Child Care Quality Matters: How Conclusions May Vary with Context

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Faculty Publications, Department of Child, Youth, and Family Studies Child, Youth, and Family Studies, Department of

7-1-2003

Child Care Quality Matters: How Conclusions May Vary With Context

John M. Love Mathematica Policy Research

Linda Harrison Charles Sturt University

Abraham Sagi-Schwartz University of Haifa

Marinus H. van IJzendoorn Leiden University

Christine Ross Mathematica Policy Research

See next page for additional authors

Follow this and additional works at: https://digitalcommons.unl.edu/famconfacpub

Part of the Family, Life Course, and Society Commons

Love, John M.; Harrison, Linda; Sagi-Schwartz, Abraham; van IJzendoorn, Marinus H.; Ross, Christine; Ungerer, Judy A.; Raikes, Helen; Brady-Smith, Christy; Boller, Kimberly; Brooks-Gunn, Jeanne; Constantine, Jill; Kisker, Ellen Eliason; Paulsell, Diane; and Chazan-Cohen, Rachel, "Child Care Quality Matters: How Conclusions May Vary With Context" (2003). Faculty Publications, Department of Child, Youth, and Family Studies. 41. https://digitalcommons.unl.edu/famconfacpub/41

This Article is brought to you for free and open access by the Child, Youth, and Family Studies, Department of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Faculty Publications, Department of Child, Youth, and Family Studies by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Authors John M. Love, Linda Harrison, Abraham Sagi-Schwartz, Marinus H. van IJzendoorn, Christine Ross, Judy A. Ungerer, Helen Raikes, Christy Brady-Smith, Kimberly Boller, Jeanne Brooks-Gunn, Jill Constantine, Ellen Eliason Kisker, Diane Paulsell, and Rachel Chazan-Cohen

This article is available at DigitalCommons@University of Nebraska - Lincoln: https://digitalcommons.unl.edu/ famconfacpub/41 Published in Child Development 74:4 (July/August 2003), pp. 1021–1033. Copyright © 2003 by the Society for Research in Child Development, Inc. Published by Blackwell Publishing, Inc. DOI 0009-3920/2003/7404-0004 Used by permission. “This article may not exactly replicate the ﬁ nal version published in the SRCD journal. It is not the copy of record.”

Child Care Quality Matt ers: How Conclusions May Vary With Context

John M. Love, Linda Harrison, Abraham Sagi-Schwartz, Marinus H. van Ĳ zendoorn, Christine Ross, Judy A. Ungerer, Helen Raikes, Christy Brady-Smith, Kimberly Boller, Jeanne Brooks-Gunn, Jill Constantine, Ellen Eliason Kisker, Diane Paulsell, and Rachel Chazan-Cohen*

Three studies examined associations between early child care and child outcomes among families diff erent from those in the National Institute of Child Health and Human Development (NICHD) Early Child Care Research Network study. Results suggest that quality is an important infl uence on children’s development and may be an important moderator of the amount of time in care. Thus, the generalizability of the NICHD fi ndings may hinge on the context in which those results were obtained. These studies, conducted in three national contexts, with diff erent regulatory climates, ranges of child care quality, and a diversity of family characteristics, suggest a need for more complete estimates of how both quality and quantity of child care may infl uence a range of young children’s developmental outcomes.

* John M. Love, Mathematica Policy Research; Linda Harrison, Charles Sturt University; Abraham Sagi-Schwartz, University of Haifa; Mari- nus H. van Ĳ zendoorn, Leiden University; Christine Ross, Mathematica Policy Research; Judy A. Ungerer, Macquarie University; Helen Raikes, Society for Research in Child Development visiting scholar, Administration for Children and Families, U.S. Department of Health and Human Services; Christy Brady-Smith, Columbia University; Kimberly Boller, Mathematica Policy Research; Jeanne Brooks-Gunn, Columbia Universi- ty; Jill Constantine, Ellen Eliason Kisker, and Diane Paulsell, Mathematica Policy Research; Rachel Chazan-Cohen, Administration for Children and Families, U.S. Department of Health and Human Services. Harrison and Ungerer of the Sydney Family Development Project gratefully acknowledge the contributions made by Brent Waters and Bry- anne Barnett in designing the Sydney Family Development Project, and the continual inspiration provided by Robyn Dolby. The research was supported by grants from the Australian National Health and Medical Research Council, the Australian Research Council, Macquarie Univer- sity, University of New South Wales, and The University of Melbourne. Writing was supported by grants from Charles Sturt University. Sagi-Schwartz and van Ĳ zendoorn of the Combined Haifa–National Institute of Child Health and Human Development (NICHD) Studies gratefully acknowledge the following support: The data on child center care in the United States were derived from the data set made available by the NICHD Early Child Care Research Network supported by NICHD through a cooperative agreement that calls for scientifi c collabo- ration between the grantees and the NICHD staff . We are grateful to Sarah Friedman for her support. Pieter Kroonenberg and Marielle Linting helped prepare the data for analysis. The contribution of Marinus van Ĳ zendoorn was supported by a grant from the Dutch Ministry of Social Aff airs and Employment to the NCKO (Dutch Consortium for Research in Child Care). The contribution of Abraham Sagi-Schwartz was supported by the dean of research at the University of Haifa. The Early Head Start fi ndings reported here are based on research conducted as part of the national Early Head Start Research and Evalu- ation Project funded by the Administration for Children and Families (ACF), U.S. Department of Health and Human Services under contract 105-95-1936 to Mathematica Policy Research, Princeton, NJ, and Columbia University’s Center for Children and Families, Teachers College, in conjunction with the Early Head Start Research Consortium. The consortium consists of representatives from 17 programs participating in the evaluation, 15 local research teams, the evaluation contractors, and Administration on Children, Youth, and Families. Research institutions in the consortium (and principal researchers) include ACF (Rachel Chazan-Cohen, Judith Jerald, Esther Kresh, Helen Raikes, and Louisa Tarul- lo); Catholic University of America (Michaela Farber, Lynn Milgram Mayer, Harriet Liebow, Christine Sabatino, Nancy Taylor, Elizabeth Tim- berlake, and Shavaun Wall); Columbia University (Lisa Berlin, Christy Brady-Smith, Jeanne Brooks-Gunn, and Allison Sidle Fuligni); Harvard University (Catherine Ayoub, Barbara Alexander Pan, and Catherine Snow); Iowa State University (Dee Draper, Gayle Luze, Susan McBride, Carla Peterson); Mathematica Policy Research (Kimberly Boller, Jill Constantine, Ellen Eliason Kisker, John M. Love, Diane Paulsell, Christine Ross, Peter Schochet, Cheri Vogel, and Welmoet van Kammen); Medical University of South Carolina (Richard Faldowski, Gui-Young Hong, and Susan Pickrel); Michigan State University (Hiram Fitzgerald, Tom Reischl, and Rachel Schiff -man); New York University (Mark Spell- mann and Catherine Tamis LeMonda); University of Arkansas (Robert Bradley, Mark Swanson, and Leanne Whiteside-Mansell); University of California, Los Angeles (Carollee Howes and Claire Hamilton); University of Colorado Health Sciences Center (Robert Emde, Jon Korfmach- er, JoAnn Robinson, Paul Spicer, and Norman Watt ); University of Kansas (Jane Atwater, Judith Carta, and Jean Ann Summers); University of Missouri-Columbia (Mark Fine, Jean Ispa, and Kathy Thornburg); University of Pitt sburgh (Beth Green, Carol McAllister, and Robert McCall); University of Washington School of Education (Eduardo Armĳ o and Joseph Stowitschek); University of Washington School of Nursing (Kath- ryn Barnard and Susan Spieker); and Utah State University (Lisa Boyce and Lori Roggman). The content of this publication does not necessari- ly refl ect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or or- ganizations imply endorsement by the U.S. government. Correspondence concerning this article should be addressed to John M. Love, Mathematica Policy Research, 600 Alexander Park, Princeton, NJ 08540 (609-275-2245); to Linda Harrison, School of Teacher Education, Charles Sturt University, Bathurst, NSW 2795, Australia (612-6338- 4872); and to Abraham Sagi-Schwartz, Center for the Study of Child Development, University of Haifa, 6035 Rabin Complex, Haifa 31905, Isra- el (972-4-8240196). Electronic mail may be sent to [email protected] , [email protected] , and [email protected] . 1021 1022 Love, et al. in Child Development 74 (2003)

