Download - Diversity, child care quality, and developmental outcomes · backgrounds beneﬁted from sensitive and stimulating caregiving as measured on a standard assessment tool, ECERS, especially

Early Childhood Research Quarterly 18 (2003) 401–426

Diversity, child care quality, and developmental outcomes

Margaret R. Burchinal∗, Debby Cryer

Frank Porter Graham Child Development Center, University of North Carolina, 216 Sheryl Mar,CB #8185, Chapel Hill, NC 27599-8185, USA

Abstract

It is widely accepted that high quality child care enhances children’s cognitive and social development,but some question whether what constitutes quality care depends on the child’s ethnic and culturalbackground. To address this question, secondary analysis of data from the two largest studies of childcare experiences in the United States, Cost, Quality, and Outcomes Study and the NICHD Study ofEarly Child Care, tested whether standard measures of child care quality were less reliable or valid forAfrican-American and English-speaking Latino children than for white children. Widely used measuresof child care quality showed comparably high levels of reliability and similar levels of validity for white,African-American, and Latino children. Analyses tested whether cognitive and social skills were relatedto child care quality, the match between child’s and caregiver’s ethnicity, and the match between themother’s and caregiver’s beliefs about child-rearing. Results indicated children from all three ethnicgroups showed higher levels of cognitive and social skills on standardized assessments shown to predictschool success when caregivers were sensitive and stimulating. Children’s skills were not consistentlyrelated to whether the child’s and caregiver’s ethnicity matched or whether the mother’s and caregiver’sbeliefs about child-rearing were similar. These two large studies suggest that children from all threeethnic groups benefit from sensitive and stimulating care on child outcomes related to school success.The results are interpreted as indicating that the global dimension of quality may be reflected in verydifferent types of practices that reflect cultural differences.© 2003 Elsevier Inc. All rights reserved.

Keywords:Ethnic diversity; Child care quality; Developmental outcomes

One of the most consistent findings in developmental research links higher quality child carewith young children’s developing skills (Lamb, 1998; Vandell & Wolfe, 2000). Responsiveand stimulating care in child care, as well as at home, is linked theoretically (Bronfenbrenner& Morris, 1998) and empirically (Lamb, 1998) to better cognitive and social outcomes for

∗ Corresponding author. Tel.:+1-919-966-5059; fax:+1-919-966-0862.E-mail address:[email protected] (M.R. Burchinal).

0885-2006/$ – see front matter © 2003 Elsevier Inc. All rights reserved.doi:10.1016/j.ecresq.2003.09.003

402 M.R. Burchinal, D. Cryer / Early Childhood Research Quarterly 18 (2003) 401–426

young children. While there is general consensus that child care quality is linked to suchchildoutcomes, there is less agreement about whether these outcomes are what is most importantfor all children, or whether the definition of high quality care applies across cultures (Moss &Pence, 1994).

Child care quality is of concern to parents and policy makers because infants and preschool-ers need responsive and stimulating interactions with adults to enhance social, cognitive, andlanguage development in early childhood (Bronfenbrenner & Morris, 1998; Sameroff, 1983).Adults who take turns in interactions with young children, share periods of joint focus, andexpress positive affect provide young children with the linguistic scaffolding needed to facil-itate language and cognitive development and the secure base needed to develop social skills(Bradley et al., 1989; Bronfenbrenner & Morris, 1998; Howes, 2000; Katz & Snow, 2000;Tomasello & Farrar, 1986). Further, it is believed that children need opportunities to inter-act with stimulating materials in a safe environment at home or in child care (Bradley et al.,2001; Harms & Clifford, 1980). Accordingly, measures of child care quality were developedto assess the extent to which child care providers provide sensitive and responsive interactionswith children and create stimulating environments with the belief that such experiences shouldbe related to their cognitive and social development (Harms & Clifford, 1980; NICHD EarlyChild Care Research Network, 1996).

Widely used measures of child care quality measure the sensitivity of the care provider andamount of simulation in the care environment. These measures include the Early ChildhoodEnvironmental Rating Scale (ECERS;Harms & Clifford, 1980) and the Profile (Abbott-Shim& Sibley, 1992) as measures of center classroom, the Family Day Care Environmental RatingScale (FDCRS;Harms & Clifford, 1989) as a measure of the child care home, and the Ob-servational Record of the Childcare Environment (ORCE;NICHD Early Child Care ResearchNetwork, 1996) and the Caregiver Interaction Scale (CIS;Arnett, 1989) as a measure the re-lationship between the caregiver and children in any setting. All measures assess the extent towhich care providers meet the developmental needs of children by creating a comfortable andsimulating environment and being sensitive and responsive in interactions with children.

Extensive research has shown that child care quality as measured by these instruments isrelated to children’s development. Child care quality has been consistently linked to children’scognitive and less consistently to social skills in all large, multi-site, observational studies usingthe ECERS and CIS (Howes, Phillips, & Whitebook, 1992; Peisner-Feinberg & Burchinal,1997; Zill, 1999) and the ORCE (NICHD Early Child Care Research Network, 1998, 2000)and most smaller studies using the ECERS or FDCRS (Burchinal et al., 1996, 2000; Dunn,1993; Kontos, 1991; McCartney, 1984; Schliecker, White, & Jacobs, 1991). Furthermore, thelink between child care quality and child outcomes has been observed in child care homesand relative care as well as in child care homes (Clarke-Stewart et al., 2002; Kontos, Howes,Shinn, & Galinsky, 1995; NICHD Early Child Care Research Network, 1998). Some studiessuggest that these measures of quality are linked more strongly to child outcomes for childrenof color (Bryant, Peisner-Feinberg, & Clifford, 1993; Burchinal, Peisner-Feinburg, Bryant,& Clifford, 2000; Burchinal, Ramey, Reid, & Jaccard, 1995) or children from low-incomefamilies than for middle-class children (Bryant, Burchinal, Lau, & Sparling, 1994; Caughy,DiPietro, & Strobino, 1994; Peisner-Feinberg & Burchinal, 1997; Vandell & Corasaniti, 1990).For example, a retrospective study of Seattle families reported that both vocabulary skills and

M.R. Burchinal, D. Cryer / Early Childhood Research Quarterly 18 (2003) 401–426 403

behavior problems in middle-childhood were more strongly related to child care experiencesfor African-American than white children (Burchinal et al., 1995) and a small meta-analysissuggested that child care quality was more strongly related to early childhood vocabularyskills for African-American than white children (Burchinal et al., 2000). Nevertheless, someprofessionals and parents worry that these standards reflect white, middle-class values and maynot be appropriate for children from other cultures.

There are three separate schools of thought regarding child care quality and ethnic and eco-nomic diversity. Based in part on the early intervention literature, high quality center-basedchild care is viewed by some professionals and policy makers as a means to enhance de-velopment among children experiencing risk factors such as low parental education, singleparent homes, and poverty (Lamb, 1998). Experimental early intervention studies (Campbell& Ramey, 1994; Lazar & Darlington, 1982; Schweinhart, Barnes, & Weikart, 1993) demon-strated that high quality child care improves children’s cognitive development for these at-riskchildren, and that such effects impact the children’s academic success and adjustment to adult-hood. These findings and those from observational studies in which child care quality wasmore strongly related to child outcomes for at-risk children support the contention that childcare quality is more strongly linked to outcomes when discrepancies between care at homeand child care are greater. However, several recent observational studies failed to find evidencefor moderating effects when examining parent education, family income, quality of home en-vironment, or child sex (Burchinal et al., 2000; NICHD Early Child Care Research Network,1997, 2000; Stipek, Feiler, Daniels, & Milburn, 1995).

In contrast, other professionals questioned whether mainstream measures of quality arerelevant for children from diverse backgrounds (Moss & Pence, 1994). They worry that devel-opment will be impaired if children experience discontinuities between home and child carein child-rearing beliefs and practices (Brophy & Statham, 1994;Garcia-Coll, 1990; GarciaColl et al., 1996; New, 1994). They argue that what constitutes high quality child care may bedifferent for children of color than for white/non-Hispanic children because of differences incultural practices and the history of societal discrimination. For example, some have arguedthat practices rooted in culture such as an emphasis on communal rather than individualistgoals, music, and activities are important for the success of African-American children (Allen& Boykin, 1992). Boykin and co-workers identified five cultural themes consistent with Afro-cultural ethos: movement, expressiveness, verve, affect, orality, and communalism. They foundthat African-American children performed better in school if their elementary classrooms wererated higher on these dimensions (Boykin & Bailey, 1999).

Other developmentalists offer a middle ground. They argue that the dimensions assessedby child care quality measures are important for all children, but those dimensions can be ex-pressed very differently depending on the child’s culture (Wishard, Shivers, Howes, & Ritchie,2003). Carollee Howes and co-workers demonstrated that at-risk children from diverse culturalbackgrounds benefited from sensitive and stimulating caregiving as measured on a standardassessment tool, ECERS, especially when child care quality was reflected in practices rootedwithin ethnic communities. All the programs had similar ECERS scores, but demonstratedvery different types of practices. They argue practices are the particular strategies used to teachor influence children, and are best viewed as within the overall cultural context; “Practices,more than quality, appear to deeply embedded within value and belief systems that rooted


in ethnicity, community, and social class” (Wishard et al., 2003, p. 5). The key elements ofpractice include in which activities the child engages, who participates, how the activities areorganized, and what are the goals. Very different types of practices can be judged as highquality if those practices reflect sensitive and stimulating caregiving.

