et al., 2002). All of these measures have excel- lent psychometric properties and predict child developmental outcomes. Another approach to the assessment of child care quality is consider- ation of structural characteristics, such as child-adult ratio, and caregiver characteristics, such as caregivers' specialized training as indicators of child care quality. Structural-caregiver characteristics are the only indicators of quality in studies such as the National Longitudinal Survey of Youth, the National Child Care Survey (Hofferth et al., 1991), and the National House- hold Education Survey (Hofferth et al., 1998). Both structural-caregiver characteristics and process quality measures were collected in the NICHD
108 WORKING FAMILIES AND GROWING KIDS Study of Early Child Care, in the Cost, Quality and Outcome Study, and the Child Care Staffing Study. Structural-Caregiver Characteristics, Process Quality, and Child Outcomes The conceptual model that has guided much of the research on the effects of child care quality on children's developmental outcomes is shown in Figure 5-1 (Blau, 2001; Lamb, 1998; NICHD Early Child Care Research Network, 2002b; Vandell and Wolfe, 2000). This schema posits that pro- cess quality is directly related to child developmental outcomes, whereas structural-caregiver characteristics are posited to affect child outcomes in- directly through their impact on process quality. In this model, appropriate structural-caregiver characteristics are seen as providing necessary, but not sufficient, conditions for high-quality care. Also reflected in the model is the recognition of the importance of family factors for child developmental outcomes and selection into child care. For the most part, individual studies (see Annex Tables 5-1, 5-2, and 5- 3) have focused on one or another component of the overall model, while positing that the other pathways exist. One report, however, has formally tested the overall model (NICHD Early Child Care Research Network, 2002b). In that study, structural equation modeling (SEM) was used to test relations between structural-caregiver characteristics and process quality as predictors of child developmental outcomes. Two main findings were found: (1) process quality measured by the ORCE predicted children's cognitive competence and social competence at 41/2 years, controlling for Family Structure Family Process Child Outcome Structural-Caregiver Process Quality Characteristics FIGURE 5-1 A conceptual model of relations among structural-caregiver chateris- tics, process quality, and child outcomes. SOURCE: Blau (2001).
EFFECTS OF CHILD CARE 109 family income, maternal education, and parenting quality; and (2) there was a connection between structural qualities such as caregiver training and child-staff ratio and cognitive competence and social competence that was not accounted for by family variables. As noted above, other research studies have focused on one or another pathway in the overall model. In the following sections, we review these studies, starting with studies that consider relations between structural-caregiver characteristics and process quality. Then we examine research pertaining to process quality and child developmental outcomes. Finally, we consider relations between struc- tural-caregiver characteristics and child developmental outcomes. Structural-Caregiver Characteristics and Process Quality Blau (2001) and Vandell and Wolfe (2000) reviewed the research ex- amining associations between structural-caregiver characteristics and pro- cess quality. (These studies also are summarized in Annex Table A5-1.) Four structural-caregiver characteristics--child-adult ratio, group size, spe- cialized training, and general education level--have been the focus of much of the research. Although early research studies were criticized for various methodological limitations (small samples, single site, simple bivariate analyses), recent research (Burchinal et al., 2002; NICHD Early Child Care Research Network, 1997b, 2000a) has examined relations between struc- tural-caregiver characteristics and process quality using multivariate tech- niques, multiple study sites, and large samples. As shown in Annex Table A5-1, many studies have reported associa- tions between child-adult ratios and process quality. When child-adult ratios are lower, caregivers spend less time managing children in their classrooms, children are less apathetic and distressed (Ruopp et al., 1979), and caregivers are more stimulating, responsive, warm, and supportive (Clarke-Stewart et al., 1994; NICHD Early Child Care Research Network, 1996, 2000a; Phillipsen et al., 1997). Lower child-adult ratios also are associated with higher process quality scores on the ECERS and the ORCE (NICHD Early Child Care Research Network, 1996, 2002b,c). For the most part, these findings are derived from correlational research designs, although one study (Ruopp et al., 1979) used a random assignment experi- mental design to assess the effects of varying child-adult ratios. The number of children in the group (or group size) also is associated with process quality. In multivariate analyses that included ratio, group size, caregiver training, and caregiver education, caregivers appeared more responsive, more socially stimulating, and less restrictive when there were fewer children in the group (NICHD Early Child Care Research Network, 1996, 2000a; Ruopp et al., 1979). Process quality is higher in child care
110 WORKING FAMILIES AND GROWING KIDS homes that are in compliance with recommended group sizes dependent upon the age of the children (Clarke-Stewart et al., 2002). Caregivers' education--both formal education and specialized training pertaining to children or early education--is related to process quality. When caregivers have more formal education and more specialized train- ing, the care they provide is more likely to be stimulating, warm, and supportive, to offer better organized materials, and to provide more age- appropriate experiences (Howes, 1997; NICHD Early Child Care Research Network, 1996, 2000a; Phillipsen et al., 1997). Analyses based on the Cost, Quality, and Outcome (CQO) dataset also indicate that ratio and caregiver education are related to observed quality. The chapter in the CQO Technical Report by Mocan et al. (1995) showed consistent findings when hierarchical regressions (preferred by psycholo- gists) and econometric analyses were conducted. Both sets of analyses showed ratio, caregiver education/training, and caregiver wages (especially wages for lower paid employees or employees with less education) to pre- dict observed quality. Associations between caregiver wages and process quality also have been reported in the Child Care Staffing Study (Howes et al., 1992) and the Three-State Study (Phillips et al., 2000). There is evidence that low wages are associated with high turnover rates, another indicator of poorer quality care (Whitebook et al., 1998, 2001). These relations among wages, process quality, and turnover are worri- some because wages for child care staff tend to be low. For example, whereas the incomes for kindergarten, elementary, middle, and secondary teachers ranged from $37,610 to $42,080 in 2000, the median earnings for prekindergarten teachers were $17,810 (U.S. Bureau of Labor Statistics, 2003:203). Furthermore, the salaries of child care teachers tend to be even lower than prekindergarten teachers. Finally, there is some evidence that relations between structural-caregiver characteristics and process quality vary by child age. In a study that observed children from 6 though 36 months, group size and child-adult ratios were stronger predictors of process quality for infants, whereas caregiver educa- tional background and training were stronger predictors of process quality for preschoolers (NICHD Study of Early Child Care, 1996, 2000a). Blau (1997, 2000) has argued that the research evidence pertaining to structural-caregiver characteristics and process quality must be viewed with caution because unobserved characteristics may be confounded with the structural-caregivers characteristics, causing biased estimates. Thus, it seems likely that better structural-caregiver characteristics can facilitate improved process quality, but research to date leaves uncertainty about the magnitude of the effects.
EFFECTS OF CHILD CARE 111 Process Quality and Child Outcomes Since the publication of the report, Who Cares for America's Children (National Research Council, 1990), a substantial body of research has found significant associations between process quality and child develop- mental outcomes (see Blau, 2001; Love et al., 1996; Vandell and Wolfe, 2000). Findings from these studies are summarized in Annex Tables 5-2 and 5-3. First is an examination of relations between process quality and child functioning in the child care setting. Then process quality and concurrently assessed child functioning in other settings is considered, including laboratory assessments, standardized tests, and maternal re- ports. Finally, we review research evidence pertaining to relations be- tween process quality and subsequent child developmental outcomes and relations between structural-caregiver characteristics and child develop- mental outcomes. Children's Behaviors in the Child Care Setting. Nine studies examined relations between process quality and child behavior in the child care set- ting (see Annex Table A5-2). Five of these studies had relatively small samples (less than 100 children), and five studies did not include controls for family background. Within these constraints, a consistent pattern of findings is evident. Children were more likely to display positive affect and to appear securely attached to their caregivers in child care settings in which process quality is higher (Elicker et al., 1999; Hestenes et al., 1993; Howes et al., 1992; Howes and Smith, 1995). Children appear more prosocial and positively engaged with peers when their caregivers are sensitive and posi- tive to them (Holloway and Reichhart-Erickson, 1988; Howes et al., 1992; Kontos and Wilcox-Herzog, 1997). Finally, children are rated as more cognitively competent during free play in child care settings that offer more opportunities for art, blocks, and dramatic play (Kontos and Wilcox- Herzog, 1997). Relations between process quality and child functioning were more consistently reported in studies that used assessments with strong psychometric properties (Vandell and Pierce, in press). Child Functioning in Other Settings. A total of 17 studies examined concurrent relations between process quality and child functioning in labo- ratory assessments, maternal reports, and teacher reports (see Annex Table A5-2). Five of these studies had relatively small samples (less than 100 children), and seven studies did not include controls for family background. The research evidence across the 17 studies, however, suggests that higher process quality is associated with better cognitive and social-emotional development. In particular, children whose child care is higher in process quality score higher on standardized language and cognitive tests (Burchinal
112 WORKING FAMILIES AND GROWING KIDS et al., 1996; Dunn et al., 1994; Goelman, 1988; McCartney, 1984; NICHD Early Child Care Research Network, 2000a, 2002c; Peisner-Feinberg and Burchinal, 1997; Schliecker et al., 1991). Higher process quality also has been related to fewer behavior problems (Hausfather et al., 1997; Peisner- Feinberg and Burchinal, 1997) and to higher social competence (Phillips et al., 1987b), as reported by mothers and teachers. Longer Term Associations Between Process Quality and Child Function- ing. Relations between process quality and children's subsequent develop- ment also have been reported (see Vandell and Wolfe, 2000, and Annex Table A5-3). Significant associations are more evident in research studies that assessed process quality at several time periods (Burchinal et al., 2002; NICHD Early Child Care Research Network, 2000a, 2002c) and less appar- ent in studies that relied on a single assessment (Chin-Quee and Scarr, 1994; Deater-Deckard et al., 1996). Because children in the United States typically experience multiple arrangements and because caregivers change as children move from one class to another and as caregivers leave the program, it is not surprising that effects are less evident in studies that measured child care on only a single occasion or at a single age. Multiple assessments are more likely to yield more reliable indicators of children's care. In the NICHD Study of Early Child Care (2000a, 2002c), cumulative scores of process quality (the ORCE positive caregiving composite) ob- tained from child care observations at 6, 15, 24, 36, and 54 months pre- dicted children's performance on standardized cognitive and language as- sessments at these same ages, controlling for other aspects of child care (quantity and type) and family factors (maternal IQ, family income, ob- served quality of the home environment, and observed maternal stimula- tion). In the Cost, Quality, and Outcomes Study, process quality (assessed with the ECERS) predicted cognitive, language, and social development during the early grade school years, controlling for maternal education and ethnicity (Peisner-Feinberg et al., 2001). Children who had closer relation- ships with their preschool teachers appeared more sociable in kindergarten. Children who attended higher quality child care demonstrated better math skills prior to school entry, in kindergarten, and in 2nd grade. The children had better receptive language skills during the preschool period. These relations were stronger for children whose mothers had less education. Other longitudinal analyses also have reported longer term associations between process quality and social developmental outcomes. In analyses of the Cost, Quality, and Outcome sample, Howes (2000) reported that preschoolers who attended centers in which there were closer teacher-child relationships were subsequently rated by their 2nd grade teachers to be more socially competent with peers, controlling for family factors and pre- vious child functioning. Other research also has found relations between
EFFECTS OF CHILD CARE 113 process quality and children's social-emotional outcomes. Children who had caregivers who were involved and invested in positive socialization practices during the first three years were reported by their kindergarten teachers to have fewer behavior problems and better verbal IQs (Howes, 1990). Both economists and developmental psychologists (Blau, 1999b; Burchinal et al., 1995, Duncan and Gibson, 2000) have worried, however, that the relations between process quality and child functioning are biased estimates because of omitted variables. To more definitively address the question of causality, random assignment experiments are needed, but ran- domly assigning some children to low-quality child care is unethical and unlikely to be acceptable to parents, institutional review boards, or re- searchers. Consequently, investigators have turned to a variety of statistical strategies to address this problem. One strategy has been to expand the list of family covariates in an effort to better capture omitted variables. In the NICHD Study of Early Child Care (2002c), for example, child care effects were tested in models that had nine covariates (such as maternal education, ethnicity, family structure, income-to-needs ratio, maternal depressive symptoms, observed parenting quality, observed home environment quality) and then 15 family covariates (the 9 covariates plus measures of child temperament, maternal psychological adjustment, maternal report of social support, maternal sepa- ration anxiety, and maternal beliefs about the benefits of maternal employ- ment). The effects associated with child care did not differ in the two models, suggesting that omitted family variables did not account for the child care effects. In these analyses, effect sizes between process quality and preacademic skills and between process quality and language skills were 0.24 and 0.17, respectively. A second strategy has been to test effects of early child care experiences after controlling for concurrent quality (NICHD Early Child Care Research Network, in pressb). In these analyses, the quality of care from 6 to 36 months was found to predict child cognitive outcomes at 41/2 years when process quality at 41/2 years was statistically controlled. Social outcomes, in contrast, were predicted by concurrent child care quality, but not earlier quality. A third strategy has been to examine changes in child functioning associated with changes in child care quality (NICHD Early Child Care Research Network and Duncan, 2002). Modest but statistically significant impacts of process quality were observed when children were 6-24 and 36- 54 months. An increase of 0.5 point on the 4-point ORCE scale (represent- ing an increase of one standard deviation in child care quality) between 36 and 54 months was associated with an increase of 2.4 points on standard- ized cognitive tests. Effects of quality between 6 and 24 months were found
114 WORKING FAMILIES AND GROWING KIDS to have had an independent and additive impact on outcomes at age 54 months that were of similar magnitude, suggesting that higher quality care between 6 and 54 months was associated with a 5.0 point increase in cognitive scores. Effect sizes were considerably higher for children with low early cognitive scores. Structural-Caregiver Characteristics and Child Outcomes Consistent with the conceptual model delineated in Figure 5-1, a third body of research has examined relations between structural-caregiver char- acteristics and children's developmental outcomes (see Annex Tables 5-2 and 5-3). In one study (Ruopp et al., 1979), effects of structural-caregiver characteristics on child outcomes were tested both quasi-experimentally and experimentally. Children assigned to classrooms in which teachers had more education and training displayed greater gains in cooperative behav- ior, task persistence, and school readiness over a nine month period relative to children assigned to classrooms in which teachers had less education and training. Other studies have adopted correlational designs to examine relations between structural-caregiver characteristics and child outcomes. Infants who attended programs with smaller observed child-adult ratios had better receptive and expressive language skills (Burchinal et al., 1996; Vernon- Feagans et al., 1997), higher cognitive scores (Burchinal et al., 1996), and better social knowledge and social skills (Holloway and Reichhart-Erickson, 1988) than children who were attending programs with higher observed child-adult ratios. When teachers were better educated and had more recent child care training, children displayed better expressive language skills (Burchinal et al., 1996; Howes, 1997) and higher scores on cognitive inventories (Dunn, 1993; Clarke-Stewart et al., 2000). A limitation of much of this research, however, is that it relied on small samples (less than 100) and did not control for family background. Child developmental outcomes also have been examined in relation to aggregated scores of structural-caregiver characteristics. Some programs, for example, seek to trade off teacher training and group size by having better trained teachers care for more children. In other programs, more highly trained teachers are assigned fewer children. And in still other programs, centers fail to meet any recommended standards. In one report (NICHD Early Child Care Research Network, 1999b), centers were scored in terms of the age-based guidelines for group size, child-adult ratio, caregiver training, and caregiver formal education recommended by the American Public Health Association. If a particular guideline was met, a point was awarded. If a guideline was not, no point was given. The number of points was then summed. Controlling for family income and
EFFECTS OF CHILD CARE 115 maternal sensitivity, children who attended centers that met more recom- mended guidelines had fewer behavior problems at age 2 and 3 years and higher school readiness and language comprehension scores at 3 years. Compliance with specific guidelines also was related to child scores. Children displayed fewer behavior problems and more positive social be- haviors when centers were observed and met the recommended child-adult ratio at 2 years. Children exhibited fewer behavior problems and obtained higher school readiness and language comprehension scores at 3 years if caregivers reported levels of specialized training and education that met the recommended guidelines for specialized training and formal education. Longer term relations between structural-caregiver characteristics and children's subsequent social and cognitive development also have been re- ported (see Love et al., 1996; Vandell and Wolfe, 2000). Howes (1988), for example, examined structural-caregiver characteristics (amount of teacher train- ing, child-adult ratio, group size, a planned curriculum, and space) when chil- dren were 3 years of age in relation to the children's functioning in 1st grade. During the intervening period, the children attended the same university lab school, meaning that they experienced classes with the same or similar struc- tural and caregiver characteristics. Controlling for maternal work status, fam- ily structure, and maternal education, children whose early care met more structural-caregiver guidelines had fewer behavior problems and better work habits compared with children whose early care had met fewer guidelines. In a different sample, Howes (1990) found significant relations using a composite score of structural-caregiver indicators (child-adult ratio, caregiver training, caregiver stability) at 18, 24, 30, and 36 months. Con- trolling for family sociodemographic factors and socialization strategies, children with a history of poor-quality structural-caregiver indicators were rated as more difficult by their preschool teachers and as more hostile by their kindergarten teachers. Recent research from the Otitis Media Study also has focused on specific structural and caregiver characteristics in rela- tion to subsequent child developmental outcomes (Burchinal et al., 2002). Children whose child care classrooms met recommended guidelines for child-staff ratios exhibited better receptive language and communication skills, controlling for child gender, family poverty, and cognitive stimula- tion and emotional support in the home. Girls whose caregivers had at least 14 years of education (with or without early childhood training) had better cognitive and receptive language skills over time. Blau (1997, 1999b, 2000) has been critical of the fact that research in this area has relied on nonrepresentative samples and has failed to control adequately for family factors and other child care features. To address these concerns, random assignment experiments are needed to establish whether the reported associations between caregiver characteristics, struc- tural characteristics, and child social and cognitive functioning are causal.