ccumulated evidence suggests that for chil- The child care sett ings of the three studies also ex- dren in child care, the quality of that care ist within diff erent regulatory contexts: both regulat- Ais important for their development (Lamb, ed (moderately high quality) and nonregulated set- 1998; Love, Schochet, & Meckstroth, 1996), but fi nd- tings in Sydney, homogenous but lower standards in ings diff er with respect to the nature of the relation- Haifa, and homogenous and high standards in Ear- ship of quality and quantity of care with various ly Head Start. The three studies include diverse mea- developmental outcomes. Although higher quality sures of child care quality as well as a range of out- care has been associated with improved cognitive come measures that span children’s cognitive and and language skills across a range of studies (e.g., social-emotional domains (from att achment security see Campbell, Pungello, Miller-Johnson, Burchinal, to language to aggressive behavior problems). & Ramey, 2001; National Institute of Child Health Together our three perspectives capitalize on this and Human Development [NICHD] Early Child diversity, and together they investigate how quantity Care Research Network, 2000b, 2003b), associations and quality of child care may relate to children’s de- between care quality and social-emotional devel- velopment. We begin with the Sydney study, move opment have been more mixed. The latest report to the merged Haifa–NICHD data, and then turn to from the NICHD Early Child Care Research Net- the United States to look at child care in the federal work (2003b), consistent with its previous reports, Early Head Start program. concludes that more time spent in ‘‘any of a variety of nonmaternal care arrangements’’ leads chil- An Australian Perspective on Quality, Quantity, dren to display more externalizing behavior prob- and Stability in Child Care lems, and that this relationship holds regardless of quality and other factors (NICHD Early Child Care The fi rst perspective is based on a 6-year longitudi- Research Network, 2003a). Given the potential im- nal study of the use and eff ects of child care in an plications of such fi ndings for understanding chil- Australian sample of 147 primiparous mothers. Like dren’s social-emotional development and contrib- the study conducted by the NICHD Early Child Care uting to child care policy, it is critical to assess the Research Network, the SFDP used a correlational de- extent to which the NICHD fi ndings might general- sign to assess the relationships of type, amount, and ize to other child care contexts. stability of children’s child care experiences to de- In this article, we bring together research on this velopmental outcomes at key points: infant–moth- issue from three perspectives: (a) the Sydney Family er att achment at 12 months, behavior problems at 30 Development Project (SFDP), (b) the Haifa–NICHD months and 5 years, and teacher-rated adjustment to merged data, and (c) the Early Head Start program school at 6 years. The results, however, indicate a dif- evaluation in the United States. These perspectives ferent relationship between child care and develop- contribute to the child care debate by extending ment than that reported by the NICHD study. We be- the levels of observed child care quality beyond the lieve this is due, in part, to diff ering levels of child more restricted range of the NICHD study, expand- care quality in the Australian and U.S. contexts. ing the diversity of families included in the research Like the United States, Australia has achieved a and breaking the correlation between quality of care high level of workforce participation among wom- and socioeconomic status (SES) found in the NICHD en of child-bearing age (46% with children under 5 study. The Haifa study adds a sample of families years, Australian Bureau of Statistics, 1998; 65% with- using much lower quality child care across all SES in 18 months aft er giving birth, Glezer, 1988); howev- groups; the Australian study adds a sample of fami- er, unlike the United States, this has been linked to lies from diff erent SES groups using generally high- government child care policies that have actively en- er quality government-regulated care relative to the couraged mothers to return to the workforce. Signifi - average quality of care in the United States; and the cant funds at both state and federal levels are direct- Early Head Start study adds a sample of low-income ed to child care services for children from birth to 12 families experimentally off ered care of higher quali- years, and to preschool programs for children aged ty than is generally available to families with infants 3 to 5 years. State regulations require child care cen- and toddlers in the United States, and who are more ters to meet minimum child-staff ratios (e.g., New diverse than the NICHD study families. South Wales, 5:1 for children under 2 years, 8:1 for Child Care Quality Matt ers: How Conclusions May Vary With Context 1023

