Associations between continuity of care in infant-toddler classrooms and child outcomes

Early Childhood Research Quarterly 42 (2018) 105–118

Contents lists available at ScienceDirect

Early Childhood Research Quarterly journal homepage: www.elsevier.com/locate/ecresq

Associations between continuity of care in infant-toddler classrooms and child outcomes

MARK

⁎

Diane M. Horma, , Nancy Fileb, Donna Bryantc, Margaret Burchinalc, Helen Raikesd, Nina Forestieric, Amy Encingerd, Alan Cobo-Lewise a

Early Childhood Education Institute, University of Oklahoma-Tulsa, United States University of Wisconsin-Milwaukee, United States c Frank Porter Graham Child Development Institute, University of North Carolina-Chapel Hill, United States d University of Nebraska-Lincoln, United States e University of Maine, United States b

A R T I C L E I N F O

A B S T R A C T

Keywords: Continuity of care Infant-toddler center-based care Child outcomes

Ensuring that young children, especially infants and toddlers, experience consistency in child care providers over time is a practice endorsed by multiple professional organizations. This practice, commonly referred to as continuity of care (CoC), is recommended for center-based group settings to provide infants and toddlers with the sensitive, responsive care needed to promote early development. Despite widespread endorsement, there has been limited empirical examination of CoC. This study examines the extent to which CoC experienced in infanttoddler center-based care is associated with social-emotional and language development. Associations of CoC with children’s social-emotional development during the infant-toddler period and with later social-emotional and language outcomes at age 3 were investigated in a large sample of children attending high-quality early childhood programs designed for young children growing up in poverty. During the infant-toddler years, CoC was related to higher teacher ratings of self-control, initiative, and attachment, and lower ratings of behavior concerns. In addition, a classroom quality × CoC interaction indicated that CoC differences were larger in higher, than lower, quality infant-toddler classrooms. In contrast, CoC in infant-toddler classrooms was not related to rates of change in teacher ratings of social skills during the infant-toddler years nor to children’s vocabulary development or ratings of social skills after they transitioned to preschool. Neither were there quality × CoC interactions at preschool. These findings do not provide clear support for the current widespread recommendations for CoC, but suggest a need for additional research. The need for future research to more fully understand associations with child outcomes as well as to examine potential impacts of CoC on teachers, families, and peers is highlighted.

1. Introduction Infant-toddler care is the fastest growing and most sought-after form of child care in the U.S., according to the National Association of Child Care Resource and Referral Agencies (2008). Currently, approximately 50% of U.S. children birth to age 3 experience a regular child care arrangement, with the percentage enrolled in center-based care increasing with age from 9% of children under 12 months to 20% of 2- to 3-year-olds (NSECE Project Team, 2015). Based on developmental and attachment theories (e.g., Ainsworth, Blehar, Waters, & Wall, 1978; Sroufe, 1988; Thompson, 2000), multiple professional organizations have endorsed the provision of continuity in nonparental caregivers, commonly referred to as continuity of care (CoC), as a recommended

⁎

Corresponding author. E-mail address: [email protected] (D.M. Horm).

http://dx.doi.org/10.1016/j.ecresq.2017.08.002 Received 22 August 2016; Received in revised form 28 July 2017; Accepted 8 August 2017 0885-2006/ © 2017 Elsevier Inc. All rights reserved.

practice with the intent to provide infants and toddlers with the sensitive, responsive care needed to promote early development (Sosinsky et al., 2016). Despite this widespread endorsement, there has been limited empirical examination of CoC (Sosinsky et al., 2016). For this reason, this study uses an existing dataset to examine the extent to which CoC experienced in infant-toddler group settings is associated with social-emotional and language outcomes in a network of programs that all endorsed the goal of providing CoC for young children. Specifically, the research focused on whether low-income children experiencing CoC during their infant-toddler center-based care show more advanced social-emotional development in the short- and longer-term than children without CoC, with an added examination of receptive vocabulary at age 3. Given the research relating early care to child

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

child to a new classroom when a developmental milestone was attained resulted in transitions occurring as often as every 6 months for infants (Cryer et al., 2005).

outcomes in middle childhood and adolescence (e.g., NICHD ECCRN, 2005, 2006; Vandell et al., 2010), the growing number of infants and toddlers experiencing center-based care, and the emphasis on CoC in professional recommendations, careful examination of the relative and interactive contributions of CoC and quality in a large sample is crucial.

1.1.2. Attachment theory and research Attachment theory (Ainsworth et al., 1978) posits that infants’ early development depends on frequent, warm, consistent interactions with their primary caregivers. Those interactions promote close, secure, and nurturing relationships between the caregiver and infant that serve as the foundation for children’s subsequent language and socio-emotional development (Sroufe, 1988). Applied to infant-toddler care, high rates of teacher turnover and frequent changes in caregivers were thought to place the early development of infants and toddlers in jeopardy (Thompson, 2000). Research indicated that infants formed attachment relationships with their child care providers as well as their parents (Howes & Hamilton, 1992; Van IJzendoorn, Sagi, & Lambermon, 1992). Raikes (1993) reported that a higher percentage of children had secure attachments with teachers when they had been with the same teacher for more than one year. Howes and Spieker (2008) noted that extended interactions with a child may increase a caregiver’s emotional investment in the child and promote responsive caregiving. Additionally, more time and experience with a child may help a caregiver read and appropriately respond to a child’s unique cues and needs. Ahnert, Pinquart, and Lamb (2006) reported that children were more likely to receive sensitive and responsive care and to develop secure relationships with their caregivers when the dyads were together for a longer period of time. This responsive and individualized care, in turn, has been found to support children’s development in a variety of areas. For example, higher levels of language development and greater levels of peer play have been associated with sensitive and responsive caregiving experienced in group care (NICHD ECCRN, 2005).

1.1. Continuity of care CoC is an intentional grouping and staffing pattern implemented in infant-toddler center-based care to enhance consistency in caregivers (Sosinsky et al., 2016). CoC for infants and toddlers occurs when the child remains with the same caregiver across multiple years (Cryer, Hurwitz, & Wolery, 2001), preferably until 36 months of age (Program for Infant/Toddler Care, n.d.). It can occur when a caregiver is assigned to a group of similarly-aged infants, and that group remains together until the children all move to preschool at the same time (McMullen, Yun, Mihai, & Kim, 2016; Sosinsky et al., 2016). This arrangement is referred to as looping. It can also occur in mixed-age infant-toddler classrooms when infants enter at various times and stay with the teacher until they reach the age to move to the preschool classrooms (Sosinsky et al., 2016; Theilheimer, 2006), and are replaced within their former classroom with young infants. As noted by McMullen et al. (2016), furnishings, equipment, materials, supplies, room location, and peers may change in CoC, but the central characteristic is that teachers/ caregivers, and thus the relationship between an individual child and teacher, remain intact for a given child over an extended period of time. The premise, regardless of strategy, is that if the child and his/her caregiver stay together for an extended period of time, the caregiver’s knowledge and understanding about that child and his/her family will deepen, resulting in trust and security between and among all within the group – caregiver, child, and family members (Essa, Favre, Thweatt, & Waugh, 1999; Sosinsky et al., 2016; Theilheimer, 2006). Based on attachment theory, the infant should be able to develop a secure attachment with this caregiver if that person is responsive and sensitive, and that attachment serves as a secure basis for the development of subsequent social and cognitive skills (Sroufe, 1988). CoC became more prevalent in the U.S. in the 1990s during a period when a number of factors converged: rates of maternal employment for mothers of infants and toddlers increased; debate about the potential effects of maternal employment and the associated increase in use of out-of-home care for infants and toddlers reached a high point; attachment principles and results of studies of secondary attachment became better known; and infant-toddler child care was of notably poor quality. Findings related to infant-toddler center-based child care and associations with attachment and caregiver consistency provide the rationale and justification for CoC (Sosinsky et al., 2016) and thus are discussed in more detail below.

1.1.3. Practices in infant-toddler out-of-home settings Over the past several decades some infant-toddler group care in the U.S. implemented practices consistent with attachment theory and research, seeking to provide infants and toddlers with caregivers/teachers who offer children a secure base and a secondary attachment during their time away from primary caregivers. For example, the Program for Infant/Toddler Care (PITC, Lally & Mangione, 2009), a widely-used infant-toddler approach (Horm, Goble, & Branscomb, 2011), provides a philosophy and strategies for promoting responsive, relationship-based infant-toddler center-based care. PITC and similar programs advocate practices consistent with attachment research by keeping children and teachers together throughout the infant-toddler period until age 3 (Program for Infant/Toddler Care, n.d.). In fact, CoC is one of PITC’s anchor policies (Mangione, 2006). Several professional organizations and groups including Zero to Three, the National Association for the Education of Young Children (NAEYC), the National Head Start Association (NHSA) and Early Head Start, and the Center for Law and Social Policy (CLASP) have also endorsed CoC as a recommended practice with infants and toddlers (Sosinsky et al., 2016). For some, maintaining the same caregiver during the infant-toddler period is viewed as a key indicator of highquality infant care (Copple, Bredekamp, Koraleck, & Charner, 2013; Zero to Three, 2008). However, in early research infant-toddler teachers were reported to be less enthusiastic about the practice; for example, fewer than half of 273 early childhood teachers in one study agreed that infants and toddlers should have the same teacher until age 3 (Cryer et al., 2001). On the other hand, De-Souza (2012) in a dissertation study using in-depth interviews with 21 center directors, found that directors supported the concept but had concerns about whether it could be implemented in their own programs due to staff issues. Thus, philosophical support for CoC may be more wide-spread than the practice, although there is little research documenting the implementation of CoC.