116 WORKING FAMILIES AND GROWING KIDS EFFECTS OF DIFFERENT TYPES OF EARLY CHILD CARE Research that has considered the effects of different types of child care has taken several forms. One set of studies has used experimental designs and quasi-experimental designs to evaluate the effects of high-quality early education programs on children in families with low incomes. The primary purpose of these programs is the improvement of the developmental out- comes of children who are at risk for school failure, although the programs also can function as child care, especially if they are full-day programs. Well-known examples of such programs include Head Start, which pro- vides preschool educational services, nutrition, a health component, social services, and parent involvement for families with 3- and 4-year-old chil- dren (Zigler and Styfco, 1993), and the Chicago Parent-Child Centers (Reynolds, 2000). Other types of early child care that have been studied include comparisons of the effects of child care centers, child care homes, and relative care. In this section, we first review the research pertaining to the effects of high-quality early education programs and then turn to con- sideration of the effects of different types of child care typically used by families, including child care centers, family child care, in-home care (i.e., nannies), and grandparent care. Early Center-Based Intervention Programs Early childhood intervention programs have been categorized accord- ing to three overall models: those that provide educational services to chil- dren; those that provide parenting, vocational, and other supports to par- ents; and those that combine these two components (Brooks-Gunn et al., 2000; Seitz, 1990; National Research Council and Institute of Medicine, 2000; St. Pierre et al., 1995; Yoshikawa, 1995). In this section, we do not consider programs that provide services to parents only, because the focus of this chapter is on child care; we review evidence on the two other forms of early childhood intervention. We do not aim to provide a comprehensive review of effects of each type of intervention program, but rather to sum- marize patterns of effects by type of program, review evidence on these programs' coverage of working parents in the United States, and consider the quality of these programs. We exclude programs serving solely families with children with disabilities. Child-Focused Programs The provision of educational services to low-income, preschool-age children, most often in center-based programs, is the main focus of this set of early childhood interventions. Beginning with evaluations of educa-
EFFECTS OF CHILD CARE 117 tional preschools prior to the 1960s, this kind of program has demon- strated consistent short-term benefits on measures of cognitive develop- ment and early school performance in randomized evaluations (Barnett, 1995; Bryant and Maxwell, 1997). However, few of these experimental studies have been of large-scale programs; the vast majority of studies have been conducted in single-site, demonstration-quality programs. For ex- ample, of 21 large-scale programs reviewed by Barnett (1995), none was evaluated using random assignment designs. Similarly, although state prekindergarten programs have expanded tremendously in the recent decade, no randomized evaluations of these programs yet exist. Head Start is the most well-known example of this type of intervention. Although a variety of quasi-experimental studies have shown short-term gains from Head Start in areas of cognitive ability and school performance, with some showing some longer term effects into early adolescence (e.g., Currie and Thomas, 1995; Garces et al., 2000), no experimental evalua- tions of Head Start exist. A random assignment evaluation of the Head Start program is now under way. Some recent work on the Chicago Parent-Child Centers, a large-scale preschool program tied to follow-on transition services through 3rd grade, attempted to adjust for selection biases by utilizing sample selection mod- els. This work demonstrated long-term effects of the program in reducing rates of delinquency and raising levels of school achievement (Reynolds, 2000). Another program, which enhanced Head Start with a social skills component and parent training, has shown short-term effects in reducing aggressive behaviors among Head Start children in a randomized evalua- tion (Webster-Stratton et al., 2001). One of the best known early intervention projects is the Carolina Abecedarian Project (Campbell and Ramey, 1995; Ramey et al., 1998, in press). This clinical trial began at 6 weeks postpartum and included: (1) a randomized control group (n = 54) that received family support social services, pediatric care, and child nutritional supplements, and (2) an ex- perimental group (n = 57) that received the services of a high-quality center- based intervention for the first five years. The center operated 5 days a week for 50 weeks a year. Child-adult ratios were 3:1 for infants and 6:1 for older children. The curriculum emphasized language development. IQ scores at ages 8 and 15 years were significantly higher for the children who received high-quality center-based care during the first five years (4.5 point difference at 8 years and 4.6 point difference at 15 years). The effect sizes, calculated as the difference in means between treatment and control sub- jects divided by the standard deviation of the control group, were similar at the two ages (0.36 and 0.35, respectively) (Campbell and Ramey, 1995). Children who had participated in the early intervention program also had higher scores on tests of reading and mathematics achievement at age 15
118 WORKING FAMILIES AND GROWING KIDS and were less likely to be retained in grade by age 15 (31.2 compared with 54.5 percent, p < 0.02), and they were less likely to be placed in special education (12 compared with 47 percent). The most recent follow-up report from this research team (Ramey et al., 1999) includes findings to age 21. Intervention children were reported to be older, on average, when their first child was born and to have been more likely to attend a four-year college. Another demonstration project that has reported beneficial effects of a high-quality early education program is the Perry Preschool Project (Schweinhart et al., 1993), which involved 123 black children who were followed to age 27. The experimental group consisted of 45 children who entered the preschool program at age 3 and an additional 13 who entered at age 4, attending a half-day center-based program and receiving teacher home visits. The researchers report that the experimental group had a somewhat lower probability of ever being arrested by age 27 (57 com- pared with 69 percent), but a larger difference in the average number of lifetime arrests by age 27 (2.3 compared with 4.6). Differences in the proportion receiving public assistance by age 27 were also large--15 com- pared with 32 percent. Mean earnings were far higher for the experimen- tal group than the control group at age 27--monthly reported mean earnings were $1,219 for the experimental group and $766 for the con- trol group. Participation in the Chicago Child-Parent Centers (CPC) also has been related to long-term beneficial effects (Reynolds et al., 2000). This project has followed the educational and social development of 1,539 black (93 percent) and Hispanic (7 percent) children as they grew up in high-poverty neighborhoods in the central city of Chicago. Some of the children (n = 989) participated in government-funded (Title I) early childhood programs in 1985-1986, whereas others did not (n = 550). A rich array of data, including surveys from teachers, parents, school administrative records, standardized tests, and the children themselves has been collected since that time. Reynolds (1994) reported (at the end of the program in 3rd grade), that extended program participation was related to one half of a standard deviation improvement in reading and math scores over the comparison group, even after controlling for family risk status, child gender, and later program participation. Reynolds and Temple (1998) obtained similar ef- fects of extended program participation on reading and math performance at age 13. At age 20, participants in the CPC were more likely to have completed high school (49.7 compared with 38.5 percent, p < 0.01) and to have lower rates of juvenile arrests (16.9 compared with 25.1 percent, p < 0.001) (Reynolds et al., 2001). Even though only a few studies have followed children into adulthood, it is notable that all find some evidence of long-term gains. For instance,
EFFECTS OF CHILD CARE 119 Heckman (2000) looks at long-term effects of each education program on long-term success in the job market. In this discussion of early center-based interventions, it is relevant to explore whether these programs provide adequate coverage for low-income families. Head Start served 858,000 children in 2000, with a budget of $6.2 billion. The program served just over 40 percent of eligible families. Only a quarter of Head Start programs are full-day programs (Administra- tion for Children and Families, 2002). Experimental data from welfare-to- work evaluations shows no effect of mandated employment programs on Head Start use, while the use of other forms of center care appeared to increase (Chang et al., 2002); this difference in effects may be due to the high proportion of part-day Head Start programs or the fact that Head Start centers may already be full. Data from a random sample of 40 Head Start programs (comprising 518 classrooms) found that the average ob- served quality of services in classrooms was in the "good" range of the ECERS, with under 2 percent of classrooms in the minimal-to-inadequate range and 17 percent of classrooms in the excellent range (a score of 6 or 7; Zill et al., 1998). Researchers noted that this was a somewhat higher average than in most studies of center care for low-income families, with fewer classrooms at the bottom of the distribution on the measure (i.e., fewer classrooms in the "minimal" or "inadequate" ranges of the ECERS). In 1999, state prekindergarten funds served over 4.5 million children in 32 states (Gilliam and Zigler, 2000). A higher percentage of these programs are full-day (just under 50 percent). However, the quality of these programs is largely unknown. One study in Michigan found its sample of preschools to fall in the range of medium to high quality on a measure developed from the state's performance standards; subscales in the areas of administration, supervision, program philosophy, and use of fund- ing were related to child development ratings in kindergarten in predicted directions (Florian et al., 1997). Gilliam and Zigler's (2000) review of state-sponsored prekindergarten evaluations discusses serious shortcomings with the designs of all the exist- ing evaluations. For example, only the evaluation in New York used a reasonable method of selecting a control (using other children from the waiting list), and none used random assignment. Parent-Focused Programs and Combination Child- and Parent-Focused Programs The combination of services to parents (whether in job training, adult education, parenting information and support, or other social services) with child-focused enrichment has been termed the "two-generation" ap- proach (Smith, 1995). The most prevalent forms of parent-focused services
120 WORKING FAMILIES AND GROWING KIDS in this program model are parenting support and vocational/educational support. In parenting support models, professional or paraprofessional staff (most often home visitors) provide informational support about parenting and child development, as well as some general support for voca- tional and educational goals, to low-income parents. As of 1999, a conser- vative estimate was that 550,000 children were participating in these pro- grams (totaling across selected programs with national offices only; Gomby et al., 1999). However, few of these programs combine home visiting with a part- or full-day, educational, child-focused component in a two-genera- tion approach. No national estimates exist regarding the numbers of chil- dren served overall in two-generation programs. Demonstration-quality programs that have combined these approaches have shown impressive effects in experimental evaluations, not only on child school performance, but also on antisocial behavior in late childhood and adolescence and some measures of parenting and parent employment (Bryant and Maxwell, 1997; Yoshikawa, 1995; National Research Council and Institute of Medicine, 2000). However, it is unclear to what extent these same programs, if scaled up, would retain these effects. The largest federal initiative taking this program approach is the Early Head Start program, which in 2000 served 55,000 children ages 0 to 3 in 664 communities. This program, which provides parent education and educational child care, is implemented in center-based, home-based, and combination versions, depending on the site (each must adhere to a set of performance standards). A three-year follow-up, in a randomized evalua- tion of families in 17 sites, documented an overall pattern of impacts at age 3. For children, positive impacts were found in cognitive development, language development, and social-emotional development. For parents, positive impacts were in parenting, home environment, participation in education and job training, and subsequent births (mothers less likely to have a second child). There were also some positive effects specifically related to fathering and father-child interactions, but not all of the pro- grams had a father-oriented component and participated in father studies. All effects were in the small range (between 0.10 and 0.20 of a standard deviation). An eight-site experimental evaluation was conducted on a program with a somewhat similar combination of parent-focused home visits begin- ning at birth and high-quality child care beginning at 12 months, the Infant Health and Development Program (Brooks-Gunn et al., 1992, 1994; McCormick et al., 1993). This program showed positive effects on IQ and other cognitive measures among a sample of low-birthweight infants when they were 36 months old. However, assessments at ages 5 and 8 showed that the control group had improved in its levels of cognitive outcomes, and no experimental effect was found on cognitive or behavioral measures of
EFFECTS OF CHILD CARE 121 child development (McCarton et al., 1997), although positive cognitive outcomes have been found for children receiving higher dosages (i.e., at- tending more days) (Hill et al., 2003). This was an unusually large-scale demonstration program, rather than a randomized evaluation of an exist- ing large-scale program. What constitutes quality in two-generation programs and what the effects of variation in program quality are on child development remain largely unstudied. The Early Head Start evaluation, rating each of the 17 programs on level of implementation, found that experimental effects were largest in the group that was judged to have achieved the highest level of implementation (Love et al., 2002). Evaluations of a model home visiting program that was developed by David Olds and colleagues have shown that the proportion of home visit time focused on parenting issues was positively associated with HOME scores of parenting quality and observed measures of mothers' empathy toward children (Korfmacher et al., 1998). It is unknown how the most widely available two-generation programs score on such measures of quality. Child Care Centers, Child Care Homes, and Grandparent Care In addition to the evaluations of the early childhood intervention pro- grams, investigators have studied the effects of participating in different types of child care, such as child care centers, child care homes, in-home care (including nannies), and grandparents. As noted by Clarke-Stewart et al. (1994), these care arrangements differ in a myriad of ways, including their physical facilities, the number and kinds of materials, the flexibility of the daily schedule, the number of children in the arrangement, the child- adult ratio, and caregivers' education, training, and reasons for providing care. In comparison to child care homes and nannies, center-based care typically offers more highly educated caregivers, larger group sizes, more time spent in lessons, more structured activities, and more child-oriented materials, activities, and toys. Caregivers are more likely to have a more professional orientation and less likely to provide care as a favor for the family. In child care homes, children spend more time in free exploration, casual learning, and watching TV than in centers. Consistent with the more educational focus of the centers, Clarke-Stewart et al. (1994) found that center-based care was associated with higher scores on standardized cogni- tive assessments, controlling for family demographic characteristics and observed parenting. Children in center-based care also were more compe- tent with strangers and independent of mothers in a laboratory playroom. Research conducted by the NICHD Early Child Care Research Net- work (2000a, 2002b) also has considered the effects of center-type experi- ence on child developmental outcomes. Children who had more experience
122 WORKING FAMILIES AND GROWING KIDS in center-based care received higher cognitive and language scores at age 2 years (2000) and 3 years (2000), and higher language and memory scores at 41/2 years (2002), controlling for quality and quantity of child care and for family background characteristics. Effect sizes in these analyses ranged from 0.21 to 0.43, indicating expected differences in standard deviations on cognitive outcomes between children who attended centers and those who did not. The NICHD Early Child Care Research Network and Duncan (2003) report that children who attended centers between the ages 27 and 54 months (but not earlier) scored 4.1 points higher on cognitive tests than children who never attended centers during this period, controlling for family factors and previous child performance (effect size = 0.27). The developmental consequences of other relative care arrangements are more equivocal. Dana cares for her young grandson, and they spend much of their time watching television (Newman, 1999:203). Grandma Dana is not particularly attentive to Anthony's emotional needs, even though she keeps him fed and safe. He is never left alone, he does not run the streets, and his clothes are clean. Participation in child care homes does not appear to confer similar cognitive advantages. In a study of low-income families in the National Longitudinal Survey of Youth (NLSY), Yoshikawa (1999) found that num- ber of months in child care homes in the first five years of life was associ- ated with lower standardized math and reading achievement at ages 7 and 8, controlling for months in relative care and center care as well as family background characteristics. Evidence of center-type effects also was found in the area of social devel- opment. Children whose care settings contained more children were re- ported by caregivers to be more sociable but also to have more negative interactions with their peers (National Institute of Child Health and Human Development, 2001a, in pressd). Children's experiences in child care homes were less consistently related to cognitive and language development. An early study by Baydar and Brooks-Gunn (1991) is one of the few studies to consider explicitly the effects of grandmother care on child devel- opmental outcomes. In a study of white children in infancy whose mothers were employed, grandparent care was associated with fewer behavior prob- lems at age 4 years than children who attended child care homes and centers. Additional research is needed to clarify the conditions under which relative and other informal care arrangements are supportive of children's development.