2-to 3-year-olds, 10:1 for 3-to 5-year-olds) and to em- The informal and formal sectors of Australia’s ploy specialist staff with early childhood qualifi ca- child care system provide a useful dichotomy for de- tions (e.g., 3-to 4-year university degree, 2-to 3-year scribing child care quality and comparing the eff ects technical college diploma). Similar standards oper- of home-based care of variable quality with centers ate in Australian Family Day Care services, where and Family Day Care homes of uniformly moderate the child-adult ratio is 5:1 for children under 5 years, to high quality. As such, Australian research has the and caregivers receive regular training and supervi- capacity to extend what has been reported for quali- sion by qualifi ed early childhood staff . At the fed- ty versus quantity of child care in the NICHD study. eral level, the Quality Improvement and Accredita- tion System requires centers to meet criterion-based Commonalities and Diff erences Between the SFDP and standards of care for families to receive government the NICHD Study subsidies for the cost of care (National Childcare Ac- Participants in the SFDP diff ered from those in the creditation Council [NCAC], 1993). Fee reductions NICHD study in that all the SFDP children were fi rst- apply to most families, on an income-to-needs ba- born, and mothers were selected to be representative sis, and provide a major incentive for child care op- of a larger community sample (n = 453) on broad in- erators to become accredited. Australia’s national dexes of personality functioning (see Harrison & Un- system of formal, government-regulated child care gerer, 2002a). The range of mothers’ educational levels services arguably achieves a uniformly higher lev- (less than high school to postgraduate) was compara- el of quality than is found in the United States (e.g., ble to the range in the NICHD sample (NICHD Early Wangmann, 1995) and, accordingly, could be expect- Child Care Research Network, 1997b). ed to show a diff erent relationship with children’s Families’ use of child care was similar to that of development than has been reported in studies in the NICHD families. By age 12 months, 72% of chil- the United States. dren were receiving regular nonmaternal care, 18% Despite these provisions, however, not all child att ended fewer than 10 hr per week, 32% att ended care is provided through the formal system, espe- part-time (11–30 hr per week), and 22% att ended full- cially during the fi rst 2 years of life, when informal time (> 30 hr per week). At 30 months, 86% were in care with relatives, friends, babysitt ers, or nannies is regular care (24% informal, 62% formal). By 3 years, more typical (e.g., from birth to 1 year, 37% infor- 97% of children were in care, and many (46%) had mal vs. 8.5% formal; age 1 to 2 years, 46% informal entered preschool. Cross-age correlations showed vs. 24% formal; Australian Bureau of Statistics, 1999). that average weekly hours during the fi rst 12 months Although similar informal arrangements have been were consistent with weekly hours at 30 months, 3 noted in U.S. studies (NICHD Early Child Care Re- years, and 4 years, rs = .50 to .63, p < .001. search Network, 1997a), the reasons may not be the Assessments of children’s developmental out- same, perhaps because American parents do not re- comes matched those in the NICHD longitudinal ceive government fi nancial support for the full cost of study. We included security of infant–mother att ach- child care and may choose informal care as a cheaper ment at 12 months (Strange Situation; Ainsworth, Ble- option (Scarr & Eisenberg, 1993). Australian families har, Waters, & Wall, 1978), mother-reported behavior are eligible for fi nancial assistance, which suggests problems at 30 months and 5 years (Child Behavior that the decision to use informal care arrangements Checklist [CBCL]; Achenbach, 1991), and teacher rat- may be based on personal preference rather than ings of adjustment to school at 6 years. Teacher–child cost. In regard to quality, Australia requires infor- confl ict was assessed (Student–Teacher Relationship mal child care providers to be registered but as yet Scale: Pianta, 1990) along with other measures of be- has no regulatory systems to support or supervise havioral and social competence (Teacher–Child Rat- caregivers. As a result, child-adult ratios, caregiver ing Scale: Hightower et al., 1989; Classroom Behav- qualifi cations, and other factors potentially infl uenc- ior Inventory: Schaefer, Edgerton, & Aaronson, 1978; ing quality are determined solely by the caregivers Teacher Rating Form: Prosocial Behavior Scale: Klein themselves. Without regulatory standards, informal & Abu Taleb, 1993). sett ings are likely to be more variable in quality than The SFDP research questions focused on quality are formal services and, therefore, less predictive of (formal vs. informal care) and quantity of early child outcomes for children. care (i.e., from birth to 30 months of age), and sta- 1024 Love, et al. in Child Development 74 (2003)

bility of care over time (i.e., from birth to 6 years). adjustment to the learning demands of school (task Changes in care arrangements at three periods (birth orientation, creativity, intelligent behavior, distractibil- to 12 months, 12 to 30 months, 30 months to 6 years) ity). Relationships between child care factors and child were used to construct a measure describing consis- outcomes were tested using hierarchical regression tent patt erns of more or less stable care. To assess the analyses, controlling for family and child characteris- predictive relationships between child care factors tics (Harrison & Ungerer, 2002b). Results showed that and child outcomes, eff ects were assessed aft er con- teacher– child confl ict was associated with patt erns trolling for family SES, maternal psychological well- of more unstable care over time (ΔR2 = .039, p < .05), being, marital relationship quality, social support, but that it was not related to quantity or type of early and child gender and temperament characteristics. care. Ratings of social-emotional adjustment were also For att achment security, maternal sensitivity also related to stability of care (ΔR2 = .031, p < .05), being was included as a predictor (see Harrison & Unger- lowest in the group of children whose care had been er, 2002a). As the questions related to the experience consistently more unstable and highest in the group of nonmaternal care, children who were in exclusive whose care had a more stable patt ern. Personal adjust- maternal are were not included in the analyses. ment ratings were higher when children had att ended formal in contrast to informal care during the fi rst Longitudinal Findings From the SFDP: Eff ects of 30 months (ΔR2 =.061, p < .01). Competence in learn- Quality, Quantity, and Stability of Care ing was predicted by type of care (higher ratings for Infant–mother att achment security at 12 months. formal vs. informal care) and by quantity of care (low- Secure versus insecure att achment outcomes were er ratings for longer hours of care), together explaining compared for the 85 infants who received regular 6.3% of the variance (overall R2 = .199, p < .01). child care during the fi rst year—52 in informal set- tings versus 33 in formal care. (Sixty children in ma- How and Why SFDP and NICHD Early Child Care ternal care were not included in these analyses.) Se- Research Findings Diff er curity was associated with formal care (Family Day Longitudinal results from the SFDP, based on Care: 100% secure; center care: 63% secure) rather teachers’ and mothers’ reports, provide a useful com- than informal care (56% secure). Of the children in parison with fi ndings reported by the NICHD Ear- part-time or full-time care, 70% had secure att ach- ly Child Care Research Network. Although there are ments, whereas only 39% of children att ending care many similarities between SFDP and NICHD chil- for fewer than 10 hr per week were secure (Harrison dren’s experiences of care over time, the relation- & Ungerer, 1997). The association between security ships of dimensions of care with developmental out- and formal care was confi rmed by logistic regression comes diff er. We att ribute this to diff erent systems of tests, controlling for hours of care, maternal educa- child care provision and regulation. The present re- tion, social support, and child diffi cult temperament port from NICHD att ributes poorer outcomes for so- (Wald coeffi cient = 5.24, p = .02; model χ2 = 19.69, N = cial-emotional development to longer hours of care 85, df = 5, p = .001; with maternal sensitivity, Wald co- in general and, in particular, to more time in cen- effi cient = 3.53, p = .06; model χ2 = 23.87, N = 78, df = 6, ter-based care. In contrast, the SFDP study found p = .001). Quantity of care was not a signifi cant pre- no relationship between quantity of care and moth- dictor in these analyses. er-reported problem behavior, social adjustment to Behavior problems at 30 months and 5 years. Moth- school, or teacher–child confl ict. Rather, it was stabil- ers’ scores for the child’s internalizing, externalizing, ity of care over time that contributed to social-emo- and total behavior problems at age 30 months and 5 tional aspects of development. Children who had ex- years showed no associations with type or quantity perienced a consistent patt ern of more changes in of early care, rs(115) = –.01 to .08, ns. care were rated by teachers as having more conduct Adjustment to school at 6 years. Teacher ratings pro- problems and less eff ective social skills. From an at- vided a measure of teacher–child relationship con- tachment perspective (Bowlby, 1969/1978), repeat- fl ict and summary ratings of children’s social-emotion- ed changes of care, which place additional demands al development (acting out, hostility, considerateness, on children to form new relationships and create the prosocial behavior), personal adjustment (outgoing stress of losing existing relationships, are a poor ba- with peers, extrovert, shy or anxious, introvert), and sis for developing emotional resourcefulness or so- Child Care Quality Matt ers: How Conclusions May Vary With Context 1025