Discontinuity between home and child care is believed to be confusing to the child, therebyimpairing adjustment, according to developmental theories such as Bronfrenbrenner’s eco-logical framework (for review, seeShpancer, 2002), but this tenet has not been examinedextensively. This is, in part, because the concept of continuity between home and child carehas been broadly defined, ranging from consistency in specific behaviors and beliefs aboutchild-rearing practices to whether child and caregiver share the same ethnicity. Whether con-tinuity itself is inherently beneficial has been questioned when discontinuity allows for theopportunity that children experience more optimal caregiving in one setting than the other.For example, Provost and co-workers reported that children’s adjustment and play skills werehigher among those who experienced higher levels of discontinuity in terms of types and num-bers of play objects at home and in child care (Provost, 1994) andErwin, Sanson, Amos, andBradley (1993)reported that children showed fewer behavior problems when caregivers andparents reported greater discrepancy on beliefs about child-rearing.

These issues have important policy implications as state and federal governments investlarge amounts of money in early care and education to ensure that children enter school readyto learn. Whether standard measures of quality relate to developmental outcomes differentlydepending on the ethnicity of the child, the match between the child’s and caregiver’s eth-nicity, and the match between parent’s and caregiver’s beliefs about child-rearing were ex-amined in secondary data analysis of the two largest child care studies. These two studieswere selected because they are the only studies that both measured child care quality andoutcomes and included a moderate to large number of children of color as well as a largenumber of white/non-Hispanic children. These are the Cost, Quality, and Child OutcomesStudy and the NICHD Study of Early Child Care. Neither study is ideal for addressing thisissue because neither recruited representative or large samples of ethnically diverse children,especially Hispanic children. We chose to examine this issue with these studies because webelieve it is an important issue and these data provide the best source of information thatis available, not because either project provides representative samples of ethnically diversechildren.

The Cost, Quality, and Child Outcomes study examined child care centers in each of fourstates (CA, CO, CT, and NC) selected to represent variation in economic and regulatory cli-mates, and child outcomes in a subset of those centers. Preschool-aged children in observedclassrooms were recruited and their cognitive, language, and social development was assessedfor 757 children (Peisner-Feinberg & Burchinal, 1997). Child care quality, a composite formedfrom four widely used observational measures of quality, was associated, albeit modestly, withhigher vocabulary and math scores on standardized tests and to the child care provider’s rat-ings of the child’s cognitive skills. An interaction between maternal education and care qualityindicated the quality composite was related to children’s reading skill only for children withmother’s who had less education. Child care quality was correlated (.15 < r < .30) with theseoutcomes prior to adjusting for covariates, with somewhat smaller associations after adjustmentfor site, maternal education, gender, and ethnicity.


NICHD Study of Early Child Care prospectively examined the association between childcare quality and development from birth through 3 years in 10 sites for over 1,100 children.Families were recruited at the child’s birth, and the child care settings chosen by the familieswere observed. Quality of care for children in a wide variety of settings (centers, child carehomes, babysitters in child’s home, and care by relative including grandparents and fathers)was measured with a standard rating scale when the children were 6, 15, 24, and 36 months ofage. Children’s cognitive and language development at 15, 24, and 36 months (NICHD EarlyChild Care Research Network, 2000) and social skills at 24 and 36 months (NICHD EarlyChild Care Research Network, 1998) was related to observed quality in analyses that adjustedfor family and child characteristics.

The purpose of this study was to examine the Cost, Quality and Outcomes Project (CQO) andNICHD SECC data more closely to determine whether child care quality related to outcomesdifferently for children from diverse ethnic backgrounds. Of particular interest was determin-ing whether children’s cognitive and social development was enhanced by having a child careprovider of the same ethnic background and whether mainstream measures of quality were lesspredictive of children’s outcomes for children of color than for white/non-Hispanic children.We also examined discrepancies in maternal and caregiver attitudes about child-rearing and itsassociation for child outcomes with the data from one study, the NICHD SECC. These analysesextend those reported previously by including information about the ethnicity of the child careprovider and testing whether the global measures of quality predict outcomes differently de-pending on the child’s ethnicity, the match between the child and child care provider’s ethnicity,and the match between the mother’s and child care provider’s attitudes about child-rearing.

1. Cost, Quality, and Outcomes study

1.1. Methods

1.1.1. ParticipantsThe Cost, Quality and Outcomes Project was designed to examine the relations among child

care costs, quality, and longitudinal outcomes for children in full-time care in community childcare centers. Included in this paper are child care and child outcomes data collected during thefirst year of the study in 1993 (seePeisner-Feinberg & Burchinal, 1997for details). Child carecenters were randomly selected from centers providing full-time care for at least 11 months ayear in four regions—Los Angeles County in California, the Hartford-New Haven corridor inConnecticut, the Frontal range in Colorado, and the Piedmont region in North Carolina. Theoutcome component of the study included a subsample of 177 of the 521 observed preschoolclassrooms in 170 of the 401 centers participating in the general study. Classrooms wereeligible if they served at least one child in the next-to-last year of preschool (i.e., eligible forkindergarten in fall of 1994).

Table 1lists descriptive statistics for these classrooms. About half (54%) were in non-profitcenters, the average observed staff–child ratio was approximately 1:8, and the average groupsize was about 14 children. In comparison, in the original sample 50% were non-profit centers,staff–child ratios averaged 1:6, and group size averaged 14.

406M

.R.B

urch

ina

l,D.C

ryer/E

arly

Ch

ildh

oo

dR

ese

arch

Qu

arte

rly1

8(2

00

3)

40

1–

42

6

Table 1Cost, Quality, and Outcomes Study—descriptive statistics by child’s ethnicity

White (n = 410) African-American (n = 105) Hispanic (n = 31)

M SD Range M SD Range M SD Range

Family and child characteristicsMaternal education 14.61 2.23 10 to 20 13.18 1.71 10 to 20 12.52 1.61 10 to 17Income (in thousands)a 54 24 2.4 to 84 27 23 2.4 to 84 21 17 4.8 to 72

Poverty (%) 6 34 50Working poor (%) 11 32 25Middle-class (%) 83 34 25

GenderMale (%) 51 51 42Female (%) 49 49 58

Child care settingType of care

Center (%) 100 100 100

Ethnicity of caregiverMatch 77 47 42No match (%) 23 53 58

Positive caregiving 0.27 1.59 −4.4 to 3.0 −0.20 1.47 −3.0 to 2.9 −0.86 1.64 −3.7 to 2.9ECERS total 4.55 1.05 1.7 to 6.6 4.15 0.92 2.7 to 6.4 3.93 1.07 2.4 to 6.3CIS total 3.07 0.56 1.4 to 3.8 2.97 0.47 1.7 to 3.7 2.70 0.56 1.6 to 3.8AIS prop responsive 0.30 0.26 0 to 1 0.31 0.21 0 to 1 0.18 0.22 0 to 0.61Child-centered 0.16 0.59 −1.5 to 1.1 −0.04 0.60 −1.2 to 1.0 −0.18 0.62 −1.3 to 0.92

Child outcomesPPVT receptive language 99.16 16.31 40 to 144 80.21 17.68 46 to 130 76.84 17.65 41 to 124

WJ-RAcademic readiness 102.96 10.62 79 to 142 95.60 10.91 67 to 130 91.35 1.41 60.5 to 111Letter–word identification 100.57 12.59 63 to 161 96.10 11.92 68 to 140 90.81 8.80 65 to 107Applied math 105.36 12.16 55 to 136 94.72 12.90 56 to 129 91.90 14.52 56 to 126

Class Behavior InventoryPositive Behavior Scale 3.79 0.68 1.4 to 5.0 3.47 0.78 1.4 to 5.0 3.42 0.73 1.8 to 5.0Problem behaviors 2.42 0.86 1 to 4.9 2.60 0.92 1 to 4.8 2.53 0.81 0.8 to 4.2

a Income reported in thousands of dollars. Poverty was indicated if income was less than federal poverty threshold, working poor if income was between100 and 200% of poverty threshold, and middle-class if income was greater than 200% of the threshold. Many parents did not provide income. Sample sizesweren = 366 for white,n = 67 for African-American, andn = 24 for Hispanic families so this variable was not included in analyses.


About 4 months after the quality data were collected and after the classroom child careprovider agreed to participate in the child outcomes portion of the study, consent forms weresent home to parents of all potentially eligible children, with up to 12 children randomlyselected from each classroom. The four criteria for inclusion were: (1) children eligible toenter kindergarten in the fall of 1994, (2) children were enrolled in the classroom during thequality observations; (3) parents expected their child to continue at that center the followingyear; and (4) the primary language spoken at home was English. Refusal rates were low; forexample, of those eligible to participate in NC, 7% of the parents or children refused. Most ofthe sample was European-American (68%), but the sample also included African-Americans(15%), Asian-Americans (4%), Hispanics (6%), Native-Americans (1%), and children of theother or mixed ethnic backgrounds (8%).

