Deborah Lowe Vandell and Barbara
Wolfe(1)
Institute for Research on Poverty
University of Wisconsin–Madison
Child care has become the norm for young children in the United States. In 1995, 59 percent of children who were 5 years or younger were in nonparental care arrangements on a regular basis (Hofferth, Shauman, Henke, and West, 1998). This care typically began at early ages and lasted substantial hours: with 44 percent of infants under the age of 1 year were in nonparental care for an average of 31 hours a week. In the late preschool years, 84 percent of 4- to 5-year-olds were recorded as being in child care for an average of 28 hours per week. The use of nonparental care in the United States is expected to grow even further as welfare reform is fully implemented (Vandell, 1998).
It is within this framework of widespread and early-age use that questions about child care quality have been raised. Among child care researchers, the established view is that child care quality contributes to children’s developmental outcomes, higher quality care being associated with better developmental outcomes and poorer quality care being associated with poorer outcomes for children (Clarke-Stewart and Fein, 1983; Phillips, 1987). This view is reflected in Michael Lamb’s (1998) comprehensive critique of child care research that was published in the Handbook of Child Psychology. Lamb concluded, based on extant research, that:
“Quality day care from infancy clearly has positive effects on children’s intellectual, verbal, and cognitive development, especially when children would otherwise experience impoverished and relatively unstimulating home environments. Care of unknown quality may have deleterious effects.” (p. 104)
A similar conclusion was drawn in a review prepared for the Rockefeller Foundation (Love, Schochet, and Meckstroth, 1996):
“The preponderance of evidence supports the conclusion of a substantial positive relationship between child care quality and child well-being. Evidence for this relationship encompasses multiple dimensions of quality and diverse indicators of children’s well-being.” (p. 3)
This view, however, is not uniformly held. Some researchers and policy makers have begun to question the conventional wisdom regarding child care quality (Besharov, 2000; Blau, 1999c; Scarr, 1998). Sandra Scarr (1998), for example, has concluded that:
“Widely varying qualities of child care have been shown to have only small effects on children’s concurrent development and no demonstrated long term impact, except for disadvantaged children.” (p. 95)
A major goal of the current report is to evaluate the research evidence from which these claims and counterclaims are drawn. We then analyze the argument for public intervention to improve the quality of child care, especially for children from lower income-families.
A careful review of the literature indicates that reviewers often draw on the same research studies, but interpret findings differently. These different interpretations are based, in part, on where the reviewers have “set the bar.” Some researchers place more weight on studies that include observational assessments of child care quality and that measure psychological processes using multiple strategies (NICHD Early Child Care Research Network, 1994). These same investigators tend to place less emphasis on the necessity of large, nationally representative samples. Although the investigators believe that it is important to assess and control for selection biases, they worry more about overcontrol than undercontrol in their analyses. In contrast, others (see Besharov, 2000; Blau, 1999c, in press) have emphasized the importance of large, nationally representative samples and the need to have sufficient controls in the statistical analyses. These investigators have placed greater credence on information obtained from nationally representative surveys, even if studies lacked observational assessments of child care quality or objective measures of child performance.
An additional factor contributing to different conclusions about child care quality is how heavily reviewers weigh the importance of concurrent vs. long-term findings. As can be seen on Tables 1, 2, and 3, the research literature describing concurrent associations between child care quality and child performance is larger and findings are more consistent than the research literature that tests for longer-term effects. A number of factors may contribute to the more mixed picture for long-term effects, including measurement problems and lack of control for experiences during the intervening period. A better consensus about realistic and reasonable expectations about effect sizes also is needed (McCartney & Rosenthal, in press).
Thus, a variety of factors must be considered if we are to determine whether associations between child care quality and children’s developmental outcomes are large enough for parents, researchers, and policy makers to care about, and whether effects warrant public or private expenditures to improve quality. In an effort to address these broad issues, we pose five specific questions:
A critical issue in evaluating the research evidence is consideration of how child care quality is measured. Researchers have measured quality in various ways: by observing process, by recording structural and caregiver characteristics, by assessing health and safety provisions. Child care processes refer to actual experiences that occur in child care settings, including children’s interactions with caregivers and peers and their participation in different activities. Sometimes process measures are global scores that combine experiences across several areas that include health and safety provisions, interactions with caregivers, and age-appropriate materials. Other process measures target specific activities or experiences, such as language stimulation by caregivers. Structural and caregiver characteristics refer to features such as child:adult ratio, group class size, caregiver formal education, and caregiver specialized training related to children. Structural and caregiver characteristics are conceptualized as more distal indicators of child care quality. Health and safety provisions refer to both health-promoting practices, such as hand-washing, and safety in the classroom and on playgrounds.
One well-known process measure is the Early Care Environment Rating Scale (ECERS, Harms and Clifford, 1980). This measure is composed of 37 items that evaluate seven aspects of center-based care for children ages two and a half to five years. These areas are personal care routines, furnishings, language reasoning experiences, motor activities, creative activities, social development, and staff needs. Detailed descriptors are provided for each item and each item is rated as inadequate (1), minimal (3), good (5), and excellent (7). The ratings, according to the scale developers, are based on a minimum of a two-hour block of observation in the classroom. The Infant/Toddler Environment Rating Scale (ITERS, Harms, Cryer, and Clifford, 1990) is a related measure that assesses process quality in centers for children younger than two and a half years. The 35 items of the ITERS also are organized under seven domains and are rated on 7-point scales.
These same investigators have developed a 32-item observational measure, the Family Day Care Rating Scale (FDCRS), to assess process quality in child care homes (Harms and Clifford, 1989). Some items parallel items on the ITERS and the ECERS, but other items are unique because the instrument “tries to remain realistic for family day care home settings by not requiring that things be done as they are in day care centers” (p. 1).
As can be seen on Tables 1, 2, and 3, these measures are used widely in child care research. The measures have important strengths, including having good psychometric properties and being relatively easy to use reliably. Their widespread use means that cross-study comparisons are possible. These measures also have some limitations. The global composite score combines features of the physical environment, social experiences, and working conditions for staff. Some of these areas may well have greater influences on children’s intellectual functioning or social-emotional well-being than others. The composite score may underestimate effects relative to more targeted scales. A second limitation is that these measures are setting-specific. As a result, they cannot be used as interchangeable measures of quality, meaning that it is not possible to make simple comparisons across types of care or to combine scores in omnibus analyses that look at quality effects across different types of care. A third limitation is that these measures are not appropriate for assessing in-home care given by nannies or grandparents.
The Observational Record of the Caregiving Environment (ORCE) was developed to address these limitations (NICHD Early Child Care Research Network, 1996, in press-a). Because psychological theory and research have indicated the central role of experiences with caring adults for children’s well-being and development, the ORCE focuses on this domain. Both time sampled behavioral counts of caregiver actions (e.g., responds to vocalization, asks questions, speaks negatively) and qualitative ratings of those behaviors over time to characterize caregivers’ behavior with individual children are collected during a minimum of four 44-minute observation cycles spread over a two-day period. At the end of each 44-minute cycle, observers use 4-point ratings scaled from 1 = “not at all characteristic” to 4 = “highly characteristic” to describe caregiver behavior. A positive caregiving composite score is created by obtaining a mean score across scales over all of the ORCE cycles at a given age period. Higher scores indicate caregivers who are more sensitive and responsive to a child’s needs, who are warm and positive, who are cognitively stimulating, and who are not detached or hostile. Unlike the ECERS, ITERS, or FDCRS, the ORCE can be used in all types of child care and with children across the first five years. Age-appropriate behavioral descriptors for caregivers’ behaviors with infants, toddlers, and preschoolers are provided.
Another commonly used process measure is the Caregiver Interaction Scale (Arnett, 1989) that rates teachers’ sensitivity during interactions with children. This 26-item measure yields three scores (sensitivity—warm, attentive, engaged; harshness—critical, punitive; detachment—low levels of interaction, interest, or supervision) which are combined to create an overall caregiver quality score. The ratings are made after two 45-minute observations conducted on two separate occasions by two separate observers.
The Assessment Profile (Abbott-Shim & Sibley, 1992a, 1992b) assesses different aspects of quality, namely features related to health and safety, physical facilities, and individualized child services. Different forms of the instrument are available for child care homes and centers. These forms list individual items that are viewed as exemplars of (a) healthy, safe settings, (b) rich physical environments, and (c) settings that meet the needs of adult staff. Individual items are scored using a yes/no format, with “yes” designating items that were either observed or reported by staff. These items can be scored reliably (see NICHD Early Child Care Research Network, 1996). Caregivers have been observed to offer more positive caregiving in settings that receive higher Profile scores (NICHD Early Child Care Research Network, 1996, in press-a).
The CC-HOME Inventory is a measure of process quality that uses a checklist approach to create a quality score across multiple domains, including the health and safety of the physical environment, variety of experiences, and materials (NICHD Early Child Care Research Network, 1996). Derived from Bradley and Caldwell’s well-known assessment of the quality of the home environment, 45 items are scored on a yes/no basis and then summed (alpha = .81). In one study, children who attended better-quality child care homes as measured by the CC-HOME Inventory obtained higher Bayley scores at 24 months and higher school readiness and language comprehension scores at 36 months, in comparison to children who attended poorer-quality child care homes (Clarke-Stewart, Vandell, Burchinal, O’Brien, and McCartney, 2000).
Other measures have been less successful in providing reliable and valid assessments of process quality. For example, Lamb and colleagues failed to find concurrent associations between child care quality and child functioning in their study of child care in Sweden (Broberg, Hwang, Lamb, and Bookstein, 1990). One factor that likely contributed to the lack of significant relations was problems with their quality measure. The Belsky-Walker Checklist (Broberg et al., 1990) asks observers to check off if 13 positive events (e.g., caregiver provided verbal elaboration, caregiver gives heightened emotional display; signs of positive regard ) and 7 negative events (e.g., child cries; child aimless; caregivers in non-child conversations) occur at least once during 3-minute observation intervals. This 3-minute time observation frame was substantially longer than the 10- to 30-second intervals recommended for recording social interactions (Yarrow and Zahn-Waxler, 1979). Consequently, the checklist may have failed to detect meaningful distinctions in caregiver behavior because the time interval was too long to detect meaningful differences. This checklist underscores the challenge of designing and assessing process quality. Detecting relations between process quality and child outcomes requires robust measures.
A second approach to describing child care quality is in terms of their structural and caregiver characteristics. Characteristics such as child:adult ratio, group class size, caregiver formal education, and caregiver specialized training are viewed as more distal contributors to quality environments. Structural and caregiver characteristics are the only quality indicators obtained in survey studies such as the National Child Care Survey (Hofferth, Brayfield, Deich, and Holcomb, 1991), the National Household Education Survey (Hofferth et al., 1998), and the National Longitudinal Survey of Youth (Blau, 1999-c). Structural and caregiver characteristics have been collected in addition to process-oriented measures in studies such as the Cost, Quality and Outcome Study, thereby permitting relations between these characteristics and process quality to be evaluated.
Relations between structural and caregiver characteristics and process quality are well-documented in the research literature. Table 1 is a compilation of the studies conducted in the United States that have considered this issue. The table includes information regarding sample size, type of care setting, the structural and caregiver characteristics that were measured, the process quality measures that were collected, and findings that were obtained. As indicated in Table 1, some studies have considered bivariate relations between structural and caregiver characteristics, and process quality using Pearson correlations and t-tests. Other studies (Blau, in press; NICHD Early Child Care Research Network, 1996, in press-a; Phillipsen, Burchinal, Howes, and Cryer, 1997) utilized multiple regression techniques in an effort to isolate the relative impact of different characteristics. As documented on the table, the multivariate results are consistent with the bivariate and global composite analyses. As is evident in Table 1, studies have considered both global composites of structural and caregiver characteristics and individual factors in relation to process quality (Howes, 1990; Vandell and Powers, 1983).
When child:adult ratios are lower, caregivers spend less time managing children in their classrooms and children appear less apathetic and distressed (Ruopp, Travers, Glantz, and Coelen, 1979). When child:adult ratios are lower, caregivers offer more stimulating, responsive, warm, and supportive care (Clarke-Stewart, Gruber, and Fitzgerald, 1994; Howes, 1983; NICHD Early Child Care Research Network, 1996, in press-a; Phillipsen et al., 1997; Volling and Feagans, 1995). Ratios also are associated with global process quality scores (Burchinal, Roberts, Nabors, and Bryant, 1996; Howes, Phillips, and Whitebook, 1992; McCartney, et al., 1997; Scarr, Eisenberg, and Deater-Deckard, 1994; Whitebook, Howes, and Phillips, 1990). For example, in a study of 414 children who resided in three states, Howes et al. (1992) determined that “good” and “very good” scores on the ITERS and ECERS were more likely in infant classrooms with ratios of 3:1 or less, in toddler classrooms with ratios of 4:1 or less, and in preschool classrooms with ratios of 9:1 or less. More than half of the infant classrooms with ratios higher than 4:1 and preschool classrooms with ratios higher than 5:1 received scores that were categorized as “inadequate.”
Group size also has been considered in relation to process quality. In simultaneous multiple regressions that included group size, ratio, caregiver education, and caregiver specialized training, the NICHD Study of Early Child Care (1996; in press-a) determined group size to be uniquely associated with positive caregiving. Similarly, Ruopp et al. (1979) reported group size to predict caregiver behavior even when child:adult ratio was controlled. In these studies, caregivers were more responsive, socially stimulating, and less restrictive when there were fewer children in their classrooms. These relations also are observed in child-care homes (Elicker, Fortner-Wood, and Noppe, 1999; Stith and Davis, 1984).
Caregivers’ formal education and specialized training also are related to quality of care. Caregivers who have more formal education (NICHD Early Child Care Research Network, 1996; Phillipsen et al., 1997) and more specialized training pertaining to children (Arnett, 1989; Berk, 1985; Howes, 1983, 1997) offer care that is more stimulating, warm, and supportive. Highly educated and specially trained caregivers also are more likely to organize materials and activities into more age-appropriate environments for children (NICHD Early Child Care Research Network, 1996). These settings are more likely to receive higher scores on the global quality scales such as the ECERS, ITERS, ORCE, and CC-HOME (Clarke-Stewart, et al., 2000; Howes and Smith, 1995; NICHD Early Child Care Research Network, 1996, in press-a).
Repeated-measure analyses conducted for children in the NICHD Study of Early Child Care at 15, 24, and 36 months ascertained that group size and child:adult ratios were stronger predictors of process quality for infants, whereas caregiver educational background and training were stronger predictors of process quality for preschoolers (NICHD Study of Early Child Care, in press-a). These relations do not appear to be an artifact of restricted ranges. The standard deviations for caregiver formal education and caregiving training were similar at different assessment points. Standard deviations for ratio and group size increased for older children. The differential patterns, then, suggest the merits of an age-related strategy for improving process quality. Ratios and group size may be more critical for infant care; caregiver training and education may be more critical for preschoolers.
Caregiver wages is another factor associated with process quality (Howes, Phillips, and Whitebook, 1992; Scarr et al., 1994). See Table 1. In the Three-State Study, Scarr et al. reported teacher wages to be the single best predictor of process quality. In analyses of the Cost, Quality, and Outcome data set, Phillipsen et al. (1997) determined lead teachers wages to significantly predict scores on the ECERS and the Arnett sensitivity scales.