cial competence. This may be especially salient as higher standards of care. Because the Haifa Study children make the transition to school, where they of Early Child Care was developed in parallel to the are required to adapt to large numbers of children NICHD project and covered non-maternal care of and proportionally fewer adults than they experi- much lower quality standards (Sagi, Koren-Karie, enced in child care. Gini, Ziv, & Joels, 2002), we can now test the gener- Quality of care, described by formal, regulated care alizability of some of the earlier NICHD fi ndings in versus informal, unregulated care, was found to be an a broader range of quality standards by comparing important predictor of child outcomes in the SFDP. and combining the two data sets. Children who had att ended formal sett ings before age 30 months (85% of whom had received some cen- Commonalities and Diff erences Between the Haifa ter care) were rated by their school teachers as more and the NICHD Studies outgoing and extroverted and less shy and anxious Both the Haifa and the NICHD studies addressed than children whose care had been in informal, non- questions concerning ecologically relevant predictors regulated arrangements. Formal care was also asso- of social-emotional adjustment among infants who ex- ciated with teachers’ higher ratings for competen- perienced various types and quality of early care, and cies in learning. These fi ndings support a theoretical in both studies, all SES groups were included. Because position that higher quality care (i.e., care programs public child care centers in Israel are part of a nation- that meet required standards for equipment, space, wide network, however, infants from both lower-and and programming, and are provided by appropriate- middle-class families are placed in the same centers, ly qualifi ed staff ) will support children’s learning and whereas SES tends to be confounded with quality of development (e.g., see National Association for the group care in the United States. In both studies the Education of Young Children, 1987; NCAC, 1993). investigators focused on mother characteristics, in- The SFDP results provided some evidence for the fant characteristics and development, mother–child negative eff ects of longer hours of child care and, in interaction, mother–father relationship, the environ- this sense, were consistent with the NICHD report. ment, and the structure and quality of various types Quantity of care was negatively associated with teach- of group care. Moreover, the samples are very large ers’ ratings for competence in learning, regardless of (N = 1,153, United States; N = 758, Israel), with a low care type. (Note, however, that competence scores att rition rate and recruited in two diff erent developed were highest for children who had att ended formal Western cultures, both of which place a high premium care for fewer hours, and lowest for children receiving on education. A major outcome variable studied was more hours of informal care.) Our results do not sup- quality of the infant–mother att achment (for more de- port the suggestion that behavior problems (linked to tails on the U.S. study, see NICHD Early Child Care quantity of care) interfere with readiness to learn, but Research Network, 1997a; for more details on the Is- they indicate that problems of att ention (task orienta- raeli study, see Sagi et al., 2002). tion, distractibility) and interest (creativity, intelligent Various child care correlates were found in the behavior) were linked to longer hours of care. NICHD study to neither adversely aff ect nor pro- mote the security of infants’ att achment to their moth- Quality of Child Center Care Is Important: A ers at the 15-month age point. However, certain child Mega-Analysis of the NICHD and the Haifa care conditions, in combination with certain home en- Study of Early Child Care vironments, did increase the probability that infants would be insecurely att ached to their mothers. More How Generalizable Is the NICHD Study? specifi cally, infants who received poor quality of care, The present report by the NICHD Early Child Care received more than 10 hr of care per week, or were Research Network (this issue) states that amount of in more than one child care sett ing during the fi rst 15 early nonmaternal care—not quality of care—is the months of life were more likely to be insecurely at- crucial predictor of later social-emotional adjust- tached only if their mothers were lower in sensitivity ment. The main question is whether this fi nding per- (NICHD Early Child Care Research Network, 1997a). tains to the restricted quality range of nonmaternal In the Haifa study, child care, especially center care provisions in the United States, and to how gen- care, increased the likelihood of the infants’ att ach- eralizable it is to sett ings with much lower or much ment insecurity to their mothers (Sagi et al., 2002). 1026 Love, et al. in Child Development 74 (2003)