For the purposes of these analyses, we included only children who were white, African-American, or Hispanic and who had complete data on maternal education, child’s ethnicity,care provider’s ethnicity, and child care quality. These procedures yielded a sample including828 children. We did not include the 23 children from whom ethnicity was missing or the 30children listed as Asian-American, or the 65 children listed as “other.” Missing data at eitherthe classroom or the child/family level (n = 71) resulted in an analysis sample of 669 children,with smaller numbers for some analyses. Comparison on any analysis variables of the 669 chil-dren included and the 159 excluded from analyses yielded significant differences on teacherratings of behavior problems (t(824) = 3.63, p < .001), pre-reading skills (t(791) = 2.64,p < .01), and proportion of time the teacher was observed as responsive (t(809) = 4.13,p < .001) that favored the children excluded from the analysis. Children from North Carolinawere significantly more likely to be included than children from other states (χ2(1) = 50.3,n = 828,p < .001).

Table 1provides a description of the sample. The mean number of participating children ineach class was 4.35 (SD = 2.38, range= 1–12). The children were an average of 4.3 yearsold at the time of the assessments. The sample was approximately evenly divided by gender(52% boys), and about one-third were children of color.

1.1.2. Measures

1.1.2.1. Child care classroom observations.Teams of six to eight assessors from each of thefour sites were trained to acceptable levels of reliability during a week-long session. Interraterreliability visits were conducted at the mid-point of data collection, and included both within-and between-state. Each classroom was observed by two assessors for approximately 3–4 hoursin a single visit.

Global classroom quality was measured by theEarly Childhood Environment Rating Scale(Harms & Clifford, 1980), which examines the developmental appropriateness of classroompractices. The ECERS contains 37 items which are rated on a 1–7 scale from inadequateto excellent. Psychometric analysis indicated that a single total score most parsimoniouslyrepresented the data (Cronbach’sα = .96). The total score was computed as the meanof the 32 child-oriented items. Interrater reliability was calculated using Pearsoncorrelation coefficients, and ranged from .83 to .98 for the total score, with a medianof .94.


Child care provider sensitivity was rated using theCaregiver Interaction Scale(Arnett,1989). Twenty-six items are rated on a 1–4 scale indicating how characteristic they are, fromnot at all to very much. Psychometric analyses suggested a single factor most parsimoniouslyrepresented these data (Cronbach’sα = .93). Interrater reliability ranged from .89 to .98 forthe four subscales.

The extent to which the teaching style is didactic versus child-centered was rated using theUCLA Early Childhood Observation Form(Stipek, 1993). Twenty-four items are scored on 3-,4- and 5-point scales, representing the continuum from didactic (low) to child-centered (high).A total score was computed by converting all items to the metric of a 5-point scale (α = .92).Interrater reliability ranged from .81 to .97 for each subscale, with median scores ranging from.91 to .95.

Data on child care provider responsiveness to children was gathered with theAdult Involve-ment Scale(Howes & Stewart, 1987). For this instrument, two children (one boy and one girl)were randomly selected in each classroom and observed for three observations of 5 minuteseach, with child care provider–child interactions coded every 20 seconds. (No additional infor-mation was gathered for these children.) For the present analyses, child care provider respon-siveness was calculated as the percentage of time the child care provider was at least minimallyresponsive to children (the four upper points of the scale). No measure of internal consistencywas computed because this value is the sum of these mutually exclusive categories. Interraterreliability was calculated using Cohen’s kappa, with median= .92 (range= .83–.96).

The four summary measures were moderately to highly correlated, with correlations from .74to .91 among the ECERS classroom environment, CIS teacher sensitivity, and ECOF teachingstyle, and from .26 to .31 between the AIS teacher responsiveness and the other measures.Therefore, a single composite quality index was computed. A principal components analysisof the four measures indicated that one factor accounted for 68% of the total variance, and thatsubsequent factors were unnecessary. The composite observed child care quality index wascalculated as the weighted mean of the four scores based on the principal component analysis.

1.1.2.2. Children and families.Site coordinators from each state were trained in the standardprocedures for administering the assessment instruments in a 3-day session. They then trainedfour to seven data collectors at each site and monitored them throughout the data collec-tion process. Any assessors who had collected classroom quality information did not collectchild outcome data from the same classroom. Individual assessments of each child were con-ducted at the child care center in a session that lasted approximately 30 minutes. Following theassessments, questionnaires were given to child care providers. The parent demographic ques-tionnaires were previously sent home with the original consent form. Pre-addressed, stampedenvelopes allowed parents and child care providers to return the forms to the investigators. Thereturn rates were high, 98% for the parent surveys and 96% for the child care provider surveys.Information pertaining to children’s cognitive and socio-emotional developmental status wasgathered from individual assessments of the children and from child care provider ratings. Inaddition, parents provided demographic information about the child and family.

1.1.2.3. Individual child assessments.Receptive language comprehension was measured usingthePeabody Picture Vocabulary Test—Revised(PPVT-R;Dunn & Dunn, 1981). The format


of this test is appropriate for young children, requiring children to point to the picture thatmatches the word spoken by the examiner. PPVT-R raw scores were converted into standardscores based on age, with a mean of 100 and a standard deviation of 15 in the norming sample.Based on the original test development, this measure has good split-half (median= .80) andtest–retest reliability (median= .82), correlates highly with other measures of vocabulary andmoderately with intelligence tests and school achievement.

Children’s achievement in pre-reading and pre-math skills was measured using two sub-tests of theWoodcock–Johnson Tests of Achievement—Revised(WJ-R)(Woodcock & Johnson,1990). Pre-reading skills were examined using the letter–word identification subtest, whichmeasures children’s ability to recognize letters and words and to associate pictures and sym-bols. Pre-math skills were examined using the applied problems subtest, which measureschildren’s abilities in understanding basic numeracy, comparing different numbers of items,and simple counting. Standard scores based on age were used, with a normed mean of 100 anda standard deviation of 15. Test development information showed high internal consistencyfor these subtests (median= .92, .91), and moderate correlations with other tests of achieve-ment. To create a school readiness score, we averaged the standard scores for letter–wordidentification and applied problems subscales.

1.1.2.4. Child care provider surveys.For each participating child, the lead child care provider inthe classroom was asked to rate the child’s social and cognitive skills. TheClassroom BehaviorInventory(Schaefer, Edgerton, & Aaronson, 1978) was used to measure child care providerperceptions of children’s social and intellectual skills across 10 scales. Forty-two items are ratedfor how typical they are of the child, using a 5-point likert scale from not at all to very much like.The 10 CBI scale scores were factor-analyzed that accounted for 76% of the variance. Thesefactors replicated other factor analyses of the CBI (Osborne, Schulte, & McKinney, 1991),yielding a pro-social factor and a problem behavior factor. The pro-social factor consisted ofthe independence, task orientation, dependence (reversed) and distractibility (reversed) scales,with internal consistency (α) of .84 in this sample. The problem behavior factor consisted ofthe distractibility, hostility, and consideration (reversed) scales, with an alpha of .77.

1.2. Results

Table 1presents descriptive statistics for all analysis variables by the child’s ethnic back-ground. Of the overall sample, the analysis sample included 410 white children, 105 African-American children, and 31 Hispanic children. These children had complete data on the familycovariates (ethnicity, gender, and maternal education), the child care variables (positive caregiv-ing and match between the child’s and caregiver’s ethnicity), and at least one child outcome.Positive caregiving was computed as the mean of 0 and standard deviation of about 1.5 inthe sample. As can be seen in this table, the white children were more advantaged than theAfrican-American children. Their families were much less likely to be poor, and much morelikely to be middle-class. Their mothers tended to have more education and their child careproviders tended to provide more positive caregiving.

First, we looked at the psychometric properties of the measures of child care quality forchildren from the three ethnic groups. The internal consistency of measures of quality and


Table 2Internal consistency of measures of observed child care quality

White (α) African-American (α) Hispanic (α)

Overall quality ECERS .96 .96 .96Teacher sensitivity CIS .90 .91 .92Child-centered ECOF .91 .91 .91Composite positive caregiving .80 .82 .84

the composite was similar for children from the three ethnic groups (seeTable 2). Overall,the internal consistency was very high for the measures of global quality (ECERS), teachersensitivity (CIS), and child-centeredness (ECOF), and high for the composite created from thetotal scores from these three instruments and the AIS measure of teacher responsiveness.

Next, correlations among the two child care experience variables and the four child outcomeswere examined as a preliminary measure of validity. Those correlations are shown inTable 3.Examination of these correlations shows positive caregiving more strongly associated withcognitive outcomes than with social outcomes. Positive caregiving was moderately correlatedwith receptive language scores for all three groups, although the magnitude of the correlationswas stronger among African-American and Hispanic children than among white children.Positive caregiving was modestly correlated with the school readiness score for all three groupsof children. This correlation was significantly different from zero only for white children onlydespite the fact that the magnitude of the association was stronger for Hispanic children dueto differences in sample size. Positive caregiving was not reliably correlated with either socialoutcome. In addition, the match between the child’s and caregiver’s ethnicity did not showconsistent positive correlations with outcomes within ethnic groups.