EFFECTS OF CHILD CARE 123 EFFECTS OF QUANTITY OR AMOUNT OF CHILD CARE A third aspect of early child care that has been subject to considerable research is quantity or amount of child care (see Lamb, 1998, and Belsky, 1999, for reviews). A related literature has considered amount (and timing) of maternal employment (see Chapter 2 of this report and Brooks-Gunn et al., 2002). One of the most hotly debated issues is whether extensive hours in early child care (or extensive hours of early maternal employment) place children at risk. In a series of papers, Belsky (1986, 1988, 1999) argued that early and extensive hours place young children at risk for insecure relationships with mothers and heightened behavior problems, including aggression and noncompliance. Other scholars (Brazelton, 1986; Egeland and Heister, 1995) made similar arguments based on the need for mothers and infants to have sufficient time to build emotionally attuned relation- ships that serve as a basis for subsequent social competencies. These views were countered by other scholars, who argued that the effects of early and extensive care might be explained by other factors, such as: (a) the quality of the child care (Phillips et al., 1987b), (b) differences in family back- ground that accounted for different amounts of child care and differences in child developmental outcomes (Richters and Zahn-Waxler, 1990; Thomp- son, 1988), or (c) a failure to distinguish between avoidance and indepen- dence in assessments of infant attachment relationships and between assertiveness and aggressiveness in the assessments of older children (Clarke- Stewart, 1989). Much of the initial research that reported effects associated with quantity of care did not control for quality of care and had only limited controls for family selection. The NICHD Study of Early Child Care, initiated in 1991, afforded the evaluation of these alternative positions. Key elements included in the design of that study were consideration of the effects of early and extensive hours in a prospective longitudinal design that had: (1) a sufficiently large sample to detect effects; (2) robust measures of child care quality, type, and quantity; (3) extensive and repeated measures of family characteristics and processes that could be used to detect family characteristics associated with child care selection as well as family characteristics that changed as a func- tion of child care; and (4) a diverse set of child developmental outcomes that were assessed at multiple ages. The NICHD Early Child Care Research Network has considered the effects of amount and timing of child care on children's social and cognitive development, including attachment to mother (1997a, in pressd), behavior problems (1998, 2002c, in pressa), social competence (1998, 2002c, in pressb), and cognitive, language, and preacademic performance (2000b, 2002b). Infants with extensive child care experience did not differ from infants with little or no child care experience in their distress during separa-
124 WORKING FAMILIES AND GROWING KIDS tions from the mother in a strange situation (1997b). There were no significant effects of amount of care on attachment security at 15, 24, or 36 months (1997b, in pressb). Amount of care also was not related to children's cognitive, language, or academic performance assessed at 24, 36, or 54 months (2000b, 2002a). Amount or hours of care were related to children's social development at age 24 months, 54 months, and in kindergarten. In particular, children who had more hours of care had more behavior problems, according to their caregivers (NICHD, 1998), and were less socially competent, accord- ing to their mothers (2002a). In kindergarten, children who had more hours in care were reported by both mothers and teachers to have more externalizing problems and, by teachers, to have more conflict relationships (in pressd). This finding held when type of care was controlled. Examination of the proportions of children with substantial behavior problems (defined as one or more standard deviations above the mean) revealed that only the group of children in care for more than 45 hours a week displayed higher than expected rates of substantial problems: 19 percent of this group had elevated behavior problems, according to kinder- garten teachers, and 21 percent had elevated behavior problems, according to mothers. In contrast, 9 percent of the children who had been in care for 0 to 9 hours were reported by kindergarten teachers to have substantial behavior problems. By definition, on this normed instrument, 17 percent of the children are expected to score one standard deviation above the mean. Additional research is needed to identify the processes or mechanisms that mediate relations between quantity of care and behavior problems. By and large, in the NICHD analyses, effects were not attenuated when the positive caregiving composite (the ORCE measure of process) and maternal sensitivity were included in the regression analyses, suggesting that the quantity findings were not mediated by the quality of caregiving provided by child care providers or mothers, at least as measured by the study investigators. Further research is needed to consider other aspects of the child care environment beyond those reflected in the measure of process quality used in the NICHD study. For example, the specific strategies that caregivers use to promote children's social skills and to handle noncompli- ance and aggression or experiences with peers may help to explain the effects associated with quantity of care. Recent research by Watamura et al. (in press) found elevated cortisol levels at the end of the day when children were in centers all day, suggesting that the experiences were taxing for them. Other findings from this labora- tory (Dettling et al., 1999) have shown that the largest increases in cortisol over the course of the child care day were observed in children who had the most difficulty regulating their negative emotions and behavior. Other issues warranting additional study pertain to the timing of early
EFFECTS OF CHILD CARE 125 child care. In general, reports from the NICHD study have found cumula- tive indicators of child care quantity are stronger predictors of child devel- opmental outcomes than age-segmented predictors (see NICHD, 2000a). One exception is that kindergarten teachers reported higher levels of behav- ior problems for children in care for more hours in the first 6 months of life, controlling for amount of care in later time periods. A recent study by Brooks-Gunn et al. (2002) also found age-specific relations in analyses involving the NICHD dataset. In analyses of European-American children in the sample, children whose mothers were employed for 30 or more hours a week by 9 months had lower preacademic skills at 36 months than children whose mothers worked less than 30 hours a week. These effects were not evident in the children's cognitive performance at 15 or 24 months or in the cognitive performance of ethnic minority children at 15, 24, or 36 months. The findings of associations between extensive maternal employ- ment in the first year and lower cognitive performance of European-Ameri- can children are, however, similar to findings from the Child Supplement of National Longitudinal Survey of Youth (Baydar and Brooks-Gunn, 1991). These findings suggest that amount as well as quality and type of child care need to be considered in relation to child developmental outcomes. While children appear to benefit cognitively (and perhaps socially) from high-quality child care and from center-type experiences, extensive hours in child care are associated with increased problem behaviors. Additional research is needed to determine why extensive hours are related to problem behaviors. EFFECTS OF CHILD CARE ARRANGEMENTS DURING MIDDLE CHILDHOOD Because parents' workdays are typically longer than the school day, needs for child care do not disappear when children begin elementary school. Families of school-age children have adopted a variety of strategies to cover the nonschool hours when parents are at work, including self-care, before- and after-school programs, extracurricular activities, and informal care by sitters and relatives. These different care arrangements are often used in combination, and children move from one type of care to another in the course of an afternoon and across the week. These arrangements vary in their opportunities for children to engage in activities that they enjoy and care about (Larson, 2000), to develop physical, social, and cognitive skills (Larson, 1994), and to be with friends, adult mentors, and parents (McLaughlin et al., 1994; Posner and Vandell, 1994). For example, children who attend after-school programs have a chance to spend more time in academic enrichment, arts, and sports activi- ties, whereas children in informal settings spend more of their after-school
126 WORKING FAMILIES AND GROWING KIDS hours watching television and hanging out with friends (Posner and Vandell, 1999). Extracurricular activities and lessons offer opportunities for sub- stantial engagement in activities that children care about. In this section, we consider research findings pertaining to effects of different types of before- and after-school care on developmental outcomes during middle childhood. Self-Care Self-care, sometimes called latchkey care (Steinberg, 1986; Woods, 1972), refers to various unsupervised circumstances, including children be- ing home alone, being cared for by older siblings, providing care for younger siblings, and hanging out with unsupervised peers (Belle, 1997; Galambos and Maggs, 1991; Vandell and Su, 1999). The notion of self-care stirs mixed reactions. The goal for many parents in the United States is for children to become independent and capable of functioning without par- ents or other adults directly supervising their activities. Toward this end, children often experience a gradual transition from direct parental supervi- sion to self-care. At the same time, it is clear that young children lack the maturity and judgment to care for themselves. Thus, an important issue is how children and families navigate the transition from close supervision to independent self-care and a determination of the circumstances under which self-care is beneficial for children's development and the conditions under which it is detrimental. Self-care is predicted by a number of child, family, and community factors. It is more likely to be used by older children versus younger children and by children who previously exhibited fewer internalizing and externalizing problems. White children are more likely than black children and Hispanic children to be in self-care (Capizzano et al., 2000b). Self-care also more likely if mothers are employed (Smith, 2002) and family incomes are higher (Capizzano et al., 2000b), reflecting perhaps the greater avail- ability of relatives and other adults in lower income households and greater concerns about the dangers of leaving children alone in low-income neigh- borhoods. Self-care is more common in suburban and rural areas than in urban areas (Hofferth et al., 2000) and more likely when parents and children perceive their neighborhoods to be safe places (Vandell and Posner, 1999). Evidence of relations between self-care and children's developmental outcomes indicates that these associations vary depending on: (1) the child's age and previous functioning, (2) family characteristics, (3) neighborhood characteristics, and (4) the amount and type of self-care (for reviews, see Vandell and Shumow, 1999; Vandell and Su, 1999; Powell, 1987). Self- care appears to be more problematic when combined with such child fac-
EFFECTS OF CHILD CARE 127 tors as previous behavior problems (Pettit et al., 1997, 1999), such family factors as poverty (Marshall et al., 1997; Pettit et al., 1997) and low paren- tal monitoring (Galambos and Maggs, 1991; Steinberg, 1987; Pettit et al., 1999), and unsafe neighborhoods (Pettit et al., 1999). Retrospective data on Anthony Hayes, "a high achiever," illustrates how children assume adult-like responsibilities when their parents work (Clark, 1983:67-68): We all had responsibilities even when I was in second or third grade. I had my own door key. My mother and father would go to work and it was up to us to come home and do what we were supposed to do. And during lunch periods, I came home and ate lunch and went back to school. When school was over, I came home and cleaned up and did whatever I had to do. Pettit and colleagues (1997) found some longer term effects of early self-care. Children whose mothers retrospectively reported more unsuper- vised care (alone or with a sibling) in 1st and 3rd grade were less socially competent in 6th grade, according to teacher reports. They also received lower grades and achievement test scores in comparison to their classmates who had experienced less self-care, even after controlling for family charac- teristics and children's functioning in kindergarten. Self-care (alone or with siblings) in 5th grade was not related to the children's functioning in 6th grade, consistent with the proposition that these forms of self-care are more problematic for younger children than for young adolescents. Pettit et al. also found interactions between self-care and previous child adjustment and between self-care and family income. The highest levels of behavior problems in 6th grade were evident in those children who had extensive self-care in 1st grade as well as high levels of behavior problems in kinder- garten. The combination of low family income and early self-care also predicted higher levels of behavior problems in 6th grade. In a subsequent report, Pettit and colleagues (1999) distinguished among three forms of unsupervised care during 6th grade--time with unsu- pervised peers, time alone, and time with siblings. Children who spent more time with unsupervised peers in 6th grade displayed more externaliz- ing problems in 7th grade, controlling for family background factors and 6th grade behavior problems. The greatest risk was found for unsupervised children who were less closely monitored by parents and who lived in less safe neighborhoods. Time alone, time with unsupervised siblings, and time with supervised peers during 6th grade did not predict externalizing prob- lems in 7th grade.