Although much of the research literature has reported significant relations between structural and caregiver characteristics, and process quality, Blau (in press) has cautioned that these associations may be the result of uncontrolled factors that are confounded with the structural and caregiver characteristics. He argues that these confounding factors might include center policies, curriculum, and directors’ leadership skills. To address this perceived shortcoming, Blau conducted secondary analyses on 274 child care centers that were part of the Cost, Quality, and Outcomes Study. In his first set of analyses, Blau conducted regressions to determine if individual structural and caregiver characteristics were associated with process quality when other factors (teacher, family, center characteristics) were controlled. His findings were consistent with other reports. When child:adult ratios were larger, ITERS and ECERS scores were lower. When caregivers had attended college or training workshops and when caregivers had college degrees in fields related to child care, ECERS scores were higher.
Blau then tested relations between structural-regulable characteristics and process quality using a more stringent fixed-effects model that included center as a control variable. This fixed-effects approach was possible because two classrooms were typically observed in each center. In centers in which there were both infants and preschoolers, one classroom of each type was observed. In centers serving only preschoolers, two preschool classrooms were selected randomly. When center was controlled along with type of classroom (infant vs. preschool), relations between structural and caregiver features and process quality were reduced. Blau interprets this reduction to mean that unobserved center characteristics account for the previously reported relations between structural factors and process quality. Our concern, however, is that the center fixed-effect control is inappropriate. As Blau himself noted, this approach requires within-center variability in the structural characteristics. It is unlikely that classrooms in the same center are highly variable in terms of caregiver training, ratio, or group size, especially given that the model also controlled for type of classroom (infant/toddler vs. preschool). The inclusion of the specific center as a control variable resulted in an underestimation of effects.
Global process quality measures such as the ECERS, CC-HOME, and Profile Assessment include health and safety indicators as a component of process quality. Research conducted in the medical and public health arenas has focused more exclusively on these indicators in relation to children’s physical health and safety. More hygienic practices by staff and children (Niffenegger, 1997; St. Sauver, Khurana, Kao, and Foxman, 1998) are associated with fewer respiratory illnesses and other infectious diseases. These practices include frequent handwashing after diapering, before meals, and after nose wiping. Child injuries in child care settings are most likely to occur on playgrounds and are most due to falls from climbing equipment (Briss, Sacks, Addis, Kresnow, and O’Neil, 1995; Browning, Runyon, and Kotch, 1996). Height of the equipment and lack of an impact-absorbing surface under the equipment have been consistently identified as the factors most highly associated with injuries that required medical treatment. The North Carolina Smart Start initiative was successful in improving the safety of child care centers with playground improvement grants (Kotch and Guthrie, 1998).
The weight of the research evidence demonstrates significant relationships between process quality, structural and caregiver characteristics, and health and safety practices. The next section uses process, structural, and caregiver measures to predict developmental outcomes for children.
There are substantial challenges for researchers and policy makers who seek to answer questions about the effects of child care quality on children’s development. One well-acknowledged difficulty is the absence of well-controlled experiments in which children are randomly assigned to child care that varies in quality. Instead, investigators have studied children whose families and child care settings are willing to participate. This examination of naturally occurring child care, as opposed to more controlled experiments, poses challenges for researchers and policy makers (Blau, 1999c; Lamb, 1998; NICHD Early Child Care Research Network, 1994; Vandell and Corasaniti, 1990). These challenges are related to family/child selection biases and to restricted variability in quality scores. Before reviewing research findings pertaining to effects of quality, we briefly describe common strategies for addressing these research challenges.
Family/Child Selection Biases. The possibility that families differ in their child care choices is a topic of interest in its own right (NICHD Early Child Care Research Network, 1997; Singer, Fuller, Keiley, and Wolf, 1998). It also is a critical issue for investigators who are interested in ascertaining the effects of child care on children (Howes and Olenick, 1986; Vandell, 1997). The problem is that ostensible “effects” of child care quality may be artifacts of family characteristics that are confounded with child care quality. As a result of this concern, it has become standard practice for researchers to incorporate family selection factors into their analyses. As is evident in Tables 2 and 3, almost all studies conducted in recent years have included controls for family characteristics.
As an example of this strategy, the NICHD Study of Early Child Care has utilized three criteria for identifying family variables that are then used as selection controls in analyses: (1) the family characteristic is significantly related to child care, (2) the family characteristic is related to the child outcome of interest, and (3) the family characteristic is not highly related to other family factors. The third criterion is applied to reduce collinearity among family characteristics.
At one level, concern about family selection bias is clearly merited. There is evidence, for example, that type and quality of child care are related to parents’ education and income (see Figure 1).
Parents who have higher incomes and more education are more likely to place their children in centers that have higher ECERS scores, lower child:adult ratios, and better-trained teachers (Blau, 1999c; Peisner-Feinberg and Burchinal, 1997). Children with more sensitive mothers are more likely to be placed in care arrangements that offer more positive caregiving experiences (NICHD Early Child Care Research Network, 1997). Children whose home environments are more cognitively stimulating and more emotionally supportive are more likely to be placed in child care settings that are stimulating and supportive (NICHD Early Child Care Research Network, in press-b). These family factors, if not controlled, may masquerade as child care effects.
At another level, however, selection effects do not appear to be as large as initially thought. In the Cost, Quality, and Outcomes Study, for example, the correlation between maternal education and the ECERS was .24; the correlation between family income and the ECERS was .09. In the NICHD Study of Early Child Care, correlations between maternal education and ORCE positive caregiving ratings were .11 at 6 months, .14 at 15 months, .22 at 24 months, and .19 at 36 months. Correlations between family income and ORCE positive caregiving were typically lower than these figures. These relatively modest associations between child care quality and family factors suggest that selection effects are not substantial, at least within the range of studies that have been conducted. In the future, selection effects may be greater as welfare reform is fully implemented and the numbers of children in child care increase.
Variability in Child Care Quality. The ability to detect child care quality effects also is dependent on obtaining sufficient variability in quality scores. Obviously, if there is no variation in quality, it is not possible to detect variations associated with quality. If quality is sampled within a truncated range, effects associated with quality are reduced and larger samples are needed to detect differences. One reason that the Swedish studies have not detected quality effects may be the restricted range of the quality scores that were sampled, coupled with relatively small sample sizes (Broberg, Wessels, Lamb, and Hwang, 1997; Lamb, Hwang, et al., 1988). These same issues are pertinent to child care research in the United States, when restricted ranges of quality are sampled and sample sizes are small.
Control for Prior Child Adjustment. A third challenge is determining when and how to control appropriately for prior child adjustment in examinations of child care effects. Some researchers have argued that stronger tests of child care quality require controls for prior child adjustment. Such controls could be used successfully in studies of after-school programs that controlled for children’s adjustment prior to entry into the programs (Vandell and Posner, 1999). Controls for prior child adjustment in studies of early child care quality are more difficult. Children typically begin child care during their first year of life, prior to the time that robust and reliable measures of child cognitive, language, and social adjustment can be administered. Using measures of child adjustment collected at some later period, after substantial child care experience has accrued, does not make sense because these measures may well be a reflection of the effects of quality to that point. By controlling for child adjustment scores that were already affected by quality, we may be eliminating (or at least minimizing) the very quality effects that are of interest. This potential confounding of child care quality and child adjustment scores means that fixed-effects models that control for prior (or concurrent) child adjustment must be applied with caution.
With these methodological challenges in mind, we turn to the conceptual framework that guides our evaluation of child care quality. This model is presented in Figure 2.
A central feature in the model is an awareness that children are not randomly assigned to child care. Child care quality is expected to be related to family characteristics including demographic, psychological, and attitudinal differences. Because these family characteristics—income, parental education, maternal sensitivity, stimulating and supportive home environments—also can predict children’s developmental outcomes, it is necessary to control for them. Otherwise, quality effects may be overestimated or underestimated. As shown in the model, research also needs to take into account other child care parameters, such as amount of care and type of care, that may be confounded with quality or that may contribute independently to child outcomes.
Children’s developmental outcomes are considered in relation to process quality and in relation to structural and caregiver characteristics. Specifically, the model posits that process quality is directly related to child developmental outcomes. Structural and caregiver characteristics are posited to be indirectly related to child outcomes, through their influence on process quality. It is expected that structural and caregiver characteristics also directly influence child outcomes in ways that are not mediated through the available measures of process quality. In the sections that follow, research findings pertaining to this model are considered in terms of concurrent relations between child care quality and children’s development, and in terms of longer-term associations between child care quality and child adjustment.
Table 2 is a summary description of results from empirical studies that examined relations between process quality and child developmental outcomes. The description includes sample size, child’s age at the time of the concurrent assessments, the measures of process quality that were used, the measures of structural quality that were used, the controls (if any) for family factors, the child developmental domains that were considered, and a summary of findings.
As is evident is Table 2, some of the available research focuses on relations between process quality measures and child behavior in the child-care setting. Other research considers relations between process quality and child behavior outside of child care. The former set of studies provide descriptions of children’s immediate reactions to caregiving experiences that are emotionally supportive and cognitively enriching versus experiences that are less supportive and enriching. These studies yield firsthand evidence about children’s reactions to care of varying quality. The latter set of studies considers whether reactions to quality experiences are evident in children’s behavior in other settings.
Process Quality and Children’s Behavior in Child Care. Several investigators have delineated systematic relations between process quality and children’s behavior in the child care setting (see Table 2). For example, controlling for child gender and family socioeconomic status, children appear happier in child care settings where activities are developmentally appropriate and caregivers are more involved (Hestenes, Kontos, and Bryan, 1993). Children show more intense negative affect when their caregivers are less involved with them. Children display closer and more secure attachment relationships with their caregivers when the caregivers are more positive and responsive to the children’s needs (Elicker et al., 1999; Howes et al., 1992; Howes and Smith, 1995).
Associations between caregiver-child interactions and children’s interactions with peers also have been reported (see Table 2). Children who have more positive interactions with their caregivers and more secure relationships with their caregivers appear more prosocial and positively engaged with their classmates (Holloway and Reichart-Erickson, 1988; Howes et al., 1992; Kontos and Wilcox-Herzog, 1997). Children who have opportunities to participate in activities such as art, blocks, and dramatic play demonstrate greater cognitive competence during their free play (Kontos and Wilcox-Herzog, 1997). Taken together, these studies suggest that experiences associated with better quality foster competent performance in the child care setting. By the same token, children are less likely to display competent behavior in child care settings characterized by lower process quality.
Process Quality and Children’s Behavior in Other Settings. The next issue is whether process quality is related to children’s behavior in other settings. Several studies (see Table 2) have found higher quality child care is associated with better performance on standardized language tests, even when family characteristics are controlled (Burchinal et al., 1996; Dunn, Beach, and Kontos, 1994; Goelman, 1988; McCartney, 1984; NICHD Early Child Care Research Network, in press-b; Peisner-Feinberg and Burchinal, 1997; Schliecker, White, and Jacobs, 1991). These relations are evident when the process measure is a global score such as the ITERS, ECERS, or FDCRS, and when the process measure focuses more narrowly on caregiver language stimulation. It is notable that associations between process quality and language performance are evident for child care that occurs in both centers and homes.
Children’s performance on standardized cognitive tests also has been linked to concurrent process quality. Infants who attend centers with higher ITERS scores receive better scores on the Bayley Mental Development Inventory than infants in poorer quality centers (Burchinal et al., 1996). Similarly, children who attend centers that have higher ECERS scores receive higher scores on the CBI intelligence scale (Dunn, 1993). The Cost, Quality, and Outcome Study reported that higher ECERS scores were associated with better scores on the reading subtest of the Woodcock-Johnson (Peisner-Feinberg and Burchinal, 1997).
Finally, process quality is related to children’s social and emotional functioning. High-quality care as measured by the ECERS is related to greater child interest and participation, whereas poorer process quality is associated with heightened behavior problems (Hausfather, Tohari, LaRoche, and Engelsmann, 1997; Peisner-Feinberg and Burchinal, 1997). The Bermuda Study (Phillips, McCartney, and Scarr, 1987) found higher ECERS scores to predict both caregiver and parent reports of children’s considerateness and sociability, and caregiver reports of children’s higher intelligence and task orientation and less anxiety.
Although the majority of studies (see Table 2) have reported significant relations between process measures of quality and concurrent child functioning, it should be noted that there are exceptions. Scarr and colleagues did not find relations between process quality and children’s social outcomes (McCartney et al., 1997). Measurement problems may have contributed to the lack of findings. For example, observers were only moderately reliable on the measures of quality, with exact agreement of 55–58 percent between sites on the ITERS/ECERS. Cross-site reliability in the classroom observations of children’s social behavior (a key dependent variable) also was poor to moderate, with kappa coefficients ranging from .40 to .76. The likelihood of detecting associations may have been hampered by unreliable measurements.
There has been a longstanding interest in structural and caregiver characteristics in relation to children’s developmental outcomes, in part because the structural and caregiver characteristics are easier to measure and to monitor than process quality. An early study—the National Day Care Study (Ruopp et al., 1979)—included a clinical trial in which 3- and 4-year-olds were randomly assigned to 29 preschool classrooms with different child:adult ratios and levels of staff education. Two levels of ratio (5.4:1 vs. 7.4:1) were contrasted along with three levels of staff education (B.A., Associate of Arts, or less than an A.A. in early childhood education). Child behaviors were assessed at the beginning of the intervention and 9 months later. Children assigned to classrooms with fewer children obtained greater gains on measures of receptive language, general knowledge, cooperative behavior, and verbal initiations, and exhibited less hostility and conflict in their interactions with others than did children assigned to classrooms with larger numbers of children. Children whose assigned teachers had more education and training achieved greater gains in cooperative behavior, task persistence, and school readiness than children whose teachers had less education and training.
Correlational studies also have reported concurrent associations between child:adult ratio and children’s language, cognitive, and social functioning. Infants who attend centers with smaller child:adult ratios are found to have better receptive and expressive language skills than children who attend centers with larger child:adult ratios (Burchinal et al., 1996; Vernon-Feagans, Emanuel, and Blood, 1997). Lower child:adult ratios also are associated with higher Bayley scores (Burchinal et al., 1996) and with better social knowledge and social behaviors (Holloway and Reichhart-Erickson, 1988).
Teachers’ education and training also are related concurrently to child performance and adjustment. Burchinal et al. (1996) report that infants have better expressive language skills when their caregivers are better educated. Preschoolers’ receptive language skills are higher when caregivers have at least an Associate of Arts degree in a child-related field (Howes, 1997). Children whose caregivers have degrees in child-related fields received higher CBI intelligence scores than children with less-educated caregivers (Dunn, 1993). Caregiver education and training in child care homes are similarly related to children’s performance on standardized cognitive measures (Clarke-Stewart et al., 2000).
Observations of children’s experiences in classrooms and child care homes suggest why these relations might occur. Children are more likely to engage in language activities, complex play with objects, and creative activities in their classrooms when teachers have bachelor degrees in child-related fields (Howes, 1997). Toddlers are more likely to talk with their caregivers and to engage in complex play when classrooms have smaller child:adult ratios (Howes and Rubenstein, 1985). Toddlers are more likely to cry and to have their actions restricted in classrooms in which group sizes are larger (Howes and Rubenstein, 1985). In child care homes, positive caregiving is more likely when group sizes are smaller, caregivers are more educated, and caregivers have more specialized training pertaining to children (Clarke-Stewart et al., 2000).
An alternative research strategy has been to consider aggregated structural and caregiver characteristics. For example, the NICHD Study of Early Child Care (NICHD Early Child Care Research Network, 1999a) assessed four structural and caregiver characteristics (child:staff ratio, group size, caregiver specialized training in child development or early childhood education, and caregiver formal education) in terms of guidelines recommended by the American Public Health Association. The investigators then summed the number of structural and caregiver characteristics that met recommended guidelines, resulting in summed scores of 0 to 4. At 24 months, 10–12 percent of classrooms met all four standards, whereas 34 percent of the classrooms did so at 36 months. At 24 months, 9 percent of the observed centers met none of the recommended standards; 3 percent of the centers met none of the standards at 36 months.