Finally, we asked whether child care experiences predicted child outcomes when familyselection factors were considered, and whether the magnitude of those associations varied byethnicity. Hierarchical mixed-effect regressions (also called hierarchical linear models) werefit to the child outcomes. Child care center was included in all HLM analyses as a randomeffect variable to adjust for potential lack of independence due to the children’s common

Table 3CQO correlations between child care experiences and child outcome

PPVT receptivelanguage (r)

WJ-R schoolreadiness (r)

CBI pro-social(r)

CBI behaviorproblems (r)

White childrenPositive caregiving .28∗∗∗ .18∗∗ .07 −.05Teacher–child ethnic match .07 −.02 .06 −.03

African-American childrenPositive caregiving .32∗∗∗ .13 .11 −.05Teacher–child ethnic match −.31∗∗∗ −.11 .02 .08

Hispanic childrenPositive caregiving .41∗∗∗ .32 .02 .30Teacher–child ethnic match −.22 .14 −.19 .16

∗∗ p < .01.∗∗∗ p < .001.


classroom experiences. HLM analyses related the child care quality composite to children’sdevelopment, adjusting for child and family characteristics. Included in the all regressionsas fixed effect predictors were the composite child care quality index, maternal education inyears, gender (1= male, 0= female), and ethnicity (represented in two dummy variables,1 = African-American, 0= otherwise; 1= Hispanic, 0= otherwise). Income and otherfamily characteristics were not included due to the extensive amount of missing data on thosevariables. Many of those questions were collected in subsequent years of data collection, butabout one-fourth of the sample changed child care providers between years one and two. Thosefamilies were lost to the study.

The analysis models were fit hierarchically. The first model included state, gender, maternaleducation, the child’s ethnicity, and positive caregiving. The second model added the interac-tion between positive caregiving and ethnicity. The third model added the match between thechild’s and caregiver’s ethnicity. The fourth model added the two-way interactions involvingmatch. The final model included match, ethncity, and positive caregiving. Results are shownin Table 4, listing the random-effects variance estimate for center, the unstandardized regres-sion coefficients for all continuous variables, and theF statistics for all predictors of interest.Coefficients were not listed for non-significant interactions.

Positive caregiving was significantly associated with both receptive language (F(1, 494) =16.83,p < .001) and school readiness (F(1, 492) = 4.28,p < .05). We computed effect sizesas the increase in the child outcome associated with one standard deviation increase in quality.

Table 4CQO regression results: relating ethnicity, positive caregiving, and child–caregiver match to child outcomes

PPVT receptivelanguage

WJ-R schoolreadiness

CBIpro-social

CBI behaviorproblems

Step 1: background variables and child care experiencea

Level 1—random effectsCenter σ2(SE) 36.6∗∗∗ (9.5) 6.73∗ (3.60) 0.08∗∗∗ (0.02) 0.04 (0.03)

Level 2—fixed effectsMaternal education B (SE) 16.1∗∗∗ (2.7) 1.15∗∗∗ (0.18) 0.03∗ (0.01) −0.05∗∗ (0.02)Positive caregiving F 16.83∗∗∗ 4.28∗ 2.15 0.03

B (SE) 1.89∗∗∗ (0.46) 0.57∗ (0.28) 0.03 (0.02) −0.00 (0.02)

Step 2: ethnicity× positive caregivingF 2.34 0.76 1.07 0.40

Step 3: child–caregiver matchF 0.03 0.05 2.20 0.52B (SE) −0.30 (1.78) −0.26 (1.12) −0.13 (0.09) 0.07 (0.10)

Step 4: match interactionsEthnicity× match F 1.21 1.19 1.40 1.02Positve care giving× match F 0.92 0.17 0.30 0.63

Step 5: ethnicity×match× positive caregivingF 0.12 1.36 0.33 0.10

a Model also includes site, gender, and ethnicity as categorical variables.∗ p < .05.∗∗ p < .01.∗∗∗ p < .001.


A standard deviation increase in positive caregiving (1.5) was associated with an increase of2.8 points on the PPVT standard score (i.e., 1.5 × 1.89; regression coefficient for quality inanalysis of PPVT= 1.89) and 0.9 points on the WJ reading and math composite. Despite asignificant negative association between ethnic match and PPVT scores for African-Americanchildren in the descriptive analyses, neither the match between the child’s and caregiver’sethnicity nor interactions involving positive caregiving, match, and ethnicity was significantfor any outcome. No evidence emerged suggesting that positive caregiving was more stronglyrelated to outcomes for white children than for children of color. Indeed, the interaction betweenethnicity and positive caregiving (F(1, 492) = 2.34,p = .09) in analysis of the PPVT providednon-significant evidence that positive caregiving was more strongly related to PPVT scoresfor African-American children (b = 3.43, SE = 1.07) and Hispanic children (b = 4.46,SE= 2.64) than for white children (b = 1.63,SE= 0.60).

Neither positive caregiving nor the child’s and caregiver’s ethnic match was reliabilityrelated to the caregiver’s rating of pro-social skills or behavior problems.

In summary, these secondary analyses suggested that children of diverse backgroundsshowed better cognitive outcomes when they experienced more sensitive and stimulating childcare. No evidence emerged to indicate that these measures of child care quality were less re-liable or valid for African-American or Hispanic children regardless of the ethnicity of theircare providers.

2. NICHD Study of Early Child Care

2.1. Participants

Participants were recruited from hospitals located in or near Little Rock, AR; Irvine, CA;Lawrence, KS; Boston, MA; Philadelphia, PA; Pittsburgh, PA; Charlottesville, VA; Morganton,NC; Seattle, WA; and Madison, WI. During selected 24-hour sampling periods in 1991, 8,986women giving birth were visited in the hospital. Of these, 5,265 met the eligibility criteriafor the study and agreed to be contacted after their return home from the hospital. Excludedfamilies included families who did not anticipate remaining in the areas or who lived morethan a hour from assessment sites, mothers who did not speak English or who were less than18 years old, and children who remained hospitalized for more than 7 days or had detectabledisabilities. When the infants were 1-month old, 1,364 families (58% of those contacted) withhealthy newborns were enrolled in the study. Families were similarly distributed across the tensites, with each site enrolling a minimum of 120 families.

Of those 1,364 children, 597 were included in these secondary data analyses. Children wereexcluded from analyses if their ethnicity was listed was not listed as white, African-American,or Hispanic (n = 66), their developmental outcomes were not assessed at 36 months (n = 145),they were not enrolled in child care at 36 months for at least 10 hours per week (n = 384),they were enrolled in child care but their child care was not observed (n = 121), or the careprovider did not list their ethnicity (n = 51). Children who were included in the analysesdiffered significantly from those excluded on maternal education (t(1361) = 6.08), family in-come (t(1362) = 10.92), maternal attitudes (t(1356) = 4.42), children’s 36-month language


(t(1153) = 3.56), school readiness skills (t(1153) = 3.00), and maternal ratings of socialskills (t(1175) = 2.93), and behavior problems (t(1173) = 2.29). Most of these differenceswere due greater loss to follow-up and less likelihood of using child care among more disad-vantaged families. Children lost to follow-up or who not in child care were significantly moredisadvantaged than children attending child care.

2.2. Measures

Data were collected in three settings: the child’s home, a laboratory, and the child’s primarychild care arrangement (the arrangement in which the child spent the most time or, if thechild spent equal time in two settings, the arrangement that was more formal). Mothers andchildren were visited in their homes when the children were 1, 6, 15, 24, and 36 months old; theprimary child care environment was observed at 6, 15, 24, and 36 months of age; and childrenand their mothers were seen in the laboratory at 15, 24, and 36 months. At each assessmentpoint, mothers responded to standardized interview questions about family demographics andother domains of family life not dealt with in this report. The families were telephoned at3-month intervals between assessment points to update information about child care and familycharacteristics.

All data collectors were highly trained and certified on data collection procedures. Theperformance of data collectors was monitored centrally to insure uniform, high-quality datacollection across the 10 sites. Details about all data collection procedures are documented inManuals of Operation of the study (NICHD Early Child Care Research Network, 1993).

2.2.1. Quality of careMeasures of child care quality were coded live using theObservational Record of the

Caregiving Environment(ORCE;NICHD Early Child Care Research Network, 1996). TheORCE was the only measure that was collected in all types of child care setting, so it is theonly measure discussed here. The ORCE was assesses the quality of caregiver–child interactionexperienced by an individual child. Observations of child care quality were made during twohalf-day visits at 6, 15, 24, and 36 months. The first three cycles consisted of 10-minuteobservation periods during which recordings of predetermined caregiver and child behaviorswere made every 30 seconds. The three cycles were separated by two 2-minute breaks duringwhich notes were made for qualitative ratings of caregiver behavior (ratings were made usinga 4-point scale). The last 10-minute cycle was devoted exclusively to qualitative ratings. (TheORCE method is described in detail inNICHD Early Child Care Research Network, 1996.)

The 36-month global quality variable, positive caregiving rating composite, was created bysumming the ratings for seven scales: sensitivity to non-distress, stimulation of cognitive devel-opment, positive regard, detachment (reversed), flatness of affect (reversed), fostering explo-ration, and intrusiveness (reversed). The composites had good internal consistency (Cronbach’sα = .83).

Each ORCE observer was trained to reach criterion using videotapes that had been codedby experts. The criterion was 60% straight match with the expert coder. The criterion was80% agreement with the expert for grouped codes (for details seeNICHD Early Child CareResearch Network, 1996). Live inter-observer reliability was also calculated three to four


times at approximately 3-month intervals throughout each data collection period. Intraclasscorrelations among partners ranged from .89 to .99.

