Table of Contents | Previous | Next |
III. QUALITY OF CHILD CARE USED BY EARLY HEAD START FAMILIES AND PROGRAM IMPACTS ON THE QUALITY OF CHILD CARE FAMILIES USED
Early Head Start program designers, as reflected in the report of the Advisory Committee on Services for Families with Infants and Toddlers, anticipated that Early Head Start programs would not only enable families to access the care the needed (as we saw in Chapter II), but also enhance the quality of that care. Thus, Early Head Start was expected to increase the use of good-quality child care by low-income families with infants and toddlers, whether that care is provided by Early Head Start programs directly or by community child care providers (U.S. Department of Health and Human Services 1994). The Early Head Start evaluation was designed so that data collected on child care quality would enable us to assess the extent to which the programs included in the research made a difference in the quality of the care the families enrolled their children in. The randomized design of the evaluation enables us to present strong evidence of the extent to which Early Head Start programs created better-quality center child care experiences for the enrolled children when compared with their randomly assigned control-group counterparts.
In this chapter, after presenting our methods and procedures, we report data that describe the levels of quality of care that Early Head Start children received, using a range of quality indicators. We then use data from a subset of the programs to examine the extent to which these Early Head Start programs made a difference in the proportion of families placing their children in good-quality center care arrangements, and in the levels of quality of center child care, when the experience of Early Head Start children is contrasted with that of the control group. Finally, we discuss regression analyses conducted within the Early Head Start sample suggesting that high levels of both child care use and child care quality may contribute to more-positive developmental outcomes for Early Head Start children.
A. MEASURING CHILD CARE QUALITY IN THE EARLY HEAD START EVALUATION
1. Child Care Settings Included in this Study
We assessed child care quality through direct observation at the time of the child assessments, when the children were 14, 24, and 36 months old. At these ages, the child settings eligible for observational assessment shared the following characteristics:
- Regular Child Care Arrangement—We
observed arrangements in which the child spent 10 or more hours
per week for at least two weeks prior to the interview outside
the child’s home (or by a nonrelative in the child’s
home). These criteria are the same as those used in the NICHD
Study of Early Child Care. The hours per week criterion ensures
that the child’s exposure to the arrangement meets a minimum
threshold for it to potentially influence his or her development.
The two-week criterion ensures that the provider and child are
minimally acquainted so that typical interactions can be observed.
- Relatives and Nonrelatives—The
focus of the observational study was care outside the child’s
home with relatives or nonrelatives, and in-home care settings
with nonrelatives. These child care arrangements provide a sufficient
distinction either in the caregiver (a nonrelative) or the place
(out of the child’s home) to warrant intensive data collection
as a distinct aspect of the child’s environment. Care provided
by a relative in the child’s own home was considered to
be very similar to parental care; therefore, we excluded these
settings from the observational study.
- “Family Child Care”—We
refer to all observed in-home care settings as family child care,
whether care was provided by a relative or nonrelative caregiver.
Because the regulation of home-based care settings varies from
state to state, and because information about these arrangements
came from parent reports, we did not collect information about
whether these home-based child care settings were registered or
licensed. Thus, our references to family child care include care
provided by relatives and nonrelatives, as well as regulated and
unregulated care.
- Primary Child Care Arrangement—If more than one child care arrangement met these criteria, the arrangement used for the most hours per week was chosen for the observational study.
The total number of child care settings that could be observed was thus influenced by several factors: (1) the response rate to the birthday-related assessment (parents who were not interviewed in a particular wave of these assessments could not be asked about child care arrangements); (2) the rates of child care use; (3) the rates of use of out-of-home or nonrelative child care (settings “eligible” for observation); (4) parents’ willingness to allow interviewers to contact their child care providers; (5) our success in locating the providers; and (6) the providers’ willingness to be interviewed and observed.
2. Response Rates
Table III.1 shows how these factors combined to produce the response rates in the observational study of child care at each of the birthday-related assessment points. At each time point, between 70 and 80 percent of the parents completed an interview. Since half or fewer of the families interviewed were using an “eligible” child care arrangement for the child, only 32 to 35 percent of the original full sample of children were using child care that could be observed at any point. Observations were completed with between 53 percent and 56 percent of the “eligible” arrangements at each point, with a much higher completion rate for center care arrangements (approximately 70 percent) than for family, relative, and other home-based care arrangements (approximately 32 percent).