Associations between the number of child care standards that were met and child outcomes were then tested, with family income and maternal sensitivity controlled (see Table 4). Children who attended centers that met more recommended guidelines had fewer behavior problems at 24 and 36 months, and higher school readiness and language comprehension scores at 36 months. There were significant linear trends between the number of recommended standards that were met and children’s concurrent adjustment.
Analyses also compared children who were enrolled in classrooms that met a given individual standard with children whose classrooms did not meet that standard (see Table 4). At 24 months, children displayed fewer behavior problems and more positive social behaviors when centers met the recommended child:adult ratio. At 36 months, children whose caregivers had specialized training or who had more formal education exhibited fewer behavior problems and obtained higher school readiness and language comprehension scores.
Investigators also have considered longer-term associations between process quality and children’s developmental outcomes. A compilation of these studies can be found in Table 3. Included in the table are studies that considered relations between earlier child care experiences and later adjustment. To our knowledge, there are no published accounts that relate early child care quality to children’s adjustment beyond middle childhood. Table 3 presents information regarding sample size, controls for family factors, descriptions of the quality measures, descriptions of the child outcomes, and specific findings.
Findings on this issue have been reported by the NICHD Study of Early Child Care. Extensive information about the children, the families and child care was collected during home visits (1, 6, 15, 24, and 36 months), child care visits (6, 15, 24, and 36 months), and laboratory assessments (15, 24, and 36 months). Phone interviews were conducted every 3 months to track hours and types of child care. Children who were in nonmaternal care for more than 10 hours a week were observed in that care. The investigators (NICHD Early Child Care Research Network, 1998; 1999b; in press-b) asked if cumulative positive caregiving (the average of ORCE positive caregiving ratings collected during visits at each observation) is related to child developmental outcomes at 24 and 36 months. Mental development at 24 months was assessed in the laboratory with the Bayley. School readiness was measured using the Bracken School Readiness Scale, a scale that assesses knowledge of color, letter identification, number/counting, shapes, and comparisons. Expressive language skills and receptive language skills were measured at 36 months using the Reynell Developmental Language Scales. Mother and caregiver reports of child behavior problems were obtained using composite scores from the Child Behavior Checklist and the Adaptive Social Behavior Inventory. Peer skills were assessed during a videotaped semistructured play situation with a friend.
Relations between cumulative positive caregiving and child development were tested in analyses that controlled for child and family factors (child gender, maternal education, family income, maternal psychological adjustment, home quality assessed by Bradley and Caldwell’s HOME scale and videotaped observations of mother-child interaction) and other aspects of child care (time in center and total hours in care from 3 to 36 months). Table 6 summarizes findings from regression analyses and resultant partial rs that indicated effect sizes. As shown, the quality of child care during the first 3 years was related to children’s school readiness, expressive language, and receptive language at 3 years. Also shown on Table 6 are comparisons of children in high-quality and low-quality child care (defined with quartile splits), using the same covariates. This extreme group approach yielded d statistics. Effect sizes using this extreme group approach were significant for measures of school readiness, expressive language, and receptive language at 36 months.
In order to evaluate the magnitude of these findings, the NICHD investigators conducted parallel analyses that tested relations between quality of the home environment during the first 3 years and the child developmental outcomes (using the same covariates), and relations between child care hours during the first 3 years and child developmental outcomes (using the same covariates). Table 6 presents these effect sizes as well. Effects associated with quality of the home environment (the cumulative composite scores created from the Bradley and Caldwell HOME scale and mother-child interaction ratings) were roughly twice the size of the child care quality score. Effects associated with child care hours were substantially smaller than effects associated with child care quality. The NICHD investigators argued that these findings suggest effects of child care quality assessed longitudinally to age 3 years were neither huge nor trivial, but were large enough to be meaningful. It also should be noted that these effect sizes are likely to be a conservative estimate because of the selective participation by higher-quality settings. If the poorest quality child care settings refused to allow observations to be conducted, the range of quality scores would be truncated, resulting in smaller effect sizes.
Longer-term findings obtained from the Otitis Media Study (Burchinal, Roberts, Riggins, Zeisel, Neebe, and Bryant, in press) are consistent with those reported in the NICHD study. In that study, hierarchical linear models were tested. Observations of classroom quality obtained annually over a 3-year period were used to predict children’s adjustment up to age 3 years. Higher-quality child care over time was associated with better cognitive development, better receptive and expressive language skills, and better functional communication skills over time, controlling for child gender, family poverty status, and home environment quality.
A limitation with both of these reports is that children were studied only to age 3. Thus, it cannot be ascertained if early effects are harbingers of later differences or if these effects dissipate by the time that children enter grade school. As additional findings from these ongoing investigations become available, they can be used to identify conditions under which early child care quality differences are maintained or dissipate.
In the meantime, the Cost, Quality, and Outcomes Study has information that is relevant to this issue (Peisner-Feinberg et al., 1999). Started in 1993, observations were conducted in child care centers located in four states—California, Colorado, Connecticut, and North Carolina—that varied in licensing standards. Centers were evenly distributed in each state into nonprofit and for-profit programs. Within the eligible programs, 509 preschool classrooms and 224 infant/toddler classrooms were studied. Process quality was rated using the ECERS or ITERS, the Caregiver Interaction Scale (Arnett, 1989), and the Teacher Involvement Scale (Howes and Stewart, 1987). Quality indicators were combined into a single process quality composite.
A subsample of children was followed through 2 years of child care and the first 3 years of formal schooling (kindergarten through second grade). Children were assessed for receptive language skills, reading ability, and math skills. Child care and school teachers rated the children’s cognitive/attention skills, sociability, and problem behaviors each year. Longitudinal hierarchical linear models examined relations between the child care quality composite collected at age 4 (Time 1) and children’s developmental outcomes through grade 2. In all analyses, selection factors (maternal education, child’s gender and ethnicity) were controlled statistically.
Children who were enrolled in higher-quality child care classrooms as preschoolers were found to have better receptive language skills. Effect sizes for receptive language were moderate for the preschool period (.60 and .51 for the 2 years preceding school entry), more modest in kindergarten (.30), and not significant in second grade. Child care quality also was related to children’s math skills. Children who were enrolled in higher-quality child care had better math skills prior to school entry and during kindergarten and second grade, with modest effect sizes across the years (.20–.29). The relation was stronger for children whose mothers had less education. In further analyses that controlled for the quality of the elementary school classroom, the relations between child care quality and children’s math skills were maintained. It is notable that a similar finding was obtained in research conducted in Sweden. Broberg et al. (1997) found that process quality assessed using the Belsky and Walker checklist at 16, 28, and 40 months predicted better math skills at age 8, even after controlling for child and family factors.
Other research has considered longer-term associations between child care quality and children’s social-emotional outcomes. Howes (1990) focused on one particular aspect of process quality, child care socialization practices, in relation to children’s subsequent developmental outcomes. Caregivers’ involvement and investment in child compliance were measured during naturalistic observations in the child care setting. Having a more involved and invested caregiver during the first 3 years was associated with kindergarten teachers’ reports that the children had fewer behavior problems and better verbal IQs.
Alternative Views. As shown in Table 3, some investigators have not found relations between child care quality and later developmental outcomes. For example, Chin-Quee and Scarr (1994) did not find evidence of long-term effects in a longitudinal follow-up of the Bermuda study. In the initial study, concurrent associations were reported between process quality as measured by the ECERS and child developmental outcomes (McCartney, 1984; Phillips et al., 1987). In the follow-up study, teachers rated social competence (peer relations and cooperative behavior) and academic achievement for 97 of the original sample of 166, when children were in grades 1 and 2 (Time 2) and grades 3 and 4 (Time 3). Associations between the quality indicators during the preschool years and competence at school were tested with hierarchical regressions in which parental values, age of entry into care, and total amount of child care before school entry were controlled. Neither the global quality score nor the specific measures of caregiver language predicted children’s social competence and academic achievement at Time 2 or Time 3.
A longitudinal follow-up of children who participated in the Three-State Study also failed to detect long-term effects (Deater-Deckard, Pinkerton, and Scarr, 1996). In this project, assessments of child care quality were first obtained in 363 classrooms located in 120 centers in three states (Georgia, Virginia, Massachusetts) when 718 study children were infants, toddlers, and preschoolers. Process quality ratings were obtained by pulling items pertaining to teacher-child interaction from the ECERS and ITERS and the Assessment Profile (a process measure scored for presence or absence of specific items). Four years later, follow-up assessments were conducted for 141 of the original sample. Multiple regressions controlled for child (child adjustment at Time 1, age at Time 2, child gender) and family characteristics (SES, a composite of parenting stress and low emotional support, maternal endorsement of harsh discipline practices). The child care quality measure was a composite of the ITERS/ECERS, the Assessment Profile that measures physical facilities, caregiver training and education, and caregiver wages. In these analyses, the child care quality composite score at Time 1 did not predict changes in children’s behavior problems or social withdrawal at Time 2.
Although Scarr (1998) has argued that these studies demonstrate that child care quality has little or no long-term impact on children’s development, the findings must be interpreted with caution. Both studies are based on the assumption that a quality assessment obtained at one point in time is an adequate and accurate representation of child care quality. Single assessments might be sufficient if care arrangements and quality are stable; however, a single observation is not adequate if care is unstable or changing. In the Bermuda sample, Chin-Quee and Scarr (1994) reported that half the children experienced one, two, or three arrangements during the intervening period, and half experienced more than three arrangement changes. In the Three-State study, no information about child care quality in the intervening four years was collected. In both studies, it is difficult to interpret the meaning of the null findings in light of no information about child care quality across early childhood. Stronger, more valid tests of the effects of child care quality need to take into account cumulative quality and the pattern of quality over time.
The lack of long-term relations in the Three State Study may also reflect limitations in the assessment of process quality. Only moderate interobserver agreement was reported across the three research sites—.58 for the ECERS and .55 for the ITERS (McCartney et al., 1997). Lower relations between process quality and child outcomes would be expected when process quality scores are less reliable.
Other studies have considered relations between structural and caregiver characteristics in relation to children’s subsequent developmental outcomes (see Table 3). Howes (1988), for example, examined structural and caregiver characteristics at 3 years in relation to children’s first-grade adjustment. Quality in 81 centers was defined in terms of five areas: teacher training, child:adult ratio, group size, a planned curriculum, and space. Higher-quality care met recognized standards in all five areas; medium-quality care met standards on three or four dimensions, and low-quality care met three or fewer standards. During the intervening period, the 87 children attended the same university lab school, meaning that they experienced classes with the same or similar structural and caregiver characteristics.
In analyses that controlled for maternal work status, family structure, and maternal education, Howes found that children who had attended higher-quality child care programs prior to enrollment in the university school had fewer behavior problems and better work habits as compared to children who had attended lower quality programs. Additionally, boys who had attended higher-quality centers received better first-grade teacher ratings of academic performance compared to other boys.
Using a different sample of 80 children who were enrolled in center-based care, Howes (1990) examined relations between a structural quality composite (child:adult ratio, caregiver training, caregiver stability) measured at 18, 24, 30, and 36 months, and children’s kindergarten adjustment. High-quality care was defined as ratios of 4:1 or less for children who were < 2 years and 7:1 for children > 2 years, caregivers with 12 units of college-level child development courses, and no more than two different primary caregivers in the prior year. Low-quality care was defined as ratios of 6:1 or higher for children who were 2 years or less and ratios of 10:1 for children who were older than 2 years, caregivers with no formal child development training, and more than two primary caregivers in the prior year.
Associations between structural quality in the first 3 years and children’s later preschool and kindergarten adjustment were tested, controlling for a family socialization composite and a family demographic composite. Children with a history of poor-quality child care during the first 3 years were rated by their preschool teachers as being more difficult and by their kindergarten teachers as being more hostile. The children also engaged in less social pretend play and displayed less positive affect in their preschool classroom.
Recent research from the Otitis Media Study has focused on specific structural and caregiver characteristics in relation to subsequent child developmental outcomes (Burchinal et al., in press). The researchers initially recruited 89 children who were 4 to 9 months of age for a study of the effects of otitis media on children’s development. Children attended 27 centers that varied in quality. Child care quality was assessed annually using the ECERS and ITERS. Children whose child care classrooms met recommended guidelines for child-staff ratios exhibited better receptive language and functional communication skills over time as compared to children whose classrooms did not meet recommended ratio guidelines, controlling for child gender, family poverty, and cognitive stimulation and emotional support in the home. Caregiver education also predicted children’s adjustment, but only for girls: 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 compared to girls whose caregivers had fewer than 14 years of education, controlling for the family factors.
Blau (1999c) also has examined structural and caregiver characteristics in relation to children’s subsequent developmental outcomes. For these analyses, he used secondary data obtained from the National Longitudinal Survey of Youth (NLSY), an ongoing nationally representative study of 12,652 youth begun in 1979. Beginning in 1986, information about children of the female respondents was collected. Mothers also provided information about their children’s primary child care arrangements—the number of children cared for in the group, the number of adult care providers in the arrangement, and whether the main caregiver had specialized training in early childhood education or child development. Blau then averaged these maternal reports of structural and caregiver characteristics through age 2 and for ages 3–5 years. Children completed the Peabody Picture Vocabulary Test (PPVT), a measure of receptive language skills, at 3 years or older. Mothers reported on children’s behavior problems at 4 years or older. Children completed math and reading subscales of the Peabody Individual Achievement Test (PIAT) at 5 years or older.
Simple correlations revealed statistically significant, but small, associations between mothers’ reports of caregiver training when the children were in infant/toddler care and the children’s later performance. Children whose mothers reported that their caregivers had more specialized training obtained higher math and receptive language scores. When type of care was controlled, these associations continued to be significant. Blau then asked if these structural and caregiver factors uniquely predicted child performance in a regression model that included 64 additional child care and family variables. These controls included number of arrangements that were used, hours per week in care, months per year in care, paid cash for care, cost of care, center care, family day care home, relative care, child gender, cognitive stimulation, emotional support, Hispanic ethnicity, black ethnicity, grandmother worked when mother was 14, mother’s education, grandmother’s education, fraction of mother’s preschool years her mother was present, fraction of mother’s high school years her father was present, month of pregnancy in which mother first received prenatal care, child’s birth order, Catholic, child received well-care visit in first quarter, mother’s age, mother’s age at birth of child, siblings in various age groups, and fraction of pregnancy during which mother worked. In ordinary least squares regression analyses, relations between maternal reports of caregiver training and children’s math and receptive language scores were no longer evident when these other variables were controlled. From these analyses, Blau concluded:
“There seems to be little association on average between child care inputs experienced during the first three years of life and subsequent child development, controlling for family background and the home environment.” (p. 20)
Blau’s conclusion does not appear warranted, for several reasons. First, his analyses relied on maternal reports of structural and caregiver characteristics. Questions can be raised about whether mothers can provide this information accurately, especially retrospectively. Unfortunately, Blau provides no evidence regarding the accuracy of these reports. In order to estimate the accuracy of mothers’ concurrent reports of structural and caregiver characteristics, we turned to the NICHD Study of Early Child Care data set, which included both mothers’ and caregivers’ reports of group size and child-adult ratio. These reports were compared to observers’ independent counts of ratio and group size during 2-day visits. The mean correlation between mothers’ and caregivers’ reports of group size for children in centers was .55 (range = .51 to .63). The mean correlation between maternal reports of child:adult ratio and observed ratios was .33 (range = .27 to .42). These figures suggest that mother concurrent reports can be viewed as moderately reliable. Maternal retrospective reports of group size and ratio appear to be considerably less reliable. In other studies, near-zero correlations were obtained between observational assessments of group size and child-adult ratio when children were age 4 years (Vandell and Powers, 1983) and maternal retrospective reports of these same structural variables 4 years later (Vandell, Henderson, and Wilson, 1988).