128 WORKING FAMILIES AND GROWING KIDS McHale et al. (2001), however, have reported evidence that both time alone and time with unsupervised peers is related to problematic develop- ment during middle childhood. Children who spent more time alone at ages 10 and 12 reported more depression than children who spent less time alone. Children who spent more time with unsupervised peers, in contrast, had lower grades and more externalizing behavior problems. After-School Programs Increases in maternal employment, beliefs about children's needs for supervision and enrichment during the nonschool hours, stories in the popu- lar press about the negative effects of self-care, and concerns about lagging academic achievement in children who are growing up in poverty have contributed to the substantial growth in after-school programs (Vandell and Su, 1999). These programs are housed at schools, community centers, and child care centers. Historically, school-based programs and day care centers have been funded by parental fees and served children of middle- income families, whereas community centers historically served children of low-income families (Halpern, 2002). The 1990s have been marked by a substantial increase in after-school programs serving children of low-income families. A number of program models have emerged, including a range of program activities such as com- munity service, academic enrichment, recreation, arts, mentoring, and child care.2 One of the best known programs is the 21st Century Community Learning Centers (CCLC), a school-based after-school program initially administered by the U.S. Department of Education. Funding for the pro- gram grew from $40 million in 1997 to $1 billion in 2002. In 2001, 1.2 million elementary and middle school students participated in programs located in 3,600 schools. Even with the growth in programming, a General Accounting Office (GAO) study estimates that as little as 20 percent of the demand for pro- grams is met in urban areas (U.S. General Accounting Office, 1997). It is estimated that only about one-third of the demand for programs is being met in rural areas (Larner et al., 1999). The GAO report and other studies (Mezey et al., 2002) indicate shortages of care in certain critical areas, including for infants and toddlers, children with special needs, older school- age children, and children of families working nonstandard hours. While the GAO study predates the expansion of child care opportunities available 2 Descriptions of many of these federal, state, and local after-school initiatives can be found at http://www.gse.harvard.edu/hfrp/projects/afterschool/mott/mott1.html.
EFFECTS OF CHILD CARE 129 through the 21st Century Community Learning Centers, in the 2000 com- petition for funding for these programs, 2,252 communities sought funds to establish or expand after-school programs, but funds were available to support only 310 grantees. Evidence pertaining to the effects of after-school programs on children's developmental outcomes is mixed, with some studies reporting positive effects (Grossman et al., 2002; Marshall et al., 1997; Pettit et al., 1997; Posner and Vandell, 1994; Vandell and Corasaniti, 1988; Welsh et al., 2002), other studies reporting no effects (Pettit et al., 1997), and still others reporting negative effects (Vandell and Corasaniti, 1988). Findings are related to family factors (Posner and Vandell, 1994; Vandell and Corasaniti, 1988; Marshall et al., 1997), program quality (Pierce et al., 1999), and dosage (Cosden et al., 2001; Vandell and Pierce, 1999). After-school pro- grams have been more consistently associated with positive effects for chil- dren from low-income families than for children from middle-income fami- lies (Grossman et al., 2002; Marshall et al., 1997; Pettit et al., 1997; Posner and Vandell, 1994; Vandell and Corasaniti, 1988; Welsh et al., 2002) and for children whose parents have limited English proficiency (Cosden et al., 2001; Welsh et al., 2002). In a large-scale evaluation of 96 programs serving low-income students in New York City (25,909 program participants and 39,780 students who did not participate in a program), Welsh et al. (2002) reported that low- achieving students, black students, Hispanic students, and English language learners were especially likely to benefit from active participation in the programs, as evidenced by greater gains in math achievement relative to their peers. In the Boston After-School Study, Marshall et al. (1997) found that children of low-income families (but not middle-income families) had fewer behavior problems if they regularly attended after-school programs. Similarly, in a study conducted in a context of low family income and unsafe neighborhoods, Posner and Vandell (1994) found that children who attended after-school programs had fewer antisocial behaviors and better reading and math grades, work habits, emotional adjustment, and peer relationships than children who were in self-care, sitter care, or parental care after school. The after-school programs appeared to serve as a safe haven for children in neighborhoods in which crime rates are high and unsupervised time after school exposed them to deviant peers and violence. The evaluation of the Extended Services School Initiative (Grossman et al., 2002) found changes in school engagement for students who regularly participated in after-school programs in comparison to students who par- ticipated less regularly in the programs. Controlling for students' baseline performance in the outcomes of interest and family background character- istics such as family income, household structure, parental education, and the quality of the parent-child relationships, students who regularly at-
130 WORKING FAMILIES AND GROWING KIDS tended the after-school programs were more likely to show positive changes in school engagement and attentiveness in class and less likely to start skipping school and drinking alcohol in comparison to students who at- tended programs less regularly. As noted by the study authors, a limitation of the dose/response analysis strategy is that participation patterns may have been the result of unmeasured factors that might be the result of self- selection, not program participation. As is the case in early child care, there is wide variation in the quality of after-school programs. In terms of structural-caregiver characteristics, child-staff ratios ranged from 4 to 1 to 25 to 1 in the National Survey of Before- and After-School Care (Seppanen et al., 1993). Staff education ranged from less than a high school diploma through graduate degrees. Although some programs reported no staff turnover during the previous year, turnover averaged 60 percent a year. Consistent with the model out- lined in Figure 5-1, effects on school-age children are related to the quality of the after-school programs. Structural-caregiver characteristics predict process quality. When child-staff ratios were higher, staff appeared more negative and hostile toward the children in the program (Rosenthal and Vandell, 1996). Staff were warmer, more sensitive, and more supportive in programs in which child-staff ratios were lower (Pierce et al., 1999). Chil- dren also spent less time waiting and in transition and more time interacting positively with staff when child-staff ratios were lower. Staff education was also associated with observations of process quality. In programs in which staff were more highly educated, staff members used more positive behav- ior management strategies and were less harsh with children (Pierce et al., 1999; Rosenthal and Vandell, 1996). Variations in process quality in after-school programs, in turn, predict child developmental outcomes (Pierce et al., 1999). Boys who attended after-school programs in which there was a positive emotional climate were reported by their 1st grade teachers to exhibit fewer problem behaviors at school in comparison to boys who attended programs with less positive climates. More negative emotional climate in the after-school programs was related to boys' poorer academic performance at school. Boys who attended programs rated as fostering autonomy and choice among activities had better social skills, according to their 1st grade teachers. Amount or dosage of program experience also is related to program effects (Cosden et al., 2001; Vandell and Pierce, 1999; Welsh et al., 2002). Educationally at-risk students who attended a three-year homework club for more sessions scored higher on reading, math, and language achieve- ment tests than children who participated less consistently (Cosden et al., 2001). Welsh et al. (2002) used a quasi-experimental design to examine program effects associated with participation in The After-School Corpora- tion (TASC) programs. Changes in reading and math achievement for
EFFECTS OF CHILD CARE 131 highly active participants (n = 12,973), active participants (n = 17,805), nonactive participants (n = 8104) and nonparticipants (n = 39,870) were examined. Students who were active participants in TASC programs for more than a year showed significantly greater gains in math achievement than did similar nonparticipating classmates. In a smaller scale study of four after-school programs located in high-crime neighborhoods, children who attended after-school programs for more days during the school year demonstrated improvements in their academic grades and work habits, whereas the performance of children who attended the programs for only a few days did not improve (Vandell and Pierce, 1999). Interestingly, many of these programs have an enrichment focus and are not limited to tutoring and homework help. In an effort to evaluate the implementation and impact of after-school programs supported by 21st Century Community Learning Center funds, the U.S. Department of Education and the Charles Stewart Mott Founda- tion provided support for a two-year evaluation of several of the CCLCs across the nation. The evaluation was conducted by Mathematica Policy Research, Inc. The initial evaluation report, released on February 3, 2003, describes first year findings from samples of elementary and middle school students assessed during the 1999-2000 school year (U.S. Department of Education, 2003). The report purports that the CCLCs had little impact on the academic or social behavior of the participants. There are, however, several notable limitations with the National CCLC Evaluation. For instance, the elementary school sample involved a small number of sites that agreed to have students randomly assigned to participate or not to participate in the CCLC's after-school programs. These schools were not representative of the larger population of elementary schools receiv- ing CCLC funds. In the middle school sample, a matching design was used to compare after-school program participants and nonparticipants. The match- ing, however, was based on limited information about the students at the initial assessment, and the resulting comparison groups were dissimilar. In particular, the after-school participant group showed heightened risk at baseline in several areas, including markedly lower achievement test scores, more behavior problems, and greater socioeconomic disadvantage. In light of these initial differences, it is interesting to note that by the end of the school year the program participants and nonparticipants were reported to have similar levels of academic and social competence. The absence of certain baseline data, treatment and comparison group contamination, and issues surrounding the evaluation's timing and measurement are also methodologi- cal concerns in the National CCLC Evaluation. An ongoing debate in the after-school arena is how best to organize and structure programs, and it centers on how academically oriented programs should be. Some contend that programs should emphasize homework help,
132 WORKING FAMILIES AND GROWING KIDS tutoring, and preparation for mandated tests. Others contend that pro- grams should emphasize extracurricular enrichment activities. Still others have argued that after-school programs should provide a safe place for youth to relax and hang out. The effects of these different approaches (or hybrids of these approaches) to after-school programming have not been systematically evaluated. Research is needed to determine if these ap- proaches are differentially associated with improvements in school atten- dance, student achievement, emotional well-being, positive youth develop- ment, and decreases in problem behaviors. Structured Voluntary Activities Another source of supervised experiences for children during the after- school hours is structured activities, a term that encompasses lessons and extracurricular activities, such as piano lessons, coached sports, and scouts. Structured activities are typically funded by fees, which are paid by partici- pants. Consequently, it is not surprising that children of higher income families are more likely than children of lower income families to partici- pate in these activities. According to the National Child Care Survey, 20 percent of the children in families with incomes over $50,000 (1990 dol- lars) are enrolled in lessons, whereas only 6 percent of children whose families earned between $15,000 and $25,000 have these experiences (Miller et al., 1997). Updated data from the 1999 SIPP reports indicated a similar discrepancy, with 8 percent of youth (ages 6 to 14) whose families earned less than $18,000 a year being reported to participate in lessons, clubs, and sports, whereas 20.4 percent of the youth whose families earned more than $54,000 a year were reported to participate in these activities Most of the research examining the effects of structured activities was conducted with adolescents (see Chapter 6). However, indications are that these types of activities also benefit school-age children (ages 6 to 12). Controlling for child prior performance, ethnicity, and gender as well as family demographics and parenting, children who consistently participated in extracurricular activities during kindergarten and 1st grade obtained higher reading and math scores at the end of 1st grade than children who sometimes or never participated in extracurricular activities (NICHD Early Child Care Research Network, 2002b). For the most part, the children were not "overprogrammed." Children typically participated in a single activity for less than three hours each week. Few children (less than 4 percent) were involved in extracurricular activities for more than five hours each week. Pettit et al. (1997) also have found moderate amounts of structured activities to be beneficial. Children who engaged in structured activities for one to three hours a week in 1st grade were more socially competent in 6th
EFFECTS OF CHILD CARE 133 grade than children who had either no structured activities or high amounts of these activities (more than four hours a week) in 1st grade, controlling for family factors and child behavior in kindergarten. In addition, extracurricular activities have been related to functioning in older school-age children. Time spent in extracurricular time during 3rd, 4th, and 5th grades predicted children's emotional well-being in 5th grade, controlling for emotional adjustment in 3rd grade and family background (Posner and Vandell, 1999). Time spent in sports activities and hobbies was associated with fewer depressive symptoms at age 10 and age 12, controlling for family factors (McHale et al., 2001). Larson's (2000) research suggests why structured activities may be ben- eficial for children. In his studies, adolescents were more likely to report concentrated effort and intrinsic motivation during structured activities, which Larson posits is particularly conducive to the development of initia- tive. In contrast, adolescents report low concentration but high choice while they are hanging out and high concentration and low choice while at school. SUMMARY Conceptual and methodological advances have contributed to a sub- stantial research literature that has considered the effects of child care quality, quantity, and type on children's developmental outcomes. This research literature has begun to specify the conditions in which child care can enhance positive developmental outcomes for children as well as the conditions in which it can be problematic and associated with poorer devel- opmental outcomes. These findings are based on several multisite projects as well as single-site studies that have utilized psychometrically strong mea- sures of child care. Cognitive, language, social, and behavioral outcomes have been assessed. Efforts to address concerns about selection bias and omitted variables have resulted in expanded lists of family factors, analyses of change scores, and controls for prior child performance. Studies of the effects of early child care quality have considered both process quality and structural and caregiver characteristics. Process quality refers to the experiences that children have with their caregivers, with other children, and with age-appropriate activities and materials, and structural and caregiver characteristics refer to such factors as child-adult ratio, the number of children in the class or group, the amount of formal education that caregivers have, caregivers' specialized training related to children, and caregivers' wages. Consistent with the conceptual model shown in Figure 5-1, many of the studies reviewed have found that: (1) structural and caregiver characteristics predict observed process quality;
134 WORKING FAMILIES AND GROWING KIDS (2) process quality predicts children's cognitive, language, and social competencies both concurrently and over longer periods of time; and (3) structural and caregiver characteristics predict children's cognitive, language, and social competencies. When process quality is higher, children display better performance on a range of cognitive, language, and social assessments. By the same token, when process quality is lower, children demonstrate poorer performance in these areas. Relevant to these findings is evidence from the NICHD study of early child care that almost 60 percent of the observed child care settings in that study were of either poor or fair quality, rather than of good or excellent quality. In comparison to their higher income peers, children of low-income families appear more likely to receive poor-quality child care and less likely to receive excellent quality child care, especially in the early years. The available research indicates that type of child care arrangement also is related to children's developmental outcomes. Both experimental and correlational studies have found that center-type experiences are asso- ciated with higher scores on cognitive and language assessments, particu- larly for 3- and 4-year-olds. Experience in child care homes and relative care, in contrast, has been less consistently related to cognitive and lan- guage performance and does not appear to convey the same benefits for preschool-age children. Quantity or amount is the third aspect of early child care that has been studied in relation to child developmental outcomes. In several studies, children who are in care for more hours per week are reported to have more behavior problems than children who are in child care for fewer hours a week. Substantial behavior problems, defined as a score of one or more standard deviations about the mean on a normed measure of problem behaviors, have been reported by both teachers and mothers for the group of children who averaged more than 45 hours a week of care up to age 41/2 years. Rates of substantial behavior problems are less than expected for children who averaged less than 10 hours per week and are at the norm for children who averaged 10 to 45 hours a week. There is some evidence that more extensive hours in the first 6 to 9 months of life is associated with heightened behavior problems and lower cognitive scores for some groups of children. Additional research is needed to determine the processes or mechanisms that mediate these effects. Studies of the effects of child care on school-age children (6- to 12-year- olds) have considered three types of before- and after-school care: self-care, programs, and extracurricular activities. This research literature is less extensive than the available evidence pertaining to early child care. None- theless, findings have emerged that begin to delineate the conditions in which the child care experiences during middle childhood serve as positive