2.2.2. Type of careThe primary care arrangement of each child was classified at each of the four child care

assessments as one of five types: (1) child care center; (2) child care home (care in someoneelse’s home by a non-relative or relative other than the child’s grandparents); (3) care at homeby some one other than parents or grandparents; (4) grandparent care, or (5) father care. Forthe purpose of data analysis, whether the child was observed in center care was coded becausecenter care was the only type of care to provide independent prediction of child outcomes(NICHD Early Child Care Research Network, 2000).

2.2.3. Cognitive and language measuresThe Bracken School Readiness Scaleconsists of 51 items grouped into five categories:

knowledge of color, letter identification, number/counting, comparisons, and shape recogni-tion (Bracken, 1984). The score analyzed was the percentile rank. TheReynell DevelopmentalLanguage Comprehension Scale(RDLS; Reynell, 1991), was administered at 36 months ofage during the laboratory session. Composed of 67-item scales the scale assesses verbal com-prehension. For example, verbal comprehension is assessed as a child is presented with a setof objects and asked, “Where’s the spoon?” The alpha was .93.

2.2.4. Behavior problems and social skillBoth mother- and caregiver-reports of total child behavior problems (24 and 36 months)

were taken from theChild Behavior Checklist-2/3(CBCL; Achenbach, Edelbrock, & Howell,1987) and of pro-social skills from theAdaptive Social Behavior Inventory(ASBI; Hogan,Scott, & Bauer, 1992). The CBCL Total Behavior Problem scores from the 99-item CBCL-2/3included: externalizing (aggressive and destructive behavior), internalizing (social withdrawaland depression), sleep problems, and somatic problems. Research indicates that the CBCL-2/3shows good test–retest reliability and concurrent and predictive validity; it discriminates be-tween clinically referred and non-referred toddlers, and it predicts problem scores over a 3-yearperiod (Achenbach et al., 1987). Pro-social behavior ratings were calculated as the mean ofthe Express and Comply scales of the ASBI. In the current sample the coefficient alphas forthese scales ranged from .76 (express-mother ratings) to .87 (comply-caregiver ratings).

2.3. Results

Table 5lists the descriptive statistics for the analysis variables by ethnic group. As with theCQO sample, the white children tend to be more advantaged than either the African-Americanor Hispanic children. The white families have more income and were much less likely tobe poor, and the mothers tended to have more years of education and less traditional, moreauthoritative attitudes about child-rearing.

Next, we looked descriptively at whether the reliability and validity of the child care qualitymeasure varied among the children from the three ethnic groups. The internal consistencyof positive caregiving measure was similar for all three groups:α = .83 for white children;

M.R

.Bu

rchin

al,D

.Crye

r/Ea

rlyC

hild

ho

od

Re

sea

rchQ

ua

rterly

18

(20

03

)4

01

–4

26

415

Table 5NICHD Study of Early Child Care—descriptive statistics by ethnic group

White (n = 480–483) African-American (n = 60–61) Hispanic (n = 38–40)


Family and child characteristicsMaternal education 15.01 2.43 7 to 21 13.18 1.89 10 to 18 13.65 2.03 8 to 18

Family income/poverty 4.44 3.36 0.2 to 28.5 2.39 2.25 0.4 to 13.6 2.93 2.92 0.3 to 16.8Poverty (%) 6 27 20Working poor (%) 17 33 23Middle-class (%) 77 40 58

Mother’s traditional caregiving attitudes 70.50 15.05 33 to 117 89.41 17.85 46 to 113 79.95 16.02 56 to 121

GenderMale (%) 50 50 50Female (%) 50 50 50

Child care settingType of care

Center (%) 42 47 39Child care home (%) 24 18 24Father (%) 13 15 21Grandparent (%) 8 15 11In-home (%) 13 5 5

Ethnicity of caregiverMatch (%) 86 70 34No match (%) 14 30 66

Positive caregiving 19.91 3.24 10.25 to 27.25 17.57 3.27 11 to 23 19.83 3.02 14.3 to 26Caregiver’s attitudes 73.95 17.14 31 to 123 84.47 20.16 39 to 118 74.09 20.02 41 to 122Difference: mother’s− caregiver’s attitudes −3.39 20.36 −66 to 65 5.77 21.11 −30 to 52 5.37 18.63 −34 to 41

Child outcomesReynell Auditory Comprehension

Receptive language 102.12 14.29 62 to 136 83.62 13.81 62 to 125 91.50 12.89 68 to 115

Bracken School ReadinessPercentile score 47.71 26.12 0.2 to 99 21.27 22.87 0.2 to 91 32.74 22.09 0.4 to 84

416M

.R.B

urch

ina

l,D.C

ryer/E

arly

Ch

ildh

oo

dR

ese

arch

Qu

arte

rly1

8(2

00

3)

40

1–

42

6

Table 5 (Continued)

White (n = 480–483) African-American (n = 60–61) Hispanic (n = 38–40)


Caregiver reportASBI pro-social scale 57.77 7.14 33 to 69 54.77 6.82 36 to 67 55.50 6.68 40 to 68CBCL total problems 26.18 18.55 0 to 114 32.37 20.79 0 to 79 32.84 21.68 0 to 96

Mother’s reportASBI pro-social scale 59.13 5.21 44 to 69 54.78 6.62 34 to 68 57.70 5.47 43 to 66CBCL total problems 34.96 16.65 1 to 97 39.27 23.58 3 to 118 35.69 19.23 6 to 95


Table 6NICHD Study of Early Child Care—correlations between child outcomes and child care experiences by ethnicgroup

ReynellReceptiveLanguage

BrackenSchoolReadiness

Caregiver report Mother’s report

ASBIpro-social

CBCLbehaviorproblems

ASBIpro-social


White childrenPositive caregiving .15∗∗∗ .21∗∗∗ .15∗∗∗ −.12∗ .11∗ −.09+

Center .15∗∗∗ .13∗∗ −.11 −.04 −.06 .04Teacher–child match .03 .00 .06 −.08 .03 −.08+

Caregiver’s attitudes −.20∗∗∗ −.14∗∗∗ −.02 .14∗∗ .02 .06

African-American childrenPositive caregiving .17 .18 −.01 .01 .05 −.06Center .02 .23+ −.12 −.14 −.14 −.06Teacher–child match −.12 −.34∗∗ .17 .05 .09 .07Caregiver’s attitudes −.15 −.31∗∗ .08 .04 −.04 .33∗

Hispanic childrenPositive caregiving .21 .17 .10 −.21 .04 .02Center .59∗∗∗ .39∗ .10 −.18 .17 −.02Teacher–child match −.35∗ −.20 −.15 .07 −.17 −.07Caregiver’s attitudes −.16 −.12 −.05 .13 −.11 .18+ p < .10.∗ p < .05.∗∗ p < .01.∗∗∗ p < .001.

α = .82 for African-American children; andα = .82 for Hispanic children. We correlated thechild care experience measures with the child outcomes for children in these three ethnic groups(seeTable 6). Positive caregiving was significantly correlated with all but one outcomes for thewhite children, and showed similar correlations for the cognitive outcome for both the African-American and Hispanic children. The extent to which the care provider endorsed traditional at-titudes about child-rearing was negatively, albeit modestly, correlated with cognitive outcomes.Neither the match between mother’s and caregiver’s attitudes nor the match between the child’sand care provider’s ethnicity showed a consistent pattern of correlation with outcomes.

Next, we tested whether these child care experiences were related to child outcomes in hier-archical regressions when we adjusted for family selection factors. The first analysis model in-cluded site, gender, ethnicity, maternal education, whether the observed child care setting was acenter, average hours of child care through 36 months, hours of care in the observed setting, andquality of child care (positive caregiving). The second model tested whether positive caregivingrelated to outcomes differently for white, African-American, and Hispanic children. The thirdmodel tested whether child outcomes varied as a function of the ethnic match between the childand care provider. The fourth and fifth models added the two- and three-way interactions, re-spectively, involving the ethnic match. Results are shown inTable 7, listing theF statistics for thepredictors of interest and the unstandardized regression coefficients for all continuous variables.