To our knowledge, there are no data available from which the accuracy of maternal reports of caregiver training can be evaluated. We suspect, based on our own personal experiences, that mothers are less likely to know about caregiver training than about group size and ratio, which they can observe. Taken together, we believe that the lack of precision in the mothers’ reports in the NLSY result in an underestimation of effects associated with structural and caregiver characteristics.
Blau also adopted a stringent, perhaps unrealistic, test for long-term effects. Child outcomes were assessed a minimum of 2 years after mothers reported structural and caregiver characteristics, and the lag appears to have averaged 5 years or more because children were reported to be, on average, 8 years of age when outcomes were assessed. Interestingly, there was some evidence of longitudinal associations when shorter time lags were considered (even though mothers’ reports were used). For example, significant relations were found between maternal reports of child:adult ratios and caregiver training during the first 3 years and behavioral adjustment and math scores for children who were less than 9 years of age. Relations were not evident for very long time period, i.e., children who were older than 9 years. Smaller group sizes during the preschool period (3–5 years) were associated with higher scores on math, reading, and language performance. Lower child:staff ratios were associated with fewer behavior problems. The long lag between the infant quality reports and the child outcome assessments is further complicated by the omission of quality reports during the older preschool years, resulting in an underestimation of effects associated with child care quality.
Conclusions. Structural and caregiver characteristics have been found to be associated with children’s academic, cognitive, behavioral, and social development. Smaller group sizes, lower child-caregiver ratios, and more caregiver training and education appear to have positive effects on these important developmental outcomes. Future work might address threshold levels for these child care characteristics, or the point at which further improvements in structural quality do not yield additional developmental benefits for children.
The traditional approach of those working in the field of developmental psychology is to use standardized regression coefficients in hierarchical regression models. These seem quite different from the methods employed by economists, but the full model used is a standard OLS regression model, and the standardized coefficients can be converted into the nonnormalized coefficients more traditional in the economics discipline. Doing so allows us to address the question of the expected change in developmental outcomes of children were quality to be improved based upon standard OLS estimates (see Hanushek and Jackson, 1977).
Using estimates based on NICHD data, we perceive that the quality of child care can indeed make a difference. Table 7 reports on the cognitive and language results for the NICHD samples at ages 15 months, 24 months and 36 months (NICHD Early Child Care Research Network, in press-b). Three outcome measures are used: for the younger two ages, the Bayley, CDI vocabulary production, and CDI vocabulary comprehension tests are used; for the oldest age group (36 months) the Bracken school readiness, Reynell expressive language and Reynell verbal comprehension tests are used. (See above for a more detailed discussion of these tests.) In addition to the measures of child care quality, the model includes measures of parental background, quality of the home, the child care setting, and time spent in child care. These are an attempt to minimize the role of parental selection of child care in order to capture the effects of child care quality differences on measures of child development.
Two models are presented for each age group and outcome measure. The first tests for the effects of the child care quality using a cumulative score of positive caregiving rating, while the second adds a specific measure of language stimulation. Caregivers’ behaviors were assessed during four 44-minute observations over two half-days at 6, 15, 24, and 36 months. These were combined into the cumulative positive caregiving rating.
We converted the standardized coefficients reported in NICHD (in press-b), into nonstandardized coefficients. Table 7 only reports those for child care quality that are statistically significant at the 5 percent level. Combining these with the measures of quality and reinterpreting the standardized coefficients indicates the following effects.
The expected improvement in the CDI vocabulary production test for toddlers aged 15 months, when their care quality shifts from one standard deviation below the mean to one standard deviation above, is nearly 7 points, or 24 percent; if the shift is from the minimum score (5) to the maximum (20) in caregiver rating, the estimated gain is 18 points. (Note that the standard deviation is 2.9, within a range of 5–20.) For the vocabulary comprehension score, shifting from one standard deviation below the mean leads to an expected increase of 8 points; and moving a child from the minimum score to the maximum score in caregiver rating is expected to increase a child’s score by 21.6 points, or 55 percent of the mean.
At age 24 months, statistically significant changes are registered for the Bayley test and the sentence comprehension test. The Bayley estimates produce an expected increase of about 5 points when the shift is from one standard deviation below the mean to one above. The sentence comprehension results produce an expected increase of 7 points. A shift from the lowest caregiver to the highest increases expected performance on the vocabulary test by about 40 percent relative to the mean, but only about 13 percent in the case of the Bayley test. (Note that for the 24 month olds the standard deviation is 2.9 as well, and the range is the same as that for those 15 months old, 5–20.) The expected changes are larger for those aged 36 months than for those aged 24 months.
For the Bracken school readiness test at age 36 months, a shift from one standard deviation below the mean on the caregiving rating to one above is expected to lead to an increase of 6.9 points. (Note that the caregiver rating has a larger range for those aged 36 months, 7–28, while the standard deviation is 3.3.) The same shift in caregiver quality is expected to lead to a 5-point increase in the expressive language score and an 8.6 point increase in the verbal comprehension score, relative to the mean. A shift from the lowest rating to the highest for caregiver rating is expected to result in a shift of about 50 percent relative to the mean for each of these three outcomes. These estimates give us some sense of the magnitude of the possible changes in children’s outcomes as a result of improvements in positive caregiver rating.
We can do a similar exercise with the second measure of quality, language stimulation. This measure is added to the regression in an alternative specification. In most cases the addition of this measure reduces the estimated impact of positive caregiving. (Because language stimulation is a major component of caregiving quality, this is not surprising.) We simulate the impact of language stimulation only for ages 15 months and 24 months, where it is statistically significant for all three of the child development outcome measures.
For children 15 and 24 months of age, having a child care arrangement in which more language stimulation is provided can play a small but significant role in improving all three of the outcome measures. For children of 15 months, simulating an improvement in caregiver language stimulation from one standard deviation below the mean to one above increases performance on the Bayley test by nearly four points, by 12 points on the CDI vocabulary production test and by about 9 points on the CDI sentence comprehension test. At 24 months of age, a child exposed to a level of language stimulation one standard deviation below the average is expected to gain about 5 points on the Bayley test, or about 4 percent relative to the mean, were the child moved to an arrangement with a rating one standard deviation above the norm, or by 61 units (about one third of the full range of 0–180). The same child would be expected to gain 12 points on the CDI sentence comprehension test, or nearly 30 percent relative to the mean. And the same child would be expected to gain about 9 points on the CDI vocabulary production test, or about 20 percent relative to the mean.
The NICHD group also estimated models with both concurrent and lagged quality measures (NICHD, in press-b). Their results suggest that the cumulative impact of child care quality may be far greater than the concurrent impact. In Table 8 we provide two examples, one for children at 24 months of age using caregiver quality measures at 15 and 15–24 months, and one at 36 months of age using quality measures from 15–24 months and 36 months. Only those quality measures which are statistically significant at the 5 percent level are reported. Similar measures for language stimulation are chosen as the cognitive outcome measure. Both sets of results suggest that there is a lagged effect of language stimulation and that the cumulative effect of exposure to higher levels of language stimulation appears to be greater than the concurrent impact alone. In the case of those aged 24 months, the converted results suggest that a child who was in a child care arrangement over this entire period that was one standard deviation below the mean (calculated at both 15 and 24 months) is simulated to gain about 30 points on a CDI vocabulary production test if moved to one above the mean. Looking at only the concurrent language stimulation, an increase of 11 points, or about one-third of the full cumulative gain, is simulated. In the case of 36 those aged months, we find that both language stimulation and overall caregiver quality are significantly associated with performance on the Reynell vocabulary comprehension test measured at 36 months, and that only the lagged measure of language stimulation is statistically significant among the two lagged measures of quality. Among these 36-month-olds, we simulate a combined gain of 11 points on the Reynell vocabulary test if the child moved from one standard deviation below the mean on both quality measures to one above the mean over the entire period from 15 months to 36 months. This far exceeds the simulated increase if only the contemporary measures were modified. The respective means of the outcome measures are shown on the table.
The magnitudes of simulated change are large, but the resulting changes are also rather large, suggesting that caregiver quality can significantly influence these outcomes. All of these estimates are subject to the usual caveat that the underlying estimates may not be causal. We should also note that the measures of quality are based on limited observation, so that the effect captured is likely to underestimate the true effect of quality.
We now briefly review the findings from a small set of early childhood intervention studies that look at such long-term outcomes as criminal activity, earnings, and the use of cash welfare assistance. The Syracuse Family Development Research Program was a small program that enrolled slightly more than 200 children and followed them for 5 years. It was one of the earlier programs, beginning in 1969. The intervention included infant and preschool enrollment in Syracuse University’s preschool program as well as direct provision of information on raising children, nutrition, etc. to the parents. The follow-up analysis found that by age 15, 6 percent of the experimental group had been referred to probation as compared to 22 percent of the controls. Through based on very small numbers, the very large differences provide some evidence that the combined interventions had a positive effect on the reduction of crime (Lally, Mangione and Honig, 1988).
Perhaps the best-known early intervention project is the Carolina Abecedarian Project (Campbell and Ramey, 1995; Ramey, Campbell, and Blair, 1998; Ramey et al., in press). This clinical trial began at 6 weeks postpartum and included (1) a randomized control group (n = 23) that received family support social services, pediatric care, and child nutritional supplements, (2) an experimental group (n = 25) that received the services of a high-quality center-based intervention for the first 5 years and additional educational support services from kindergarten to grade two, (3) an experimental group (n = 24) that received only the early intervention, and (4) an experimental group (n = 24) that received only the K-2 educational support. IQ scores at 8 years and 12 years were significantly higher for preschool participants than for other children. Furthermore, children who had participated in the preschool program had higher scores on tests of reading and mathematics achievement at 8 and 12 years. They were less likely to be retained a grade at ages 8, 12, and 15, and they were also less likely to be placed in special education. The most recent follow-up report from this research team (Early Learning, Later Success: the Abecedarian Study, 1999) included findings to 21 years. 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.
The Perry Preschool Project (Schweinhart, Barnes, Weikart, et al., 1993) involving 123 black children has reported long-term follow-up to 27 years. 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 versus 69 percent), but a larger difference in the average number of lifetime arrests by age 27 (2.3 versus 4.6). Differences in the proportion receiving public assistance by age 27 were also large: 15 compared to 32 percent. Mean earnings were far higher for the experimental group than the control group at age 27: monthly reported mean earnings were $1,219 for experimentals, $766 for controls.
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 African-American (93 percent) and Hispanic (7 percent) children as they grew up in high-poverty neighborhoods in central city 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 have been collected since that time. Children who participated in the CPC preschool programs obtained significantly higher math and reading achievement test scores at 5, 8, and 14 years, even after controlling for family risk status, child gender, and later program participation. At age 20, participants in the CPC were more likely to have completed high school and to have low rates of juvenile crime.
Even though only a few studies have followed children into adulthood, it is notable that all find some evidence of long-term gains.
In much of the existing literature linking parental employment and child care, the primary issue is the affordability of care and the elasticity of response to child care costs. In this sense one can see the potential for a trade-off between quality of care and labor force participation, in that higher-quality care is likely to be more costly. A parent facing that higher cost may decide to forgo or limit employment or to elect lower-quality and less costly care (Scarr, 1998). Maume (1991) found that a $10 increase in the weekly cost of child care was associated with a 1.6 percent increase in the probability of exiting employment within a year. A slightly earlier study by Blau and Philip (1989) also provided evidence that an increase in the cost of care was associated with an increased probability of a mother leaving the labor force.
Quality of care may influence employment in several ways: parents may be reluctant to leave their child in a low-quality, unsafe environment or with adults who do not provide a stimulating or warm environment for their child. This may be a particular problem for lower-income families, who have more limited choices of providers. In contrast, a safe, warm, stimulating environment may encourage employment and longer hours of work. Parents may also be more effective employees if they do not have concerns about the environment in which their children spend a good part of each working day. Having well-cared-for children may also lead to employees with higher productivity than those whose children are left in less satisfactory environments. Parents may also be more likely to be on time to work and less likely to miss time from work if their children are cared for in a safe, warm and stimulating environment.
Evidence. There is limited evidence on whether higher-quality care has positive impacts on parental employment, because there have been few studies. The available evidence suggests that among low-income women, higher-quality child care may increase employment, stability of employment, and hours of work. See Table 8 for detail on these studies.
Meyers (1993) has explored how a mother’s perception of the quality of her child’s care and the convenience of the child care arrangement affected her labor market progress in the JOBS program. The results show that a mother’s perception of the safety of her child’s care arrangement and the trustworthiness of the provider were significant predictors of the mother’s continued participation in the JOBS program and in labor force participation more generally. Mothers also responded to the ratio of children to staff: those mothers who reported that the ratio of the children to adults in their care arrangement exceeded professionally recommended standards were more than twice as likely to drop out of the program than mothers who reported that their child’s care arrangement met the standard.
Additional evidence that quality plays a role in women’s labor force participation comes from another experiment: the Teenage Parent Demonstration. As reported in Ross and Paulsell (1998), among the group of first-time teenage welfare recipients whose members were randomly assigned to the program, which required employment, job training or schooling, nearly a third reported that the quality of child care was a problem that led them to stop working or to change hours and/or activities. Ross and Paulsell interpret this to mean that “Mothers who are required to work as a condition of receiving welfare benefits may try to manage with lower-quality child care than they would in the absence of such a requirement, but this low-quality care may be the reasons that mothers interrupt their employment activities.” (p. 40)
Brooks-Gunn et al. (1994) evaluated an early intervention program, the Infant Health and Development Program, in terms of its impact on mothers’ employment. This experiment focused on low-birthweight infants and used random assignment to an intervention program that provided center-based child care when the child was 2 to 3 years of age. The authors report that mothers in the intervention group were significantly more likely to be working than women in the control group. Effects were particularly pronounced for mothers with a high school degree or less schooling. Similar beneficial effects of high- quality child care on mothers’ subsequent educational achievement were evident in the high-risk sample that participated in the Abecedarian clinical trials (Ramey et al., in press). Benasich, Brooks-Gunn, and Clewell (1992) report qualitatively similar results for several other early intervention programs (pp. 41–42).
Other evidence (see Hofferth, 1999) that child care quality can influence labor market participation can be found in research studies that differentiate formal arrangements from informal arrangements. These analyses show that parents miss work more often and are late more often if they use informal arrangements. The big difference seems to be the stability of the arrangement—an element of quality.
The evidence reviewed above indicates that child care quality has meaningful effects on children and their parents. Our next question considers the quality of the care that is available in the United States. One part of this question is a determination of whether high-quality care (of the sort that fosters positive developmental outcomes) is the norm or the exception. The flip side of this question is a determination of the likelihood that children are in poor-quality care that can impair development. Unfortunately, at the current time it is not possible to provide a definitive response to these questions, because observations of process quality have not been conducted for a nationally representative sample of children. In the absence of such a report, we must reply on existing data from multisite studies that provide suggestions about the distribution of quality of care in the United States.