1.1.1. Historical context Waldman (1983) reported that in 1970 participation in the labor force among married mothers whose youngest child was less than a year old was 24% and by 1983 that number had increased to 45%. In March 2012, the U.S. Department of Labor reported that 55% of all mothers with children under 3 years of age were in the labor force (Bureau of Labor Statistics, 2014). Increasing maternal employment has meant that increasing numbers of very young children experience outof-home care. Increased use of out-of-home care attracted research attention (e.g., NICHD ECCRN, 1996, 2005), and questions were raised about potential harmful effects of nonmaternal care, particularly in regard to children’s early attachments to primary caregivers (Belsky & Eggebeen, 1991; Han, Waldfogel, & Brooks-Gunn, 2001). Concerns grew with studies documenting that over 40% of infant-toddler center-based care was of poor quality (Cost Quality and Child Outcomes Study Team, 1995); that the average annual teacher turnover rate in early childhood programs was 30% (Whitebook & Sakai, 2003); and that common infant-toddler classroom practices such as moving a 106

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

NICHD ECCRN, 2006; NICHD ECCRN & Duncan, 2003). Given the existing extensive quality literature, it is important to investigate the potential moderating effect of overall classroom quality when examining CoC and relations with child outcomes. In previous studies, researchers recognized the importance of classroom quality and accounted for it in their studies of CoC by attempting to hold quality constant through restricting their sample to NAEYC-accredited centers (Owen et al., 2008) or by investigating QRIS ratings and structural indicators of quality such as child:staff ratios, both of which were higher in CoC versus non-continuity classrooms (Ruprecht et al., 2016). Neither of these studies, however, asked whether higher quality classrooms might intensify the hypothesized positive impacts of CoC or could potentially buffer the hypothesized negative effects of discontinuity.

Despite its plausible theoretical underpinnings and endorsement by professional associations and groups, there is relatively little empirical research on CoC (Owen, Klausli, Mata-Otero, & O’Brien-Caughy, 2008; Ruprecht, Elicker, & Choi, 2016; Sosinsky et al., 2016). As noted above, while research on caregiver attachment and instability strongly implies that CoC should have positive impacts on child outcomes, little research has examined CoC, per se, as an intentional care practice (Sosinsky et al., 2016). The emerging research has focused on different aspects of CoC, for example, the implementation of CoC including barriers and supports (e.g., Ackerman, 2008; Aguillard, Pierce, Benedict, & Burts, 2005; Cryer et al., 2001). Some studies have examined teacher behaviors and found that programs promoting CoC have teachers who are more responsive and engaged with the children (Owen et al., 2008: Raikes, 1993Owen et al., 2008: Raikes, 1993; Ruprecht et al., 2016). Other studies have examined CoC as a means to promote stronger parent-caregiver relationships and greater parental involvement (Owen et al., 2008). Only a few studies have examined CoC and child outcomes, and the results of these few studies are inconclusive. Early studies documented positive relations between stability in caregivers and child outcomes through 30 months of age (Howes & Hamilton, 1992, 1993; Hamilton, 1992, 1993). In more recent research, toddlers who experienced CoC in their classrooms were rated as having fewer behavior problems, but not more social competence (Ruprecht et al., 2016). Preschool children in CoC classrooms were more engaged with their caregivers than similar children attending comparison centers, with reported effects greater for African-American than Latino children, in a study of preschool children attending programs serving low-income families (Owen et al., 2008). However, this study reported no consistent benefits of CoC for child receptive vocabulary development, school readiness, or problem behaviors. In summary, to date no studies have examined a variety of child outcomes related to CoC during infancy, and mixed findings define the current research base. The goal of this study is to examine, with a larger sample than used in previous research, CoC and its association with a range of developmental outcomes in infants and toddlers—the age group the background literature on attachment and caregiver stability would predict may be most impacted by CoC. Based on the theory and research in this area, the logic model is that sensitive and responsive care is the mechanism through which teacher-child attachments are developed and undergird positive child outcomes. Thus, CoC effects could be large with sensitive and responsive caregivers due to their creation of the secure base necessary for infants and toddlers to develop social and cognitive skills. However, given the importance of classroom quality to child outcomes, it could be that CoC has a smaller impact when quality is high due to the protective role that classroom quality has been found to play in relation to child outcomes in center-based settings. This study examines these possibilities.

1.3. This study The current study takes advantage of a large database of independently collected child, parent, and classroom quality data to examine child outcomes at semester intervals during the infant-toddler period and at age 3 (after attendance in the infant-toddler program) for children who did or did not experience CoC in infant-toddler classrooms. The data were collected for the Implementation Study of the Educare Learning Network (ELN), an association of 21 schools that describe their programs as enhanced Early Head Start and Head Start programs. In a randomized clinical trial the ELN has documented that children enrolled in Educare classrooms experience moderately high to high-quality early care and education, and had better language and social skills at two years than a control group (Yazejian et al., 2017). An earlier observational study of all children enrolled in Educare through 2013 showed that children who entered earlier and stayed longer had higher language outcomes than those who entered later (Yazejian, Bryant, Freel, Burchinal, & the Educare Learning Network Investigative Team, 2015). To participate in the ELN, all schools committed to deliver high-quality early care and education, with CoC as an expected practice. Yet, data show that schools are differentially successful in achieving CoC in the staffing of their classrooms. Therefore, secondary analysis of the Educare data should provide much-needed empirical evidence regarding whether CoC during infancy is related to social and language development in programs with no pre-existing difference in terms of their commitment to using CoC. Thus, the purpose of this study was to investigate associations between CoC and child outcomes over and above quality. Specifically, the research questions were: 1. Does continuity of care experienced in infant-toddler center-based classrooms relate to: a.) children’s social-emotional development during the infanttoddler period; and b.) children’s social-emotional and language skills at entry to preschool? 2. Is continuity of care more strongly (or weakly) related to early development when infants and toddlers experience higher quality classrooms?

1.2. Quality of infant-toddler center-based care CoC is but one indicator of high-quality infant-toddler care practices. Other classroom indicators, such as regulatable (e.g., class size, child:staff ratio) and global quality (e.g., observational measures of the environment), have been addressed in past and recent empirical research investigating the impact of center-based programs on young children’s development, including infants and toddlers living in poverty (IOM & NRC, 2012; National Scientific Council on the Developing Child, 2004; NICHD ECCRN, 1996; Phillips & Lowenstein, 2011; Vandell et al., 2010). Whether global quality of care moderates associations between CoC and early development has not been examined. Quality of infanttoddler care is thought to be the mechanism through which early care impacts infant-toddler development (IOM & NRC, 2012), with considerable experimental and observational evidence linking quality to infant-toddler development (Burchinal, Roberts, Nabors, & Bryant, 1996; Campbell, Pungello, Miller-Johnson, Burchinal, & Ramey, 2001;

2. Method 2.1. Participants Data from the first 11 Educare schools (Educare uses “school” to highlight the intentional learning opportunities provided in their program model) that joined the ELN were used in this study to ensure that enough years of data were available to assess continuity over a period of at least 4 years. The 11 Educare schools were located in Chicago, Denver, Kansas City, Miami, Milwaukee, Oklahoma City, Omaha (two schools), Seattle, and Tulsa (two schools). These center-based programs 107

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

variables were averaged over time with two exceptions: if a disability plan was ever reported, the child was considered to have a disability; and children who were ever reported to be a DLL were always coded DLL.

serve children from low-income families from 6 weeks to public school entry at age 4 or 5. All families meet the Early Head Start (EHS) and Head Start (HS) eligibility criteria, including living at or below the federal poverty line. Educare schools meet all Head Start performance standards and implement additional features including full-day and year-round programming, bachelor-degreed lead teachers in all classrooms, use of data to inform programming, and CoC within EHS and separately within HS classrooms (see Yazejian, Bryant, & Kennel, 2013 for a complete description). While it is expected that CoC is implemented as a program component and appropriate for every child, schools’ ability to deliver continuity varies somewhat. Teacher turnover has been the typical reason for non-continuity, although relative to other child care, teacher turnover in Educare is much less, at about 15% a year versus a reported high of 27% for for-profit chains in 2012 (Whitebook, Phillips, & Howes, 2014). To be included in the analyses, children had to have been enrolled in Educare for at least 18 months in the EHS years (birth – age 3) in the time period covering school years 2006–07 through 2014–15 and to have transitioned into a Head Start classroom after age 3. Eighteen months was used to define the minimal EHS enrollment period for study eligibility because it exceeds the 12-month age groupings typically seen in childcare settings (e.g., separate 1-year-old and 2-year-old rooms). Eighteen months also is about half the maximum amount of time that children could have attended EHS had they enrolled as very young babies. Data are collected each semester in these year-round programs so it was possible to determine if a child was with the same teacher during each semester; 18 months corresponds to 3 semesters. Our dataset did not allow us to determine which model (looping or mixed-age group) was used to implement CoC, but we were able to determine if a child had a continuous relationship with the same caregiver over time. To be included in the analyses, children also had to have attended the Educare Head Start program into the fall of their first HS year, had a socio-emotional rating by their HS teacher, and an independent language assessment. Of the 3189 children in the dataset who entered these Educare EHS programs during the specified timeframe, 851 children met these selection criteria. Table 1 presents their demographic data. Sixteen percent of these 851 were enrolled for 3 semesters, 39% for 4, 11% for 5, 32% for 6, and 2% for 7 or 8 semesters.

2.2.2. Continuity of care measures Continuity in the EHS years was the independent variable in this study. The identity of adults and children within the EHS (birth-age 3) classrooms was recorded twice per year, once at the beginning of the fall and once near the end of the spring. We operationalized CoC in four different ways: 1. Continuity with Lead Teacher (Y/N): A categorical variable coded “yes” if the same adult was a child’s lead teacher throughout all of their EHS semesters, “no” otherwise. 2. Percentage Continuity with Lead Teacher: The percentage of time, measured by semesters, child was enrolled in EHS with the same lead teacher. 3. Continuity with Any Teaching Staff Member (Y/N): A categorical variable coded “yes” if one consistent adult was present in a child’s classroom throughout all of their EHS semesters, regardless of what role that adult filled (teacher, assistant teacher, teaching aide). 4. Percentage Continuity with Any Teaching Staff Member: The percentage of time, measured by semesters, child was enrolled in EHS with one consistent adult in the classroom, regardless of what role that adult filled. For the third and fourth ways of operationalizing continuity noted above, the calculation was made for the adult who maximized the measure of continuity. Table 1 presents descriptive data on these predictor variables. The four methods of measuring continuity tended to classify children similarly. Overall, 79% of children were classified into the same CoC category regardless of whether we examined continuity with the lead teacher or any teacher (note: 61% of those having continuity with any teacher had continuity with the lead teacher). Little extra information was gained when examining proportion of semesters with continuity as indicated by correlations between classification of continuity and proportion of semesters with lead teacher (r = 0.89) and any teacher (r = 0.92). Accordingly, we focused our analyses on whether the child experienced continuity with the lead teacher in the infant-toddler years, but then examined the other continuity indices in sensitivity analyses.