More specifi cally, a signifi cantly larger proportion emphasized. The Israeli data for children in low-qual- of insecure-ambivalent infants was found in center ity nonmaternal care suggest that maternal behav- care (46%) than in each of the following groups: fam- ior was less salient for these infants, who might be so ily child care (28%), paid individual care (27%), indi- overwhelmed by the low quality of the center care set- vidual care with a relative (19%), and maternal care ting that they may no longer experience their mothers’ (26%). Furthermore, the data clearly showed that it child-rearing behavior as good enough. is the very high infant:caregiver ratio (average of 8:1) Based on the structural characteristics of cen- that accounted for this increased level of att achment ter care in the two studies (e.g., child-caregiver ra- insecurity among center care infants when compared tio), it is safe to conclude that center care in the Haifa with other professional care (viz., family or paid in- study represented considerably lower quality of care dividual care). Because of the large sample size, it (which was confi rmed by direct observations) rela- was possible to examine the sole eff ects of type and tive to center care in the NICHD Early Child Care quality of care by controlling for a vast array of po- Research Network (1997a; Sagi et al., 2002). In the tentially intervening maternal and child characteris- NICHD study, the social structure as well as more tics in parallel with the NICHD study. None of these defi ned state restrictions concerning certifi cation variables was found to mediate or to minimize the of center care facilities may have prevented the re- negative eff ects that were discovered only for center searchers from investigating facilities with extreme- care. Thus, early infant center care in Israel, with its ly low quality. Some child care facilities in deprived very high infant:caregiver ratio, challenged infants’ and dangerous inner-city areas may have been inac- security of att achment to their mothers. cessible to the researchers, and noncertifi ed centers In an additional analysis of the Haifa data (Aviezer, may have been hesitant to make themselves avail- Sagi, & Koren-Karie, in press), the expected link be- able for government-subsidized research. Although tween maternal sensitivity and infant att achment se- Israel is a developed Western country with a high curity was found only for infants in individual care education level, the early care system has received but not for infants in center care, despite the fact that inappropriate public att ention, resulting in a very- mothers of children in individual care arrangements low-quality system of center care for infants. Thus, and mothers of children in center care were equally in combining the two studies for the purpose of sec- sensitive. This analysis suggests that the lack of associ- ondary analyses, we create a broader continuum of ations between maternal sensitivity and infant att ach- quality of center care and thus make it less likely that ment security might contribute to the lower security restriction of range will infl uence the generalizabili- rates found for center care children. Thus, childrear- ty of the fi ndings. ing context may override the expected infl uence of maternal sensitivity. Indeed, in center care, the pro- A ‘‘Mega-Analysis’’ on Combined Data from the Haifa portion of sensitive mothers with insecurely att ached and NICHD Studies infants was similar to the proportion of insensitive Combining data from the two studies, we focused mothers with insecurely att ached infants, whereas the on amount of care and child-caregiver ratio as im- proportion of sensitive mothers with secure infants in portant indicators of early care. Amount of care and individual care sett ings was signifi cantly higher than child-caregiver ratio were defi ned identically in both the proportion of sensitive mothers with insecure in- studies (for details, see NICHD Early Child Care Re- fants. Hence, maternal sensitivity did not predict at- search Network, 1997a; Sagi et al., 2002). Child-adult tachment security for center care children. ratio is a distal index of quality of care. We were not It should be noted that the moderating infl uence of able to use a more proximal assessment of quality of center care on the formation of att achment relations in care here because the measure used in the NICHD the Haifa study is diff erent from the moderating ef- study for assessing qualitative aspects in caregiver– fect of center care that was found in the NICHD study infant interaction (Observational Record of the Care- (NICHD Early Child Care Research Network, 1997a). giving Environment; ORCE) was a unique tool de- In the NICHD study, maternal behavior was more sa- veloped by the NICHD Early Child Care Research lient for children in low-quality facilities of nonmater- Network, which became accessible to the scientifi c nal care, and the expected associations between ma- community only at a later stage. The Haifa study had ternal sensitivity and infant att achment were more to develop its own quality assessments. In the Hai- Child Care Quality Matt ers: How Conclusions May Vary With Context 1027 fa study, however, child-caregiver ratio is strongly amount of care did not correlate signifi cantly with at- linked to quality of care (Koren-Karie, Sagi, & Egoz- tachment security (secure vs. nonsecure as a dichoto- Mizrachi, 1998; Sagi et al., 2002). Also, the NICHD mous variable). Only in the combined sample did we Early Child Care Research Network (2000a) has re- fi nd a signifi cant correlation between child-caregiver cently found that positive caregiving was associat- ratio and att achment security (r = –.13, p < .05), indi- ed with smaller child-caregiver ratios. The NICHD cating that a larger ratio was associated with less in- study assessed positive caregiving in fi ve types of fant att achment security. Because amount of care was care: centers, child care homes, and care provided by also signifi cantly related to child-caregiver ratio, we in-home sitt ers, grandparents, and fathers (it should included both predictors in a logistic regression on be noted that in most publications the NICHD net- att achment security. In the combined sample, this re- work presents paternal care as another type of non- gression proved to be signifi cant (χ2 = 6.02, n = 267, df maternal care, in the same league as professional = 2, p = .049). Only child-caregiver ratio contributed caregiving arrangements). Thus, child-caregiver ra- signifi cantly to this regression equation (Wald coef- tio, which is a standard index in both the Haifa and fi cient = 4.09, df = 1, p = .043). Amount of care did not the NICHD studies, is a useful index of quality of signifi cantly contribute to the logistic regression. In care in our secondary analysis. Our analyses on the the separate NICHD and Haifa samples, the logistic combined Haifa and NICHD studies are restricted to regressions with amount of care and child-caregiver infant–mother att achment as an outcome variable, ratio did not signifi cantly explain the variance in at- and no data are available to conduct a longitudinal tachment security. secondary data analysis with the combined sample. In a broader range of center care quality, we found that a higher child-caregiver ratio was indeed associ- Summary of Major Findings ated with less att achment security, whereas amount The present analysis is based on a combined sam- of care was not a signifi cant predictor. In the sepa- ple (n = 294) of all center care cases in both studies (n = rate NICHD and Haifa samples, a similar result 143, United States; n = 151, Israel). From Table 1 it can failed to emerge. We suggest that the generalizabili- be derived that the child-caregiver ratio in the Haifa ty of the NICHD fi ndings hinges on the specifi c con- study was twice as large as the ratio in the NICHD text in which these results have been obtained, and study. In both studies, most infants were involved in on the resulting restriction of range for crucial vari- center child care full time, with an average of about ables. The NICHD fi nding of a small association be- 5.5 hr per week longer involvement in the Israeli case. tween center care quality and child outcomes should In Israeli center child care, the percentage of securely be limited to the specifi c population of centers from att ached infants was signifi cantly lower (54%) than which the NICHD study sample was derived—until in the NICHD centers (67%). In the NICHD study as empirically demonstrated otherwise. Although our well as in the Haifa study, child-caregiver ratio and fi ndings in the combined sample are limited in time 1028 Love, et al. in Child Development 74 (2003)

(infancy) and in number and type of variables (child- nical reports, which also include detailed descriptions caregiver ratio, amount of care, security of att ach- of the study design, instruments, data collection pro- ment), our case illustrates the risks of premature gen- cedures, and analytic methods. eralizations to other contexts, in particular when the Important for the context of this research are the causal processes leading to a signifi cant association high standards the federal Head Start Bureau sets for between quantity of care and child outcomes, such child care quality. The Head Start Program Perfor- as security or aggression, still are obscure. Counter- mance Standards established a clear set of expecta- intuitive fi ndings that are not based on a priori pre- tions for the quality of center-based child develop- dictions from a theoretical framework should be in- ment services. The standards require (a) a child-staff terpreted with caution, although such fi ndings may ratio of 4:1 and a maximum group size of eight in- turn out to be among the most important and excit- fants and toddlers in center-based child care sett ings, ing in our fi eld of inquiry. But let us fi rst establish and (b) child care staff to have a Child Development their truth value in other contexts and construct an Associate credential within 1 year of being hired as adequate theoretical framework to account for them, an infant-toddler teacher (U.S. Department of Health before jumping to (policy) conclusions. The absence and Human Services, 1996). These standards exceed of evidence for a signifi cant contribution of child those reported for the sett ings of the Australian and care quality to children’s social development, espe- Israeli studies, as well as for standards set by most cially aggression, should be considered a fi nding in states within the United States, which is the context search of replication and explanation. for the NICHD research.

Infant and Toddler Child Care in the Context Method of Early Head Start: Quantity, Quality, ACF selected 17 programs to participate in the na- and Children’s Development tional evaluation; these sites span all regions of the country and are in both urban and rural sett ings. A The third perspective on the eff ects of child care on total of 3,001 families applying to these Early Head children’s development comes from the national eval- Start programs between July 1996 and September uation of the federal Early Head Start program in the 1998 were randomly assigned either to the program United States. Like the NICHD study, the Early Head or to a control group, which could access all servic- Start evaluation examined the developmental progress es in the community except Early Head Start. Ear- of children during the fi rst 3 years of life. Many of the ly Head Start families were diverse: 63% were His- children were in nonparental child care during those panic, African American, or other non-White groups; early years. In contrast to the NICHD study, however, 48% had not earned a high school diploma; 38% the Early Head Start study focused exclusively on chil- were teenage parents; and 23% were neither em- dren from low-income families. Moreover, because of ployed nor in school. Parent interviews and assess- the Early Head Start intervention, many of the chil- ments of children’s development were conducted dren received good-quality, center-based child care, when the children were approximately 14, 24, and which few of the NICHD children from low-income 36 months old. Assessments when children were 24 families experienced. Early Head Start was launched months old included measures of cognitive develop- in 1995 by the Administration for Children and Fami- ment (the Bayley Scales of Infant Development Men- lies (ACF), in the U.S. Department of Health and Hu- tal Development Index, BSID– MDI; Bayley, 1993), man Services and was designed as a two-generation language development (the MacArthur Communi- program serving low-income pregnant women and cative Development Inventory language production families with infants and toddlers up to the age of 3. scale, CDI; Fenson et al., 1993), and aggressive be- Early Head Start grantees design programs to achieve havior (the Child Behavior Checklist, CBCL; Achen- benefi ts for both children and their parents by provid- bach, 1992); see Administration on Children, Youth, ing home-or center-based child development services, and Families (2001, chap. 4) for details. At 36 months combining these approaches, or implementing other of age, child assessments included the BSID–MDI, locally designed options. Detailed fi ndings about the the Peabody Picture-Vocabulary Test–Third Edition programs’ implementation (Administration for Chil- (PPVT–III; Dunn & Dunn, 1997), and the CBCL ag- dren and Families [ACF], 2002b) and eff ects through gressive behavior scale (Achenbach & Rescorla, 2000; age 3 (ACF, 2002a) can be found in the project’s tech- see ACF, 2002a, chap. 5). Child Care Quality Matt ers: How Conclusions May Vary With Context 1029