The Cost, Quality, and Outcomes Study (Helburn et al., 1995) provides a perspective on center-based care. ECERS assessments were conducted in 398 centers located in four states that varied in economic health and child care regulations. In that study, 12 percent of the centers received ECERS scores lower than 3, indicating care that was less than minimal quality, and 15 percent received ECERS scores higher than 5, indicating good-quality care. The remainder of the centers were evenly divided between those receiving scores in the 3s (37 percent) and scores in the 4s (37 percent). This distribution of quality scores in the observed settings, however, may be an optimistic view. The observed centers represented only 52 percent of the eligible centers; the remainder declined to participate. It seems likely that the nonobserved settings offered care that was lower in quality.
The Relative and Family Day Care Study (Kontos, Howes, Shinn, and Galinsky, 1995) provides a perspective on quality of care in homes. FDCRS scores were obtained in 226 child care homes and relative care settings in three communities. Minority race, low-income, and nonregulated home settings were over sampled so that the investigators could study the effects of these factors on observed quality. In that study, 34 percent of the child care homes received FDCRS scores of less than 3 and were described as “inadequate,” 58 percent were “adequate/custodial,” and 8 percent were “good.” These unadjusted quality estimates are probably negatively biased, because two of the three states (Texas and North Carolina) have less stringent regulations for child care homes than other states and because nonregulated and low-income settings were over sampled.
Perhaps the best available estimate of process quality for children 3 years or younger is provided by the NICHD Study of Early Child Care. Observations were conducted in nine states (Arkansas, California, Kansas, Massachusetts, North Carolina, Pennsylvania, Virginia, Washington, Wisconsin) and included urban, suburban, and rural communities. The distribution of child care regulations in those states paralleled those in the United States. Observations were conducted in all types of nonmaternal care settings, including grandparents, in-home caregivers, child care homes, and centers. A total of 612 child care settings were assessed at 15 months, 630 child care settings at 24 months, and 674 child care settings at 36 months.
The study sample of 1,364 families was drawn from hospitals at the 10 research sites and included ethnic minorities (24 percent), mothers without a high school education (10 percent), and single-parent households (14 percent) as well as white, middle-class and two-parent households. At 15 months, 17 percent of the households had incomes below established poverty levels (income-to-needs ratio < 1.0). An additional 18 percent of the sample had incomes near poverty (income to needs ratio 1.0 –1.99) (NICHD Early Child Care Research Network, 1997). The sampling plan yielded a large and diverse sample, but it is not nationally representative. The sampling plan also did not include adolescent mothers (3.8 percent of the potential families in the hospitals), mothers who did not speak English (4.4 percent), and infants of multiple births, with obvious disabilities, or extended hospital stays postpartum (7.7 percent of the births).
Results from the NICHD Study of Early Child Care observations (NICHD Early Child Care Research Network, in press-a) are summarized in Table 9. ORCE ratings less than 2 indicate poor-quality care. Scores of 2 to less than 3 indicate fair-quality care. Scores equal to 3 but less than 3.5 indicate good-quality care, and scores of 3.5 or higher indicate excellent-quality care. Care was most often judged to be only fair in quality. Relatively little care was observed at the extremes, with 6 percent of the settings offering poor quality care and 11 percent of the settings offering excellent care. Poor-quality care was more likely in centers serving infants and toddlers than in centers serving older children (10 percent versus 4 percent).
An extrapolation to the quality of care in the United States was derived by applying NICHD observational parameters, stratified by maternal education, child age, and care type to the distribution of American families documented in the National Household Education Survey (1998).(2) This stratification was needed because the NICHD investigators determined that variations in process quality were associated with these three factors. Based on the numbers of children of particular ages using specific different types of care, positive caregiving was estimated to be of poor quality for 8 percent of children under 3 years in the United States, fair quality for 53 percent, good quality for 30 percent, and excellent quality for 9 percent. These distributions suggested to the investigators that care is “neither outstanding nor terrible, but plenty of room for improvement [remains].”
The quality of child care in the United States also can be estimated based on reports of structural and caregiver characteristics. Drawing on empirical research and advice from professionals in the field, organizations such as the American Academy of Pediatrics and the American Public Health Association (1992) have established age-based guidelines for group size and child:adult ratio. For example, the recommendations for child:adult ratios are 3:1 for children from birth to 24 months, 4:1 for children from 25 to 30 months, 5:1 for children from 31 to 35 months, 7:1 for 3-year-olds, and 8:1 for 4-year-olds.
Table 10 lists regulations for child:adult ratio and group size for each of the 50 states as compiled by the Center for Career Development in Early Care and Education (1999). It is clear that very few states have regulations as strict as those recommended by professional organizations. For example, only three states have the recommended 3:1 ratio for infants, and only one state has the recommended 3:1 ratios for 18-month-olds. Two states have ratios consistent with the recommended 5:1 ratio for 3-year-olds. Some states are at substantial odds with the recommended standards. For example, eight states have child:adult ratios of 6:1 for infants. There is a similar failure to meet recommended group size standards, with 20 states having no regulations pertaining to group size.
Another way of estimating the quality of care in the United States is to consider reports of structural and caregiver characteristics. One nationally representative survey, the Profile of Child Care Settings (Kisker, Hofferth, Phillips, and Farquhar, 1991), obtained this information in 1990 from child care centers, early education programs, and licensed child care homes. According to the Profile, the average child:adult ratio was 4:1 for infants under 1 year of age, 6:1 for 1-year-olds, and 10:1 for preschoolers. This report indicates that the average center and child care home in 1990 did not meet standards for child:adult ratios that have been linked to higher quality. In contrast, the Profile of Child Care Settings found that caregivers tended to be well educated and to have specialized training pertaining to children. Nearly half of all teachers reported that they had completed college (47 percent) and an additional 13 percent reported a two-year degree. Most of the remaining teachers had a Child Development Associate (CDA) (3)credential (12 percent) or some college experience (15 percent). Only 14 percent did not have any education beyond high school. Ninety percent of the teachers in child care centers reported that they had received at least 10 hours of in-service training.
The Profile survey found that regulated child care home providers had less formal education and training than teachers in centers. Approximately 11 percent of regulated home providers reported that they had completed college; 34 percent had no schooling beyond high school. About two-thirds had received specialized in-service training. This study represents the best available information regarding structural and caregiver characteristics from nationally representative samples. The survey is dated, however, in that the data were collected in 1990, so the reports may not reflect current structural and caregiver characteristics.
Published reports from two additional national surveys are less useful for this issue. The National Child Care Survey,1990, and the National Household Education Survey, 1995, collected information from parents regarding child:adult ratios and caregiver training. The published reports from these surveys (Hofferth et al., 1991; Hofferth et al., 1998), however, did not present ratio and group size information separately by children’s age. As a result, it is not possible to use the reports to evaluate the percentage of child care settings that meet (or fail to meet) standards for infants, toddlers, and preschoolers that are set differently. A second limitation of these reports is that parents may not be accurate respondents of these quality parameters.
A third source of evidence pertaining to structural and caregiver characteristics is the NICHD Study of Early Child Care (NICHD Early Child Care Research Network, 1999a). In that study, child:adult ratios were observed at regular intervals and caregivers reported their educational background and specialized training. The percentage of center classrooms that met the AAP and APHA recommendations for child:adult ratio and group size is shown on Table 12. Also shown are the percentage of classrooms in which caregivers had at least some college and specialized training. As indicated, 36 percent of the infant classrooms were observed to have the recommended child:adult ratios of 3:1. Fifty-six percent of caregivers in infant classrooms had received specialized training during the preceding year; 65 percent of infant caregivers had some college courses. Proportions were similar for toddler care (the 15- and 24-month-olds). When compared to figures reported in the Profile of Child Care Settings, the NICHD figures suggest that there has been some decline in the educational background and training of child care staff during the 1990s.
The decrease in caregiver education and training may be related to the generally low wages in the child care field (see Figure 3).
In 1997, child care teachers averaged $7.50–$10.85 per hour, or $13,125–$18,988 per year, when they were employed for a 35-hour week and a 50-week year. Wages for assistant teachers were $6.00 to $7.00 an hour (or $10,500–$12,250 per year). Figure 1 shows salaries for lower-paid and higher-paid child care workers relative to the median salaries of women 25+ by level of education for both 1992 and 1997. The figure shows the low salaries of child care workers relative to other occupations and indicates that there has not been any improvement in terms of the relative salaries over the 1992–1997 time period for most levels of education. While high school graduates who were child care teachers or assistants could only earn between 73 and 85 percent of the salaries they might expect to receive elsewhere, the relative salaries were far lower relative to the median for women with more schooling. A child care teacher with a B.A. degree could expect to earn between 52 and 75 percent of the median salaries across all occupations. Current child care salaries are not consistent with attracting and keeping providers who have the level of education and training that research suggests is needed to structure emotionally supportive and cognitively stimulating learning environments.
The generally low salaries earned by child care staff also appear to be a factor contributing to high staff turnover in the child care field (see Figure 4).
In 1997, 27 percent of teachers and 39 percent of assistants left their jobs during the previous year (Figure 5 and Whitebook, Howes, and Phillips, 1998).
Twenty percent of centers reported losing half or more of their staff. Centers that offer higher wages have lower turnover rates than centers that offer lower wages (Whitebook et al., 1998).
Market failure is defined as a “a situation in which a market left on its own fails to allocate resources efficiently” (Mankiw, 1998, p. 10). Wherever market failure exists, public sector intervention may improve the performance of that sector of the economy. In the child care sector, market failure stems from two sources: lack of information and the existence of externalities (effects beyond the primary consumers). Regarding the first, a major problem is the absence of information on the part of parents, including information on the quality of child care, sometimes on the availability of care, and often on the net costs of alternative arrangements. Related to this lack of information is the difficulty in capturing process quality by measuring observable differences in structural quality. Even information on the structural quality differences is not easily obtained a priori, and the cost of acquiring information can be high.
One difficulty in providing information to parents is that much of this market is made up of small providers. Parents may know something about the child care used by their neighbors, but very little about other types or providers of care. Parents report being unsure about how to go about evaluating child care quality. Child care convenience also is of considerable concern to parents limiting their search to care in particular small geographical areas. This way of thinking about market failure is quite similar to that of medical providers. The problem may be particularly acute for lower-income families and for families who need care for evening or weekend employment (Vandell, 1998).
The second major cause of market failure is the existence of externalities. The benefits of quality care accrue not just to the parent(s) but also to the child and to society more generally. Parents may take all or part of the benefit to their child into account, but not benefits that are external to the family. Such benefits include lower costs for subsequent schooling (reduced probability of grade retention and special education, for example), future reductions in crime, increased productivity that results in higher productivity for others, payment of higher taxes, and possibly lower costs for social services. Improving child care quality may affect grade school classrooms by increasing the proportion of children in the class who have strong language and cognitive skills. By the same token, poor-quality child care may undermine grade school classrooms by increasing the numbers of children with academic and social deficits. Unsafe and unhealthy child care may result in reduced productivity for others, when parents are absent from their jobs, caring for injured or ill children. Adding these benefits to the parent’s demand for higher-quality care should shift the demand curve for quality care to the right (that is, increase demand for higher-quality care at every price).
There may also be a third cause of market failure, an imperfect capital market. Parents of young children tend to have low incomes relative to their permanent income, but may face borrowing constraints that reduce their ability to pay for high-quality care.
Justification for government intervention may also be based on distributional or equality-of-opportunity goals. This may be especially relevant today, in view of the requirement that most low-income single parents work. The core argument here is that if high-quality child care can provide gains in cognitive ability, school readiness, and social behavior, children in low-income families should be given an opportunity to benefit from such experiences just as high-income children benefit. Parents with limited earnings do not have the private means to purchase high-quality child care for their children. Government subsidies are necessary if equal opportunity for high-quality care is to be afforded children in low-income families. The other side of this argument is that if subsidies are not provided, parents with limited incomes will use poor-quality care, including multiple arrangements, which may be detrimental to the safety of their children, may increase family stress, and may result in children with reduced opportunities. A subsidy (or direct provision of care) for children in low-income families could also complement the Earned Income Tax Credit and serve as an employment-related income subsidy (see Council of Economic Advisers, 1997).
There are additional issues regarding availability of child care for families with very low incomes and/or unusual and nonflexible hours of work. According to several studies:
“The structure of low-wage work and its lack of fit with the structure of more formal child care options restricts the access of low-income families to child care centers and many family day care homes. Data from the National Child Care Survey indicate that one-third of working-poor mothers (incomes below poverty) and more than one-fourth of working-class mothers (incomes above poverty but below $25,000) worked weekends” (Hofferth, 1995).
Yet only 10 percent of centers and 6 percent of family day care homes reported providing weekend care. Almost half of working-poor parents worked on a rotating or changing schedule, further restricting these families’ child care options to more flexible arrangements made with relatives, friends, and neighbors.
The features of low-wage work appear to promote reliance on multiple providers as a way of patching together child care to cover parents’ nonstandard and shifting work hours (Siegel and Loman, 1991). Meyers added that irregular and unpredictable work schedules led to disruptions in child care for the families in her study of the California GAIN program (Meyers, 1993). Deborah Phillips wrote (1995)
“In sum, when selecting child care, many working-poor and low-income families must choose from a seriously constrained set of options. They face a set of obstacles that derive primarily from the structure of low-wage jobs and from the meager incomes that these jobs provide. Their low incomes enable them to afford only free care by relatives and friends or very inexpensive care; their nonstandard and often rotating work hours restrict them to arrangements with flexible and weekend or evening hours of operation. These factors may also lead to greater reliance on multiple providers and expose young children to shifting child care arrangements.”
This lack of stability and frequent changing is itself a measure of poor quality of care.(3)
The market failure components argue for government intervention in improving child care. They all lead the authors of this report to believe that the demand for high-quality care is too low. And because demand is too low, compensation is too low, resulting in better-trained providers tending to seek employment in other spheres. This results in a decline in quality unless intervention occurs. Intervention can take many forms.
The arguments behind the need for a role for government (including subsidies for child care) is quite similar to those for primary schooling. Traditionally, the amount of schooling provided has heavily depended on the public sector. For children up to the age of 16, or older in some states, schooling is mandatory, and is provided by the public sector. In the cases of elementary and secondary education, public colleges and universities, the price charged tends to be far below the marginal cost of schooling. Evaluation of the appropriate level of public investment in education requires an analysis of all returns to schooling, including nonmarket and external effects. For example, greater education may lead to social cohesion and may enable one to use new technologies; it may reduce the probability of criminal acts, reduce the probability of application and receipt of transfers, and it may increase savings rates. (For more on this see Wolfe and Zuvekas, 1997 and Michael, 1982.)
Many of the benefits of child care are like those of primary schooling, because child care is early childhood education. These early childhood educational experiences affect children’s readiness for primary schooling in the same way that primary schooling affects children’s readiness for secondary schooling. In both cases many benefits are external to the child and family. The community at large would benefit from the cognitive, language, and behavioral competencies that are associated with higher-quality child care. The argument for equality of opportunity is similar as well.
A high-quality child care system also is needed if welfare reform is to succeed. The recent change in welfare policy, establishing work requirements, means that more parents, particularly single parents, are working, because work is their only potential source of income. Requiring work means that more parents must find child care for their children. Given this increase in demand, the issue of child care quality becomes even more important. Unfortunately, as described in earlier sections, much of the child care in the United States is not of high quality. Over 60 percent of children under the age of 3 are receiving care in which positive caregiving is not characteristic. Only 10 percent are in care settings that are described as excellent.
A wide variety of approaches might be used to improve child care. Figure 7 provides a path model that attempts to identify the various links between interventions and quality, taking into account parents’ resources.