EFFECTS OF CHILD CARE 135 influences on developmental outcomes and conditions in which experiences are problematic. These findings are consistent with those reported for young children in that three aspects of care--quality, type, and amount-- are important considerations. Effects also appear to vary by family circum- stances, neighborhood safety, and children's previous adjustment. Self-care is increasingly common for children as they move through middle childhood. Self-care is not associated with poor academic or social outcomes if it occurs in limited amounts, takes place in safe neighborhoods, is accompanied by parental monitoring, and is used by competent children who are emotionally ready for the experience. Self-care appears more prob- lematic when used by younger school-age children, when unsupervised time is spent in the company of peers, when neighborhoods are unsafe, when children have previous behavioral problems, and when families have low incomes. The effects of after-school programs on children's developmental out- comes also are variable. Positive effects on children's emotional well-being, academic performance, and peer relationships are more evident when pro- gram quality is high and the children attend regularly. Children of low- income families and children who reside in unsafe neighborhoods appear to derive greater benefit from participation in after-school programs than do children of more affluent families, perhaps because their opportunities to participate in fee-based extracurricular activities and lessons are so limited. An issue that requires additional research is the determination of how best to organize and structure after-school programs to meet the developmental needs of school-age children. Studies that have examined the effects of voluntary structured activities (i.e., extracurricular activities) on child developmental outcomes suggest a model for after-school programs that may be particularly effective. Time spent in extracurricular activities, such as organized sports, music, and art, is associated during middle childhood with positive developmental out- comes, including higher grades and higher standardized test scores, even when family factors and previous child adjustment are controlled. These findings are consistent with findings that extracurricular activities during adolescence (discussed in Chapter 6) were particularly conducive to posi- tive youth development. In summary, the quality of child care is likely to have important conse- quences for the development of children during the early years and middle childhood. The research presented in this chapter shows that the impact of child care quality on child development depends on such variables as the activities children experience in care, caregiver training and education, type of setting, and amount of time in care. ANNEX TO CHAPTER 5 FOLLOWS
136 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Relations Between Structural-Caregiver Characteristics and Process Quality Structural-Caregiver Citationa N Type of Care Variables Arnett (1989) 59 Centers CG training: (1) no training; (2) two courses Bermuda college; (3) four-course training program; (4) four-year college degree in ECE Berk (1985) 37 Centers CG formal education and CG specialized training Blau (1997) 5 cities; 204 centers; Centers 37 center characteristics 1,094 teachers; 567 including group size, classrooms ratio, caregiver (reanalysis of the education, different National Child Care types of training Staffing Study) Blau (2000) 548 classrooms Centers Group size, ratio, (reanalysis of CQO CG experience, job data) tenure, ethnicity, formal education, specialized training Burchinal, Total = 244 Family child care CG education, formal Howes, and Florida Child Care and informal training Kontos (1999) Study = 144 experiences, experience California Licensing as a child care provider, Study = 100 group size, business practices Points (sum of number of children, weighted by age of children)
EFFECTS OF CHILD CARE 137 Process Qualityb Analysis Findings Parental Modernity ANCOVA CG with half or all the Bermuda Scale, CIS (positive College training less authoritarian in interaction, childrearing attitudes than CG with no punitiveness, training, rated higher on positive detachment, interaction and lower on detachment in permissiveness) interactions with children. CG group with 4-year ECE degree differed from other 3 groups: childrearing attitudes less authoritarian, interactions with children rated higher on positive interaction and lower on punitiveness and detachment. Observations of ANOVAs and College-educated caregivers had more caregiver behavior correlations encouraging behaviors, more suggestions, less restrictive actions. Ratings on Arnett Ordinary least OLS regressions indicate that formal Scale (sensitivity, squares regressions; education associated with caregiver harshness, detached), fixed effects sensitivity and appropriate caregiving. appropriate analyses Effects were reduced and caregiving, nonsignificant in many of the FE developmentally analyses. appropriate activities ECERS, ITERS Pearson Simple correlations and regressions correlations, that did not include the fixed effect-- regressions with center control found lower group size, and without a fixed lower C:A ratio, and more CG training effect control for to be related to better ECERS scores. center ID These relations were substantially reduced when the center fixed effect control was added to the model. FDCRS, CIS Pearson CG education and experience better correlations predictors of CC quality than C:A ratios. CG with more education more sensitive and rated higher on global quality. More experienced CG slightly more detached and provide lower-quality care. Regression CG with more education tended to have settings with higher global quality ratings. (continued)
138 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables Burchinal, 79 Centers Director and observer Roberts, reports of group size Nabors, and and C:A ratio; teacher Bryant (1996) report of training and experience Burchinal et 27 Centers C:A ratio, CG education, al. (2000) group size Clarke- 15 months = 133 Child care homes Group size, group size Stewart et al. 24 months = 146 points, CG education, (2002) 36 months = 131 amount of specialized training, recent training Dunn (1993) 30 Day care centers CG education, child major, training, center experience, field experience, CG age, group size, C:A ratio, ECERS
EFFECTS OF CHILD CARE 139 Process Qualityb Analysis Findings CG experience was negatively related to observed quality in the licensed Family Child Care Study. Group size or ratio not related to observed quality of care. ITERS Pearson Higher observed and reported C:A correlations ratios were associated with lower ITERS scores. Higher CG training was associated with higher ITERS scores. ITERS, ECERS Pearson Higher C:A ratios were related to Correlations lower global quality at 12, 24, and 36 months. Higher group size was related to lower global quality at 24 and 36 months. Higher teacher education was related to higher global quality at 12 and 36 months. ORCE-Positive Correlations Both correlational analyses and HLM Caregiving analyses indicated overall quality of CC-HOME care measure by CC-HOME and by ratings of observed CG, behavior was higher when CG was more highly educated, had more specialized training pertaining to children, and had HLM received training in the past year, with the strongest effects evident at 36 months. CG exhibited more positive caregiving when age-adjusted group sizes were smaller. Play space, variety, Pearson CG with more experience in the field divergent/elaborative correlations and larger group sizes was positively interact, praise/ related to higher ECERS scores. nurturance/ Larger group size was positively redirection, clear related to more variety in classes. limits, total limits Higher ECERS scores were related to more divergent/elaborative interactions and less total limits. (continued)
140 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables Dunn et al. 30 Day care centers Group size, C:A ratio, (1994) CG education, CG experience in field, CG experience in centers, CG certification Elicker, 23 Family day care Group size, C:A ratio Fortner-Wood, and Noppe (1999) Goelman 74 Center day care Caregiver education (1988) Family day care Holloway and 15 Preschools and Group size, C:A Reichhart- day care centers ratio Erickson (1988) Howes (1983) 40 Center day care C:A ratio, group size, and family day number of adults, CG care years experience, training child development
EFFECTS OF CHILD CARE 141 Process Qualityb Analysis Findings Language/reasoning Pearson Only one structural quality variable (ECERS), correlations, correlated with quality of developmentally simultaneous environment. CG who held some form appropriate activities regression of teacher certification provided (ECERS), variety, classes rated higher on literacy quality literacy activities, scale. literacy quality Caregiver-Infant Pearson Smaller group size and fewer children Involvement-AQS correlations per adult more CG-child involvement. CG years of experience, CG educational level, income, overall work satisfaction, work-related stress, control over work schedule, work and family conflict not significantly correlated with CG-child involvement or infant-CG attachment. Learning activities, Pearson Higher CG education correlated with social development, correlations higher total quality scores in both language development, family day care and center day care. creative activities, total quality ECOI Pearson correlations Smaller group sizes were related to higher ratings on the Interaction Quality Composite and accommodation of varied groups. C:A ratio was not related to any ECOI Indicators. CG behavior Pearson Caregivers in both settings with fewer (facilitative social, correlations children in their care, who worked express positive shorter hours, with less housework affect, negative responsibilities engaged in more affect, restrictiveness, facilitative social stimulation, responsivity) expressed more positive affect, were more responsive, and less restrictive and negative. Family day care caregivers who worked in spaces specifically designed to be safe and appropriate for children were less restrictive of toddler activity. (continued)
142 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables Howes (1997) Total = 1,065 Child care C:A ratio, CG Cost, Quality, centers education, CG ECE Outcome Study training (CQOS) = 655 Florida Quality Improvement Study (FQIS) = 410 Howes, 143 Child care C:A ratio, group size Phillips, and centers Whitebook (1992) Howes and Total = 50 Home, center C:A ratio, group size Rubenstein Home = 23 day care, family (1985) Center day care = 11 day care Family day care = 16 Howes and 150 Child care CG characteristics Smith (1995) centers (years of education + specialized training in ECE), C:A ratio, group size Howes, 1,300 Child care CC experience, Phillips and centers specialized training, Whitebook, education (1992)
EFFECTS OF CHILD CARE 143 Process Qualityb Analysis Findings CIS, AIS ANOVA CQOS: CG with BA or higher degree in ECE rated more sensitive than CG with AA degrees in ECE, who were more sensitive than CG with other backgrounds. CG with at least AA degree less harsh than CG in other backgrounds. CG in classes in compliance with ratio standards rated more sensitive, less harsh, and less detached. FQIS: CG with at least BA in ECE rated more sensitive than CG with CDA training who were rated as more sensitive than all other CGs. Caregivers with most advanced education most effective. Appropriate Chi-square Classrooms with higher child:adult caregiving, ratios were more likely to be rated as developmentally inadequate in caregiving and activities. appropriate activities Large group sizes also were more likely to be rated as inadequate in caregiving and activities. Smaller group sizes were more likely to be rated as developmentally appropriate. Caregiver-child Pearson Lower C:A ratio was associated with interaction (talk correlations, higher quality of CG-child interactions. and play, one-way ANOVA restrictiveness and cry, touch and laugh) ITERS, ECERS Pearson Classes with more educated and trained correlations teachers had higher ITERS and ECERS scores. ECERS, ITERS, Pearson CG formal education and specialized Arnett teacher correlations, training was associated with behavior. sensitivity measure multiple regression Formal education better predictor than specialized training. Infant/toddler CGs need more college-level specialized training than preschool teachers to be competent teachers. (continued)
144 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables Iutcovich Total = 675 Center, group CG education, CG years et al. (1997) Center = 561 home, family in field, CG salary, CG Group home = 70 long-term educational Family = 44 goal, training characteristics, organizational climate Kontos, Training group = 130 Family day care CG training, C:A ratio, Howes, and Regulated providers group size Galinsky = 112 (1997) NICHD Early 576 Center, child care Group size, C:A ratio, Child Care homes, in-home physical environment Research sitters, CG characteristics Network grandparents, (formal education, (1996) fathers specialized training, child care experience, beliefs about childrearing) NICHD Early 612 Center, child-care C:A ratio, group size, Child Care homes, in-home CG education, CG Research sitters, specialized training, Network grandparents, CG beliefs, (2000a) fathers CG experience Phillipsen, Total = 749 Child care CG background Burchinal, 228 = I/T centers (educational level and Howes, and 521 = P experience), class Cryer (1997) structure (C:A ratio)
EFFECTS OF CHILD CARE 145 Process Qualityb Analysis Findings ITERS, ECERS, Pearson Higher CG salary higher ITERS and FDCRS correlations ECERS scores. Younger CG, CG with more long-term educational goals, evaluating appropriateness, and evaluating usefulness higher FDCRS scores. CGs with higher long-term educational goals, more likely to evaluate appropriateness and usefulness higher FDCRS scores. CG higher ratings of professional growth, clarity, reward system, goal consensus, and task orientation higher ECERS scores. Process quality: Chi square, Training group and comparison group Arnett Scale of t-test were similar on structural, process, Provider Sensitivity, and global quality. Providers in Adult Involvement comparison group cared for slightly Scale more children per adult than training Global quality: group. FDCRS Effects of training no changes on Arnett scale but increases on the FDCRS. ORCE: caregiver Pearson Caregivers rated as providing more interactions correlations and positive caregiving when group sizes multiple regression and C:A ratios were smaller and when analyses (backward CG held less-authoritarian beliefs elimination about childrearing. Seen in all five procedure) types of care. ORCE (positive Pearson Positive caregiving ratings higher when caregiving frequency, correlations and CG had more child-centered beliefs positive caregiving multiple (all ages), higher levels of education, rating) regressions and more experience providing care Global quality (at 24 and 36 months), and more rating specialized training (15 months), and when lower C:A ratio and smaller group sizes (15 and 24 months). ITERS, ECERS, TIS, MANOVA Structural measures predicted process CIS Hierarchical quality more strongly in preschool regressions than in infant/toddler classes. Infant/ toddler process quality higher in (continued)
146 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables (C:A ratio and group size) CG education x A:C ratio, lead CG wages, center structure, direct background, economic characteristics of center, state, and sector Ruopp, National Day Care Centers C:A ratio, group size, Travers, Study = 57 CG years of education, Glantz, and preschool day care child-related training, Coelen (1979) centers in 3 cities; education, physical Infant/Toddler environment Substudy = 74 caregivers in 38 centers Scarr, 363 Child care C:A ratio, group size, Eisenberg, centers CG training in child and Deater- development and child Deckard (1994) care, CG education, highest wage paid to a CG in the center, staff turnover Stallings and 303 Child care Observed C:A ratio Porter (1980) homes, included sponsored, regulated, and unregulated homes
EFFECTS OF CHILD CARE 147 Process Qualityb Analysis Findings classes with moderate experience and better paid teachers and more experienced directors. Preschool process quality higher in classes with CG having more education, moderate amount of experience, and higher wages. Better C:A ratios, lower center enrollment, and lower proportion of infant/toddler and subsidized children in center also predicted higher process quality for preschool. Teacher wages strongly related to process quality in infant/toddler classes. Caregiver behaviors, Multiple For 3-, 4-, and 5-year-olds, smaller including regression groups consistently related to more management, social socially active children. Smaller C:A interaction; child ratios = less time in child behavior aimless wandering management. More child-related training associated with more teacher-child interaction. For children < 3 years: group size and ratio are strongly related to process quality. ITERS, ECERS, Pearson Highest CG wages were highly APECP correlations correlated with process measures of quality (ITERS/ECERS and Profile Score). Lower C:A ratios, more teacher education, and more teacher training were correlated with higher process measures of quality, however, less correlated with process quality criteria. Specific caregiver Pearson Higher child:adult ratios associated behaviors including correlations with less caregiver teaching, playing teaches, plays, Multiple with child, and facilitating child directs, converses, regressions activities, and more efforts to control comforts, tends to child behavior. Relations stronger physical needs, when children < 35 months. housekeeping, not involved (continued)
148 WORKING FAMILIES AND GROWING KIDS TABLE A5-1 Continued Structural-Caregiver Citationa N Type of Care Variables Stith and 30 Employed Group size Davis (1984) mothers, substitute CG unemployed moms Vandell and 53 Center Structural composite Powers (C:A ratio and toys (1983) accessible + CG education + space allotment per child) Volling and 36 Center Group size Feagans C:A ratio (1995) a For full references, see the report reference list. b Process quality measures alphabetized by acronym: AIS: Adult Involvement Scale; APECP: Assessment Profile for Early Childhood Programs; AQS: Attachment Q-Set; CIS: Caregiver Interaction Scale; ECERS: Early Childhood Environment Rating Scale; ECOI: Early Child- hood Observation Instrument; FDCRS: Family Day Care Rating Scale; HOME: Home Obser- vation for Measurement of the Environment; ITERS: Infant/Toddler Environment Rating
EFFECTS OF CHILD CARE 149 Process Qualityb Analysis Findings Yarrow, Rubenstein, Pearson Larger group sizes less expression of and Pedersen's correlations positive affect and less contingency of (1975) infant responses to distress. environment observational scale Positive and negative ANOVA Higher quality structural composite behavior with adults, related to more positive interactions total adult-directed with adults. behavior Positive adult-child Pearson Smaller group sizes and C:A ratios interaction, correlations related to more time in positive staff- nonsocial activity, child interactions and less time in positive peer nonsocial activities. Larger C:A ratios interaction, negative related to more frequent negative peer interaction interactions with peers. Scale; ORCE: Observational Record of the Caregiving Environment; TIS: Teacher Involve- ment Scale; CG = caregiver; ECE = early childhood education; C:A ratio = child:adult ratio; CC: child care. NOTE: Vandell and Wolfe (2000) served as the source for studies prior to 2000.