2.2. Measures 2.2.1. Parent interview Educare family support or local evaluation staff conducted parent interviews in the fall and spring in the 2006–2009 school years and only in the fall from 2009 to 2015. The interview included questions about children’s health, family demographics, and other family characteristics. These measures included whether the child came from a single-parent household, the number of years of education of the child’s primary caregiver, and whether the child’s biological mother was under age 20 when the child was born. The interview also assessed the family’s food insecurity in an index calculated from responses to two questions from the USDA’s annual household food survey: how often the family ran out of food during the past year and how often they were worried about running out of food during that time (Nord, Andrews, & Carlson, 2007). A project-created life events scale consisted of 20 major life events, and parents indicated those that had occurred in their family in the past year. Mothers’ depressive symptoms were assessed with a 3-question modified version of the 6question Rand depression screener (Rand Health, 1994) except that in the first year of the study, it was assessed using the Center for Epidemiologic Studies Depression Scale-Revised (Radloff, 1977). The primary caregiver rated the child’s health status on a 1–5 scale of poor (1), fair, good, very good, or excellent (5) and reported whether the child had a disability plan (IFSP). The parent interview also included questions about languages spoken in the home, the child’s first language, and the child’s strongest language to determine whether the child was a dual language learner (DLL). For children with multiple parent interviews, scores on these

2.2.3. Global classroom quality Global quality was measured in the middle of each school year in each classroom using the Infant-Toddler Environment Rating ScaleRevised (ITERS-R; Harms, Cryer, & Clifford, 2003); following the authors’ guidelines for classrooms with a majority of older 2s and 3s, the Early Childhood Environment Rating Scale – Revised (ECERS-R; Harms, Clifford, & Cryer, 1998) was also used. Data collector reliability on these measures was assessed annually through visits with a gold-standard assessor. Reliability visits were conducted for 20% of the classroom observations over the study period. Item-level inter-rater reliability with the gold standard assessor averaged 89.8% agreement within one point over the study period (site range over the years, 83%–95.9%). The ITERS-R or ECERS-R scores were included as time-varying predictors (see Table 2 for descriptive statistics) in longitudinal analyses of the infant-toddler social outcomes and as an aggregated (mean) score in the cross-sectional analysis of the 3-year-old outcomes. 2.2.4. Child measures In Educare, children’s social-emotional development is assessed each fall and spring by having their teacher complete the Devereux Early Childhood Assessment (DECA; LeBuffe & Naglieri, 1999), a process supervised by the local independent evaluator. For the children included in this study, we used all ratings by infant-toddler EHS 108

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Table 1 Descriptive Statistics for Child and Family Characteristics and Primary Outcomes by Continuity of Care Group. Continuity of Care with Lead Teacher over EHS Years No

Yes

Characteristic

N

%

MN (SD)

N

Educare experience Age entered Educare (yrs.) Number of EHS semesters Years of EHS

437 437 437

Child characteristics Child gender (male = 1) Child race: White/nonHispanic African-American Hispanic Other Child is dual language learner Child has IFSP or IEP Child heath (1–5 rating)

437 437 437 437 437 437 437 434

55 7 48 37 8 32 17

Parent characteristics Single parent Parent education (yrs.) Teenage mother Parent depressive symptoms Family food insecurity Family total life events

433 435 430 434 434 434

56

0.95 (0.58) 4.77 (1.14) 2.38 (0.57)

414 414 414 51 5 50 39 6 30 16

4.31 (0.69)

414 414 414 414 414 414 414 411

53

2.88 (1.78)

412 412 412 411 412 412

Continuity of care indices % of semesters–w lead teacher % of semesters–w any teacher Number of lead teachers Number of teachers, any type

437 437 437 437

59 79

100 100

2.32 (0.71) 5.39 (1.76)

414 414 414 414

Classroom quality in EHS First ERS observation score Mean of each child’s multiple ERS score

320 385

5.51 (.80) 5.77 (0.52)

290 411

5.71 (.73) 5.70 (0.50)

Preschool outcomes (1st sem. HS) PPVT standard score DECA Initiative t score DECA Self-control t score DECA Attachment t score DECA Behavioral concerns t score

340 367 367 367 357

92.14(13.74) 47.75 (9.01) 47.89 (9.32) 48.44 (8.99) 52.47 (9.20)

318 360 360 360 355

93.00 47.42 48.95 49.18 52.65

12.26 (2.05) 17 13 42

%

MN (SD)

t

1.11 (0.60) 4.51 (1.13) 2.25 (0.66)

p < 0.001 p < 0.001

4.33 (0.70)

12.23 (2.06) 13 16 38 2.66 (1.67)

1.05 (0.24) 3.53 (1.04) p < 0.01

(14.58) (9.33) (9.33) (9.14) (8.96)

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R).

Table 2 Longitudinal Assessments of Infant-Toddler Social-Emotional Development by Continuity of Care Group. CoC (Yes/No)a

Child’s Semester in Early Head Start 1st

DECA Scale t scores Initiative Attach-ment Self- control Behavior concerns Time-varying Covariates Age in months Classroom quality–ERS

2nd

3rd

4th

5th

N

M

(SD)

N

M

(SD)

N

M

(SD)

N

M

(SD)

N

M

(SD)

N Y N Y N Y N Y

434 408 434 408 150 185 17 13

49.76 51.56 49.06 49.93 48.08 50.05 51.65 52.92

(9.43) (9.08) (9.91) (9.54) (10.1) (9.62) (6.82) (9.81)

422 402 422 402 243 278 62 61

51.09 52.60 49.79 51.89 48.10 51.01 54.74 51.90

(9.46) (8.25) (9.68) (8.98) (9.15) (8.97) (7.41) (8.02)

382 358 382 358 313 319 106 118

51.84 53.39 51.18 53.42 49.36 50.92 54.91 52.90

(8.83) (9.34) (9.44) (9.44) (9.57) (9.26) (8.06) (7.88)

330 310 330 310 327 309 158 174

52.94 53.95 53.02 54.16 49.69 50.77 54.53 53.64

(9.69) (9.22) (9.76) (9.04) (9.90) (9.15) (8.33) (7.54)

179 129 179 129 178 129 98 65

53.59 53.98 53.69 54.44 49.28 51.77 54.95 53.05

(9.90) (8.78) (9.71) (9.12) (9.07) (7.89) (7.56) (7.27)

N Y N Y

434 408 322 296

14.63 16.38 5.51 5.71

(7.81) (7.68) (0.81) (0.73)

422 402 320 286

20.26 21.93 5.51 5.75

(7.93) (7.37) (0.79) (0.70)

382 358 344 301

24.74 26.30 5.87 5.98

(6.87) (6.64) (0.58) (0.51)

330 310 294 261

29.70 31.61 5.91 6.05

(6.21) (6.21) (0.59) (0.48)

179 129 179 128

31.20 31.23 6.05 6.21

(4.09) (3.37) (0.66) (0.44)

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). a CoC = continuity of care with lead teacher over the Early Head Start semesters, N = no, Y = yes.

109

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

teachers and the first and second ratings made by the new preschool teacher in the fall and spring of the year that children transitioned to Head Start classrooms. The DECA measures within-child protective factors that consist of behaviors related to resiliency. Measures of selfregulation, such as the behavioral concerns and self-control scales of the DECA, have been closely associated with school readiness and academic achievement (Ursache, Blair, & Raver, 2012). Included in this analysis were three DECA scales describing positive adjustment, the DECA Initiative scale (alpha = 0.90 for teacher raters), the Self-control scale (alpha = 0.90), and the Attachment scale (alpha = 0.85), and one problem behavior scale, Behavioral Concerns for children 3–5 years (alpha = 0.93). The DECA has different versions for infants, toddlers, and preschoolers. The infant and toddler versions are quite similar, but the preschool version is somewhat different, so the longitudinal analysis models include a covariate for preschool version. Table 2 presents descriptive data on the DECA at each time administered by CoC group. The mean scores are well within the range of “typical” development, defined as scores of 40–60 by the DECA authors. Because the children in these analyses had to have been enrolled in EHS at an Educare program at least 3 semesters, almost all children had at least 3 DECA ratings. Some children had 4 or more ratings because they entered Educare as young infants. The Initiative and Attachment scales are included in all versions of the DECA so these ratings were made on children across the age range. The Behavioral Concerns scale is not included until the preschool version, so children were most likely to have this rating in their 3rd, 4th, or 5th semester. Children’s receptive vocabulary was assessed at about age 3 and in the spring of each year in preschool with the Peabody Picture Vocabulary Test (PPVT-4; Dunn & Dunn, 2007). Because the study language assessment schedule was age-based, PPVTs were administered to some children who turned 3 while still in EHS classrooms and to others shortly after they transitioned to Head Start classrooms. The assessor was a trained member of the ELN Implementation Study research team who did not know the continuity status of the children. Children were asked to point to one of four pictures that best showed the meaning of a word that was said aloud in standard American English by a trained assessor. All children were assessed in English because this was the dominant language used in Educare classrooms included in the study. This test is suitable for individuals ranging in age from 2.5 years through adulthood and has established age norms based on a national sample of 3540 children and adults. For children in the 2to 6-year-old range, the published internal consistency reliability coefficients were reported to range from 0.95 to 0.97, with test–retest reliabilities ranging from 0.91 to 0.94 (Dunn & Dunn, 2007). The manual reports high correlations of the PPVT-4 with other vocabulary measures (0.82 with EVT-2nd edition vocabulary score; Williams, 1997) and with closely aligned subtests of broader language measures (e.g., 0.77 with lexical/semantic composite of the CASL, CarrowWoolfolk, 1999). Standard receptive vocabulary scores were analyzed.