Most of the interventions would be included in the elements of the left column. These include the provision of information, licensing requirements, placement activities, subsidies to compensate child care workers, training programs for providers, tuition subsidies for students who enroll in early childhood education, increased tax credits to cover the cost of care for lower- to middle-income families, incentive payments to individual teachers and assistants who remain in the same center for a minimum of 3–4 years, and even direct provision of care. In the longer run, we need research to help better identify those factors that best improve the quality of child care. In essence, we need better understanding of the production of high-quality care, which may differ for children of different backgrounds.
The minimum roles for the public sector are as providers of information on available slots, hours of operation, structural quality features, costs of care and education, and training of personnel. The government might also establish a certification program to certify providers who met certain requirements. Minimum standards need to be strengthened in many states. Incentives might be offered to providers who meet certain requirements. Other government activities to increase the availability of high-quality care include operation of training programs and covering the cost of instructors and facilities for these programs. Information on the successful completion of such programs could be disseminated by the public sector as part of its information activities.
A more ambitious role designed to increase the pool of well-qualified individuals who enter (and remain) in the field of early childhood education would be some form of tuition subsidy for those willing to major in this field. There is a long tradition of such programs when shortages are feared; examples include nursing education and medical school. An alternative might be a college loan forgiveness program based on years spent as a child care provider following college, or Associate degree completion. Another approach to increasing the pool of qualified providers is to raise salaries. This seems especially important given the relatively low salaries in child care compared to other occupations (see Figure 1 and the discussion of the figure above). Such increases might result from increased information to parents, tax credits to parents, and the expansion of subsidy programs or direct payment to providers by the public sector. An innovative program might reward the stability of providers by paying a bonus after some specified period of years.
Current programs to improve child care quality exist and might be replicated or expanded. For example, a broad-based community initiative in North Carolina (Smart Start) has been successful in improving child care quality (Bryant, Maxwell, and Burchinal, 1999). This initiative, established by the governor of North Carolina in 1993, is a partnership between state government, local communities, service providers, and families. Twelve county partnerships were initially selected based on competitive review to receive funds for new and improved child care services. Data to evaluate the effectiveness of the initiative were obtained in 1994 and 1996 from over 180 child care centers in 12 counties. Local quality improvement activities (in 1996) were distributed as follows: training workshops (83 percent), funds to attend training activities (53 percent), on-site consultation or technical assistance (58 percent), higher child care subsidy rates (35 percent), increased subsidies for meeting higher standards (29 percent), funds to improve quality by purchasing new equipment (70 percent), funds to improve quality by purchasing new educational materials (63 percent), funds to achieve higher licensing level (26 percent), funds to achieve national accreditation (13 percent), funds to improve services for children with disabilities (11 percent), teacher substitute pool (20 percent), transportation services (18 percent), lending library (51 percent), and provider compensation programs (35 percent). The mean number of improvement activities per center was 5.3 in 1994 and 5.8 in 1996, with a range of 0 to 14 activities each year. In both years, ECERS scores were significantly related to the number of local quality improvement activities in which individual centers participated. In addition, process quality was significantly higher in 1996 than in 1994. Only 14 percent of centers were rated as good quality in 1994. In 1996, this figure was 25 percent. Participation in the Smart Start initiative also was related to a significant increase in the percentage of centers that obtained the higher-level Associate of Arts licensure credential.
Other efforts to improve quality might be to mandate certain minimum requirements. These can take the form of reducing child:adult ratios, reducing group sizes, establishing and enforcing safety regulations, and education and training. An example of such regulations is a model standard that applies to small family home caregivers. The National Health and Safety Performance Standard (American Public Health Association and American Academy of Pediatrics, 1992) states that one small family caregiver who does not have an assistant “shall not care for more than six children, including no more than two children under age 2. These numbers include the caregiver’s own children under the age of six. If any child under age 3 is in care, there shall be no more than four children, including the caregiver’s own children under the age of six. If only children under age 2 are in care, there shall be no more than three children, including those of the caregiver” (Chapter 1, Staffing).
We also have training programs for providers of child care. The Department of Labor, Bureau of Apprenticeship and Training, West Virginia Child Care Development Specialist Registered Apprenticeship Program offers child care apprentices 4,000 hours of supervised on-the-job training and 300 hours of classroom instruction. The child care providers earn their salaries while they are in the program and receive incremental wage increases as their skill, ability, and knowledge increase. The DOL reports that employers report almost no turnover among child care providers and that providers remain highly satisfied with their careers (www.dol.gov/dol/wb/public/childcare/child3.pdf).
Private employers may sometimes directly or indirectly provide resource and referral services. Some employers contract with private agencies to assist employees in learning about the range of child care options. The government could encourage such activities by providing subsidies or tax credits to firms if they provide such assistance. An example of this is the program at the Virginia Mason Medical Center which offers child/family resource and referral services.
Recent reports suggest that although some federal funds are available to improve access to higher-quality care among lower-income families, some states have not made these funds available or have set up programs that result in low take-up rates. According to “Access to Child Care for Low-Income Working Families” released in October 1999, 1.5 million children of 14.7 million (about 10 percent) in low-income households were receiving child care subsidies. The major source of funds is the Child Care and Development Fund (CCDF), a federal program that provides funds to states to subsidize child care. According to federal law, children living in families with incomes up to 85 percent of median income in the state could be eligible, but states have set lower limits (http://www.acf.dhhs.gov/news/ccreport.htm). The levels set and the take-up rates differ across states. The report details coverage in each state and take-up rates. Expanding eligibility to the federal maximum level and/or encouraging take-up of the existing subsidies through outreach are ways to increase the demand for high-quality care.
Data aggregated from state agencies indicate that the child care arrangement most frequently chosen by parents receiving direct federal subsidies is center-based care (Phillips, 1995), see Figure 8.
This evidence indicates that families do respond to available subsidies. The profile of arrangements used by these low-income parents is strikingly different from that used by all low-income families. Siegel and Loman (1991) found, as well, that AFDC families in Illinois showed different distributions of child care use based on whether they received a child care subsidy and which subsidy they received. Families receiving a child care benefit that reimbursed them for their child care costs 30 to 60 days after they had incurred these costs were discouraged from using child care options they could not afford with their disposable income. Their patterns of use were similar to those of unsubsidized, low-income families. In contrast, families who received subsidies through programs that either subsidized providers directly or enabled the families to pay providers when fees were due showed substantially higher rates of reliance on center-based and formal family day care arrangements.
Subsidies can help low-income families gain access to the same range of quality options that are available to higher-income families. The Urban Institute’s survey of resource and referral agencies in six communities provides evidence that the quality of care received by subsidized children was as high as that for higher income, fee-paying children, as perceived by resource and referral staff (Phillips, 1995).
Another approach that could be combined with subsidies for very young children would be to provide universal coverage for child care (or coverage for children whose parents worked more than 20 hours per week) for pre-school-age children. Such a program could expand existing prekindergarten programs to a full day and include after-school care. These programs could be established for 3- and 4-year-olds through a combination of direct provision through a local school district, existing community-based programs, and vouchers that would be accepted by certified providers. Part of the costs of this care would be offset by eliminating tax credits and current government subsidies for 3- and 4-year-olds. States and local communities would decide on the details of the provision of care, and financing would be shared across levels of government. Part of a coordinated, high-quality child care system for toddlers also might include community- and school-based centers and family day care networks. Because of the necessary low child:adult ratios (and their attendant expense), part of a coordinated child care policy for infants might include vouchers that would allow a parent to stay home and care for the infant during the first year.
As we suggested, analyses that shed light on the developmental benefits of child care characteristics for children is an important element of the answer to our question. A related task involves determining the levels of investment necessary to achieve improved quality. Although this topic has not received the same level of attention in the literature as the overall relationship between quality of care and child outcomes, several studies consider the financial costs of increasing structural measures of quality. Though subject to limitations that affect the extent to which one may apply these findings to the current system of child care, this research provides useful information on the relationship between the quality of child care and cost.
The U.S. General Accounting Office (GAO, 1990) report and Powell and Cosgrove’s (1992) summary of and addendum to this GAO research include estimated cost functions for the provision of child care as well as a model of the factors that affect wages of care providers. The modeled cost function allows the authors to examine the direct influences of input prices; center characteristics, including structural measures of quality; and geographical location on center total variable cost. This use of multiple regression analysis allows the authors to estimate the effect of changing such factors as the child-adult ratio on center total costs, after controlling for other relevant factors such as group size, location, and average education of providers. By also modeling the elements of wage determination, the authors capture the indirect effect of changes in child care characteristics on the wages of providers. The results from both models are then combined to examine direct and indirect effects of quality changes on total variable costs.
Data on child care centers from a 1989 GAO survey of 265 early childhood education centers accredited by the National Associate for the Education of Young Children (NAEYC) were used in both studies.(4) The survey, administered to program directors via questionnaire, included questions of center characteristics, costs, value of in-kind donations received, staff characteristics and compensation. The survey includes several structural measures that allow the authors to estimate the costs associated with changing the following quality measures: the average number of children per teaching staff member; average number of children in groups of four-year-olds; average education of staff in years; average experience of staff in years; and the staff turnover rate.(5)
The authors find statistically significant relationships between center total cost and structural measures of child care quality. For the adult-child ratio, they find that an improvement in the quality of care achieved by decreasing the average ratio by one, for example from 11:1 to 10:1, is associated with increased costs of roughly 4.5 percent. The authors note that the average center in their data, one with 50 children and an annual per child cost of $6,500, which reduces the child-to-staff ratio from 11:1 to 10:1, faces an increase in annual cost per child of $306.(6) A related measure, average group size, had a small but statistically insignificant effect on total costs.
Changes in cost resulting from improving staff education and experience are estimated using the estimated cost and wage functions to capture total direct and indirect effects. The authors find that a one-year increase in average education of the teaching staff is statistically significantly associated with a 3.4 percent increase in total costs, which includes a 5.8 percent increase in wages. Similarly, increasing average teacher experience by one year is significantly associated with a reduction in center total costs of 0.6 percent, including an estimated increase in wages of 2.3 percent. This implies that increasing staff experience is associated with an increase in quality as well as a slight reduction in average center total costs. There is also evidence of a relationship between turnover rates and total costs—the departure of an additional 10 percent of the center’s teaching staff increases costs by 6.8 percent.
These studies have several limitations. First, they rely on data that are more than ten years old. If the relationships estimated in their model have changed since 1989, these estimates will differ from what we would expect to find using current data. Changes in quality made now may affect total costs in a different manner.
Also, the data include only accredited centers. If these centers have cost functions that systematically differ from unaccredited centers, the estimated results will not apply to those centers. If, for example, one is most concerned with improving poor quality child care, and such care is more likely to be found in unaccredited centers, these estimates may not provide an accurate depiction of the costs of doing so. The sample also only includes data for care of 4- and 5-year-old children. Because of differences in providing care for other age groups, the relationship between improving quality and costs for centers that provide care for other age groups may vary.(7) Finally although the sample includes data from many states, 70 percent of the centers surveyed were in the South and the Midwest. The results therefore may not be nationally representative.
Helburn (1995) estimates a similar cost function using data collected as a part of the Cost, Quality, and Child Outcomes in Care Centers (CQO) study. The CQO study, conducted in 1993–1994, included data from 401 child care centers (749 classrooms) in four states. The data, collected through classroom observation and interviews with program directors, include information on center characteristics, program quality and staff qualifications, compensation, and turnover. Helburn estimates center total costs as a function of staff wages at different education levels, hours of child care provided, physical size of the center, volunteer hours, region of the country, for-profit status, and quality measures.(8)
This research also indicates statistically significant relationships between cost and quality. Increasing center quality by 25 percent (from mediocre to good) is associated with increases in total variable costs of approximately 10 percent, or $346 per child per year. The authors define quality using an index similar to the ECERS measure of quality of care. Using the ECERS 7-point scale, the mean value of this index in their sample is 4.0, which ranks between fair and good. They estimate that the increase in total variable costs associated with increasing the measure of quality by 25 percent to 5.0, a rating of good quality, will be approximately 10 percent. Given an average size of 60 children per center, the increase in quality is expected to result in an increase in costs of approximately $20,700 per year.(9)
Helburn notes that this analysis assumes that wages are held constant during the quality-changing process. The model specifically implies that wages for particular quality-levels of staff (here measured by years of education) are set in the labor market and centers pay the going wage. If, however, high-quality centers pay higher than market wages to attract the most qualified providers, or to increase productivity and/or lessen the chance of turnover among existing workers, then estimated costs will understate true costs.
Although this study considers all types of centers, not accredited centers only, it has similar limitations to the GAO and Powell and Cosgrove studies. That is, the data used are not as recent as would be desired, and the results from the four-state sample are not nationally representative.
Although all three studies have limitations, the estimates do provide information on the nature of the relationship between child care costs and quality. As such this research serves as a useful starting point for further consideration of this topic. The importance of developing cost estimates suggests that future work incorporating current and nationally representative data will serve as an important source of information for evaluating public policy strategies designed to improve the quality of child care. We have discovered one likely source of such estimates in research currently being conducted by Richard Brandon at the University of Washington and Lynn Kagan at Yale University. The authors are developing a detailed simulation model to estimate costs of improving child care using varied measures of quality. This research, which the authors estimate will be completed by the fall of 2000, will provide valuable additional information
None of these studies include the investment that is likely to be the least expensive approach to improving quality: caregiver training, including in-service training. There are a variety of training programs that have been offered to formal and informal caregivers. The evidence above suggests that better-trained caregivers give higher-quality care. We do not know enough about the content or length of these programs to be able to definitively discuss their cost. Curriculum and materials are readily available (see for example the materials available from Teaching Strategies, Inc.) One program suggests that a 12-workshop series would enhance caregiver’s skills in areas that range from safety through cognitive development and communication skills.
We conclude this report by returning to primary questions raised in the report: Does child care quality matter? Does child care quality need to be improved, and can it be improved? Is there an economic justification for public intervention to improve the quality of child care, especially for children from lower-income families? Our answer to each of these questions is “yes.”
Our review of the research literature indicates that child care quality “matters” at several levels. In terms of children’s everyday experiences, children appear happier and more cognitively engaged in settings in which caregivers are interacting with them positively and in settings in which child:adult ratios are lower. There also is evidence of concurrent relations between child care quality and children’s performance in other settings. Children who attend higher-quality child care settings (measured by caregiver behaviors, by physical facilities, by age-appropriate activities, and by structural and caregiver characteristics) display better cognitive, language, and social competencies on standardized tests and according to parents, teachers, and observers. Finally, there is evidence that child care quality is related to children’s subsequent competencies. The relationship is more evident when cumulative measures of child care quality are analyzed, rather than one-time assessments, and when quality and child outcome measures have strong psychometric properties.
Two general approaches to measuring child care quality were described in this report. Process quality refers to children’s experiences in child care settings. Some process measures focus specifically on caregivers’ behaviors with children. Others include global ratings that incorporate physical facilities and age-appropriate child activities as well as caregiver behavior into their evaluation. Multisite studies suggest considerable need for improvement of process quality in the United States. Although less than 10 percent of process quality has been categorized as “inadequate” or “poor,” most settings have been characterized as only “fair” or “minimal.” These observations indicate the need for systematic efforts to improve a substantial portion of child care in the United States.