We asked parents about their use of child care four center-based programs, for example, the per- (along with other services) at several points aft er pro- centage in some type of child care for at least 30 hr gram enrollment. Child care quality was assessed by per week increased from 66% at 14 months to 74% at trained observers in 2-to 3-hr visits to the child care 36 months. Moreover, Early Head Start increased the sett ings children were in around the time that the use of child care relative to the control group at all 14-, 24-, and 36-month interviews were completed three ages by 7 to 10 percentage points (all diff erenc- with the children’s parents. Sett ings observed were es signifi cant at p < .01) and increased the percentage those that children were in for at least 10 hr per week of children in good-quality care even more (see Table for at least the 2 weeks preceding the interview and 2). Thus, the overall positive impacts of Early Head that were not in the child’s own home (unless that Start on children’s development occurred while sub- in-home care was provided by a nonrelative). When stantial numbers of children were enrolled in child children were 14 and 24 months old, observations of care. (Details of the child care analyses described center-based care were conducted using the Infant- here can be found in a special policy report prepared Toddler Environment Rating Scale (ITERS; Harms, by the authors; see ACF, 2003.) Cryer, & Cliff ord, 1990). When children were 36 The favorable impacts across a variety of dimen- months old, we used the Early Childhood Environ- sions of children’s development cannot be att ributed ment Rating Scale–Revised (ECERS–R; Harms, Clif- solely to the children’s child care experiences, how- ford, & Cryer, 1998). These scales consist of 35 items ever, because not all children in Early Head Start that assess the quality of care. (The shortened ver- were in child care sett ings. Furthermore, the Early sion of the ITERS we used excluded three items from Head Start intervention included family-and child- the adult needs category—opportunities for profes- development services, as described earlier, and the sional growth, adult meeting area, and provisions favorable overall impacts of this random-assignment for parents.) Item scores range from 1 to 7, in which study must be att ributed to the full package of ser- 1 is described as inadequate care, 3 as minimal care, 5 as vices received. Nevertheless, good-quality child care good care, and 7 as excellent care. Child-teacher ratios was an important aspect of the services received and and group sizes were also recorded. thus is responsible for a share of the favorable impacts of the program. Findings: The Overall Impacts of Early Head Start on Children Findings: The Contribution of Good-Quality Child Care The national evaluation found that the program to the Early Head Start Impacts on Children had favorable impacts on a wide range of outcomes We took two approaches to identifying the eff ects for children (as well as their families). Early Head of good-quality child care on the development of Start improved cognitive and language development Early Head Start children. Neither approach has the of children at 24 and 36 months of age, with program methodological strength of the overall random as- children scoring signifi cantly higher than control chil- signment design, but both suggest that, rather than dren on the BSID–MDI and the CDI or PPVT–III. (We doing harm, experience in good-quality child care are reporting only those program-control diff erenc- can play a role in improving outcomes for children es that were statistically signifi cant in the regression- from economically disadvantaged families. adjusted impact analyses that controlled for a large First, we focus on the four Early Head Start pro- number of family background characteristics; see grams that off ered full-day, full-year, center-based ACF, 2002a.) Early Head Start also produced favor- child development programs. Children typically at- able impacts on aspects of social-emotional develop- tended these centers operated by Early Head Start for ment at 36 months, broadening the range of impacts substantial periods each week, ranging from 51% at on these behaviors that were found at 24 months. At 24 months to 68% at 36 months att ending for 30 hr or both 24 and 36 months, levels of aggressive behav- more per week. (The children experienced even great- ior were signifi cantly lower for Early Head Start chil- er amounts of time in all types of nonparental care.) dren than for control-group children. The quality of child care in the four Early Head At the time these positive program impacts were Start center-based programs was higher than is typi- becoming manifest, most Early Head Start children cal for center-based infant-toddler care in the United were in regular child care arrangements, and child States, and for low-income children more specifi cally. care use increased as the children got older. In the Observational ratings showed that Early Head Start 1030 Love, et al. in Child Development 74 (2003)

children at these sites were in classrooms that scored Because of diff erent measures, we cannot compare an average of 4.8 on the ITERS for 14-month-olds these quality ratings directly with the NICHD study and 4.9 for 24-month-olds. The classrooms scored 4.7 classrooms, but national studies typically have found on the ECERS–R for 36-montholds. (Note that these substantially lower ITERS scores than we observed averages for Early Head Start children include some in Early Head Start centers. The scores have ranged children who, for a variety of reasons, were in com- from 3.2 to 3.6 in infant and toddler classrooms in munity centers not operated by the program. Ob- several sites across the United States (Cost, Quality, servations of all classrooms operated by Early Head and Child Outcomes Study Team, 1995; Whitebook, Start centers across all sites showed a mean ITERS Howes, & Phillips, 1989). or ECERS–R rating of 5.0, 5.2, and 5.1 for 14-, 24-, The evaluation found that, when impacts on chil- and 36-month-old children, respectively.) The Early dren’s development were estimated separately by Head Start program quality was signifi cantly high- the program’s approach to child development servic- er than the quality that control group children ex- es, center-based programs were as eff ective as pro- perienced in the same communities. Control group grams off ering home-based or mixed-approach ser- children using center care in these communities were vices (ACF, 2002a). This evidence certainly suggests in classrooms rated signifi cantly lower—an average that the large amount of good-quality center-based of 3.8 to 3.9 for 14-and 24-month-olds on the ITERS, child care off ered by Early Head Start was not detri- and 4.1 for 36-montholds on the ECERS–R (see Ta- mental to children and, indeed, contributed to posi- ble 2). In addition, as Table 2 shows, child-adult ra- tive outcomes found for the program as a whole. tios were 2.8 at 14 months, 3.2 at 24 months, and 5.6 Our second approach to the child care–child de- at 36 months, again signifi cantly bett er than ratios in velopment link focuses on children in the Early Head classrooms att ended by control group children. Ear- Start program group who received center care. All ly Head Start children in the four Early Head Start children in this sample received the full Early Head sites off ering center-based child development servic- Start intervention and were in center care during at es clearly received higher quality care as a result of least one of the three periods (14, 24, or 36 months of their enrollment in the program. age). The analyses relate child care quality and inten- Child Care Quality Matt ers: How Conclusions May Vary With Context 1031