As shown inTable 7, regression analyses that adjusted for site, maternal education, gen-der, ethnicity, type of care, and amount of child care revealed that positive caregiving was

418M

.R.B

urch

ina

l,D.C

ryer/E

arly

Ch

ildh

oo

dR

ese

arch

Qu

arte

rly1

8(2

00

3)

40

1–

42

6Table 7NICHD SECC regression results: predicting child outcomes from child care experiences




ASBIpro-social


ASBIpro-social


Step 1: background variables and child care experienceMaternal education B (SE) 2.01∗∗∗ (0.24) 3.68∗∗∗ (0.42) 0.28∗ (0.13) −1.02∗∗ (0.37) 0.34∗∗ (0.10) −1.49∗∗∗ (0.32)Center care B (SE) 4.77∗∗∗ (1.10) 7.66∗∗∗ (1.98) −1.58∗∗ (0.62) −2.36 (1.71) −0.60 (0.46) 0.67 (1.51)Total child-caring hours B (SE) 0.07 (0.05) 0.09 (0.09) 0.00 (0.03) 0.09 (0.08) 0.01 (0.02) 0.04 (0.07)Hours in setting B (SE) −0.02 (0.05) 0.05 (0.09) 0.05 (0.03) 0.05 (0.08) 0.01 (0.02) 0.03 (0.07)Positive caregiving F 14.04∗∗∗ 24.96∗∗∗ 7.92∗∗ 5.06∗ 2.45 0.81

B (SE) 0.63∗∗∗ (0.17) 1.51∗∗∗ (0.20) 0.27∗∗ (0.10) −0.60∗ (0.26) 0.11 (0.07) −0.21 (0.23)

Step 2: positive caregiving× ethnicityF 0.46 0.76 0.69 1.21 0.07 0.14

Step 3: child’s and caregiver’s ethnic matchF 0.07 0.00 0.36 1.88 0.20 3.26B (SE) 0.38 (1.46) 0.02 (2.65) 0.50 (0.83) 3.10 (2.26) 0.28 (0.62) −3.64 (2.01)

Step 4: interactions with matchEthnicity× match F 0.44 0.88 0.78 0.07 1.40 0.17Positive care× match F 0.27 0.04 0.78 0.11 7.18∗∗ 0.72

Positive care: match B (SE) 0.23∗∗ (0.08)Positive care: not matching B (SE) −0.23 (0.15)

Step 5: ethnicty× positive caregiving× matchF 0.09 0.13 1.14 1.59 1.60 0.65

∗ p < .05.∗∗ p < 0.01.∗∗∗ p < .001.


significantly related to receptive language (F(1, 560) = 14.04,p = .0002), school readiness(F(1, 563) = 24.96,p < .0001), the care provider’s report of pro-social skills (F(1, 523) =7.92, p = .005), and the care provider’s report of behavior problems (F(1, 516) = 5.06,p = .02), but not to mother’s report of pro-social skills (F(1, 564) = 2.45, p = .12) or themother’s report of behavior problems (F(1, 563) = 0.81,p = .37). A one standard deviationincrease in quality (SD= 3.23) was associated with an increase of 2.03 points on the ReynellReceptive Language standard score, 4.85 percentile points on the Bracken School Readinessscale, and 0.87 points on the care providers’s ASBI pro-social scale (ranging from 1 to 5), anda decrease of 1.94 points on the care provider’s CBCL behavior scale total score. No evidenceemerged that positive caregiving related to children’s outcomes differently depending on thechild’s ethnicity.

The ethnic match between the child and the caregiver was not related to children’s outcomes,but did interact with positive caregiving for one outcome. Positive caregiving was significantlyrelated to the mother’s rating of pro-social skills when there was an ethnic match betweenthe child and caregiver (B = 0.23), but negatively related when there was not (B = −0.23).None of the other two or three-way interactions was significant. Ethnic match was negativelycorrelated to school readiness scores for African-American, but inclusion of family and childcare characteristics in the regression analysis resulted in non-significant main effects and in-teractions.

Finally, we used these data to test directly whether child outcomes were affected by dis-crepancies in beliefs about child-rearing between the mother and care provider. The final table,Table 8, lists the results of hierarchical regressions in which maternal education, gender, eth-nicity, maternal traditional childrearing attitudes, and amount, type, and quality of child carewere included as covariates, and the difference between the mother’s and caregiver’s atti-tudes was the predictor of interest. Follow-up analyses, described as Step 2 inTable 8, testedwhether there were ethnic differences in the association between child outcomes and maternalattitudes, positive caregiving, or the discrepancy between the mother’s and care provider’sattitudes. As shown inTable 8, the extent to which mothers endorsed traditional attitudesabout child-rearing was significantly related to the cognitive outcomes and to her ratings of thechild’s behavior, but the discrepancy between her attitudes and the child care provider’s atti-tudes was not reliably related to any outcome. Furthermore, no evidence of differences amongthe white, African-American, and Hispanic children in the association between child outcomesand positive caregiving, maternal attitudes, or mother–caregiver discrepancies in attitudes.

2.4. Discussion

These secondary data analyses provide further evidence that standard measures of child carequality provide reliable and valid assessments for children of varying ethnic background. Whilewe related quality to standardized measures of child outcome that were developed primarilyfor white, middle-class children, these measures are linked to school success for all children(Pianta & Cox, 1999). The evidence supporting the use of these measures for all childrenis strengthened because the findings obtained in separate analysis of two large studies, eachof which included moderate to large numbers of children of color and standard measures ofquality. No evidence indicated that measures of sensitive and stimulating caregiving in child

420M

.R.B

urch

ina

l,D.C

ryer/E

arly

Ch

ildh

oo

dR

ese

arch

Qu

arte

rly1

8(2

00

3)

40

1–

42

6

Table 8NICHD SECC regression results: Predicting child outcomes from child care experiences and discrepancies in beliefs about child-rearing between mother andcaregiver




ASBIpro-social


ASBIpro-social


Step 1: background variables and child care experienceF 18.07∗∗∗ 13.60∗∗∗ 3.28∗∗∗ 2.42∗∗∗ 5.15∗∗∗ 2.86∗∗∗

Maternal education B (SE) 1.47∗∗∗ (0.25) 2.67∗∗∗ (0.47) 0.18 (0.14) −0.73 (0.39) 0.20 (0.11) −1.17∗∗∗ (0.35)Mother’s attitudes B (SE) −0.19∗∗∗ (0.05) −0.25∗ (0.09) −0.04 (0.03) 0.06 (0.08) −0.05∗ (0.02) 0.20∗∗ (0.07)Center care B (SE) 4.44∗∗∗ (1.16) 8.05∗∗∗ (2.13) −1.70∗∗ (0.66) −2.09 (1.76) −0.83 (0.49) 2.31 (1.58)Total hours B (SE) 0.07 (0.05) 0.07 (0.09) 0.00 (0.03) 0.05 (0.08) 0.01 (0.02) 0.06 (0.07)Hours in setting B (SE) −0.03 (0.05) 0.03 (0.09) 0.04 (0.03) 0.06 (0.08) 0.01 (0.02) 0.01 (0.07)Positive caregiving B (SE) 0.66∗∗∗ (0.17) 1.43∗∗∗ (0.32) 0.22∗ (0.10) −0.52 (0.27) 0.10 (0.07) −0.07 (0.24)Mother’s− caregiver’s attitudes B (SE) 0.01 (0.03) 0.01 (0.06) 0.01 (0.02) −0.04 (0.05) −0.02 (0.01) −0.08 (0.05)

Step 2: interactions with ethnicityPositive caregiving F 0.42 0.94 0.49 1.34 0.09 0.59Mother’s attitudes F 0.60 0.45 0.00 0.10 1.35 1.86Mother’s− caregiver’s attitudes F 0.71 0.04 2.28 2.53+ 0.24 2.06

∗ p < .05.∗∗ p < .01.∗∗∗ p < 0.001.


care were appropriate, at least as predictors of outcomes linked to school success, only forwhite children as some have worried.

Standard measures of child care quality showed comparable reliability and validity for white,African-American, or English-speaking Hispanic children in the data from the Cost, Quality,and Outcomes Study and the NICHD Study of Early Child Care. All of the child care qualitymeasures showed very good internal consistency for all three groups of children in both studies.Positive and stimulating child care was consistently and positively related to cognitive outcomesacross the two studies and to caregiver ratings of social skills in the NICHD SECC. We did notobserve stronger association between child care quality and child outcomes for white childrenthan for other children. That is, because quality× ethnicity interactions were non-significantin all analyses and because the within-group associations between quality and outcomes weretypically as strong or stronger for African-American and Hispanic children as for white chil-dren, we concluded that these quality measures appear to provide reliable and valid measuresof the quality of their child care environments regardless of the child’s ethnicity. In addition, thevalidity of the child care quality measures appeared to be independent of whether the caregiverand the child had the same or different ethnic backgrounds. The interaction between child carequality and child–teacher ethnic match was not significant in any of the analyses.

Furthermore, we saw little evidence that children’s outcome were impaired if they experi-enced discrepancies between home and child care in terms of the ethnicity of the child andprimary care provider. Cognitive and social outcomes did not differ reliably when the child andcaregiver had the different ethnic background, regardless of the child’s ethnicity. The one ex-ception occurred in analysis of pro-social skills in the NICHD SECC. Positive caregiving wasrelated to one child outcome (mother ratings of pro-social skills) more strongly when the childwas with a caregiver of the same ethnic background, regardless of the quality of the classroom.It is difficult to interpret this finding without observing similar findings in analyses of simi-lar outcomes, but might suggest that children’s social skills are enhanced when families andcaregivers share a common cultural background. Alternatively, it might suggest that mothersassume that their children are being better socialized when the teacher is of the same ethnicity.

Finally, discrepancies in child-rearing beliefs between home and child care were not signif-icantly related to children’s cognitive or social development at 3 years of age in the NICHDSECC. Children whose mothers expressed more authoritarian attitudes about child-rearingscored lower on language and school readiness test and were rated by the mother as hav-ing fewer pro-social skills and more behavior problems. However, whether the caregiver hadsimilar or different beliefs about child-rearing as the mother was not related to children’s devel-opment. In contrast, the degree to which the caregiver provided sensitive and stimulating carewas related to the cognitive outcomes and to one social outcome. These findings are consistentwith other studies that suggest that positive caregiving is beneficial, even if it is inconsistentwith family caregiving styles (Shpancer, 2002). None of these findings varied among childrenfrom the three ethnic backgrounds.