A second way of measuring child care quality is in terms of structural and caregiver characteristics, such as child:adult ratio, group sizes, teacher formal education, and teacher specialized training. There is an extensive research literature linking structural and caregiver characteristics to process quality. A review of regulatory standards in the 50 states shows that few states have adopted standards that are consistent with the recommendations of professional organizations. Furthermore, reports from nationally representative surveys indicate that average group sizes and ratios exceed recommended standards. Recent evidence suggests a decline in the educational background of staff during the 1990s, perhaps as a result of low wages. Thus, it appears that child care structural and caregiver characteristics are in need of improvement. They can be improved if additional resources are allocated. This could occur through a combination of increased subsidies for care, especially to low-income families; federal standards and/or increased state standards for both physical settings and caregiver training and child:staff ratios, improved information to parents on the quality of providers, and/or direct provision or expansion of child care in schools.
Market failure, the presence of externalities, and an argument for equality of opportunity all call for public sector intervention in the child care market. The primary form of market failure is the lack of information for parents regarding quality of care which is tied to the difficulty in measuring quality, the lack of availability of high quality care, and the need for child care for irregular hours such as weekend and late shifts.
Benefits of quality child care accrue to other members of society, including all children in schools with children who had child care; taxpayers, who are likely to save in costs of future schooling by reduction in special education and grade retention; employers, who benefit from more productive employees; and citizens, who gain in terms of future reductions in crime and use of transfer programs. Subsidizing child care for low-income families is also consistent with the goals of the 1996 welfare reform and an ideology that wishes to encourage and reward work. Finally, to the extent that high-quality child care provides benefits to children and their families, there is an argument for providing equal opportunity for such programs to children in low-income families.
Abbott-Shim, M., and A. Sibley. 1992a. Assessment Profile for Early Childhood Programs. Quality Assist, Inc., 368 Moreland Ave. NE Suite 210, Atlanta, GA.
Abbott-Shim, M., and A. Sibley. 1992b. Assessment Profile for Family Day Care. Quality Assist, Inc. 368 Moreland Ave. NE Suite 210, Atlanta, GA.
“Access to Child Care for Low-Income Working Families” released in October 1999, http://www.acf.dhhs.gov/news/press/1999/ccreport.htm
American Public Health Association and American Academy of Pediatrics. 1992. National Health and Safety Performance Standards: Guidelines for Out-of-Home Child Care Programs. Arilington, VA: National Center for Education in Maternal and Child Health.
American Public Health Association, American Academy of Pediatrics. 1992. Caring for Our Children: National Health and Safety Performance Standards: Standards for Out-of-Home Child Care Programs. Ann Arbor, MI. author.
Arnett, J. 1989. “Caregivers in Day-Care Centers: Does Training Matter?” Journal of Applied Developmental Psychology 10: 541–552.
Barnett, W. S. 1993. “Benefit-Cost Analysis of Preschool Education: Findings from a 25-Year Follow-Up.” American Journal of Orthopsychiatry 63: 500–508.
Benasich, April, J. Brooks-Gunn, and B. Clewell. 1992. “How Do Mothers Benefit from Early Intervention Programs?” Journal of Applied Developmental Psychology 13: 311–326.
Berk, L. 1985. “Relationship of Educational Attainment, Child Oriented Attitude, Job Satisfaction, and Career Commitment to Caregiver Behavior toward Children.” Child Care Quarterly 14: 103–129.
Besharov, D. 2000. “Possible CED Role in Early Child Care and Education.” Unpublished manuscript.
Blau, D. M. 1999a. “Does Nonrandom Assignment of Children to Child Care Arrangements Matter for Estimating the Effects of Child Care on Children?” Paper presented at the Biennial Meeting of the Society for Research in Child Development, April, Albuquerque, NM.
Blau, D. M. 1999b. “The Effect of Income on Child Development.” Review of Economics and Statistics 81 (2): 261–276.
Blau, D. M. 1999c. “The Effects of Child Care Characteristics on Child Development.” Journal of Human Resources 34 (4): 786–822.
Blau, D. M. In press. “The Production of Quality in Child Care Centers: Another Look.” Applied Developmental Sciences.
Blau, D., and R. Philip. 1989. “Fertility, Employment and Child Care Costs.” Demography 26: 287–299.
Borge, A. I. H., and E. C. Melhuish. 1995. “A Longitudinal Study of Childhood Behavior Problems, Maternal Employment, and Day Care in a Rural Norwegian Community.” International Journal of Behavioral Development 18: 23–42.
Brayfield, A., S. G. Deich, and S. Hofferth. 1993. Caring for Children in Low-Income Families: A Substudy of the National Child Care Survey, 1990. A National Association for the Education of Young Children Study conducted by the Urban Institute. Washington, DC: Urban Institute Press.
Briss, P. A., J. J. Sacks, D. G. Addiss, M. J. Kresnow, and J. O’Neil. 1995. “Injuries from Falls on Playgrounds: Effects of Day Care Center Regulation and Enforcement.” Archives of Pediatrics & Adolescent Medicine 149: 906–911.
Broberg, A. G., C. P. Hwang, M. E. Lamb, and F. L. Bookstein. 1990. “Factors Related to Verbal Abilities in Swedish Preschoolers.” British Journal of Developmental Psychology 8: 335–349.
Broberg, A. G., H. Wessels, M. E. Lamb, and C. P. Hwang. 1997. “Effects of Day Care on the Development of Cognitive Abilities in 8-Year-Olds: A Longitudinal Study.” Developmental Psychology 33: 62–69.
Brooks-Gunn, J., M. McCormick, S. Shapiro, A. Benasich, and G. Black. 1994. “The Effects of Early Education Intervention on Maternal Employment, Public Assistance and Health Insurance: The Infant Health and Development Program.” American Journal of Public Health 84: 924–931.
Browning, K. S., C. W. Runyan, J. B. Kotch. 1996. “A Statewide Survey of Hazards in Child Care Centers.” Injury Prevention 2: 202–207.
Bryant, D. M., K. L. Maxwell, and M. Burchinal. 1999. “Effects of a Community Initiative on the Quality of Child Care.” Early Childhood Research Quarterly 14: 449–464.
Burchinal, M. R., J. E. Roberts, L. A. Nabors, and D. M. Bryant. 1996. “Quality of Center Child Care and Infant Cognitive and Language Development.” Child Development 67: 606–620.
Burchinal, M. R., J. E. Roberts, R. Riggins, S. A. Zeisel, E. Neebe, and D. Bryant. In press. “Relating Quality of Center Child Care to Early Cognitive and Language Development Longitudinally.” Child Development.
Campbell, F. A., and C. T. Ramey. 1995. “Cognitive and School Outcomes for High Risk African American Students at Middle Adolescence: Positive Effects of Early Intervention.” American Educational Research Journal 32: 743–772.
Center for Career Development in Early Care and Education. 1999. “Child Care Centers: Child Staff Ratio and Group Size Requirements.” Wheelock College, Boston, MA.
Chin-Quee, D. S., and S. Scarr. 1994. “Lack of Early Child Care Effects on School-Age Children’s Social Competence and Academic Achievement.” Early Development and Parenting 3: 103–112.
Clarke-Stewart, K. A., and G. G. Fein. 1983. “Early Childhood Programs.” In Handbook of Child Psychology, Volume II: Infancy and Developmental Psychobiology, ed. P. H. Mussen, (M. M. Haith and J. J. Campos, series eds.). New York: John Wiley and Sons.
Clarke-Stewart, K. A, C. P. Gruber, and L. M. Fitzgerald. 1994. Children at Home and in Day Care. Hillsdale, NJ: Erlbaum.
Clarke-Stewart, K. A., D. L. Vandell, M. Burchinal, M. O’Brien, and K. McCartney. 2000. “Do Features of Child Care Homes Affect Children’s Development?” Unpublished paper, University of California, Irvine.
Connelly, R., and J. Kimmel. 2000. Marital Status and Full-Time/Part-time Work Status in Child Care
Choices. Kalamazoo, MI: W.E. Upjohn Institute.
Council of Economic Advisers. 1997. The Economics of Child Care. December. http://www.whitehouse.gov/WH/EOP/CEA/html/childcare.html#econ
Deater-Deckard, K., R. Pinkerton, and S. Scarr. 1996. “Child Care Quality and Children’s Behavioral Adjustment: A Four-Year Longitudinal Study.” Journal of Child Psychology and Psychiatry and Allied Disciplines 37: 937–948.
Dunn, L. 1993. “Proximal and Distal Features of Day Care Quality and Children’s Development.” Early Childhood Research Quarterly 8: 167–192.
Dunn, L., S. A. Beach, and S. Kontos. 1994. “Quality of the Literacy Environment in Day Care and Children’s Development.” Journal of Research in Childhood Education 9: 24–34.
Early Learning, Later Success: The Abecedarian Study. 2000, Chapel Hill, NC. http:www.fpg.unc.edu/~abc/embargoed/executive_summary.htm
Elicker, J., C. Fortner-Wood, and I. C. Noppe. 1999. “The Context of Infant Attachment in Family Child Care.” Journal of Applied Developmental Psychology 20 (2): 319–336.
Field, T. M. 1991. “Quality Infant Day-Care and Grade School Behavior and Performance.” Child Development 62: 863–870.
Fosburg, S. 1981. Family Day Care in the United States: National Day Care Home Study, Volume I: Summary of Findings. DHHS Publication No. OHDS 80-30282. Washington, DC: U.S. Department of Health and Human Services.
Fuller, B., et al. 1999. “Who Selects Formal Child Care? The Role of Subsidies as Low-Income Mothers Negotiate Welfare Reform.” Paper presented at the Biennial Meeting of the Society for Research in Child Development, April, Albuquerque, NM.
Goelman, H. 1988. “The Relationship between Structure and Process Variables in Home and Day Care Settings on Children’s Language Development.” In The Practice of Ecological Research: From Concepts to Methodology, ed. A. Pence and H. Goelman.
Goelman, H., and A. R. Pence. 1987. “Effects of Child Care, Family, and Individual Characteristics on Children’s Language Development: The Victoria Day Care Research Project.” In Quality in Child Care: What Does Research Tell Us? ed. D. Phillips. Washington, DC: National Association for the Education of Young Children.
Hagekull, B., and G. Bohlin. 1995. “Day Care Quality, Family and Child Characteristics and Socioemotional Development.” Early Childhood Research Quarterly 10: 505–526.
Hanushek, E., and J. Jackson. 1977. Statistical Methods for Social Scientists. New York: Academic Press.
Harms, T., and R. M. Clifford. 1980. Early Childhood Environment Rating Scale. New York: Teachers College Press.
Harms, T., and R. Clifford. 1989. Family Day Care Rating Scale. New York: Teachers College Press.
Harms, T., D. Cryer, and R. M. Clifford. 1990. Infant/Toddler Environment Rating Scale. New York: Teachers College Press.
Hausfather, A., A. Toharia, C. LaRoche, and F. Engelsmann. 1997. “Effects of Age of Entry, Day-Care Quality, and Family Characteristics on Preschool Behavior.” Journal of Child Psychology and Psychiatry and Allied Disciplines 38: 441–448.
Helburn, S., et al. 1995. Cost, Quality, and Child Outcomes in Child Care Centers. Public Report, Economics Department, University of Colorado at Denver.
Helburn, S., ed. 1995. Cost, Quality and Child Outcomes in Child Care Centers: Technical Report. Denver, CO: Dept of Economics, Center for Research in Economic and Social Policy, University of Colorado.
Hestenes, L. L., S. Kontos, and Y. Bryan. 1993. “Children’s Emotional Expression in Child Care Centers Varying in Quality.” Early Childhood Research Quarterly 8: 295–307.
Hofferth, S. 1995. “Caring for Children at the Poverty Line.” Children and Youth Services Review 17(1/2): 1–31,
Hofferth, S. 1999. “Child Care, Maternal Employment and Public Policy.” Annals of the American Academy 563 (May): 20–38.
Hofferth, S. L., A. Brayfield, S. Deich, and P. Holcomb. 1991. National Child Care Survey, 1990, Report 91-5. Washington, DC: Urban Institute Press.
Hofferth, S. L., K. A. Shauman, R. R. Henke, and J. West. 1998. Characteristics of Children’s Early Care and Education Programs: Data from the 1995 National Household Education Survey. Report No. 98-128. Washington, DC: U.S. Department of Education, National Center for Education Statistics.
Holloway, S. D., and M. Reichhart-Erickson. 1988. “The Relationship of Day Care Quality to Children’s Free Play Behavior and Social Problem-Solving Skills.” Early Childhood Research Quarterly 3: 39–53.
Howes, C. 1983. “Caregiver Behavior in Center and Family Day Care.” Journal of Applied Developmental Psychology 4: 99–107.
Howes, C. 1988. “Relations between Early Child Care and Schooling.” Developmental Psychology 24: 53–57.
Howes, C. 1990. “Can the Age of Entry into Child Care and the Quality of Child Care Predict Adjustment in Kindergarten?” Developmental Psychology 26: 292–303.
Howes, C. 1997. “Children’s Experiences in Center-Based Child Care as a Function of Teacher Background and Adult:Child Ratio.” Merrill-Palmer Quarterly 43: 404–425.
Howes, C., and M. Olenick. 1986. “Family and Child Care Influences on Toddlers’ Compliance.” Child Development 57: 202–216.
Howes, C., and J. Rubenstein. 1985. “Determinants of Toddlers’ Experience in Day Care: Age of Entry and Quality of Setting.” Child Care Quarterly 42: 140–151.
Howes, C., and E. W. Smith. 1995. “Relations among Child Care Quality, Teacher Behavior, Children’s Play Activities, Emotional Security, and Cognitive Activity in Child Care.” Early Childhood Research Quarterly 10: 381–404.
Howes, C., and P. Stewart. 1987. “Child’s Play with Adults, Toys, and Peers: An Examination of Family and Child Care Influences.” Developmental Psychology 23: 423–530.
Howes, C., D. A. Phillips, and M. Whitebrook. 1992. “Thresholds of Quality: Implications for the Social Development of Children in Center-Based Child Care.” Child Development 63: 449–460.
Iutcovich, J., R. Fiene, J. Johnson, R. Koppel, and Langan. 1997. Investing in Our Children’s Future: The Path to Quality Child Care through the Pennsylvania Child Care/Early Childhood Development Training System. Erie, PA: Keystone University Research Corporation.
Jacobs, E. V., and D. R. White. 1994. “The Relationship of Child-Care Quality and Play to Social Behavior in the Kindergarten.” In Children’s Play in Child Care Settings, ed. H. Goelman and E. V. Jacobs. Albany, NY: State University of New York Press.
Kisker, E. E., S. L. Hofferth, D. S. Phillips, and E. Farquhar. 1991. A Profile of Child Care Settings: Early Education and Care in 1990, Volume 1. Princeton, NJ: Mathematica Policy Research.
Kontos, S. J. 1991. “Child Care Quality, Family Background, and Children’s Development.” Early Childhood Research Quarterly 6: 249–262.
Kontos, S., and A. Wilcox-Herzog. 1997. “Influences on Children’s Competence in Early Childhood Classrooms.” Early Childhood Research Quarterly 12: 247–262.
Kontos, S., C. Howes, and E. Galinsky. 1996. “Does Training Make a Difference to Quality in Family Child Care?” Early Childhood Research Quarterly 11: 427–445.
Kontos, S., C. Howes, M. Shinn, and E. Galinsky. 1995. Quality in Family Child Care and Relative Care. New York: Teachers College Press.
Kotch, J., and C. Guthrie. 1998. “Effect of a Smart Start Playground Improvement Grant on Child Care Playground Hazards.” Smart Start Evaluation Report. Chapel Hill, NC: Frank Porter Graham Child Development Center.