(in semesters). Number of semesters was used to index time in both the longitudinal analyses and in the cross-sectional analyses of the preschool outcomes in the first semester (fall) of Head Start. Only 5 children were missing data on one or more covariates, not enough to justify imputation; these 5 cases were dropped from analyses. 2.4. Data analysis Descriptive analyses were conducted first, and then hierarchical linear models (HLMs) tested whether children who did and did not experience CoC during their EHS years differed on social outcomes during infancy and on social and receptive vocabulary outcomes during preschool. Descriptive analyses showed the distribution of child outcomes and covariates for children who did and did not experience CoC with the lead teacher in their infant-toddler class and the correlations between CoC indices and child outcomes. The first set of HLMs tested whether the repeated assessments of social skills varied in level or rate of change between children with and without CoC in EHS classrooms. Separate 3-level longitudinal HLM analyses of the DECA ratings were conducted. Level 1 described individual patterns of change over time and as a function of quality during that school year and whether infant, toddler, or preschool version of the DECA was used at that time point for that child. Level 2 described the extent to which the individual intercepts and slopes varied as a function of that child’s CoC in EHS, age at entry to Educare, and child and family characteristics. These covariates include gender, race, dual language learner, IEP, child health, single parent, teenage mother, food insecurity, and stressful life events. Level 3 described the extent to which intercepts and slopes varied as a function of the school, and accounted for the nesting of children within the Educare centers. The model is shown below, describing the development over EHS for the ith child on the jth occasion in the kth school.

Level 1: Level 2:

Level 3:

Yijk = B0ik + B1ik Timeijk + B2ik Qualityijk + B3ik Timeijk × Qualityijk + B4ik Deca versionijk + B5ik Timeijk × Deca versionijk + eijk B0ik = δ00κ + δ01κ CoCik + δ02κ Entry Ageik + δ03κ CoCik × Entry Ageik + δ04κMaleik + … + δ013κ Life Eventsik + εoik B1ik = δ10κ + δ11κ CoCik + δ12κ Entry Ageik + δ13κMaleik + … + δ112κ Life Eventsik + ε1ik B2ik = δ20κ + δ21κ CoCik; B3ik = δ30κ + δ31κ CoCik; B4ik = δ40κ; B5ik = δ50κ δ00κ = π00 + ζ ok = π01 + ζ 1k = π01; .., δ013κ = π013; δ11κ = π11; .., δ112κ = π112 δ20κ = π02; δ21κ = π21; δ30κ = π30; δ31κ = π31;δ40κ = π40; δ50κ = π50 δ10κ

δ01κ

2.3. Covariates Several child-level, parent/family-level, and class-level covariates were used. As noted above, for children with multiple parent interviews, the parent- and class-level covariates were computed as the across-time mean for that child. Child-level covariates included child’s gender, race/ethnicity (White Non-Hispanic, Black Non-Hispanic, Hispanic, Other), whether the child was a DLL, whether the child had a disability plan (IFSP or IEP) at any time during Educare attendance, and parent rating of child’s health. Covariates also included the following parent/family variables: maternal education level, single parent status, teen parent status, life events, food insecurity, and depression. For all these covariates we used parent report when available and used school administrative data if parent report was unavailable. The final set of covariates described the child’s Educare experience. These covariates included the age of entry into Educare EHS and duration of attendance

The second set of HLM analyses examined preschool outcomes during the year after the children left the infant-toddler EHS classrooms for the preschool HS classrooms. The outcomes include the PPVT and DECA ratings of social-emotional functioning. Three-level random-intercepts HLM analyses tested the extent to which outcomes, after the transition to preschool, differed as a function of continuity with the infant-toddler lead teacher. Children (level 1) were nested within classrooms (level 2), and classrooms were nested within sites (level 3) to account for clustering. The same child and family covariates were included, but unlike the prior longitudinal analyses that included entry age and EHS semesters, these cross-sectional analyses included the total number of semesters in the Educare infant-toddler program and the mean global classroom quality during the EHS years. The model is listed below, for the ith child in the lth classroom in the kth school. 110

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Level 1: Level 2:

Level 3:

Table 3 Correlations between Measures of Continuity and Indices of Educare Participation and Child Outcomes (N = 745).

Yikl = B0kl + B1kl CofCikl + B2kl Time in Educarekl + B3kl Male + …+ B12ik Life Eventsikl + eikle B0kl = δ00κ + δ01κEHS Qualitykl + εokl

Continuity of Care

B1kl = δ10κ + δ11κEHS Qualitykl B2k = δ2; B3k = δ3; ... B14k = δ14 δ00κ = π00 + ζok δ01κ

Educare indices Age entered Educare (yrs.) Number of EHS semesters Classroom quality/ ERS

= π01; ... δ14 = π14

3. Results

With Lead Teacher

With Any Teachers

With Lead Teacher

With Any Teachers

0.15***

0.14***

0.14***

0.13***

−0.13***

−0.10**

−0.12**

−0.06

0.10**

0.11**

0.12**

0.09*

0.05 0.12** 0.04 −0.09

0.12*** 0.16*** 0.11** −0.10

0.10** 0.15*** 0.06 −0.07

0.04

0.02

0.03

−0.01

−0.02

−0.01

−0.02

0.03

−0.02

0.01

0.07

0.03

−0.06

0.01

−0.08

DECA scores at 3rd semester in EHS Initiative t score 0.09* Attachment t score 0.12** Self-control t score 0.08* Behavioral −0.13+ concerns t score

These analyses examined relations between continuity of care (CoC) and children’s social-emotional development during the infant-toddler period while in EHS and child outcomes in social-emotional and receptive vocabulary development after the child transitioned from EHS classrooms and teachers to HS classrooms and teachers. Analyses focused on whether the child experienced continuity with the lead teacher in infant-toddler classrooms, but sensitivity analyses examined the other indices of CoC (percentage continuity with lead teacher, continuity with any teacher, and percentage continuity with any teacher). Sensitivity analyses also examined child outcomes at the end of the their first year of HS, after the transition from infant/toddler classrooms to preschool classrooms.

Child outcomes in fall of HS PPVT standard 0.03 score DECA Initiative t −0.02 score 0.04 DECA Attachment t score DECA Self-control 0.06 t score DECA Behavioral 0.01 concerns t score

3.1. Descriptive analyses

Proportion Semesters

Note: *p < 0.05; **p < 0.01; ***p < 0.001.

First, descriptive statistics were computed. Table 1 presents descriptive statistics for children who did and did not experience CoC with the lead teacher in their infant-toddler class including statistics about their EHS experience, demographic characteristics of the children and families, continuity of care indices, and preschool outcomes from the fall of Head Start. Compared to the Early Head Start randomized study sample of 3000 children in 17 rural and urban programs (Love et al., 2005), our urban sample was proportionally more African-American (49% v. 34%) and more Hispanic (38% v. 23%) and more likely to be a dual language learner (31% v. 21%). Fewer of the mothers in the Educare sample were younger than 20 at the birth of their child (15%) compared to 39% of the EHS study mothers, although about 50% of the mothers in both studies had a high school education or more. Table 2 presents descriptive statistics by CoC group on the teacher ratings of social-emotional adjustment during the infant-toddler period and classroom quality scores. To address concerns that children with and without CoC or their parents might have differed at entry to Educare, comparisons of the two groups were conducted on Educare experiences, demographic characteristics, and teacher ratings from the child’s first semester of enrollment. Analyses compared the two groups, using t-tests for continuous variables and Chi-square tests for categorical variables. Results indicated that no child or parent demographic characteristics differed between continuity groups, although some Educare experiences differed significantly. The CoC group entered Educare at a slightly older age (t(840) = 3.91, p < 0.001) and thus attended for slightly less time (t(840) = 3.46, p < 0.001), were in classrooms with slightly higher ERS scores (t(608) = 3.26, p < 0.01), and were rated as having higher social skills on the DECA Initiative (t(840) = 2.83, p < 0.01) and Self Control (t(333) = 1.82, p < 0.10) scales at the child’s first assessment, typically collected soon after children entered the Educare school. Also, the sites differed significantly in the proportion of children who experienced continuity in their infant toddler years (X2(10) = 75.84, p < 0.0001). Accordingly, we included all of the Educare experience and demographic characteristics in analyses to reduce any selection bias due to these factors. In summary, comparisons of the children who stayed with the same lead teacher during the infant/toddler years and children who changed lead teachers during this time indicated that CoC

children entered center-based care older, with higher levels of social skills according to their teachers, and had higher quality classrooms. Table 3 reports the correlations for the entire sample between measures of continuity and indices of the child’s Educare experience, infant-toddler outcomes when the child had spent about 18 months in EHS (at 3 semesters), and preschool outcomes in the fall of the child’s first Head Start year. These correlations reflect the baseline comparisons, also indicating that CoC children entered Educare at slightly older ages and had slightly less time in infant-toddler care. Children who were with the same lead teacher throughout their EHS years also experienced slightly higher quality care. Children with CoC, defined as having the same lead teacher in their infant-toddler years, were rated as having higher levels of initiative, attachment, and self-control in the child’s third semester of EHS, but not in the first assessment of the fall after they transitioned into Head Start. 3.2. Continuity and infant-toddler outcomes Three-level longitudinal HLM analyses of child outcomes were conducted to test the extent to which patterns of change over the infanttoddler period in teacher ratings of social-emotional development varied as a function of CoC with the lead teacher. Individual intercepts and slopes with respect to time in Educare are estimated at level 1, with time-varying covariates of classroom quality and DECA version. Level 2 includes child-level measures of CoC, entry age, and child and family covariates. Level 3 accounts for the nesting of children in schools. Ideally, we would have accounted for nesting in classrooms but many of these infant/toddler classrooms were too small to allow for estimation of random classroom intercepts. Results of the HLM analyses indicated that CoC with the lead teacher was related to higher overall levels of DECA scores during the infant-toddler years from the time children entered care until they left the infant-toddler program for the preschool program. However, CoC with lead teacher was not related to rates of change in DECA scores over time. The results are shown in Table 4, listing the coefficients and 111

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Table 4 Longitudinal HLM Results: Continuity with Lead Teacher and Infant-toddler Social-emotional Development in Early Head Start. Behavioral Concerns

Self-control

Initiative

Attachment

B

(se)

B

(se)

B

(se)

B

(se)

Intercept Continuity Entry age Quality (ERS) Time (Semesters) Continuity × quality Continuity × entry age Continuity × time Time × quality Time × entry age Continuity × time × quality

53.83 −1.52* −1.61* 0.07 −0.62* −1.18 0.07 0.32 −0.15 0.44 −0.68

(0.35) (0.62) (0.63) (0.50) (0.24) (0.98) (1.21) (0.45) (0.33) (0.36) (0.65)

50.27 1.93*** 1.98*** −0.11 0.94*** 2.30** 0.72 −0.56+ −0.43 0.14 −0.55

(0.30) (0.56) (0.53) (0.36) (0.16) (0.73) (1.04) (0.31) * (0.21) (0.26) (0.41)