The pattern of response rates and the number of observations varied considerably by site, in part reflecting the mix of child care arrangements in each site and the relative difficulty of completing observations in family child care settings.1 In some sites, the level of non-response was substantial, and in general, nonresponse was quite high among the in-home providers.
| Description of Sample | 14-Month Child Care Observations |
24-Month Child Care Observations |
36-Month Child Care Observations |
|---|---|---|---|
| Number of Children in the Sample | 3,001 | 3,001 | 3,001 |
| Number of Families Responding to Parent
Interview/Child Assessment |
2,344 | 2,166 | 2,110 |
| Percentage of all Children | 78.1 | 72.2 | 70.3 |
| Number of Children in an Eligible Child Care Arrangementa | 962 | 976 | 1,060 |
| Percentage of all Children | 32.1 | 32.5 | 35.3 |
| Number Whose Provider
Was Located, Agreed to Participate, and Completed the Observation |
509 | 547 | 596 |
| Percentage of All Children | 17.0 | 18.2 | 19.9 |
| Percentage of Children
with Eligible Arrangements Who Had a Complete Observation |
52.9 | 56.1 | 56.2 |
|
70.4 | 72.9 | 69.4 |
|
32.4 | 33.7 | 30.1 |
| In the Subset of Sites
Included in the Impact Analysis, the Percentage of Children with Eligible Arrangements Who Had a Complete Observation |
66.9 | 65.9 | 68.5 |
|
81.4 | 79.4 | 82.1 |
|
85.1 | 85.5 | 84.2 |
|
72.1 | 66.1 | 78.7 |
|
35.4 | 31.2 | 27.7 |
| Source: Parent Interviews and observations of child care arrangements
conducted when children were approximately 14, 24, and 36 months
old.
Note: Sites included in the impact analysis of child care quality include all four sites with center-based Early Head Start programs and four mixed-approach sites at all three time periods and an additional mixed-approach site at 36 months. aEligible arrangements include care outside the child’s home (with a relative or nonrelative) and care in the child’s home with a nonrelative. The child must have been in the child care arrangement for at least 10 hours per week and have 2 or more weeks’ experience in that arrangement. If the child was in more than one child care arrangement that met these criteria, the arrangement used for the most hours per week was chosen for the observation. |
Therefore, we focused the analysis of child care quality in sites with higher response rates overall. Not surprisingly, a higher proportion of families in center-based program sites and some mixed-approach sites were using child care, and care that met the study’s eligibility criteria.
Response rates were high in these sites for center care arrangements, though not for family child care arrangements. Table III.1 shows that, in the four center-based sites and the four (at age 2) and five (at age 3) mixed-approach sites included in our analysis of the impact of Early Head Start on child care environments, between 79 and 82 percent of the eligible center child care arrangements were observed, while between 28 and 31 percent of the family child care arrangements were observed. Therefore, our characterization of the average quality of care experienced by children in the study is more reliable for children in centers than in home-based or family child care. Accordingly, we focused the impact analysis on center child care arrangements in this subset of sites, for a total of 315 to 390 child care observations. By program-control status, interviewers observed a somewhat higher proportion of center child care arrangements for Early Head Start program children than for control-group children in these sites, possibly reflecting a greater ease of access to Early Head Start-sponsored child care settings.
3. Procedures and Instruments Used
To conduct the observational assessments of child care, trained observers visited the child care settings for two to three hours in the morning. Interviewer/observers conducted interviews with center directors and providers and observed the classroom or home in order to complete several structured observation protocols. The observational measures used are described in Box III.1, and our procedures for training and achieving inter-observer reliability are summarized in Appendix B.