sity at 14, 24, and 36 months to three key child out- children’s development and that quality may be an comes at 24 and 36 months of age. Child care quali- important moderator of the amount of time in care, ty was measured by the ITERS or ECERS– R and by particularly when the child care contexts diff er from the child-adult ratio. Intensity was measured by av- those of the NICHD research. Taken together, these erage hours in center child care. We relate these mea- studies point to a limited generalizability of the sures of child care to children’s cognitive development NICHD Early Child Care Research Network’s fi nd- (BSID–MDI scores), language development (CDI at 24 ings. Although some of our results corroborate the months or PPVT–III at 36), and aggressive behavior NICHD study fi ndings, other outcomes raise ques- (CBCL aggressive behavior problem scores). Using or- tions as to whether the associations between ear- dinary least squares regression analyses, mean child ly child care quality and social-emotional develop- care quality and intensity scores at 14 and 24 months ment reported by the NICHD network would hold were used to predict 24-month outcomes, and mean in a more diverse sample of children and families, quality and intensity scores at 14, 24, and 36 months and in a wider range of child care sett ings. were used to predict 36-month outcomes. All regres- In the SFDP, children in care for longer periods sion analyses controlled for child gender, child age at before 30 months of age were rated lower by their time of assessment, mother’s race or ethnicity, moth- teachers on adjustment to the learning demands of er’s education and marital status, whether mother school at age 6, which appears consistent with the was teenage (under 19 years of age) at the time of the current NICHD fi nding that extensive early experi- child’s birth, and whether the site was urban. ence in child care may be associated with later be- Among the Early Head Start children who at- havior problems. In the SFDP, however, being in for- tended child care centers, those in higher quality mal (i.e., higher quality) care was associated with center-based care showed enhanced developmen- higher ratings on learning competencies, suggesting tal outcomes. Mean child care quality over time pre- that quality of care may balance the risk associated dicted higher scores on the 24-month BSID– MDI with time in care. In contrast to the NICHD report, and 36-month PPVT–III. Mean child-adult ratio over social-emotional problems were related to stabili- time did not signifi cantly predict child outcomes. ty rather than quantity of care. We suggest that, in Mean hours in center care over time predicted high- a context in which standards for good-quality care er scores on the 24-and 36-month BSID–MDI and are enforced through government regulatory mech- the 36-month PPVT–III. Neither the quality nor the anisms, the risk for behavior problems may be ex- amount of child care predicted child aggressive be- plained by factors other than time in care. havior at 24 or 36 months. In the Haifa study, child care, especially center Consistent with previous research, these fi ndings care, increased the likelihood of the infants’ att ach- demonstrate that the quality of the child care centers ment insecurity to their mothers. We suggest that that Early Head Start children att ended was positive- this is because early infant child care, with its very ly associated with children’s cognitive and language high infant-caregiver ratio, interfered with the tradi- development. Moreover, spending more time in cen- tional link oft en reported in the literature between ter-based child care was associated with higher cog- maternal sensitivity and infants’ security of att ach- nitive and language scores at 24 and 36 months. We ment to their mothers. Analyses of the combined found no evidence that more time in child care was Haifa–NICHD data showed a signifi cant association associated with lower child wellbeing or higher rates between child-caregiver ratio (as an index of quality) of aggressive behavior (ACF, 2003). and children’s att achment security but no relation between amount of care and att achment security. Con- Summary and Conclusions trary to the suggestion that, in the case of social-emotional development, only child care quantity matt ers, This article has described three recent studies that evidence was found for the signifi cance of child care examine associations between early child care and quality. We suggest that, when the range of quality child outcomes among families who are diff erent of care is broadened—either upward or downward, from those profi led in the NICHD Study of Ear- as was the case in the combined Haifa–NICHD data ly Child Care. The three studies suggest that qual- set—quality of child care becomes more salient than ity of child care is an important factor infl uencing time in care in infl uencing children’s development. 1032 Love, et al. in Child Development 74 (2003)

Finally, in the Early Head Start study, with its ex- References perimental design, children from low-income families (who would be expected to be at higher risk Achenbach, T. (1991). Manual for the Child Behavior Checklist/4- for behavior problems than those in the NICHD 18 and 1991 profi le. Burlington: University of Vermont. sample) experienced a wide range of positive im- Achenbach, T. M. (1992). Child Behavior Checklist for ages 2-3. pacts, including reduced aggressive behavior prob- Burlington: University of Vermont. lems. These positive gains in cognitive, language, Achenbach, T. M., & Rescorla, L. A. (2000). Manual for the ASE- BA preschool forms and profi les. Burlington: University of and social-emotional development occurred for Vermont. children enrolled in a program that provided high Administration for Children and Families (2002a). Making a levels of good-quality center child care. The experi- diff erence in the lives of infants and toddlers and their families: mental evidence raises serious questions about con- The impacts of Early Head Start. Washington, DC: U.S. De- cluding that an increased amount of child care is partment of Health and Human Services. detrimental for children’s development, at least in Administration for Children and Families (2002b). Pathways the fi rst 3 years of life. In addition, for the sample to quality and full implementation in Early Head Start. Wash- of Early Head Start children in center-based child ington, DC: U.S. Department of Health and Human Ser- care, spending more time in center-based child care vices. was associated with higher cognitive and language Administration for Children and Families (2003). The role of Early Head Start programs in addressing the child care needs scores at 24 and 36 months. We found no evidence of low-income families with infants and toddlers: Infl uences on that more time in child care was associated with child care use and quality. Washington, DC: U.S. Depart- higher rates of aggressive behavior problems. We ment of Health and Human Services. suggest that this may be a function of both the sam- Administration on Children, Youth and Families. (2001). ple characteristics (100% low income) and the gen- Building their futures: How Early Head Start programs are en- erally good levels of quality care the children expe- hancing the lives of infants and toddlers of low-income families. rienced. Volumes I–II. Washington, DC: U.S. Department of Health and Human Services. In summary, the three perspectives provided by our research provide strikingly consistent evidence Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Pat- terns of att achment. Hillsdale, NJ: Erlbaum. for the importance of child care quality in the devel- Australian Bureau of Statistics (1998). Australian social trends opment of young children. We suggest that the gen- 1998. Canberra: Commonwealth of Australia. eralizability of the NICHD study fi ndings hinges on Australian Bureau of Statistics. (1999). Child care, Australia, the specifi c context in which these results have been 1999. Canberra: Commonwealth of Australia. obtained. Our data stress the need to take into ac- Aviezer, O., Sagi, A., & Koren-Karie, N. (in press). Ecological count the potentially restricted range for child care constraints on the formation of infant–mother att achment quality in each investigation. The NICHD Study of relations: When maternal sensitivity becomes ineff ective. Early Child Care is the most impressive investiga- Infant Behavior and Development. tion on child care to date, and the consortium mem- Bayley, N. (1993). Bayley Scales of Infant Development, second bers should be highly commended for their care- edition: Manual. New York: Psychological Corporation. ful, creative, and painstaking work on one of the Bowlby, J. (1978). Att achment and loss: Vol. 1. Att achment. Mid- most crucial issues in child development. Natural- dlesex, England: Penguin (Original work published ly, such seminal research triggers scientifi c debate, 1969). which underlines the importance of this unprece- Campbell, F. A., Pungello, E. P., Miller-Johnson, S., Burchi- nal, M., & Ramey, C. T. (2001). The development of cogni- dented collaborative eff ort. We hope that the cu- tive and academic abilities: Growth curves from an early mulative eff ect of the data reported here—collect- childhood educational experiment. Developmental Psychol- ed in diff erent countries, with diff erent ranges of ogy, 37, 231–242. child care quality, in diff erent regulatory contexts, Cost, Quality, and Child Outcomes Study Team (1995). Cost, and with a diversity of family characteristics—is to quality, and child outcomes in child care centers; Public report. provide more complete estimates of the manner in Denver: University of Colorado at Denver. which both quality and quantity of child care may Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabu- infl uence a range of young children’s developmen- lary Test–third edition. Circle Pines, MN: American Guid- tal outcomes. ance Service. Child Care Quality Matt ers: How Conclusions May Vary With Context 1033