53.09 1.24* 0.43 −0.05 1.34*** −0.82 1.26 −0.40 0.01 1.95*** 0.43

(0.26) (0.51) (0.45) (0.31) (0.13) (0.63) (0.84) (0.26) (0.18) (0.23) (0.37)

52.51 1.89*** 0.30 −0.18 1.52*** 1.27+ 0.78 −0.08 −0.19 1.09 0.03

(0.27) (0.53) (0.47) (0.33) (0.14) (0.66) (0.91) (0.28) (0.19) (0.25) (0.38)

Covariates DECA version Male Race (F3,105) Dual language learner IEP/IFSP Child health Single parent Parent education Teenage mother Food insecurity Life events Depressive symptoms Time (Semesters) × DECA version Time × male Time × race (F3,105) Time × DLL Time × IEP/IFSP Time × health Time × single parent Time × parent ed Time × teenage mother Time × food insecurity Time × life events Time × depress sympt

−2.58** 2.80*** ns −1.83+ 2.10* 0.50 0.33 −0.05 1.02 0.57 0.08 0.75 −1.47* 0.45 ns −0.60 −0.03 0.04 0.33 0.02 −0.64 −0.08 −0.02 0.36

(1.02) (0.61)

−3.96** −2.31*** ns 0.87 −1.11 −0.46 −0.23 −0.21 −1.72* −1.00+ −0.21 −0.55 0.51 −0.12 ns 0.13 −0.30 −0.21 −0.22 0.04 −0.07 −0.10 −0.03 −0.19

(1.42) (0.56)

−4.59** −1.98*** ns −0.91 −3.69*** 0.18 0.79 0.01 −1.04 −1.97*** 0.09 −0.52 1.19 −0.40+ ns 0.40 −0.60+ −0.22 0.48+ 0.05 −0.08 0.13 0.09 −0.36

(1.72) (0.49)

−1.05 −0.52 * −1.13 −3.16*** −0.11 0.98+ −0.05 −1.47+ −1.31* 0.02 −1.07 −0.30 −0.57* ns 0.11 −0.45 −0.21 0.26 0.09 0.10 0.25 0.00 −0.46

(1.78) (0.52)

Random Effects Residual Child intercept Child time slope Child covariance School intercept School time slope

(0.97) (0.90) (0.45) (0.64) (0.16) (0.92) (0.65) (0.19) (0.93) (0.68) (0.38) (0.59) (0.53) (0.27) (0.40) (0.10) (0.56) (0.40) (0.12) (0.58)

(0.90) (0.78) (0.40) (0.58) (0.14) (0.81) (0.58) (0.17) (0.85) (0.71) (0.27) (0.45) (0.38) (0.20) (0.28) (0.07) (0.40) (0.29) (0.08) (0.42)

(0.80) (0.70) (0.36) (0.52) (0.13) (0.72) (0.52) (0.15) (0.76) (0.73) (0.24) (0.39) (0.35) (0.18) (0.25) (0.06) (0.34) (0.26) (0.07) (0.38)

σ2

σ2

σ2

σ2

26.77 0 0 −0.39 26.22 1.17

45.45 0 0.98 −2.41 35.69 0

46.87 13.14 1.79 1.49 14.24 1.07

51.70 0 3.54 −1.40 29.96 0

(0.84) (0.74) (0.38) (0.54) (0.13) (0.75) (0.54) (0.16) (0.80) (0.76) (0.26) (0.41) (0.37) (0.19) (0.27) (0.07) (0.37) (0.27) (0.08) (0.40)

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). Note: +p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

3.3. Continuity and preschool outcomes

standard errors from the analysis of each DECA outcome in separate columns. Results indicated CoC with the lead teacher was related to overall levels of DECA scores during the infant-toddler years. Children with CoC were rated by the infant-toddler teacher as showing lower levels of behavioral concerns and higher levels of self-control, initiative, and attachment. The association between CoC and self-control was moderated by observed classroom quality, suggesting that CoC more positively related to self-control ratings when classroom quality was higher (effect size = 0.30 for CoC when ERS was one SD above the mean) and was not related when quality was lower (effect size = 0.02 for CoC when ERS was one SD below the mean). Longer enrollment was also related to all child outcomes, a finding previously reported for Educare children (Yazejian et al., 2015). Behavioral concerns scores declined and self-control, initiative, and attachment scores increased as children were enrolled longer in Educare, and these effects were comparable for both the CoC and non-CoC groups. In addition, there was no reliable evidence that children in higher quality classrooms showed higher social-emotional skills overall or greater gains over time. Neither did the children in higher quality classrooms show greater gains over time if they experienced CoC.

The next set of analyses indicated that CoC experienced during center-based infant-toddler care was not reliably related to either social or receptive vocabulary outcomes after the children transitioned to a different teacher in preschool. The HLM analyses investigated whether CoC with the infant-toddler teacher was related to child outcomes in preschool when the child had graduated to a Head Start classroom and had a new teacher. Three-level random-intercept HLM analyses tested the extent to which outcomes after the transition to preschool differed as a function of continuity with the infant-toddler lead teacher. Children were nested within preschool classrooms, and classrooms within schools to account for clustering. The same child and family covariates were included, and these cross-sectional analyses also included the total number of semesters in the Educare infant-toddler program and the mean classroom quality during the EHS years. Results are shown in Table 5. Continuity with the infant-toddler teacher was not reliably related to any of the outcomes in the fall of the child’s first year in Head Start. No evidence emerged indicating that the effect of CoC on preschool outcomes was moderated by overall classroom quality in the EHS years. 112

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Table 5 Continuity of Care and Child Outcomes in the Fall of First Head Start Year. PPVT

Behavioral concerns

Self-control

Initiative

Attachment

B

(SE)

B

(SE)

B

(SE)

B

(SE)

B

(SE)

Intercept Continuity ERS quality Continuity × quality

74.66 1.42 2.11 −0.19

(10.47) (11.67) (1.56) (2.03)

52.76 0.10 −1.36+ 0.43

(0.48) (0.67) (0.74) (1.25)

48.43 0.95 0.34 0.41

(0.47) (0.70) (0.76) (1.31)

48.08 −0.29 −0.86 0.47

(0.55) (0.66) (0.75) (1.23)

49.28 0.66 −0.03 0.36

(0.59) (0.66) (0.76) (1.23)

Covariates Time (No.of EHS semesters) Male Race (F3,562) DLL IEP Child health Single parent Parent education Teen mother Food insecurity Life events Parent depression

0.69 −3.15** *** −6.14*** −4.02** 0.01 −0.62 1.06*** −1.23 −1.51 0.50 1.19

(0.46) (1.03)

−1.11* 4.27*** ns −1.30 2.05* 0.42 0.88 −0.07 1.40 0.14 −0.00 0.46

(0.45) (0.65)

−0.49 −3.43*** ns 2.30* −0.78 −0.36 −0.09 −0.33 −1.07 0.53 −0.06 −0.75

(0.30) (0.69)

0.19 −3.43*** ns −0.47 −3.68*** 0.35 −0.16 −0.19 −1.96* −0.23 0.38+ 0.61

(0.30) (0.64)

−0.26 −2.69*** ns −0.45 −2.58** 0.44 −0.55 0.03 −0.56 0.32 −0.03 1.12

(0.30) (0.64)

(1.81) (1.50) (0.76) (1.11) (0.28) (1.50) (1.08) (0.31) (1.54)

(1.11) (0.89) (0.48) (0.71) (0.18) (0.95) (0.68) (0.20) (1.00)

(1.16) (0.93) (0.50) (0.74) +(0.18) (0.99) (0.71) (0.21) (1.04)

(1.09) (0.88) (0.47) (0.69) (0.17) (0.93) (0.67) (0.20) (0.98)

(1.09) (0.88) (0.47) (0.69) (0.17) (0.93) (0.67) (0.20) (0.98)

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). Note: +p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

fact, children in the CoC classrooms also received higher socio-emotional ratings from their teachers from the start of their enrollment in the program, when all teacher-child relationships were still new. No evidence emerged indicating that CoC was related to improvements in social-emotional development during the infant-toddler period or to higher levels of social-emotional or vocabulary development after the children transitioned to preschool classrooms. Limited evidence indicated CoC was more strongly related to ratings of children’s selfcontrol when classroom quality was higher and less related when classroom quality was lower. Findings are interpreted in the context of prior research, and potential implications for recommendations regarding CoC are discussed.

3.4. Sensitivity analyses The first set of sensitivity analyses indicated that similar findings were obtained using different measures of CoC. The analyses described above were conducted using three other indices of CoC to ensure that we were not ignoring important associations between continuity and child outcomes because we focused on the lead teacher. The other measures of CoC examined were: proportion of semesters with the lead teacher, continuity with any teaching staff member (yes/no) and proportions of semesters with any teaching staff. Results from the longitudinal analyses of infant-toddler ratings of social-emotional skills are shown in Appendix A and from the cross-sectional analyses of 3-yearold outcomes in Appendix B. Proportion of semesters with the lead teacher showed an almost identical pattern of results as reported above from analyses of continuity with the lead teacher in analyses of both DECA scores during the infant-toddler period (Appendix A) and the preschool outcomes (Appendix B). CoC with any teaching staff member was also not consistently or reliably related to child outcomes, neither in tests of the main effect nor in interactions with quality. Again, CoC interacted with classroom quality in longitudinal analyses of self-control. No evidence emerged suggesting that children showed larger gains in social-emotional skills when they experienced CoC. The second set of sensitivity analyses indicated that CoC during the infant-toddler years was not related to receptive vocabulary or social skills at the end of the first preschool year. This analysis was conducted to address concerns that potential findings in the fall might be obscured by the transition from the infant-toddler classroom to the preschool classroom. Results are shown in Appendix C. No evidence indicated that CoC experienced during infant-toddler center-based care was related to these later preschool outcomes, although higher observed classroom quality in EHS was significantly related to lower levels of behavioral concerns and higher scores on the PPVT administered near the end of the first year in HS.