Early Childhood Environment Rating Scale-Revised (ECERS-R; Harms, Clifford, and Cryer 1998; 36 months) – measures the global quality of child care for preschoolers in center settings. Items measure the quality of space and furnishings, personal care routines, language and reasoning, activities, interaction, program structure, and provisions for parents and staff. (Four items on parents/staff were omitted from the version used in this study.) Items are coded on a seven-point scale from inadequate (1) and minimal (3) to good (5) and excellent (7). The score is the average across all 39 items and can range from 1 to 7. Family Day Care Rating Scale (FDCRS; Harms and Clifford 1989; 14, 24, and 36 months) – measures the global quality of child care for infants, toddlers, and preschoolers in family child care settings. Items measure the quality of space and furnishings, basic care, language and reasoning, learning activities, social development, and adult needs. Items are coded on a seven-point scale from inadequate (1) and minimal (3) to good (5) and excellent (7). The total score is the average across the 31 items we used and can range from 1 to 7. Child-Adult Ratio (14, 24, and 36 months) – Observer’s count of the number of children and caregivers in the classroom at the time of the observation. The number used in our analysis was the average of up to six observations over the 2-hour observation period. Arnett Caregiver Interaction Scale (Arnett 1989; 14, 24, and 36 months) – measures the quality of the caregiver’s interactions with children in both center and family child care settings. Items are scored based on a 2.5 hour observation of the primary caregiver in the child care setting, and measure the extent to which the caregiver spoke warmly, seemed distant or detached, exercised rigid control, or spoke with irritation or hostility. Items are coded on a 4-point scale from “not at all” characteristic of the caregiver (1) to “very much” characteristic of the caregiver (4). We conducted factor analyses at each time point, but since no clear set of similar subscales emerged across time, we report our findings based on the full Arnett score, the average rating across all 26 items. Child-Caregiver Observation System (C-COS; Boller, Sprachman, and the Early Head Start Research Consortium 1998; 24 and 36 months) – measures the types of caregiver interaction and child activities specifically pertaining to the focus child based on six 5-minute observations. During each 5-minute observation, observers watched the focus child for 20 seconds and then indicated whether a specific set of child and caregiver behaviors occurred (the recording phase lasted 10 seconds). Over the 2-hour observation, a total of 60 20-second child-caregiver observations were made.
|
Central to the structured observations was a global assessment of the quality of the child care setting using a widely used family of measures that vary by the age of the child and the type of setting. When children were 14 and 24 months of age, observations of center care were conducted using a slightly shortened version of the Infant-Toddler Environment Rating Scale (ITERS; Harms, Cryer, and Clifford 1990). For children at age 3, we used the Early Childhood Environment Rating Scale-Revised (ECERS-R; Harms, Clifford, and Cryer 1998). At all three age points, we observed the quality of family or home-based child care using the Family Day Care Rating Scale (FDCRS; Harms and Clifford 1989). In all settings, observers also recorded child-teacher ratios and group sizes, to obtain more reliable data than would be obtained from provider self-reports.
The observation protocol also included another frequently used global quality measure, the Arnett Caregiver Interaction Scale (Arnett 1989). Observers completed the 26-item rating scale at the end of the observation period, based on their observations of the primary caregiver’s behavior toward children in the classroom throughout the observation period.
To supplement these standard, widely used global measures of quality, we developed a new measure that provided child-level data for specific teacher-child interactions: the Child-Caregiver Observation System (C-COS; Boller, Sprachman, and the Early Head Start Research Consortium 1998). The C-COS drew upon, and included features of, two existing procedures: (1) the Observational Record of the Caregiving Environment (ORCE; NICHD Early Child Care Research Network 1997) and (2) the Adult Involvement Scale (Howes and Smith 1995). C-COS was designed to capture the experiences of individual children by time-sampling aspects of caregivers’ interactions with the Early Head Start sample child in the center classroom or family child care home. As described in Box III.1, interactions were coded during the same time period that observers were completing the ITERS, ECERS-R, FDCRS, and adult-child counts. The C-COS was collected only when children were 24 and 36 months old.