150 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Concurrent Associations Between Child Care Quality and Child Developmental Outcomes Process Quality Structural Quality Citationa N Age Measureb Measureb Burchinal, 79 12 months ITERS Group size, C:A Roberts, ratio, CG training, Nabors, and CG experience Bryant (1996) Clarke-Stewart, 15 months = 15-36 ORCE Group size adjusted Vandell, 242 months CC-HOME for child age; CG Burchinal, 24 months = education, specialized O'Brien, and 248 training, recent McCartney 36 months = training (2002) 201 Dunn (1993) 60 51.85 ECERS Group size, C:A months Goals, strategies, ratio, CG education, and guide child's CG center experience, emotional CG field experience, development CG age Dunn, Beach, 60 51.85 ECERS CG education and Kontos months Language and training, certification, (1994) reasoning experience, C:A environment, ratio, group size physical environment and available learning activities Elicker, Fortner- 41 14.8 FDCRS CG experience Wood, and months caring for infants Noppe (1999) and toddlers, group size, income Goelman (1988) 105 CDC = 50.5 ECERS LFDC = DCHERS 38.3 COF UFDC = 39.8
EFFECTS OF CHILD CARE 151 Child Developmental Family Controls Outcomesc Quality Findings None MDI: (cognitive) ITERS related to better cognitive SICD-R and CTBS: development, language and (language skills) communication skills. Lower C:A ratio related to higher Bayley scores, more advanced receptive language development and communication skills. Better educated CG children higher on expressive language. Family income, observed Bayley MDI, Bracken Controlling for income and sensitivity, maternal sensitivity School Readiness, higher ORCE and CC-HOME scores Reynell language, mother related to higher cognitive scores, better and CG report of social language comprehension, and more competence, mother and cooperation. CG education and training CG report of behavior associated with better cognitive and problems language scores, controlling for family income and education. Child age, SES, parental CBI: (Socand Controlling for child age, DC history, age and education, day Intelligence) SES, parent age and education, higher care history PBQ: (social ECERS, CG child major, less experience competence) in the center higher CBI intelligence. PSI; (Cog) PPS (soc play) CPS and POS (cog play) SES CBI: (language) Controlling for SES, process quality PSI (cognitive) predicted children language development. Literacy environment predicted significant portion children's language development controlling for SES. None AQS: (attachment) Smaller group size and smaller C:A ratio Adult-child IRS: predicted more infant-CG interactive (CG-child involvement. involvement) Higher global CC quality related to better infant-CG attachment security, but not interactive involvement. PPVT-R Higher global quality in family day care EOWPVT (language) (DCHERS) significantly predicted higher children's PPVT and EOWPVT scores. (continued)
152 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Continued Process Quality Structural Quality Citationa N Age Measureb Measureb Hausfather, 155 55 months ECERS ECOS Toharia, ECOS LaRoche, and Engelsmann (1997) Hestenes, 60 52 months ECERS C:A ratio, group Kontos, and teacher engagement size Bryan (1993) Holloway and 55 53 months Early childhood Class size, C:A ratio, Reichhart- observation, process number of hours of Erickson (1988) composite substitute care Howes (1997) 760 4.25 years CIS, AIS C:A ratio, group Study 1 size, ECE training, CG education Howes (1997) 410 CIS, AIS CG background in Study 2 ECE
EFFECTS OF CHILD CARE 153 Child Developmental Family Controls Outcomesc Quality Findings SCS: (soc comp) Low-quality DC significantly contributes PBC: (behavioral to children's anger and defiance. HMR: problems) additive risk for aggressive behavior (early entry to DC, low-quality stress in parenting, males, stressful life events). High quality no relation with behavior problems. HMR: high quality, early attendance, favorable family circumstances children's level of interest and participation. Quality of care mediates positive or negative effects of age of entry. Gender, BSQ: (emotional MR: DC quality predicted measure of SES expression, effect acting for temperament (controlling temperament) for SES and gender). In DC centers with more appropriate caregiving, children displayed more positive effect. Neither structural related to effect. High level CG engagement children had higher intensity positive effect. Lower level CG engagement children display more intense negative effect. SES SSPS (social problem Children in high-quality interaction with solving) CG more prosocial responses and mentioned more prosocial categories. In larger classes, children gave more antisocial responses and used more antisocial categories. Children in classes with larger C:A ratios spent less time in solitary play. Controlling for SES, most still remained signficant. Language, preacademic, CG with at least AA in ECE higher social development PPVT-R scores, children in classes complying with C:A ratio higher prereading. Cognitive play, peer CG with BA or child development play associate degree greater child language, play and most complex play with peers, most language activity. CG with BA ECE children engaged in most complex play with objects and more creative activities. (continued)
154 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Continued Process Quality Structural Quality Citationa N Age Measureb Measureb Howes and 89 18, 24, 30, Low quality (higher Olenick (1986) and 36 C:A ratios, no months formally trained CG, < 2 primary CG) Howes, Phillips, 414 14-54 ECERS C:A ratio, group and Whitebook months Infant-Toddler size (1992) Environment Rating Scale, Developmentally appropropriate activities Howes and 840 34.07 ECERS, ITERS, AIS, Smith (1995) months Attachment Howes and 55 20.2 Family Day Care C:A ratio, group Stewart (1987) months Rating Scale, Adult size Play with Child Scale Kontos (1991) 138 53 months Overall C:A ratio, group environmental size, CG training, quality, COFAS, child development ECERS program evaluation indicator check
EFFECTS OF CHILD CARE 155 Child Developmental Family Controls Outcomesc Quality Findings Compliance, control High-quality centers children more compliant and less resistant, and children more likely to self-regulate. M.R.: for girls compliance best predicted by combination of high quality DC, low life complexity, and low parental involvement. Task-resistance best predicted by combination of low quality DC, high life complexity, and high parent involvement. CC quality best predicted self-regulation in boys. AQS- (attachment) Peer CG who practiced more appropriate Play Scale caregiving child more secure with CG. (social orientation, CG engaged in more developmentally interaction with peers) appropriate activities children were more socially oriented with CG. Regulatable quality on social competence mediated through process quality variables and through children's relationship with adults and peers. Process mediated through children's relationship with adults and peers rather than direct influence on peer competence. Cognitive Activity Scale HMR: (1) positive social interact with CG, attachment, and play activity (2) ECERS or ITERS. Classroom quality did not result in significant change. Quality indirect effect. Family characteristics Peer Play Scale Girls: controlling for family (nurturance and support, Play with Objects Scale characteristics (nurturance and support, restrictiveness and stress) restrict and stress), higher quality CC higher level play with peers, objects, and adults. Boys: controlling for family characteristics: higher quality care higher play with objects. Child age, child care Language, intelligence, Higher quality CC poorer intelligence, history social, behavior and poorer language. problems HMR (child age, CC history controls): quality did not predict language or intellect; family background did. HMR (child age, CC history control): higher quality CC (CDPE-IC: structural measure) children better socially adjusted and more sociable. (continued)
156 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Continued Process Quality Structural Quality Citationa N Age Measureb Measureb Kontos and 114 51.7 months CG responsive Wilcox-Herzog involvement (1997) CG verbal stimulation McCartney 166 36-68 DCEI, ECERS (1984) months McCartney, 166 2 years ECERS, verbal C:A ratio Scarr, Phillips, interact with CG and Grajek (1985) McCartney, 718 Infant = ECERS, ITERS, Scarr, 14.7 months CG-C interaction Rocheleau, Toddler = Phillips, and 27 months Abbott-Shim, Preschool = (1997) 47.9 months NICHD Early 97 6 months None C:A ratio, observed Child Care 118 15 months group size, CG Research 163 24 months training, CG Network (1999a) 250 36 months education
EFFECTS OF CHILD CARE 157 Child Developmental Family Controls Outcomesc Quality Findings Child age Cognitive competence, Controlling for child age, more CG social competence involvement lower cognitive competence, but not social competence even when controlling for age. MR: More contact with CG and more CG involvement higher social competence. Less contact with CG and more involvement in high yield activities higher cognitive competence. Child age, parent as PPVT-R, PLAI, ALI, HMR: Controlling for child age, values educator interview experimental conformity, and values social, higher total (values conformity, communication task quality of center care scores (ECERS) values social) children had higher PPVT, PLAI, ALI scores and performed better on communication task. Quality of DC positive effect on language development. Controlling for total number of functional utterances by CG to child, family background and group care experience, more verbal interaction with CG higher PLAI, ALI scores and better performance on communication task. PPVT-R and ALI: Intervention center highest quality rating. (intellect, language) Intervention center higher language, IQ, CBI and PBQ: and social ratings than other centers. (social skills) Mother's education AQS and Separation- Partial correlations, controlling for Reunion Quest: mother's education, more CG-C (attachment) CBS Q-sort interaction related to more social bids (social behavior, (toddlers and preschoolers), more solitary behavior problems; play (preschoolers) and fewer CG ratings Harter: (competence of negative separation/reunion for and social acceptance) toddlers. HMR: CG-C interactions not related to child outcomes. Income to needs, maternal Bayley MDI Outcomes (cognitive, language, and social) education, concurrent Bracken School better when children attended classes, single-parent status, child Readiness meeting recommended C:A ratio at 24 gender, maternal Reynell Dev Lang months and CG training and CG sensitivity CBCL, ASBI education at 36 months. More standards (social behavioral) met, better school readiness, language comprehension, and less behavior problems at 36 months. Older children more likely to be in classes meeting recommended standards. (continued)
158 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Continued Process Quality Structural Quality Citationa N Age Measureb Measureb Peisner- 757 4.3 ECERS, CIS, AIS, Feinberg, and years UCLA ECOF Burchinal (1997) Phillips, 166 36-68 ECERS, DCEI C:A ratio, director's McCartney, months years of experience and Scarr (1987b) Ruopp, Travers, Natural 3- and 4- Observations of C:A ratio, group Glantz, and study = 64 year-olds staff-child size, staff education, Coelen (1979) centers interactions; training Experiment = observation of child 57 centers behavior Schliecker, 100 4 years ECERS White, and Jacobs (1991) Vernon-Feagans, 67 24 months High and low Emanuel, and quality defined by a Blood (1997) composite of C:A ratio, group size, and CG training
EFFECTS OF CHILD CARE 159 Child Developmental Family Controls Outcomesc Quality Findings Mother's education, PPVT-R, WJ-R Controlling for child and family ethnicity, and child (prereading, pre-math), characteristics, the observed quality index gender CBI (social skills) and the STRS CG-child closeness score related to better PPVT-R scores (both quality indices), better WJ-R prereading scores (individually, observed quality index), better CG ratings of child's cognitive/attention skills on CBI (individually, CG rating of closeness), and fewer behavior problems (individually, CG rating of closeness), and higher sociability ratings. Higher quality CC better language, preacademic, sociability, and fewer behavior problems. CBI, PBQ: (social Higher overall quality higher social development) competence ratings. Better C:A ratio higher social competence ratings, but lower social adjustment (anxious). More CG-C interaction better social competence ratings. Looked at changes in Preschool Inventory Children had larger gains on PSI and child performance over (PSI), Peabody Picture PPVT when groups were smaller. Centers time as a function of Vocabulary Test with higher proportions of caregivers with systematic changes in (PPVT-Revised) child-related training had greater gains on ratio and staff training the PSI. SES PPVT-R (verbal) Controlling for SES, higher center quality higher PPVT. Family structure analyses: 2-parent families: Controlling for mother's education, mother's and father's age, and occupational prestige, children whose fathers have more prestigious occupations and are enrolled in high-quality DC have higher PPVT-R scores. 1-parent families: Controlling for mother's age, education, and occupational prestige, children whose mothers were older and are enrolled in high-quality DC have higher PPVT-R scores. All middle-income, dual Sequenced Inventory of Poor-quality child care associated with earner, white households Communiction poorer expressive language scores. Poorest Development (SICD) scores obtained when poor-quality care coupled with chronic otitis media. (continued)
160 WORKING FAMILIES AND GROWING KIDS TABLE A5-2 Continued Process Quality Structural Quality Citationa N Age Measureb Measureb Volling and 36 18-24 C:A ratio, group Feagans (1995) months size a For full references, see the report reference list. b Quality measures alphabetized by acronym: COF: Child Observation Form; DCEI: Day Care Environ- ment Interview; DCHERS: Day Care Home Environment Rating Scale; ECERS: Early Childhood Environ- ment Rating Scale; ECOI: Early Childhood Observation Instrument; ECOS: Early Childhood Observation Scale; FDCRS: Family Day Care Rating Scale; ITERS: Infant-Toddler Environmental Scale. c Child developmental outcome measures alphabatized by acronym: AQS: Attachment Q-Set; Adult- Child IRS: Howes and Stewart's Adult-Child Involvement Rating Scale; ALI: Adaptive Language Inven- tory; BSQ: Behavior Style Questionnaire; CBI: Classroom Behavior Inventory-Preschool Form; CBS Q- Sort: Child Behavior Survey, Q-Sort version; CPS: Cognitive Play Scale; CTBS: Comprehensive Test of Basic Skills; EOWPVT: Expressive One-Word Picture Vocabulary Test; Harter: Pictorial Scale of Perceived
EFFECTS OF CHILD CARE 161 Child Developmental Family Controls Outcomesc Quality Findings Child's age, age of entry, IBQ: (Temp) Controlling for child's age and age of hours/week in care TBAQ and Vandell and entry, higher C:A ratios predicted more Powers Quest: nonsocial play and less positive adult (social competence) interactions. Controlling for child's age and hours/week in care predicted more nonsocial play and less positive adult interactions. Child's temperament (social fear) interacts with quality of care. High-quality care may act as a buffer for socially fearful children in positive peer interactions and nonsocial play with peers. Competence and Social Acceptance for Young Children; MDI: Mental Developmental Index; PBC: Pre- school Behavior Checklist; PBQ: Preschool Behavior Questionnaire; PLAI: Preschool Language Assessment Instrument; POS: Play with Objects Scale; PPS: Peer Play Scale; PPVT-R: Peabody Picture Vocabulary Test-Revised; PSI: Preschool Inventory-Revised; SCS: Social Competence Scale; SICD: Sequence Inventory of Communication Development; SSPS: Spivack and Shure's Social Problem Solving Skills; TBAQ: Toddler Behavior Assessment Questionnaire; CG = caregiver; ECE = early childhood education; C:A ratio = child:adult ratio; CC: child care; DC = day care; SES = socioeconomic status. NOTE: For articles published before 2000, Vandell and Wolfe (2000) was used as the source.