4.1. Infant-toddler findings This study’s findings that infant-toddler social-emotional skills were rated higher when they experienced CoC than when they had two or more lead teachers differs from the findings from some prior studies. Prior studies indicated CoC was related to lower levels of infant-toddler behavioral concerns in one study (Ruprecht et al., 2016), but not in another study (Owen et al., 2008). In addition, prior studies did not find associations beween CoC and infant-toddler social competence (Ruprecht et al., 2016) or preschool language skills (Owen et al., 2008). This study suggests several possible explanations for these apparent discrepancies. First, it is possible that teachers who expect to have an infant in their classroom for 2 or more years judge the children in the CoC classrooms more positively from the time they entered centerbased care than did teachers who expect that child to leave in a year or less. In this study, baseline differences related to CoC were detected in the first social-emotional ratings collected within a few months of enrollment. It is not possible these baseline differences could be due to longer relationships between CoC teachers and children because both the CoC and the other children had spent about the same amount of time with their teachers. As such, perhaps the findings from this study are not as discrepant from the null findings reported by others. Second, the significant quality × CoC interaction in this study may be consistent with the findings relating CoC to infant social competence in Ruprecht’s recent study (2016). Third, it is possible that teachers new to a classroom (e.g., replacing the teachers who had caused a discontinuity in caregiving) implemented practices that reduced the potential negative effects of instability of care, especially given the emphasis on CoC in Educare.

4. Discussion The extent to which CoC improved early social-emotional and language development among infants and toddlers in center-based care was examined in this secondary analysis of a program providing highquality care to children from low-income families. Findings suggested that teachers’ ratings of self-control, initiative, and attachment were higher when CoC was maintained during the infant-toddler period. In 113

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

4.3. Features of this study

An advantage of our multi-site, coordinated data collection is the resulting large longitudinal database consisting of multiple measures in programs of similar levels of quality and among sites in which all programs endorsed CoC as a policy. Thus, we were able to control for many variables in the analyses. When we examined the number of semesters that children were enrolled in EHS, we found a pattern of relationships similar to the findings for CoC. That is, children who were enrolled for more time were rated by their infant-toddler teachers as higher in self-control, initiative, and attachment, and lower in behavioral concerns, compared to children enrolled for less time. In other words, CoC did not contribute more to children’s infant-toddler outcomes if children attended the program for more semesters. The use of teacher report raises the question of what an independent observer might have noted during the infant-toddler years. However, we must consider whether teacher perception alone could benefit young children. That is, if teachers who are with children on a continuous basis are also likely to rate their initiative, self-control, and attachment higher over time, is this likely to create more favorable contexts for the children, particularly as they navigate the toddler years? As noted in the literature review, a premise of CoC is that extended time together allows teachers to develop deeper knowledge and understanding about individual children and thus, in turn, build more positive relationships (Essa et al., 1999; Theilheimer, 2006). Regarding our second research question about whether classroom quality moderated associations between CoC and child outcomes, we found that during the EHS years, classroom quality was generally not related to children’s DECA scores in these schools with overall high levels of quality. However, one interaction effect was obtained. In higher quality classrooms, CoC was more highly related to self-control, compared to lower quality classrooms. While we believe it is vital to include quality in questions regarding children’s classroom-based experiences, the restricted range of quality in Educare classrooms may have impacted these analyses.

We must emphasize that Educare represents a unique set of schools. Generally speaking, classrooms are of higher quality, with lower caregiver turnover. The classrooms tend to be well-resourced and have more optimal caregiver to child ratios. Finally, the network of schools actively promotes the philosophy of CoC. Thus, while our database is extensive, it is representative of a distinctive context. That said, even these well-resourced programs were challenged to provide CoC at the desired level, possibly explaining why findings regarding CoC are few and mixed in the extant literature. Drawing conclusions from observational studies of early care and education is always difficult due to concerns about selection bias (Burchinal, Magnuson, Powell, & Hong, 2015). The analyses showed that children who experienced CoC were more advantaged in some ways, and yet the results for CoC were modest at best. As must always be cautioned, the results could be related to variables not included in our existing dataset. Our lack of significant findings at age 3 does not preclude the potential importance of CoC experienced during infanttoddler programming to longer-term outcomes in the preschool years or beyond, as our data were not collected specifically to address this question. As explained in our methods, we operationalized CoC in several ways. Sensitivity analyses revealed few differences in the results for these varying methods, which included a binary measure (yes/no) and a proportional measure, as well as considering only the lead teacher or any of the teaching staff. Because infants and toddlers most frequently experience care from a team of at least two adults, we thought it was important to consider both lead and assistant teachers. However, in these analyses, the findings are no different. Perhaps when a primary caregiving arrangement is utilized, wherein children are assigned to a particular staff member, it is that person’s continuous presence that should be measured. Unfortunately, this variable was not available in the existing dataset used for our analyses. The scant existing CoC literature contains studies with widely varying length of time of continuity. Indeed, because our dataset was collected over the entire length of a child’s enrollment in EHS, many children in our non-continuity group may have been considered to have experienced continuity in other published studies. One recent study (Ruprecht et al., 2016) contrasted the experiences and outcomes of 115 toddlers in classrooms implementing continuity or non-continuity practices, in which the average duration of care for toddlers with specific caregivers in continuity rooms was 14 months in contrast to 5 months in non-continuity rooms. The NICHD study found that children often experienced more than three new arrangements of childcare, on average, across their first 3 years (NICHD ECCRN, 2005). We adopted a more stringent sample inclusion criterion of at least 18 months of enrollment in the infant-toddler center in order to require that continuity extend beyond a single year. Clearly, ongoing research about CoC must be framed within both the exceptional and the typical experiences of children. Several areas are left unexplored within this study. We do not know why CoC did not occur for children in this study. Non-continuity experiences could have resulted from teacher turnover, movement of a child or children to maintain mandated ratios and group sizes, local policies that move children to different classrooms based on developmental milestones (e.g., crawling) or the calendar (e.g., changing classrooms each fall), or other practices. Factors that impact continuity differ in regard to how much control centers have over these practices, and thus what we might expect from best practice recommendations (Sosinsky et al., 2016). Although we do not know what contributed to lack of continuity, we did find that even in schools striving to implement CoC as part of their program model, roughly half of the sample of children who attended at least three semesters did not experience 100% continuity with their lead teacher. Our findings show CoC was difficult to maintain, becoming less likely as children spent more time in center-

4.2. Preschool findings In this study, the outcome measures in the infant-toddler years were ratings completed by the very teachers who provided more, or less, continuity in the children’s experiences. In light of this potential confound, we examined the social-emotional outcomes as rated by each child’s preschool teacher, one who was independent of the prior relationships. As noted, no significant associations between CoC and child outcomes resulted at this point in time. Classroom quality, averaged for the infant-toddler years, did not moderate the relation between CoC and children’s age 3 social-emotional and receptive vocabulary outcomes. However, when we examined these outcomes at the end of the first year of preschool, we found that infant-toddler classroom quality was indeed related to children’s receptive vocabulary and level of behavioral concerns. These associations were as expected, with children who had experienced higher quality classrooms scoring higher on the PPVT and being rated as having fewer behavioral concerns. The question is why these relations were not apparent at the first preschool data collection point. We speculate that the transition to preschool is an important process that bears closer examination. Generally, children’s DECA scores dropped from the last infant-toddler data point to the first preschool data point. Children moved to classrooms with approximately twice as many children (from 8 to 17), and new expectations and routines. This is in addition to having new classroom teachers. Thus, it appears that effects of transition points such as this call for closer scrutiny. In summary, while CoC was associated with positive outcomes during the infant-toddler years, there was no relation between this practice and the children’s outcomes at the start of preschool or even with gains in the outcomes over time during the infant-toddler years. Because similar findings resulted for the relation between number of semesters enrolled and children’s outcomes during the infant-toddler years, support for CoC practices remains equivocal in this study. 114

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

based infant-toddler care. These findings add to the literature that continuity of care is difficult to achieve, whatever the multiple factors that contribute to discontinuity. Also, we do not have a measure of precisely how CoC was implemented in each center. We do know that mixed-age grouping was utilized by some and that others looped children of a more restricted age range with their teachers from infancy through the toddler years. As noted by Sosinsky et al. (2016), each model requires unique skills from teachers, necessitating that some teaching teams become adept at responding to a wide variety of developmental needs at once while others adapt to changes over time and the sudden shift to picking up a new group of infants once an established group moves out of the infanttoddler program. These nuances are largely unexplored in the published literature.

Specifically, future research must continue to address the question of how to define and operationalize CoC. The issue of necessary or required timeframes for continuity must be addressed. Operational constructs should contribute to our understanding of both children’s average experiences and practices that extend CoC into much longer timeframes. The potential interaction of CoC with teacher attitudes and behaviors should also be considered. We see evidence of association with teacher behaviors in several studies showing teachers are more sensitive and interactive (e.g., Owen et al., 2008; Ruprecht et al., 2016) and this might be responsible for reports of lower behavior problems. Centers that implement CoC may attract teaching staff who are skilled in relationship-building or are otherwise “high-ability” (Raikes, 1993). The possibility exists that some teacher-child dyads might have less positive interactions due to temperamental mismatch, a less skilled teacher, or other factors, and thus continuity would not have the predicted effect, and in the worst case, could be detrimental. Again, these relationships and possibilities have not been adequately investigated to date. Future research should also include assessment of the potential effects of CoC on families and teachers. The Owen et al. (2008) finding that CoC was associated with closer parent-caregiver relationships as reported by parents provides a foundation for this work. Newer qualitative research (McMullen et al., 2016) supports this as well. Reports of closer parent-teacher relationships and of teacher sensitivity (Owen et al., 2008) and interactive involvement (Ruprecht et al., 2016) suggests that CoC could improve later child outcomes through these mechanisms. Future research efforts should examine child outcomes at a later point in time through methods like cross-lag panel analyses. Another possibility is that it is the involved adults—the caregivers and family members—who benefit the most from CoC. Future research should examine this possibility as well as consistency in peers which is another related and important, yet understudied, feature of infanttoddler classrooms. The research literature, including this study, does establish that CoC is a complex practice, implemented in any number of ways, and for varying lengths of time in programs. Because it makes so much sense on a theoretical and philosophical basis, we advocate for ongoing research to add to our understanding of this practice. Prospective, longitudinal studies designed to investigate the complexities of CoC are needed to inform practice and policy.