B. QUALITY OF CHILD CARE USED BY EARLY HEAD START FAMILIES
This section presents descriptive data on the quality of the child care arrangements Early Head Start children were in, both in centers and family child care homes. All observations of the quality of child care experienced by children in the research sample are based on care arrangements that were determined to be eligible for observation, as described earlier. First, we present data on quality obtained using standard measures of global quality (ITERS and ECERS-R for center care, FDCRS for family child care, Arnett Caregiver Interaction Scale in both types of settings), and child-adult ratios. Then, in the second part of this section, we present the caregiver-child interaction data from the C-COS. The third part compares the quality of care Early Head Start children received in Early Head Start centers with that received by Early Head Start children in community child care centers.
1. Quality of Child Care Used (Global Measures)
We first present analyses of the average quality of care experienced by Early Head Start program children observed in any center care across all 17 sites. The settings observed include center care provided by Early Head Start center-based and mixed-approach programs, care in community child care centers that Early Head Start programs partnered with, and care in community settings that Early Head Start parents selected on their own, without the assistance of the program.2 As reported in the evaluation’s implementation report, Pathways to Quality (ACF 2002c), Early Head Start program partnerships with community child care providers developed over time. Thus, we expected that as children got older parents would be increasingly likely to place their child in community centers that their programs had established partnerships with and were instilling and monitoring quality in line with the Head Start Program Performance Standards.
Early Head Start children in center care consistently experienced nearly good or good-quality care on average, as measured by the ITERS and ECERS-R classroom rating scales (Table III.2). Furthermore, quality improved slightly as children got older, rising from 4.7 on the ITERS at 14 months to 5.0 on the ECERS-R at 36 months, an increase of about one-quarter of a standard deviation.3 The range in average quality ratings was wide for each time period, however. ITERS scores ranged from a low of 1.5 in one classroom to 6.8 at another (at 14 and 24 months), and ECERS-R ratings ranged from 1.2 to 6.8 across center classrooms used by Early Head Start families.
This overall quality of center care that Early Head Start programs achieved for their families, regardless of the auspice providing the child care, is rare among large-scale programs. One widely cited national study of child care quality found that the average ITERS score across infant-toddler classrooms was only 3.4, or between 1 and 1.5 standard deviations below the Early Head Start averages for those ages (Cost, Quality, and Child Outcomes Study Team 1995). The National Child Care Staffing study found average quality ratings of 3.2 and 3.6 in centers serving infants and toddlers, respectively (Whitebook, Howes, and Phillips 1989), again substantially lower than what we observed in child care centers used by Early Head Start families. Included in this group of centers were Early Head Start centers, which we discuss specifically later in this chapter.
| Quality Measures | 14 Months | 24 Months | 36 Months |
|---|---|---|---|
| ITERS/ECERS-R | |||
| Averagea | 4.7 (1.1) | 4.9 (1.1) | 5.0 (1.1) |
| Rangeb | 1.5 - 6.8 | 1.6 - 6.8 | 1.2 - 6.8 |
| N | 274 | 290 | 316 |
| Arnett - Full Scale | |||
| Averagea | 3.4 (0.4) | 3.4 (0.5) | 3.4 (0.5) |
| Range(b) | 1.5 - 4.0 | 1.3 - 4.0 | 1.3 - 4.0 |
| N | 276 | 288 | 311 |
| Child-Adult Ratiosc | |||
| Averagea | 2.9 (1.2) | 3.5 (1.6) | 5.5 (2.6) |
| Range | 0.8 - 7.7 | 1.0 - 11.6 | 0.8 - 14.8 |
| N | 275 | 291 | 313 |
| Source: Based on observations in “eligible” care
arrangements, defined as care that occurs for at least 10 hours
per week outside a child’s home, or by a nonrelative in
the child’s home. Only one arrangement per child was observed.
Note: Individual observations were not conducted for all children at 14 months. Children in the same locations who were scheduled to be observed within three months of each other were assigned the same classroom characteristics. aStandard deviations in parentheses. bThe minimum possible average score on the ITERS and ECERS-R is 1.0 and the maximum possible is 7.0. The minimum possible average for the Arnett is 1.0 and the maximum is 4.0. cChild-adult ratios were recorded six times during each observation. The average presented here is the average of all nonmissing observations. |
Analysis of the ITERS and ECERS-R scale scores shows variation in the levels of quality across the dimensions rated. At both 14 and 24 months, centers were rated highest or second highest on the ITERS Interactions scale, while several other key dimensions also scored at or near the “good” level of 5.0 (Figure III.1). At both ages, Learning Activities received one of the lowest ratings (4.2 at 14 months and 4.4 at 24 months). Levels found with the ECERS-R scales differ somewhat from those of the ITERS, but the same general pattern was found (Figure III.2).