Lally, J. Ronald, Peter L. Mangione, Alice S. Honig, and Donna S. Wittner. 1988. More Pride, Less Delinquency: Findings from the Ten-Year Follow-Up Study of the Syracuse University Family Development Research Program.
Lamb, M. E. 1998. “Nonparental Child Care: Context, Quality, Correlates, and Consequences.” In Handbook of Child Psychology, Vol. 4: Child Psychology in Practice, ed. W. Damon, I. E. Sigel, and K. A. Renninger. New York: John Wiley.
Lamb, M. E., C. P. Hwang, A. Broberg, et al. 1988. “The Effects of Out-of-Home Care on the Development of Social Competence in Sweden: A Longitudinal Study.” Early Childhood Research Quarterly 3: 379–402.
Love, J. M., P. Z. Schochet, and A. L. Meckstroth. 1996. “Are They in Any Real Danger? What Research Does—and Doesn’t—Tell Us about Child Care Quality and Children’s Well-Being.” Report. Princeton, NJ: Mathematica Policy Research.
Mankiw, N. G. 1998. Principles of Economics. Fort Worth: Dryden Press.
Maume, D. 1991. “Child Care Expenditures and Women’s Employment Turnover.” Social Forces 70: 494–508.
McCartney, K. 1984. “Effect of Quality of Day-Care Environment on Children’s Language Development.” Developmental Psychology 20: 244–260.
McCartney, K., and R. Rosenthal. In press. “Effect Size, Practical Importance, and Social Policy for Children.” Child Development.
McCartney, K., S. Scarr, D. Phillips, and S. Grajek. 1985. “Day Care as Intervention: Comparison of Varying Quality Programs.” Journal of Applied Developmental Psychology 6: 247–260.
McCartney, K., S. Scarr, A. Rocheleau, D. Phillips, M. Abbott-Shim, M. Eisenberg, N. Keefe, S. Rosenthal, and J. Ruh. 1997. “Teacher-Child Interaction and Child-Care Auspices as Predictors of Social Outcomes in Infants, Toddlers, and Preschoolers.” Merrill-Palmer Quarterly 43: 426–450.
Meyers, M. 1993. “Child Care in Jobs Employment and Training Programs: What Difference Does Quality Make?” Journal of Marriage and the Family 55: 767–783.
Michael, Robert. 1982. “Measuring Non-Monetary Benefits of Education: A Survey.” In Financing Education: Overcoming Inefficiency and Inequity, ed. W. McMahon and T. Geske. Urbana, IL: University of Illinois Press.
National Household Education Survey. 1998. http://nces.ed.gov/nhes/.
NICHD Early Child Care Network. 1994. “Child Care and Child Development: The NICHD Study of Early Child Care.” In Developmental Follow-Up: Concepts, Domains, and Methods, ed. S. L. Friedman and H. C. Haywood. New York: Academic Press.
NICHD Early Child Care Research Network. 1996. “Characteristics of Infant Child Care: Factors Contributing to Positive Caregiving.” Early Childhood Research Quarterly 11: 269–306.
NICHD Early Child Care Research Network. 1997. “Familial Factors Associated with Characteristics of Nonmaternal Care for Infants.” Journal of Marriage and the Family 59: 389–408.
NICHD Early Child Care Research Network. 1998. “Early Child Care and Self-Control, Compliance, and Problem Behaviors at Twenty-Four and Thirty-Six Months.” Child Development 69 (4): 1145–1170.
NICHD Early Child Care Research Network. 1999a. “Child Outcomes When Child Care Center Classes Meet Recommended Standards of Quality.” American Journal of Public Health 89 (7): 1072–1077.
NICHD Early Child Care Research Network. 1999b. “Effect Sizes from the NICHD Study of Early Child Care.” Paper presented at the Biennial Meeting of the Society for Research in Child Development, April, Albuquerque, NM.
NICHD Early Child Care Research Network. In press-a. “Characteristics and Quality of Child Care for Toddlers and Preschoolers.” Applied Developmental Sciences.
NICHD Early Child Care Research Network. In press-b. “The Relation of Child Care to Cognitive and Language Development.” Child Development.
NICHD Early Child Care Research Network. Submitted for publication. Early Child Care and Children’s Peer Relationships at 24 and 36 Months.
Niffenegger, J. P. 1997. “Proper Handwashing Promotes Wellness in Child Care.” Journal of Pediatric Health Care 11: 26–31.
Peisner-Feinberg, E. S., and M. R. Burchinal. 1997. “Relations between Preschool Children’s Child-Care Experiences and Concurrent Development: The Cost, Quality, and Outcomes Study.” Merrill-Palmer Quarterly 43: 451–477.
Peisner-Feinberg, E. S., M. R. Burchinal, R. M. Clifford, M. L. Culkin, C. Howes, S. L. Kagan, N. Yazejian, P. Byler, J. Rustici, and J. Zelazo. 1999. The Children of the Cost, Quality, and Outcomes Study Go to School: Technical Report.
Phillips, D. A., ed. 1987. Quality in Child Care: What Does Research Tell Us? Washington, DC: National Association for the Education of Young Children.
Phillips, D. A., ed. 1995. Child Care for Low Income Families: Summary of Two Workshops. Washington, DC: National Academy Press. Report available at http://nccic.org/research/nrc_care/c_care.html.
Phillips, D. A., K. McCartney, and S. Scarr. 1987. “Child-Care Quality and Children’s Social Development.” Developmental Psychology 23: 537–544.
Phillipsen, L. C., M. R. Burchinal, C. Howes, and D. Cryer. 1997. “The Prediction of Process Quality from Structural Features of Child Care.” Early Childhood Research Quarterly 12: 281–303.
Pierrehumbert, B., T. Ramstein, A. Karmaniola, and O. Halfon. 1996. “Child Care in the Preschool Years: Attachment, Behaviour Problems and Cognitive Development.” European Journal of Psychology of Education 11: 201–214.
Powell, I., and J. Cosgrove. 1992. “Quality and Cost in Early Childhood Education.” Journal of Human Resources 27: 472–484.
Ramey, C. T., F. A., Campbell, and C. Blair. 1998. “Enhancing the Life-Course for High-Risk Children: Results from the Abecedarian Project.” In Social Programs That Really Work, ed. J. Crane. New York: Sage Publishing.
Ramey, C. T., F. A. Campbell, M. Burchinal, M. L. Skinner, D. M. Gardner, and S. L. Ramey. In press. “Persistent Effects of Early Childhood Education on High-Risk Children and Their Mothers.” Applied Developmental Science.
Reynolds, A. J., J. A. Temple, D. L. Robertson, and E. A. Mann. 2000. “Long-Term Benefits of Participation in Title I Chicago Child-Parent Centers.” Paper presented at the Biennial Meeting of the Society for Research on Adolescence, March 30, Chicago, IL.
Ross, C., and D. Paulsell. 1998. “Sustaining Employment among Low-Income Parents: The Role of Quality in Child Care: A Research Review.” Final Report. Princeton, NJ: Mathematica Policy Research.
Ruopp, R., J. Travers, F. Glantz, and C. Coelen. 1979. Children at the Center: Final Report of the National Day Care Study. Cambridge, MA: Abt Associates.
Scarr, S. 1998. “American Child Care Today.” American Psychologist 53: 95–108.
Scarr, S., M. Eisenberg, and K. Deater-Deckard. 1994. “Measurement of Quality in Child Care Centers.” Early Childhood Research Quarterly 9: 131–151.
Schliecker, E., D. R. White, and E. Jacobs. 1991. “The Role of Day Care Quality in the Prediction of Children’s Vocabulary.” Canadian Journal of Behavioural Science 23: 12–24.
Schweinhart, L. J., H. V. Barnes, D. P. Weikart, et al. 1993. Significant Benefits: The High/Scope Perry Preschool Study through age 27. Ypsilanti, MI: High/Scope Press.
Siegel, and Loman. 1991. Child Care and AFDC Recipients in Illinois: Patterns, Problems and Needs. Prepared for the Division of Family Support Services, Illinois Department of Public Aid. St. Louis, MO: Institute of Applied Research.
Singer, J. D., B. Fuller, M. K. Keiley, and A. Wolf. 1998. “Early Child-Care Selection: Variation by Geographic Location, Maternal Characteristics, and Family Structure.” Developmental Psychology 34 (5): 1129–1144.
Stallings, J., and A. Porter. National Day Care Home Study: Observation Component. Final Report, Volume III. Administration for Children, Youth, and Families. Department of Health and Human Services, SRI Project 6903.
Stith, S. M., and A. J. Davis. 1984. “Employed Mothers and Family Day-Care Substitute Caregivers: A Comparative Analysis of Infant Care.” Child Development 55: 1340–1348.
St. Sauver, J., M. Khurana, A. Kao, and B. Foxman. 1998. “Hygienic Practices and Acute Respiratory Illness in Family and Group Day Care Homes.” Public Health Reports 111: 544-551.
Teaching Strategies. 2000. http://www.teachingstrategies.com
U.S General Accounting Office. 1990. Early Childhood Education: What Are the Costs of High Quality Programs? HRD-90-43BR. Washington, DC: General Accounting Office.
Vandell, D. L. 1997. “Introduction: Child Care as a Developmental Context.” Merrill-Palmer Quarterly 43 (3): 333–339.
Vandell, D. L. 1998. “Child Care for Low-Income Families: Dreams and Real Life.” In Children and Families in an Era of Rapid Change: Creating a Shared Agenda for Researchers, Practitioners and Policy Makers, ed. F. Lamb-Parker, J. Hagen, R. Robinson, and Cheryl Clark. Washington, DC: Department of Health and Human Services.
Vandell, D. L., and M. A. Corasaniti. 1990. “Variations in Early Child Care: Do They Predict Subsequent Social, Emotional, and Cognitive Differences?” Early Childhood Research Quarterly 5: 55–72.
Vandell, D. L., V. K. Henderson, and K. S. Wilson. 1988. “A Longitudinal Study of Children with Day-Care Experiences of Varying Quality.” Child Development 59: 1286–1292.
Vandell, D. L., and J. K. Posner. 1999. “Conceptualization and Measurement of Children’s After-school Environments.” In Measuring Environments Across the Lifespan: Emerging Methods and Concepts, ed. S. L. Friedman and T. Wachs. Washington, DC: American Psychological Association Press.
Vandell, D. L., and C. Powers. 1983. “Daycare Quality and Children’s Free Play Activities.” American Journal of Orthopsychiatry 53: 493–500.
Vernon-Feagans, L., D. C. Emanuel, and I. Blood. 1997. “The Effect of Otitis Media and Quality Daycare on Children’s Language Development.” Journal of Applied Developmental Psychology 18: 395–409.
Volling, B. L., and L. V. Feagans. 1995. “Infant Day Care and Children’s Social Competence.” Infant Behavior and Development 18: 177–188.
Whitebook, M., C. Howes, and D. Phillips. 1990. “Who Cares? Child Care Teachers and the Quality of Care in America.” Final Report, National Child Care Staffing Study. Oakland, CA: Child Care Employee Project.
Whitebook, M., C. Howes, and D. Phillips. 1998. Worthy Work, Unlivable Wages: The National Child Care Staffing Study, 1988–1997.
Whitebook, M., L. Sakai, and C. Howes. 1997. “NAEYC Accreditation as a Stragegy for Improving Child Care Quality: An Assessment.” Final Report, National Center for the Early Childhood Work Force. Washington, DC.
Wolfe, B., and S. Zuvekas. 1997. “Nonmarket Outcomes of Schooling.” International Journal of Educational Research 27 (6): 491–502.
Yarrow, M. R., and C. Zahn-Waxler. 1979. “Observing Interaction: A Confrontation with Methodology.” In The Analysis of Social Interactions: Methods, Issues and Illustrations, ed. R. B. Cairns. Hillsdale, NJ: Lawrence Erlbaum.
1) The authors gratefully thank the following for their contributions to this paper: Josephine Chung, Kim Pierce, Scott Scriver, and Elisabeth Boehnen for their research assistance and Dawn Duren, Jan Blakeslee, Elizabeth Evanson for their editorial and typing assistance.
2) The NHES survey included an early childhood program participation component in its 1995 survey. Parents of 14,000 children from birth through third grade were asked about their use of a wide variety of childcare and early education arrangements. http://nces.ed.gov/nhes/
3) According to several studies, low-income families use more of certain types of child care and less of others than do families with more income or more education (see Figure 6).
According to the study by the National Academy (Phillips, 1995) low-income families “are more likely to rely on relatives and less likely to rely on center-based arrangements. . . Grandparents are an especially prominent source of child care for low-income, preschool-age children: 17 percent are cared for mainly by grandparents; 29 percent get some care from a grandparent. The child care arrangements of low-income families also vary greatly by household type and parental employment status. . . Single employed mothers rely to a much greater extent on non-relative arrangements (notably family day care homes and centers) than do other types of low-income families. In addition, among low-income families, about 24 percent of children under age 5 are in more than one supplemental arrangement on a regular basis (Brayfield et al., 1993). Reliance on multiple arrangements varies, however, from 14 percent of low-income preschoolers with two parents to 31 percent of those in single-mother families and 45 percent of those in employed, single-mother families”(Phillips, 1995, Chapter 2).
Connelly and Kimmel (2000) use a merged sample from the 1992 and 1993 SIPP panels which include child care questions from June through December 1994. They examine the choice of child care of full-time vs. part-time employed mothers, married and single. Table 13 shows their prediction, by marital status, of the marginal effects of the included variables on the probability of choosing among the 3 modes of child care for single employed mothers. The first panel uses a single predicted price for child care and the second includes three predicted prices of care. For single mothers, the predicted conditional probability of full-time employment is a significant predictor of child care. As has been suggested by earlier research, single mothers who are working full-time are more likely to use center-based care. This also leads to the conclusion that as more mothers enter full-time employment after leaving welfare, the utilization of center-based care over other types of care will increase. The reported results also show that women with more education are more likely to use center-based care or home-based care and that women on AFDC who receive more dollars are somewhat more likely to use center-based care and less likely to use relative care. Somewhat surprisingly, single mothers with higher predicted wages appear more likely to use relative care or home-based care than center care. The authors explain this by hypothesizing that higher-wage jobs are more likely to be positions that require more flexibility than center care can provide in terms of pick-up and drop-off times and the care of sick children.
4) The NAEYC accreditation criteria include guidelines for staff-child interaction, curriculum content, staff qualifications and training, group size and adult-child ratios, physical environment and safety, nutrition.
5) The other variables included in the model are average hourly wages for teachers and aids; rent; cost of other supplies and services per child; indicators for whether a center is for-profit, serves infants, serves children with disabilities, has operated for less than two years; and geographic location.
6) These estimates are expressed in current dollars. The original estimates in 1988 dollars are $4,500 and $207, respectively.
7) The authors do control for centers that provide infant care, but cost structures would nevertheless be expected to differ, for example, in infant-care centers that do not provide care to 4- and 5-year-olds.
8) The relationship between structural measures of quality and total variable costs are also tested, but there is no evidence of statistically significant relationships. The authors note that their index measure of quality may be a theoretically superior measure insofar as the structural characteristics capture only a portion of the “true” measure of quality. They argue that this is the case if the unobserved center quality characteristics are correlated with other independent variables.
9) These estimates are expressed in current dollars. The original estimates in 1994 dollars are $300 and $18,000 respectively.
Top of page
Contents of Full Report
Main Page of Report
Executive Summary
Home Pages:
Human Services Policy
(HSP)
Assistant Secretary for Planning and Evaluation
(ASPE)
U.S. Department of Health and Human Services
(HHS)
Last updated 05/24/00