4.4. Limitations Our study contributes to the emerging body of knowledge by utilizing a much larger database than previous research, and a greater range of measures, allowing for more control in our analyses. Our sample represented programs from several areas of the country with racially and ethnically diverse children. However, there are several limitations. First, the current landscape of infant-toddler care in the U.S. reflects rather low quality of care overall (Phillips & Lowenstein, 2011). The classrooms within the Educare network represent higherthan-average quality (Yazejian et al., 2015), limiting generalizability of the findings. Additionally, the range of quality in our sample was rather restricted. This might relate to the lack of findings for associations between quality and our child outcome measures or explain why few interaction effects were found for quality and CoC. Finally, our sample was largely urban and this characteristic limits generalizability to similar programs and samples. Second, in this study even children who did not have the same lead teacher during their enrollment in EHS (i.e., deemed no continuity), still experienced considerable continuity with their infant-toddler teachers. On average, they spent 59% of time with the same lead teacher and 79% of their time with a consistent adult, whether teacher, assistant, or aide. Thus, their experience may not be similar to other children who do not experience continuity of care practices. Third, we had no information about the actual classroom strategies used to implement CoC in our sample. Ruprecht et al. (2016) were able to document variations in the ways different programs implemented CoC, including ways of measuring continuity beyond time with a caregiver. Strategies for implementation of CoC and outcomes related to these strategies require additional study.

Acknowledgements We gratefully acknowledge the funding support for this work provided by the Buffett Early Childhood Fund and the George Kaiser Family Foundation. The content of this publication does not necessarily reflect the views or policies of the Buffett Early Childhood Fund or the George Kaiser Family Foundation. We also thank Karen Taylor and John Cashwell for valuable assistance with the data for this paper. Finally, we thank the programs for providing the data, including the Educare staff, families, and children.

4.5. Future research We noted at the start that this area of practice has been informed largely from theory and philosophy, with a thin base of empirical studies upon which to draw for evidence. Our study adds to the empirical literature but leaves many questions to be addressed by future research.

Appendix A. Longitudinal Social-emotional Development in the Early Head Start Years Using Other Indices of Continuity of Care.

Continuity–Any Teacher Continuity Entry age ERS quality Time (Semesters) Continuity × quality

Behavioral concerns

Self-control

B

(se)

B

(se)

B

(se)

B

(se)

−0.54 −1.74** 0.11 −0.66* −0.00

(0.62) (0.64) (0.51) (0.24) (1.05)

0.70 2.06*** −0.19 0.90*** 2.17**

(0.62) (0.54) (0.35) (0.16) (0.76)

0.65 0.52 −0.10 1.35*** 0.65

(0.54) (0.45) (0.30) (0.13) (0.64)

1.06+ 0.41 0.10 1.51*** 0.98

(0.57) (0.47) (0.35) (0.14) (0.67)

115

Initiative

Attachment

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Continuity × entry age Continuity × time Time × quality Time × entry age Continuity × time × qual

0.39 0.18 0.00 0.47 −0.00

(1.42) (0.51) (0.33) (0.36) (0.63)

1.60 0.13 −0.46* 0.10 0.25

(1.17) (0.35) (0.20) (0.26) (0.43)

2.11* −0.05 −0.06 1.91*** 0.42

(0.94) (0.27) (0.18) (0.23) (0.38)

1.17 0.05 −0.21 1.09*** −0.00

(1.01) (0.29) (0.19) (0.25) (0.40)

Prop Continuity–Lead Teacher Continuity Entry age ERS quality Time Continuity × quality Continuity × entry age Continuity × time Time × quality Time × entry age Continuity × time × qual

−3.13* −1.63* 0.15 −0.62* −2.20 2.75 2.01* −0.11 0.46 −136

(1.38) (0.63) (0.49) (0.24) (2.08) (2.72) (0.98) (0.32) (0.36) (1.27)

5.38*** 1.97*** −0.16 0.93*** 4.38** −2.20 −2.32*** −0.41* 0.15 −0.91

(1.24) (0.53) (0.36) (0.16) (1.48) (2.29) (0.68) (0.20) (0.26) (0.85)

3.99*** 0.38 −0.14 1.25*** 0.30 1.41 −0.79 −0.02 1.96*** 0.96

1.09 0.45 0.31 0.13 1.23 1.84 0.55 0.18 0.23 0.74

5.45*** 0.23 −0.20 1.51*** 2.70*** 0.47 −0.44 −0.20 1.10*** 0.09

1.15 0.47 0.32 0.14 1.29 1.99 0.59 0.19 0.25 0.78

Prop Continuity–Any Teacher Continuity Entry age ERS quality Time Continuity × quality Continuity × entry age Continuity × time Time × quality Time × entry age Continuity × time × qual

0.08 −1.77** −0.02 −0.66* 0.68 1.13 1.31 0.02 0.43 0.87

(1.87) (0.64) (0.50) (0.24) (2.87) (3.36) (0.88) (0.33) (0.36) (1.55)

1.99 2.08*** −0.19 0.90*** 6.22** 2.65 −0.16 −0.46* 0.11 0.70

(1.69) (0.53) (0.35) (0.16) (2.05) (3.21) (0.96) (0.20) (0.26) (1.20)

3.36* 0.47 −0.12 1.35*** 2.93+ 5.75* 0.29 −0.07 1.90*** 0.53

1.46 0.45 0.30 0.13 1.71 2.56 0.74 0.18 0.23 1.07

4.31** 0.40 −0.22 1.52*** 2.49 2.23 −0.31 −0.20 1.10*** −0.63

1.54 0.47 0.32 0.14 1.80 2.76 0.79 0.19 0.25 1.14

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). Note: *p < 0.05; **p < 0.01; ***p < 0.001. Appendix B. HLM Results: Child Outcomes in the Fall of First Year of Head Start Using Other Indices of Continuity of Care.

PPVT

Behavioral concerns

Self-control

(SE)

B

(SE)

B

(SE)

B

(SE)

B

(SE)

Continuity with any teacher Continuity 1.94 ERS quality 1.77 Time (Semesters) 0.81 Continuity × qual 2.96

(1.16) (1.10) (0.46) (2.30)

−0.07 −1.35 −0.50+ 0.44

(0.77) + (.74) (0.30) (1.47)

0.50 0.44 −0.02 −0.44

(0.80) (0.76) (0.31) (1.54)

0.30 −0.91 0.22 0.58

(0.77) (0.75) (0.30) (1.45)

0.30 0.05 −0.30 −0.14

(0.77) (0.76) (0.30) (1.46)

Prop continuity w lead teacher Continuity −1.03 ERS quality 2.13+ Time 0.62 Continuity × qual −1.74

(2.37) (1.10) (0.46) (4.37)

0.53 −1.38+ −0.47 3.21

(1.52) (0.74) (0.30) (2.74)

2.79+ 0.34 0.06 0.49

(1.58) (0.76) (0.31) (2.89)

−1.13 −0.86 0.18 2.40

(1.50) (0.75) (0.30) (2.71)

1.64 −0.03 −0.25 1.35

(1.51) (0.76) (0.30) (2.71)

Prop continuity w any teacher Continuity 3.91 ERS quality 1.82+ Time 0.76+ Continuity × qual 8.91

(3.14) (1.10) (0.45) (6.15)

−1.03 −1.34 −0.52+ 1.20

(1.99) (.74) (0.30) (3.78)

2.58 0.34 −0.00 −0.75

(2.08) (0.76) (0.31) (3.95)

1.27 −0.93 0.23 3.67

(1.98) (0.74) (0.29) (3.73)

1.36 −0.03 −0.29 1.04

(1.99) (0.75) (0.30) (3.74)

B

+

Initiative

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). Note: *p < 0.05; **p < 0.01; ***p < 0.001.

116

Attachment

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Appendix C. HLM Results: Child Outcomes in the Spring of Head Start Year as a Function of Continuity of Care with Lead Teacher in InfantToddler Years.

PPVT

Intercept Continuity ERS quality Time (Semesters) Continuity × qual

Behavioral Concerns

Self-control

Initiative

Attachment

B

(SE)

B

(SE)

B

(SE)

B

(SE)

B

(SE)

94.64 −1.50 3.75** −0.24 −2.25

(0.75) (1.18) (1.25) (0.53) (2.24)

53.05 −0.45 −2.20** −0.98* 0.06

(0.53) (0.78) (0.84) (0.36) (1.53)

49.15 0.94 −0.47 0.51 −0.09

(0.46) (0.78) (0.82) (0.35) (1.52)

49.69 −0.50 −0.80 0.36 0.56

(0.51) (0.76) (0.82) (0.34) (1.47)

49.89 −0.92 1.10 0.09 −0.62

(0.60) (0.73) (0.81) (0.33) (1.42)

Note: ERS = Environment Rating Scale, either the Infant/Toddler (ITERS) or the Early Childhood (ECERS-R). Note: *p < 0.05; **p < 0.01; ***p < 0.001.