Use of both studies in these analyses strengthens our conclusions for four reasons. First, thetwo studies focused on children in approximately the same age range, 3–4 years of age who hadat least one or more years of child care prior to entering kindergarten, but who were recruited invery different ways. The CQO study recruited child care centers and then asked children usingthose centers to participate. The SECC study recruited children, and asked caregivers of those


children to participate. Second, child care quality was measured at the center classroom levelin the CQO and at the child level in a wide variety of settings in the SECC. Third, hierarchicalanalyses adjusted for important family selection factors in both studies, and, accounted forcorrelated responses among children in the same setting using HLM. Fourth, the finding fromone study was replicated in the other study. The consistency of findings across the settings inthe SECC study and between studies also increases confidence that these child care qualitymeasures are psychometrically sound and reflect aspects of care than matter for all children.

There are several limitations. First, both studies restricted recruitment to families in whichEnglish was spoken. This restriction, the relatively small Hispanic samples, and lack of docu-mentation of the country-of-origin for the Hispanic children in the two studies limits our abilityto generalize results for these children. Accordingly, these results cannot be regarded as repre-senting levels of quality of care for Hispanic children. Furthermore, the validity of our claimsregarding whether these measures are appropriate for Hispanic children rest on the assumptionthat child care quality shows the same pattern of association with child outcomes for Hispanicchildren who were excluded from these studies as for children included in this study. We werewilling to make this assumption given that we see little evidence of substantial interactionsbetween child care quality and any variable included in the analyses. Nevertheless, these re-sults technically only demonstrate similar patterns of association with outcomes among thethree ethnic groups as they were represented in our studies. Second, only about 50% of centersapproached in the CQO study and about 50% of families in the SECC consented to participateduring recruitment. Along with non-random selection of sites, we can not assume that thesefindings are statistically representative of the country as a whole or even of the selected sites.Third, we included only family variables collected in both as family selection variables. Thus,these analyses of the SECC data include fewer family characteristics as control variables thanincluded in many prior papers. It is reassuring that similar findings regarding child care qualityobtained in these analyses as previously reported. Finally, ethnicity and income were quiteconfounded in both studies. We were not able to include income as a covariate in the CQOStudy due to extensive missing data, but recognize that our samples are clearly biased. Further,neither study likely included many of the most vulnerable children due to their geographic lo-cations and study designs. We expect that the associations between quality and outcomes wereless affected than mean levels by this omission, but recognize this issue as a major limitation.

These results have implications for policy initiatives to ensure all children enter school readyto learn. The child outcomes examined in this study predict school success (Pianta & Cox,1999). Children who enter school with better language skills and fewer behavior problems aremuch more likely to succeed academically (Alexander & Entwisle, 1988). Almost all parents,regardless of the ethnic background, want their child care providers to prepare their childrento do well in school and rank this goal as a top priority in their selection of child care (Early &Burchinal, 2001). Therefore, results from these analyses and from previous child care studiessuggest that standard measures of quality provide reliable and valid assessments for whitechildren and children of color.

These results do not imply that high quality care will look exactly the same across all childcare programs. As argued by Howes and co-workers (Wishard et al., 2003), quality child careshould reflect the cultural heritage of children (Bredekamp & Copple, 1997). They demon-strated that classrooms rated as high quality on these standard measures can utilize markedly


different practices, and that higher quality programs were more likely to use practices thatmatched the children’s cultural background (Wishard et al., 2003). This is because standardmeasures of child care quality primarily reflect the extent to which adults in the child care pro-vide frequent, stimulating, and sensitive interactions with the children and there is substantialflexibility in how this can occur.

In summary, these secondary analyses provide further evidence that widely used measuresof child care quality are positively associated with cognitive and social skills linked to schoolsuccess for white, African-American, and English-speaking Hispanic children. Widely usedmeasures of quality of center child care showed both high levels of reliability and moderatecriterion validity for white, African-American, and Latino children. This finding is not new orsurprising (cf.Lamb, 1998), but there is frequent speculation that what constitutes high qualitychild care varies depending on the child’s ethnicity and family background. Almost no evidenceemerged in these secondary analyses of the largest American child care projects to support sucha belief if a primary goal of child care is to promote academic success for all children. To theextent that ethnically diverse parents value promoting academic success, we should ensure thatthose children are given the benefit of high quality early childhood programs. These programsdo not have to employ the same or even similar practices, but they should provide the basicrequirements for responsive caregiving and developmentally appropriate stimulation known tosupport school success for all children in today’s society.

Acknowledgments

The work reported herein was supported under the Educational Research and DevelopmentCenter Program, PR/Award Number R307A60004, as administered by the Office of Educa-tional Research and Improvement, U.S. Department of Education. However, the contents donot necessarily represent the positions or policies of the National Institute on Early ChildhoodDevelopment and Education, the Office of Educational Research and Improvement, or theU.S. Department of Education, and endorsement by the Federal government should not beassumed. The Cost, Quality, Outcome study was funded by grants from the Carnegie Corpora-tion of New York, the William T. Grant Foundation, the JFM Foundation, the A. L. MailmanFamily Foundation, the David and Lucille Packard Foundation, the Pew Charitable Trusts,the USWEST Foundation, and one anonymous foundation, and was conducted by a team ofresearchers including Donna Bryant, Peg Burchinal, Richard Clifford, Debby Cryer, MaryCulkin, Suzanne Helburn, Carollee Howes, Sharon Lynn Kagan, H. Naci Mocan, John Mor-ris, Leslie Phillipsen, Ellen Peisner-Feinberg, and Jean Rustici. The NICHD Study of EarlyChild Care is directed by a Steering Committee and supported by NICHD through a cooper-ative agreement (U10), which calls for scientific collaboration between the grantees and theNICHD staff. The participating investigators are listed in alphabetical order: Mark Appelbaum,Jay Belsky, Cathryn L. Booth, Robert Bradley, Celia Brownell, Margaret Burchinal, BettyeCaldwell, Susan Campbell, Alison Clarke-Stewart, Martha Cox, Sarah L. Friedman, KathrynHirsh-Pasek, Aletha Huston, Bonnie Knoke, Nancy Marshall, Kathleen McCartney, MarionO’Brien, Margaret Tresch-Owen, Deborah Phillips, Robert Pianta, Susan J. Spieker, DeborahLowe Vandell, and Marsha Weinraub.


I am grateful to Alison Gunn for her help in preparing the manuscript, and especially to thechildren, their families, and child care staff for their participation in these projects. Requestsfor reprints should be sent to Peg Burchinal, Frank Porter Graham Child Development Center,216 Sheryl Mar, CB #8185, UNC-CH, Chapel Hill, NC 27599-8185.

References

Abbott-Shim, M., & Sibley, A. (1992).Assessment profile for early childhood programs. Atlanta, GA: QualityAssist., Inc.

Achenbach, T. M., Edelbrock, C., & Howell, C. T. (1987). Empirically based assessment of behavioral/emotionalproblems of 2- and 3-year-old children.Journal of Abnormal Child Psychology, 15, 629–650.

Alexander, K. L., & Entwisle, D. R. (1988). Achievement in the first two years of school: Patterns and processes.Monographs of the Society for Research in Child Development, 53(2, Serial No. 218).

Allen, B. A., & Boykin, A. W. (1992). African-American children and the educational process: Alleviating culturaldiscontinuity through prescriptive pedagogy.School Psychology Review, 21(4), 586–596.

Arnett, J. (1989). Caregivers in day-care centers: Does training matter?Journal of Applied Developmental Psychol-ogy, 10, 541–552.

Boykin, A.W., & Bailey, C.T. (1999).Experimental research on the role of cultural factors in school relevantcognitive functioning. CRESPAR Technical Report, Baltimore: Johns Hopkins University.

Bracken, B. A. (1984).Bracken basic concept scales. San Antonio, TX: Psychological Corporation.Bradley, R. H., Caldwell, B. M., Rock, S. L., Ramey, C. T., Barnard, K. E., Gray, C., Hammond, M. A., Mitchell, S.,

Gottfried, A. W., Siegel, L., & Johnson, D. L. (1989). Home environment and cognitive development in the firstthree years: A collaborative study involving six sites and three ethnic groups in North American.DevelopmentalPsychology, 25, 217–235.

Bradley, R. H., Corwyn, R. F., Burchinal, M., McAdoo, H. P., & Garcia Coll, C. (2001). The home environmentsof children in the United States: Part 2, Relations with behavioral development from birth through age 13.ChildDevelopment, 72, 1868–1886.

Bredekamp, S., & Copple, C. (1997).Developmentally appropriate practices in early childhood programs: Revisededition. Washington, DC: National Association for the Education of Young Children.

Bronfenbrenner, U., & Morris, P. A. (1998). The ecology of developmental processes. In W. Damon, & R. M. Lerner(Eds.),Handbook of child psychology: Vol. 1. Theoretical models of human development(5th ed., pp. 993–1028).New York: Wiley & Sons, Inc.

Bryant, D. M., Burchinal, M. R., Lau, L. B., & Sparling, J. J. (1994). Family and classroom correlates of Head Startchildren’s developmental outcomes.Early Childhood Research Quarterly, 9, 289–309.

Bryant, D. M., Peisner-Feinberg, E. S., & Clifford, R. M. (1993).Evaluation of public preschool programs in NorthCarolina: Final report. Chapel Hill, NC: Frank Porter Graham Child Development Center.