162 WORKING FAMILIES AND GROWING KIDS TABLE A5-3 Longitudinal Relations Between Child Care Quality and Child Developmental Outcomes Process Quality Structural Citationa N Age Measureb Quality Measureb Blau (1999b) N = 2,503 to Variable None Mother report of 4,031, depending group size, on outcome C:A ratio, CG training; averaged 0-2 and 3-5 Broberg et al. 84 children in Time 1 = prior Positive and None (1990) Sweden to care; M age negative events = 16 months. Belsky and Time 2 = 1 Walker Spot year later; observation Time 3 = 2 checklist years later Broberg et al. Initial sample of Composite: Composite: C:A (1997) 146 was recruited adult child ratio, number of at 12-24 months. interaction at in care per day: 123 assessed at 16, 28, and 40 16, 28, and 40 8 years. months months 123 Burchinal et al. 89 Recruited in ITERS, ECERS C:A ratio, group (2000) first year; size, teacher reassessed at education 12, 24, and 36 months.
EFFECTS OF CHILD CARE 163 Other Child Family Child Developmental Care Measures Controls Outcomesc Quality Findings Type of care, 30 items BPI (behavior Simple correlations: CG number of problems index) training related to all 4 arrangements, beginning at 4 years outcomes. hours, cost PIAT (math and OLS regressions: individual reading achievement) coefficients for the 3 collected beginning structural variables were at 5 years generally not significant. PPVT (language) collected beginning at 3 years Type of care Social status, quality Griffith's ANOVA: no care group home environment, Developmental difference in verbal abilities parents' perceived Scales-Scale C at 28 or 40 months. social support, child (verbal/linguistic PLS: no effect of child care temperament, child ability at 28 and quality nor type of care on sociability 40 months) verbal ability at 28 or 40 months. Time in child Social status, Griffith's Structural quality related at care inhibition, paternal Developmental 40 months (.30*) and 80 involvement, home Scales-(language months (.22*) with math in environment subscales) 2nd grade. Standardized Verbal in 2nd grade School predicted by (1) verbal at Readiness Test 40 and 80 months, (2) (numerical subscales) number of months in CC, (3) consistent high paternal involvement. Math in 2nd grade predicted by (1) math at 80 months, (2) structural quality, (3) inhibition scores, (4) process quality. Not predicted by number of siblings, gender, quality of home. SES not entered in equation. Child age, child Bayley (Cognitive: HLM analyses controlling gender, poverty 12, 24, 36 months) for sex, poverty, home status, home Language: receptive quality, higher process environment and express quality over time related to (vocabulary) better cognitive, receptive Communication language, expressive skills (12, 24, 36 language, and overall months communication skills. Communicative, Associated with expressive social affective, language increase with age. (continued)
164 WORKING FAMILIES AND GROWING KIDS TABLE A5-3 Continued Process Quality Structural Citationa N Age Measureb Quality Measureb Chin-Quee and 127 Recruited ECERS Scarr (1994) preschool, amount and follow-up 5-9 type of verbal years interaction of child and CG Deater-Deckard 141 Time 1 = Composite of C:A ratio et al. (1996) toddler or ITERS, ECERS, preschooler Profile, CG Time 2 = 4 education, years later wages Field (1991) 28 5-8 years Not assessed C:A ratio Study 1 in full time care All high quality CG education by 2 CG stability Field (1991) 56 6th grade (M = Not assessed C:A ratio, CG Study 2 11.5) full-time All high quality education, CG care by 2 turnover
EFFECTS OF CHILD CARE 165 Other Child Family Child Developmental Care Measures Controls Outcomesc Quality Findings symbolic skills (12, Lower C:A ratio related to 18, 24 months) higher scores over time on receptive language and overall communication skills, controlling for family factors. Teacher education: related to higher cognitive and receptive language skills for girls only. Child's Maternal education Report cards (social HMR: (1) maternal experiences in and IQ (PPVT-R), and cognitive education, maternal IQ, CC, age of entry values conform, development) values conform, values into CC, value social skills Teacher report peer social skills; (2) CC number of hours relations, experience: age began care, in attendance cooperative behavior, total time in care; (3) academic Quality: ECERS. Quality of achievement care in infant and preschool years not related to school-age outcomes. SES, child sex, child Composite scores of No significant correlations age, parenting stress, mother-reported between Time 1 process harsh parental behavior problems quality and Time 2 child discipline and social outcomes or between Time withdrawal and 1 C:A ratio and Time 2 teacher-reported child outcomes. behavior problems Also Time 1 process quality and social and C:A ratio did not withdrawal predict Time 2 outcomes in hierarchical regressions that controlled for Time 1 adjustment. Amount time in Maternal BRS: (sociability, Partial correlations care extraversion child socioemotional (maternal extraversion): outcome adjustment) amount time spent in high- Piers-Harris (self- quality, stable care and concept) later adjustment (5-8) Buck I/E scale associated with all child outcomes. Amount time in No family variables BRS (socioemotional Simple correlations: amount care associated with adjustment and of time in high-quality time in care sociability) programs. Stable care and Piers-Harris later adjustment at 6th (self-concept) grade. Peer interactive Amount of time in high- behavior quality care associated with Academic measures: all child outcomes. gifted program, language arts, math grades (continued)
166 WORKING FAMILIES AND GROWING KIDS TABLE A5-3 Continued Process Quality Structural Citationa N Age Measureb Quality Measureb Howes (1988) 87 45-57 months CG training in followed for child development, 2 years small group size, low C:A ratio, planned and individual educational program, adequate physical space Howes (1990) 80 children 45 center care, CG Composite: C:A b-4 1 year; involvement/ ratio, CG training, other full-time investment in CG stability between 1 child (toddler period) and 4 compliance (toddler period: 18, 24, 30, 36 months) Jacobs and 36 kindergartners, Kindergarten ECERS None White (1994) 4 years at recruit 32 kindergartners, not enrolled NICHD Early 1,085 24 and 36 ORCE positive Child Care 1,041 months caregiving Research rating at 6, 15, Network 24, and 36 (1998) months
EFFECTS OF CHILD CARE 167 Other Child Family Child Developmental Care Measures Controls Outcomesc Quality Findings Age of entry, Maternal education, Academic progress Controlling for family length of day, family structure, (1st grade characteristics, higher number of maternal CG report) quality predicted better different employment CBP: (behavior academic skills (for boys), arrangements problems, school better social skills for both skills) girls and boys, and fewer behavior problems in both girls and boys. Smaller number of arrangements better academic skills for boys and girls. Family social, family CBCL Process quality structure, child age CBI (longitudinal), age partialled out. Preschool: CG involvement/investment observed social play, social pretend play, positive affect, less CG rate difficult and hesitant. Kindergarten: CG involvement less parent ratings of internalizing and externalizing; less CG rate of distract, hostile, higher rate verbal IQ, consideration. SES PPVT-R: (receptive MANOVAs (same results language) with no covar and with SES SSC: (social) and PPVT cov). Children in center care higher on interest- participation than children in no care; no difference between high- and low- quality care. No care effect on cooperation-compliance. Children in high-quality center highest on receptive language, followed by no care and then low quality. Quantity, entry Income/needs, Mother-reported Children in higher-quality age, stability, psychological behavior problems child care during first 3 group type adjustment, child's and social years more compliant and gender, child's competence; cooperative during temperament caregiver report of observations; CG reported (continued)
168 WORKING FAMILIES AND GROWING KIDS TABLE A5-3 Continued Process Quality Structural Citationa N Age Measureb Quality Measureb NICHD Early 6, 15, 24, 36 Positive CG Child Care months composite, Research language Network stimulation (2000b) NICHD Early 669 24 and 36 Positive Child Care 612 months caregiving Research (ORCE) Network (2001c) Peisner- CQO Study: n = Preschool to ECERS Feinberg et al. 733 in year 1; 2nd grade CIS (2001) 499 in year 2; ECOF 399 in AIF kindergarten, 345 in 2nd grade
EFFECTS OF CHILD CARE 169 Other Child Family Child Developmental Care Measures Controls Outcomesc Quality Findings problems; laboratory fewer behavior problems. observations of compliance and negativity Quantity, type Maternal PPVT-R, Bayley MDI , Process quality significantly child gender, HOME Bracken School related to cognitive and and maternal Readiness, language outcomes at 24 stimulation Macarthur CDI; and 36 months, controlling Reynell for family factors. Developmental Language Amount of time Maternal education, Mother and More positive caregiving in CC, maternal attitude caregiver report of child more positive available other toward employment, peer competencies; sociability at 24 months, children child gender, observed peer lower proportion negative cognitive/ linguistic interaction in child interaction with peers performance at 24 care and structured observed. and 36 months, task maternal sensitivity in play, maternal psychological adjustment, family structure, number of children in home Family income, PPVT-R Higher ECERS scores education, marital WJ-R predicted higher language status, child's gender, CBI (classroom and math scores, but the child's ethnicity behavior inventory) magnitude of the STRS (teacher-child association declined over relationship) time. A significant interaction between maternal education and quality of classrooms indicated that better quality child care had stronger association when mothers had less education. Higher quality practices were significantly associated with fewer problem behaviors in Year 1, but this association declined in subsequent years. (continued)
170 WORKING FAMILIES AND GROWING KIDS TABLE A5-3 Continued Process Quality Structural Citationa N Age Measureb Quality Measureb Pierrehumbert 47 Swiss 1-5, recruited Positive contact None et al. (1996) 3-9 months (Ainsworth interactive scale) Vandell et al. 20 Observed at 4 None C:A ratio, group (1988) and 8 years size, space, materials available, CG education Vernon-Feagans, 67 Recruited Adults present Emanuel, and before age 1, (C:A ratio), Blood (1997) followed until 4 group size a For full references see the report reference list. b Quality measures alphabetized by acronym: AIS: Adult Involvement Scale; CIS: Caregiver Interaction Scale; ECOF: UCLA Early Childhood Observation Form; ECERS: Early Childhood Environment Rating Scale; ECOI: Early Childhood Observation Instrument; IEOS: Instructional Environment Observation Scales; ITERS: Infant-Toddler Environmental Scale; ORCE: Observational Record of the Caregiving Environment; STRS: Student-Teacher Relationship Scale. c Child developmental outcome measures alphabetized by acronym: ASBI: Adaptive Social Behavior Inventory; ASB: Teacher Assessment of Social Behavior; BCL: Behavior Checklist; Boehm: Test of Basic Skills; BPI: Behavior Problems Index; BRS: Behavior Rating Scale; BSQ: Behavior Screening Questionnaire; Buck I/E Scale: Buck Internalizer/Externalizer Scale; CBCL: Child Behavior Checklist; CBI: Child Behavior Inventory; CBP: Child Behavior Profile; CTBS: Comprehensive Test of Basic Skills; MacArthur CDI:
EFFECTS OF CHILD CARE 171 Other Child Family Child Developmental Care Measures Controls Outcomesc Quality Findings SES, child gender, Developmental Attach security, SES, and attachment with Quotients positive contact with CG mother, positive WPPSI predicted increase in contact with mother CBCL cognitive index between 2 and 5 years. Family structure, age PRS Controlling for social class, of entry in full-time Harter PCS positive interaction with care, family social Parent ratings adults at 4 years was class socioemotional related to more competent adjust (Santrock behavior at 8 years. and Warshak) Aimless wandering at 4 years was related to less social competence at 8 years. SICD: (receptive and Children in high-quality expressive language) centers better expressive language and receptive language. Communication Development Inventory; MDI: Mental Development Index (Bayley II); MSCA: McCartney Scale of Children's Abilities; ORCE: Observational Record of the Caregiving Environment; PBQ: Pre- school Behavior Questionnaire; PEI: Parent as Educator Interview; PIAT: Peabody Individual Achievement Test; PPS: Peer Play Scale; PPVT-R: Peabody Picture Vocabulary Test-Revised; PRS: Peer Relations Scale; RCSA: Rutter Child Scales (A and B); SCS: Social Competence Scale; SICD: Sequence Inventory of Com- munication Development; SRA: Science Research Associates Achievement Battery; TBQ: Toddler Behavior Questionnaire; WJ-R: Woodcock-Johnson Tests of Achievement-Revised; CG = caregiver; ECE = early childhood education; C:A ratio = child:adult ratio; CC: child care; DC = day care; SES = socioeconomic status. NOTE: Vandell and Wolfe (2000) was the source for articles published prior to 2000.