Stability and concordance with parental attachments. Child Development, 63(4), 867–878. Howes, C., & Hamilton, C. E. (1993). The changing experience of child care: Changes in teachers and in teacher-child relationships and children’s social competence with peers. Early Childhood Research Quarterly, 8, 15–32. Howes, C., & Spieker, S. (2008). Attachment relationships in the context of multiple caregivers. In J. Cassidy, & P. Shaver (Eds.). Handbook of attachment: Theory, research, and clinical application (pp. 317–332). (2nd ed.). New York, NY: Guilford Press. Institute of Medicine (IOM), & National Research Council (NRC) (2012). From neurons to neighborhoods: An update: Workshop summary. Washington, D.C: The National Academic Press. Lally, R., & Mangione, P. (2009). The program for infant/toddler care. In J. L. Roopnarine, & J. E. Johnson (Eds.). Approaches to early childhood education (pp. 25– 47). Upper Saddle River, NJ: Pearson. LeBuffe, P. A., & Naglieri, J. A. (1999). Devereux early childhood assessment: Technical manual. Kaplan Early Learning Company. Love, J. M., Kisker, E. E., Ross, C., Raikes, H., Constantine, J., Boller, K., ... Vogel, C. (2005). The effectiveness of Early Head Start for 3-year-old children and their parents: Lessons for policy and programs. Developmental Psychology, 41, 885–901. Mangione, P. L. (2006). Creating responsive, reciprocal, relationship with infants and toddlers. In R. Lally, P. Mangione, & D. Greenwald (Eds.). Concepts for care (pp. 25– 29). San Francisco, CA: WestEd. McMullen, M. B., Yun, N. A., Mihai, A., & Kim, H. (2016). Relationship-based care practices within infant-toddler continuity of care from the perspectives of practitioners and families. Early Education & Development, 27, 190–220. NICHD Early Child Care Research Network, & Duncan, G. (2003). Modeling the impacts of child care quality on chidlren’s preschool cognitive development. Child Development, 74, 1454–1475. NICHD Early Child Care Research Network (1996). Characteristics of infant child care: Factors contributing to positive caregiving. Early Childhood Research Quarterly, 11(3), 269–306. NICHD Early Child Care Research Network (2005). Childcare and child development: Results from the NICHD study of early childcare and youth development. New York: Guilford. NICHD Early Child Care Research Network (2006). Child care effect sizes for the NICHD Study of Early Child Care and Youth Development. American Psychologist, 61(2), 99–116. NSECE Project Team (2015). Tables on households’ ece usage and costs to parents. Chicago: NORC. National Association of Child Care Resource and Referral Agencies (2008). Covering the map: Child care resource and referral agencies providing vital services to parents throughout the United States. Arlington, VA: Author. National Scientific Council on the Developing Child (2004). Young children develop in an environment of relationships. Working paper no. 1. Retrieved July 24, 2014 from http:// www.developingchild.net. Nord, M., Andrews, M., & Carlson, S. (2007). Measuring food security in the United States: Household food security in the United States, 2001. Economic research report29. Owen, M. T., Klausli, J., Mata-Otero, A., & O’Brien-Caughy, M. (2008). Relationship focused child care practices: Quality of care and child outcomes for children in poverty. Early Education & Development, 19, 302–329. http://dx.doi.org/10.1080/ 10409280801964010. Phillips, D. A., & Lowenstein, A. E. (2011). Early care, education, and child development. Annual Review of Psychology, 62, 483–500. Program for Infant/Toddler Care. PITC’s six program policies. (n.d.). Retrieved from https:// www.pitc.org/pub/pitc_docs/policies html. Radloff, L. S. (1977). The CES-D scale a self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. Raikes, H. (1993). Relationship duration in infant care: Time with high-ability teacher and infant-teacher attachment. Early Childhood Research Quarterly, 8(3), 309–325. Rand Health (1994). Depression screener. Available from http://www.rand.org/content/ dam/rand/www/external/health/surveys_tools/depression/screener_eng.pdf. Ruprecht, K., Elicker, J., & Choi, J. Y. (2016). Continuity of care, caregiver-child interactions, and toddler social competence and problem behaviors. Early Education & Development, 27, 221–239.

References Ackerman, D. J. (2008). Continuity of care, professional community, and the policy context: Potential benefits for infant and toddler teachers’ professional development. Early Education and Development, 19(5), 753–772. Aguillard, A. E., Pierce, S. H., Benedict, J. H., & Burts, D. C. (2005). Barriers to the implementation of continuity of care practices in child care centers. Early Childhood Research Quarterly, 20(3), 329–344. Ahnert, L., Pinquart, M., & Lamb, M. E. (2006). Security of children’s relationships with nonparental care providers: A meta-analysis. Child Development, 77, 664–679. http:// dx.doi.org/10.1111/j.1467-8624.2006.00896.x. Ainsworth, M. D. S., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment: observations in the strange situation and at home. Hillsdale, NJ: Erlbaum. Belsky, J., & Eggebeen, D. (1991). Early and extensive maternal employment and young children’s socioemotional development: Children of the National Longitudinal Survey of Youth. Journal of Marriage and the Family, 53, 1083–1099. Burchinal, M., Roberts, J., Nabors, L., & Bryant, D. (1996). Quality of center child care and infant cognitive and language development. Child Development, 67, 606–620. Burchinal, M., Magnuson, K., Powell, D., & Hong, S. S. (2015). Early child care and education and child development. In M. Bornstein, R. Lerner, & T. Leventhal (Vol. Eds.), Handbook of Child Psychology and Developmental Science: Vol. IV, (pp. 223–267). Hoboken, NJ: Wiley. Bureau of Labor Statistics (2014). Women in the labor force: A databookhttp://www.bls. gov/opub/reports/cps/womenlaborforce_2013.pdf. Campbell, F., Pungello, E., Miller-Johnson, S., Burchinal, M., & Ramey, C. (2001). The development of cognitive and academic abilities: Growth curves from an early intervention educational experiment. Developmental Psychology, 37, 231–242. http:// dx.doi.org/10.1037/0012-1649.37.2.231. Carrow-Woolfolk, E. (1999). CASL: Comprehensive assessment of spoken language. American Guidance Service. Copple, C., Bredekamp, S., Koraleck, D., & Charner, K. (Eds.). (2013). Developmentally appropriate practice: Focus on infants and toddlers. Washington, DC: NAEYC. Cost Quality and Child Outcomes Study Team (1995). Cost, quality, and child outcomes in child care centers: Technical reportDenver, CO: University of Colorado at Denver. Cryer, D., Hurwitz, S., & Wolery, M. (2001). Continuity of caregiver for infants and toddlers in center-based child care: Report on a survey of center practices. Early Childhood Research Quarterly, 15, 497–514. http://dx.doi.org/10.1016/S08852006(01)00069-2. Cryer, D., Wagner-Moore, L., Burchinal, M., Yazejian, N., Hurwitz, S., & Wolery, M. (2005). Effects of transitions to new child care classes on infant/toddler distress and behavior. Early Childhood Research Quarterly, 20, 37–56. http://dx.doi.org/10.1016/ j.ecresq.2005.01.005. De-Souza, D. (2012). Child care center directors’ perceptions of continuity of care: A qualitative investigation. Available from ProQuest Dissertations & Theses Global. (1021371026). Retrieved from http://search.proquest.com/docview/1021371026? accountid=12964. Dunn, D. M., & Dunn, L. M. (2007). Peabody picture vocabulary test: Manual. San Antonio, TX: Pearson. Essa, E. L., Favre, K., Thweatt, G., & Waugh, S. (1999). Continuity of care for infants and toddlers. Early Child Development and Care, 148(1), 11–19. Han, W. J., Waldfogel, J., & Brooks-Gunn, J. (2001). The effects of early maternal employment on later cognitive and behavioral outcomes. Journal of Marriage and Family, 63, 336–354. Harms, T., Clifford, R. M., & Cryer, D. (1998). Early childhood environment rating scalerevised edition. New York, NY: Teachers College Press. Harms, T., Cryer, D., & Clifford, R. M. (2003). Infant/toddler environment rating scale-revised edition. New York, NY: Teachers College Press. Horm, D. M., Goble, C. B., & Branscomb, K. R. (2011). Infant toddler curriculum: Review, reflection, and revolution. In N. File, J. J. Mueller, & D. B. Wisneski (Eds.). Curriculum in early childhood education: Re-examined, rediscovered, renewed (pp. 105–119). New York, NY: Routledge, Taylor & Francis. Howes, C., & Hamilton, C. E. (1992). Children’s relationships with child care teachers:

117

Early Childhood Research Quarterly 42 (2018) 105–118

D.M. Horm et al.

Whitebook, M., & Sakai, L. (2003). Turnover begets turnover: An examination of job and occupational instability among child care center staff. Early Childhood Research Quarterly, 18, 273–293. Whitebook, M., Phillips, D., & Howes, C. (2014). Worthy work, STILL unlivable wages: The early childhood workforce 25 years after the National Child Care Staffing Study. Berkeley, CA: Center for the Study of Child Care Employment, University of California, Berkeley. Williams, K. T. (1997). Expressive Vocabulary Test Second Edition (EVT™ 2). Journal of the American Academy of Child & Adolescent Psychiatry, 42, 864–872. Yazejian, N., Bryant, D., & Kennel, P. (2013). Implementation and replication of the Educare model of early childhood education. In T. G. Halle, A. J. Metz, & I. MartinezBeck (Eds.). Applying implementation science in early childhood programs and systems (pp. 209–225). Baltimore, MD: Paul H. Brookes. Yazejian, N., Bryant, D., Freel, K., Burchinal, M., & the Educare Learning Network (ELN) Investigative Team (2015). High-quality early education: Age of entry or time in care differences in student outcomes for English-only and dual language learners. Early Childhood Research Quarterly, 32, 23–39. http://dx.doi.org/10.1016/j.ecresq.2015. 02.002. Yazejian, N., Bryant, D., Hans, S., Horm, D., St. Clair, L., File, N., & Burchinal, M. (2017). Child and parenting outcomes after one year of Educare. Child Development. http:// dx.doi.org/10.1111/cdev.12688. Zero to Three (2008). Caring for infants and toddlers in groups: Developmentally appropriate practice (2nd ed.). Washington, D. C: Author.

Sosinsky, L., Ruprecht, K., Horm, D., Kriener-Althen, K., Vogel, C., & Halle, T. (2016). Including relationship-based care practices in infant-toddler care: Implications for practice and policy. Brief prepared for the Office of Planning, Research and Evaluation. Administration for Children and Families, U.S. Department of Health and Human Services. Sroufe, L. A. (1988). The role of infant-caregiver attachment in development. In J. Belsky, & T. Nezworski (Eds.). Clinical implications of attachment (pp. 18–38). Hillsdale, NJ: Erlbaum. Theilheimer, R. (2006). Molding to the children: Primary caregiving and continuity of care. Zero to Three, 26(3), 50–54. Thompson, R. A. (2000). The legacy of early attachments. Child Development, 71(1), 145–152. Ursache, A., Blair, C., & Raver, C. C. (2012). The promotion of self-regulation as a means of enhancing school readiness and early achievement in children at risk for school failure. Child Development Perspectives, 6(2), 122–128. Van IJzendoorn, M. H., Sagi, A., & Lambermon, M. W. (1992). The multiple caretaker paradox: Data from Holland and Israel. New Directions for Child and Adolescent Development, 1992(57), 5–24. Vandell, D. L., Belsky, J., Burchinal, M., Steinberg, L., Vandergrift, N., & NICHD Early Child Care Research Network (2010). Do effects of early child care extend to age 15 years? Results from the NICHD study of early care and youth development. Child Development, 81(3), 737–756. Waldman, E. (1983). Labor force statistics from a family perspective. Monthly Labor Review, 16–40.

118