Fenson, L., Dale, P. S., Reznick, J. S., Thal, D., Bates, E., Har- of infant child care on infant–mother att achment security. tung, J. P. et al. (1993). The MacArthur Communicative De- Results of the NICHD study of early child care. Child De- velopment Inventories: User’s guide and technical manual. San velopment, 68, 860–879. Diego, CA: Singular. National Institute of Child Health and Human Development Glezer, H. (1988). Maternity leave in Australia: Employee and em- Early Child Care Research Network (1997b). Familial fac- ployer experiences: Report of a survey. Melbourne: Australian tors associated with the characteristics of nonmaternal Institute of Family Studies. care for infants. Journal of Marriage and the Family, 59, 389– Harms, T., Cliff ord, R., & Cryer, D. (1998). Early Childhood En- 408. vironment Rating Scale–Revised. New York: Teachers Col- National Institute of Child Health and Human Development lege Press. Early Child Care Research Network (2000a). Characteris- Harms, T., Cryer, D., & Cliff ord, R. (1990). Infant/Toddler Envi- tics and quality of child care for toddlers and preschool- ronment Rating Scale. New York: Teachers College Press. ers. Applied Developmental Science, 4, 116–135. Harrison, L., & Ungerer, J. A. (1997). Child care predictors of National Institute of Child Health and Human Development infant–mother att achment security at age 12 months. Ear- Early Child Care Research Network (2000b). The rela- ly Child Development and Care, 137, 31–46. Harrison, L. J., & tion of child care to cognitive and language development. Ungerer, J. A. (2002a). Maternal employment and infant– Child Development, 71, 960–980. mother att achment security at 12 months postpartum. De- National Institute of Child Health and Human Development velopmental Psychology, 38, 758–773. Early Child Care Research Network (2001). Before Head Harrison, L. J., & Ungerer, J. A. (2002b, December). The Sydney Start: Income and ethnicity, family characteristics, child Family Development Project: Family and child care predictors care experiences, and child development. Early Education of school adjustment at age six. Paper presented in the sym- and Development, 12, 545–576. posium, Longitudinal Studies of Early Childhood in Aus- National Institute of Child Health and Human Development tralia, J. Bowes (Chair), at the Australian Association for Early Child Care Research Network (2003a). Does amount Research in Education conference, Brisbane. of time spent in child care predict socioemotional adjust- Hightower, A. D., Work, C., Cowen, E., Lotyczewski, B., Spi- ment during the transition to kindergarten? Child Develop- nell, A., Guare, J., et al. (1989). The Teacher-Child Rating ment, 74, 976–1005. Scale: A brief objective measure of elementary school chil- National Institute of Child Health and Human Development dren’s school behaviour and competence. School Psycholo- Early Child Care Research Network. (2003b). Does quali- gy Review, 15, 393–409. ty of child care aff ect child outcomes at age 4 1/2? Develop- Klein, E. L., & Abu Taleb, T. F. (1993, March). Self concept and mental Psychology, 39, 451–469. prosocial behavior in kindergarten. Paper presented at the bi- Pianta, R. C. (1990). The Student-Teacher Relationship Scale. Un- ennial meeting of the Society for Research in Child Devel- published manuscript, University of Virginia. opment, New Orleans. Sagi, A., Koren-Karie, N., Gini, M., Ziv, Y., & Joels, T. (2002). Koren-Karie, N., Sagi, A., & Egoz-Mizrachi, N. (1998, July). Shedding further light on the eff ects of various types and Quality of infant group care in Israel. Paper presented at quality of early child care on infant–mother att achment the biennial meeting of the International Society for the relationship: The Haifa study of early child care. Child De- Study of Behavioral Development, Bern, Switzerland. velopment, 73, 1166–1186. Lamb, M. E. (1998). Nonparental child care: Context, quali- Scarr, S., & Eisenberg, M. (1993). Child care research: Issues, ty, correlates. In W. Damon, I. E. Sigel, & K. A. Renninger perspectives, and results. Annual Review of Psychology, 44, (Eds.), Handbook of child psychology, volume 4: Child psychol- 613–644. ogy in practice (5th ed., pp. 73–134). New York: Wiley. Schaefer, E., Edgerton, M., & Aaronson, M. (1978). Child Be- Love, J. M., Schochet, P. A., & Meckstroth, A. L. (1996). Are havior Inventory. Unpublished manuscript, Publisher: Uni- they in any real danger? What research does—and doesn’t—tell versity of Carolina. us about child care quality and children’s well-being. Prince- U.S. Department of Health and Human Services (1966, No- ton, NJ: Mathematica Policy Research. vember). Head Start Program: Final rule. Federal Register, National Association for the Education of Young Children. 61(215), 57186–57227. (1987). Accreditation criteria and procedures. Washington, Wangmann, J. (1995). Towards integration and quality assurance DC: Author. in children’s services. Melbourne: Australian Institute of National Childcare Accreditation Council (1993). Putt ing chil- Family Studies. dren fi rst. Quality Improvement and Accreditation System Whitebook, M., Howes, C., & Phillips, D. (1989). Who cares? handbook. Sydney, Australia: Author. Child care teachers and the quality of care in America: Final re- National Institute of Child Health and Human Development port: National Child Care Staffi ng Study. Berkeley, CA: Child Early Child Care Research Network (1997a). The eff ects Care Employee Project.