Burchinal, M. R., Peisner-Feinberg, E., Bryant, D. M., & Clifford, R. (2000). Children’s social and cognitivedevelopment and child care quality: Testing for differential associations related to poverty, gender, or ethnicity.Journal of Applied Developmental Sciences, 4, 149–165.

Burchinal, M. R., Ramey, S. L., Reid, M. K., & Jaccard, J. (1995). Early child care experiences and their associationwith family and child characteristics during middle childhood.Early Childhood Research Quarterly, 10, 33–61.

Burchinal, M. R., Roberts, J. E., Nabors, L. A., & Bryant, D. (1996). Quality of center child care and infant cognitiveand language development.Child Development, 67, 606–620.

Burchinal, M. R., Roberts, J. E., Riggins, R., Zeisel, S., Neebe, E., & Bryant, M. (2000). Relating quality of centerchild care to early cognitive and language development longitudinally.Child Development, 71, 339–357.

Campbell, F., & Ramey, C. T. (1994). Effects of early intervention on intellectual and academic achievement: Afollow-up study of children from low-income families.Child Development, 65, 684–698.

Caughy, M. O., DiPietro, J. A., & Strobino, D. M. (1994). Day-care participation as a protective factor in thecognitive development of low-income children.Child Development, 65, 457–471.


Clarke-Stewart, K. A., Vandell, D. L., Burchinal, M., O’Brien, M., & McCartney, K. (2002). Do regulable featuresof child-care homes affect children’s development?Early Childhood Research Quarterly, 17, 52–86.

Dunn, L. (1993). Proximal and distal features of day care quality and children’s development.Early ChildhoodResearch Quarterly, 8, 167–192.

Dunn, L. M., & Dunn, L. M. (1981).Peabody Picture Vocabulary Test—Revised. Circle Pines, MN: AmericanGuidance Service.

Early, D., & Burchinal, M. R. (2001). Early childhood care: Relations with family characteristics and preferred carecharacteristics.Early Childhood Research Quarterly, 16, 475–497.

Erwin, P., Sanson, A., Amos, D., & Bradley, B. (1993). Family day care and day care centers: Careers, family andchild differences and their implications.Early Childhood Research Quarterly, 86, 89–103.

Garcia-Coll, C. T. (1990). Developmental outcome of minority infants: A process-oriented look into our beginnings.Child Development, 61, 270–289.

Garcia Coll, C., Crnic, K., Lamberty, G., Wasik, B. H., Jenkins, R., Garcia, H. V., & McAdoo, H. P. (1996). Anintegrative model for the study of developmental competencies in minority children.Child Development, 67,1891–1914.

Harms, T., & Clifford, R. (1980).Early Childhood Environment Rating Scale. New York: Teachers College.Harms, T., & Clifford, R. M. (1989).Family Day Care Rating Scale. NY: Teachers College Press.Hogan, A. E., Scott, K. G., & Bauer, C. R. (1992). The Adaptive Social Behavior Inventory (ASBI): A new

assessment of social competence in high-risk three-year-olds.Journal of Psychoeducational Assessment, 10,230–239.

Howes, C. (2000). Social development, family, and attachment relationships of infants and toddlers: Research topractice. In D. Cryer & T. Harms (Eds.),Infants and toddlers in out-of-home care(pp. 87–116). Baltimore:Brookes Publishing.

Howes, C., Phillips, D. A., & Whitebook, M. (1992). Thresholds of quality: Implications for the social developmentof children in center-based child care.Child Development, 63, 449–460.

Howes, C., & Stewart, P. (1987). Child’s play with adults, toys, and peers: An examination of family and child careinfluences.Developmental Psychology, 23, 423–430.

Katz, J. R., & Snow, C. E. (2000). Language development in early childhood: The role of social interaction. In D.Cryer & T. Harms (Eds.),Infants and toddlers in out-of-home care(pp. 49–86). Baltimore: Brookes Publishing.

Kontos, S. J. (1991). Child care quality, family background, and children’s development.Early Childhood ResearchQuarterly, 6, 249–262.

Kontos, S., Howes, C., Shinn M., & Galinsky, E. (1995).Quality in family child care and relative care. New York:Teachers College Press.

Lamb, M. E. (1998). Nonparental child care: Context, quality, correlates, and consequences. In I. E. Sigel & K. A.Renniger (Vol. Eds.),Child psychology in practice, W. Damon (Ed.),Handbook of child psychology(5th ed.,pp. 73–134). New York: Wiley.

Lazar, I., & Darlington, R. (1982). Lasting effects of early education: A report from the Consortium for LongitudinalStudies.Monographs of the Society for Research in Child Development, 47(2–3, Serial No. 195).

McCartney, K. (1984). The effect of quality of day care environment upon children’s language development.Developmental Psychology, 20, 244–260.

Moss, P., & Pence, A. (1994).Valuing quality in early childhood services: New approaches to defining quality. NewYork: Teachers College Press.

New, R. (1994). Culture, child development, and developmentally appropriate practices: Teachers as researchers.In B. L. Mallory & R. S. New (Eds.),Diversity and developmentally appropriate practices(pp. 65–83). NewYork: Teachers College Press.

NICHD Early Child Care Research Network. (1993).The NICHD Study of Early Child Care: A comprehensivelongitudinal study of young children’s lives(ERIC Document Reproduction Service No. ED 3530870).

NICHD Early Child Care Research Network. (1996). Characteristics of infant child care: Factors contributing topositive caregiving.Early Childhood Research Quarterly, 11, 269–306.

NICHD Early Child Care Research Network. (1997). Child care in the first year of life.Merrill-Palmer Quarterly,43, 340–360.


NICHD Early Child Care Research Network. (1998). Early child care and self-control, compliance and problembehavior at twenty-four and thirty-six months.Child Development, 69, 1145–1170.

NICHD Early Child Care Research Network. (2000). The relation of child care to cognitive and language develop-ment.Child Development, 71, 958–978.

Osborne, S. S., Schulte, A. C., & McKinney, J. D. (1991). A longitudinal study of students with learning disabilitiesin mainstream and resource programs.Exceptionality, 2, 81–96.

Peisner-Feinberg, E. S., & Burchinal, M. R. (1997). Relations between preschool children, child care experiences,and concurrent development: The Cost, Quality, and Outcomes Study.Merrill-Palmer Quarterly, 43, 451–477.

Pianta, R. C., & Cox, M. J. (1999).The transition to kindergarten. Baltimore: Paul H. Brookes.Provost, M. (1994). Day care and social competence in preschoolers: Continuity and discontinuity among family,

teachers, and peer contributions. In A. Vyt, H. Block, & M. H. Bornstein (Eds.),Early child development in theFrench tradition(pp. 287–299). Hillsdale, NJ: Erlbaum.

Reynell, J. (1991).Reynell Developmental Language Scales(U.S. edition). Los Angeles, CA: Western PsychologicalService.

Sameroff, A. J. (1983). Developmental systems: Contexts and evolution. In P. H. Mussen (Series Ed.) & W. Kessen(Vol. Ed.), Handbook of child development: Vol. 1. History, theories, and methods(pp. 237–294). New York:Wiley.

Schaefer, E. S., Edgerton, M., & Aaronson, M. (1978).Classroom behavior inventory. Unpublished rating scale.Schliecker, E., White, D. R., & Jacobs, E. (1991). The role of day care quality in the prediction of children’s

vocabulary.Canadian Journal of Behavioral Science, 23, 12–24.Schweinhart, L. J., Barnes, H. V., & Weikart, D. P. (1993).Significant benefits: The High/Scope Perry Preschool

Study through age 27. Ypsilanti, MI: High/Scope Press.Shpancer, N. (2002). The home-daycare link: Mapping children’s new world order.Early Childhood Research

Quarterly, 17, 374–392.Stipek, D. (1993).Attitudes/perceptions of competence. Unpublished rating scale.Stipek, D., Feiler, R., Daniels, D., & Milburn, S. (1995). Effects of different instructional approaches on young

children’s achievement and motivation.Child Development, 66, 209–223.Tomasello, M., & Farrar, J. (1986). Joint attention and early language.Child Development, 57, 1454–1463.Vandell, D. L., & Corasaniti, M. A. (1990). Child care and the family: Complex contributors to child development.

In K. McCartney (Ed.),New directions in child development. San Francisco: Jossey-Bass.Vandell, D. L., & Wolfe, B. (2000).Child care quality: Does it matter and does it need to be improved?Institute

for Research on Poverty, Special Report 78 [On-line]. Available at:http://www.ssc.wisc.edu/irp/.Wishard, A. G., Shivers, E. M., Howes, C., & Ritchie, S. (2003). Child care program and teacher practices: associ-

ations with quality and children’s experiences.Early Childhood Research Quarterly, 18, 65–103.Woodcock, R. W., & Johnson, M. B. (1990).Woodcock–Johnson Psycho-Educational Battery—Revised. Allen, TX:

DLM Teaching Resources.Zill, N. (1999). Enhancing school readiness for low-income children: A national study of Head Start quality and

outcomes. In M. Lopez & L. Turillo (Chairs.),The fragile crucible: Developmental and policy dilemmas fordefining and assessing the school readiness of preschool children. Symposium presented at the biennial meetingof Society for Research in Child Development, Albuquerque, NM, April 17, 1999.

http://www.ssc.wisc.edu/irp/