172 WORKING FAMILIES AND GROWING KIDS TABLE A5-4 Distribution of Child Care Quality by Type of Care and Family Income For All Low Income Moderate Income High Income Overall Nonparental Care N (%) N (%) N (%) N (%) 6 months (n = 471) Poor quality 11 (13) 14 (9) 12 (5) 37 (8) Fair quality 39 (44) 67 (41) 89 (40) 195 (41) Good quality 31 (35) 60 (37) 74 (33) 165 (35) Excellent quality 7 (8) 21 (13) 46 (21) 74 (16) 15 months (n = 494) Poor quality 10 (9) 15 (9) 14 (6) 39 (8) Fair quality 59 (55) 74 (45) 96 (43) 229 (46) Good quality 27 (25) 55 (34) 76 (34) 158 (32) Excellent quality 11 (10) 19 (12) 38 (17) 68 (14) 24 months (n = 537) Poor quality 13 (13) 30 (15) 10 (4) 53 (10) Fair quality 62 (63) 99 (51) 126 (52) 287 (53) Good quality 19 (19) 49 (25) 76 (31) 144 (27) Excellent quality 5 (5) 17 (9) 31 (13) 53 (10) 36 months (n = 587) Poor quality 11 (7) 9 (5) 6 (2) 26 (4) Fair quality 102 (67) 122 (65) 134 (54) 358 (61) Good quality 36 (24) 50 (26) 87 (35) 173 (29) Excellent quality 3 (2) 8 (4) 19 (7) 30 (5) 54 months (n = 795) Poor quality 25 (12) 21 (7) 12 (4) 58 (7) Fair quality 90 (44) 142 (46) 112 (40) 344 (43) Good quality 65 (32) 96 (31) 113 (40) 274 (34) Excellent quality 26 (13) 48 (16) 45 (16) 119 (15) Overall (n = 2,884) Poor quality 70 (11) 89 (9) 54 (4) 213 (7) Fair quality 352 (54) 504 (50) 557 (46) 1,413 (49) Good quality 178 (27) 310 (31) 426 (35) 914 (32) Excellent quality 52 (8) 113 (11) 179 (15) 344 (12) TOTAL 652 1,016 1,216 2,884 Low Income Moderate Income High Income Overall For Father Care N (%) N (%) N (%) N (%) 6 months (n = 87) Poor quality 1 (6) 1 (2) 0 2 (2) Fair quality 5 (29) 18 (38) 4 (18) 27 (31)
EFFECTS OF CHILD CARE 173 TABLE A5-4 Continued Low Income Moderate Income High Income Overall For Father Care N (%) N (%) N (%) N (%) Good quality 8 (47) 18 (38) 13 (59) 39 (45) Excellent quality 3 (18) 11 (23) 5 (23) 19 (22) 15 months (n = 107) Poor quality 1 (3) 5 (11) 0 6 (6) Fair quality 15 (45) 13 (28) 11 (41) 39 (36) Good quality 12 (36) 18 (38) 7 (26) 37 (35) Excellent quality 5 (15) 11 (23) 9 (33) 25 (23) 24 months (n = 89) Poor quality 2 (8) 2 (6) 0 4 (4) Fair quality 15 (60) 16 (44) 10 (36) 41 (46) Good quality 4 (16) 13 (36) 11 (39) 28 (31) Excellent quality 4 (16) 5 (14) 7 (25) 16 (18) 36 months (n = 83) Poor quality 3 (9) 2 (6) 0 5 (6) Fair quality 21 (62) 12 (36) 8 (50) 41 (49) Good quality 10 (29) 16 (49) 3 (19) 29 (35) Excellent quality 0 3 (9) 5 (31) 8 (10) 54 months (n = 36) Poor quality 4 (27) 2 (13) 1 (20) 7 (19) Fair quality 8 (53) 7 (44) 1 (20) 16 (44) Good quality 2 (13) 6 (38) 3 (60) 11 (31) Excellent quality 1 (7) 1 ( 6) 0 2 (6) Overall (n = 402) Poor quality 11 (9) 12 (7) 1 (1) 24 (6) Fair quality 64 (52) 66 (37) 34 (35) 164 (41) Good quality 36 (29) 71 (39) 37 (38) 144 (36) Excellent quality 13 (10) 31 (17) 26 (27) 70 (17) TOTAL 124 180 98 402 Low Income Moderate Income High Income Overall For Grandparents N (%) N (%) N (%) N (%) 6 months (n = 94) Poor quality 0 0 0 0 Fair quality 7 (33) 6 (16) 6 (17) 19 (20) Good quality 10 (48) 22 (59) 15 (42) 47 (50) Excellent quality 4 (19) 9 (24) 15 (42) 28 (30) (continued)
174 WORKING FAMILIES AND GROWING KIDS TABLE A5-4 Continued Low Income Moderate Income High Income Overall For Grandparents N (%) N (%) N (%) N (%) 15 months (n = 77) Poor quality 1 (4) 2 (7) 1 (4) 4 (5) Fair quality 12 (48) 6 (21) 3 (13) 21 (27) Good quality 9 (36) 14 (48) 10 (43) 33 (43) Excellent quality 3 (12) 7 (24) 9 (39) 19 (25) 24 months (n = 66) Poor quality 1 (5) 2 (8) 0 3 (5) Fair quality 13 (68) 6 (23) 5 (24) 24 (36) Good quality 3 (16) 14 (54) 13 (62) 30 (45) Excellent quality 2 (11) 4 (15) 3 (14) 9 (14) 36 months (n = 61) Poor quality 3 (12) 0 0 3 (5) Fair quality 15 (58) 8 (44) 12 (71) 35 (57) Good quality 7 (27) 8 (44) 3 (18) 18 (30) Excellent quality 1 (4) 2 (11) 2 (12) 5 (8) 54 months (n = 34) Poor quality 1 (6) 0 1 (14) 2 (6) Fair quality 8 (50) 8 (73) 2 (29) 18 (53) Good quality 6 (38) 2 (18) 3 (43) 11 (32) Excellent quality 1 (6) 1 (9) 1 (14) 3 (9) Overall (n = 332) Poor quality 6 (6) 4 (3) 2 (2) 12 (4) Fair quality 55 (51) 34 (28) 28 (27) 117 (35) Good quality 35 (33) 60 (50) 44 (42) 139 (42) Excellent quality 11 (10) 23 (19) 30 (29) 64 (19) TOTAL 107 121 104 332 For in-Home Care (not parents or Low Income Moderate Income High Income Overall grandparents) N (%) N (%) N (%) N (%) 6 months (n = 83) Poor quality 1 (6) 1 (8) 3 (6) 5 (6) Fair quality 9 (56) 5 (38) 13 (24) 27 (33) Good quality 5 (31) 4 (31) 22 (41) 31 (37) Excellent quality 1 (6) 3 (23) 16 (30) 20 (24) 15 months (n = 95) Poor quality 1 (7) 0 2 (3) 3 (3) Fair quality 11 (73) 10 (48) 17 (29) 38 (40)
EFFECTS OF CHILD CARE 175 TABLE A5-4 Continued For in-Home Care (not parents or Low Income Moderate Income High Income Overall grandparents) N (%) N (%) N (%) N (%) Good quality 2 (13) 7 (33) 25 (42) 34 (36) Excellent quality 1 (7) 4 (19) 15 (25) 20 (21) 24 months (n = 78) Poor quality 3 (23) 3 (19) 0 6 (8) Fair quality 7 (54) 6 (38) 16 (33) 29 (37) Good quality 2 (15) 4 (25) 19 (39) 25 (32) Excellent quality 1 (8) 3 (19) 14 (29) 18 (23) 36 months (n = 66) Poor quality 2 (18) 0 1 (2) 3 (5) Fair quality 8 (73) 6 (50) 14 (33) 28 (42) Good quality 1 (9 ) 5 (42) 21 (49) 27 (41) Excellent quality 0 1 (8) 7 (16) 8 (12) 54 months (n = 16) Poor quality 1 (20) 0 1 (13) 2 (13) Fair quality 3 (60) 3 (10) 5 (63) 11 (69) Good quality 0 0 2 (25) 2 (13) Excellent quality 1 (20) 0 0 1 (6) Overall (n = 338) Poor quality 8 (13) 4 (6) 7 (3) 19 (6) Fair quality 38 (63) 30 (46) 65 (31) 133 (39) Good quality 10 (17) 20 (31) 89 (42) 119 (35) Excellent quality 4 (7) 11 (17) 52 (24) 67 (20) TOTAL 60 65 213 338 For Child Care Low Income Moderate Income High Income Overall Homes N (%) N (%) N (%) N (%) 6 months (n = 192) Poor quality 7 (18) 5 (7) 7 (9) 19 (10) Fair quality 19 (48) 34 (46) 37 (47) 90 (47) Good quality 13 (33) 26 (35) 23 (30) 62 (32) Excellent quality 1 (3) 9 (12) 11 (14) 21 (11) 15 months (n = 197) Poor quality 5 (11) 5 (7) 5 (6) 15 (8) Fair quality 23 (51) 34 (47) 33 (41) 90 (46) Good quality 13 (29) 26 (36) 29 (36) 68 (35) Excellent quality 4 (9) 7 (10) 13 (16) 24 (12) (continued)
176 WORKING FAMILIES AND GROWING KIDS TABLE A5-4 Continued For Child Care Low Income Moderate Income High Income Overall Homes N (%) N (%) N (%) N (%) 24 months (n = 210) Poor quality 5 (14) 10 (12) 4 (5) 19 (9) Fair quality 21 (57) 44 (52) 45 (51) 110 (52) Good quality 9 (24) 23 (27) 29 (33) 61 (29) Excellent quality 2 (5) 8 (9) 10 (11) 20 (10) 36 months (n = 168) Poor quality 3 (7) 1 (2) 1 (2) 5 (3) Fair quality 31 (72) 47 (76) 30 (48) 108 (64) Good quality 7 (16) 12 (19) 28 (44) 47 (28) Excellent quality 2 (5) 2 (3) 4 (6) 8 (5) 54 months (n = 76) Poor quality 9 (41) 3 (9) 1 (5) 13 (17) Fair quality 7 (32) 20 (57) 13 (68) 40 (53) Good quality 4 (18) 7 (20) 3 (16) 14 (18) Excellent quality 2 (9) 5 (14) 2 (11) 9 (12) Overall (n = 843) Poor quality 29 (16) 24 (7) 18 (5) 71 (8) Fair quality 101 (54) 179 (55) 158 (48) 438 (52) Good quality 46 (25) 94 (29) 112 (34) 252 (30) Excellent quality 11 (6) 31 (9) 40 (12) 82 (10) TOTAL 187 328 328 843 For Low Income Moderate Income High Income Overall Centers N (%) N (%) N (%) N (%) 6 months (n = 102) Poor quality 3 (27) 8 (21) 2 (4) 13 (13) Fair quality 4 (36) 22 (58) 33 (62) 59 (58) Good quality 3 (27) 8 (21) 14 (26) 25 (25) Excellent quality 1 (9) 0 4 (8) 5 (5) 15 months (n = 125) Poor quality 3 (14) 8 (20) 6 (10) 17 (14) Fair quality 13 (59) 24 (59) 43 (69) 80 (64) Good quality 3 (14) 8 (20) 12 (19) 23 (18) Excellent quality 3 (14) 1 (2) 1 (2) 5 (4) 24 months (n = 183) Poor quality 4 (13) 15 (22) 6 (7) 25 (14) Fair quality 21 (70) 43 (63) 60 (71) 124 (68) Good quality 5 (17) 8 (12) 15 (18) 28 (15) Excellent quality 0 2 (3) 4 (5) 6 (3)
EFFECTS OF CHILD CARE 177 TABLE A5-4 Continued For Low Income Moderate Income High Income Overall Centers N (%) N (%) N (%) N (%) 36 months (n = 292) Poor quality 3 (4) 8 (8) 4 (3) 15 (5) Fair quality 48 (67) 61 (63) 78 (63) 187 (64) Good quality 21 (29) 25 (26) 35 (28) 81 (28) Excellent quality 0 3 (3) 6 (5) 9 (3) 54 months (n = 669) Poor quality 14 (9) 18 (7) 9 (4) 41 (6) Fair quality 72 (44) 111 (43) 92 (37) 275 (41) Good quality 55 (34) 87 (34) 10 (42) 247 (37) Excellent quality 22 (14) 42 (16) 42 (17) 106 (16) Overall (n = 1371) Poor quality 27 (9) 57 (11) 27 (5) 111 (8) Fair quality 158 (53) 261 (52) 306 (54) 725 (53) Good quality 87 (29) 136 (27) 181 (32) 404 (29) Excellent quality 26 (9) 48 (10) 57 (10) 131 (10) TOTAL 298 502 571 1,371 xx