[D] |
| Source: Based on outside observations of “eligible”
care arrangements, defined as care that occurs for at least
10 hours per week outside a child’s home, or by a nonrelative
in the child’s home. Only one arrangement per child was
observed. Note: Individual observations were not conducted for all children at 14 months. Children in the same care setting who were scheduled to be observed within three months of each other were assigned the same classroom characteristics. The possible range on each subscale is 1.0 – 7.0. |
[D] |
Source: Based on outside observations of “eligible” care arrangements, defined as care that occurs for at least 10 hours per week outside a child’s home, or by a nonrelative in the child’s home. Only one arrangement per child was observed. Note: The possible range on each subscale is 1.0 – 7.0. |
The Arnett Caregiver Interaction Scale assesses the quality and content of the teacher’s interactions with children in both center and family child care settings. Two studies have found that Arnett scale scores predict teachers’ engagement with children and children’s language development and security of attachment (Helburn 1995; and Howes, Phillips, and Whitebook 1992). The quality of caregiver interactions with children, as measured by the Arnett scale, was constant across the three ages, at an average rating of 3.4 (Table III.2). The variation in these scores across classrooms was also wide, ranging from 1.3 to 4.0.4
Finally, child-adult ratios averaged 2.9 children per adult at 14 months and 3.5 to 1 at 24 months, meeting the performance standards for infants and toddlers, on average. The ratio at 36 months rose to 5.5 children per adult.5 After children turn 3, however, higher ratios might be appropriate. Even the 5.5 ratio, however, is lower than ratios found in other studies. For example, the Profile of Child Care Settings Study found an average child-staff ratio of between 6 to 1 and 7 to 1 in a nationally representative sample of centers (Kisker, Hofferth, Phillips, and Farquhar 1991).
Virtually all family child care (provided by relatives and nonrelatives) used by Early Head Start parents was found in community settings, and was not directly provided by the Early Head Start programs. Nevertheless, programs were also in the process of establishing partnerships with family child care providers throughout the evaluation period, although these partnerships were not as prevalent as ones with community centers. The evaluation observed children in these family child care settings at the same three age points, using the FDCRS. As with the centers, average FDCRS ratings rose slightly over the three time periods, from 3.4 to 3.9, but remained below the level of “good” quality, as the instrument developers describe their scale (Table III.3). The range of the quality ratings was wide for each time period, between 1.2 and 6.6.
Early Head Start children whom we were able to observe in family child care consistently experienced caregivers who were rated above 3, on average, on the Arnett Caregiver Interaction Scale. The average score was 3.2 or 3.3 for each time period, which is very close to the Arnett ratings for center teachers. The variability of Arnett ratings was less for child care homes, however: the Arnett quality ratings in family child care homes ranged from 2.0 to 4.0. Child-adult ratios averaged 4.0 or lower in all three time periods, thus meeting the Head Start performance standards for this measure.
The FDCRS also allows for analysis of quality by scales, in this case six. Across all three age points, the highest ratings of the family child care homes that Early Head Start children attended were found in the areas of Adult Needs, Supports for Social Development, and Language and Reasoning; lowest ratings (which were more than 1 point lower than the highest scales) were in Furnishings and Basic Care (Figure III.3).
| Quality Measures | 14 Months | 24 Months | 36 Months |
|---|---|---|---|
| FDCRS | |||
| Averagea | 3.4 (1.0) | 3.9 (1.2) | 3.9 (1.3) |
| Rangeb | 1.4 - 5.9 | 1.3 - 6.6 | 1.2 - 6.6 |
| N | 67 | 82 | 55 |
| Arnett - Full Scale | |||
| Averagea | 3.2 (0.5) | 3.3 (0.5) | 3.3 (0.4) |
| Rangeb | 2.2 - 4.0 | 2.0 - 4.0 | 2.1 - 4.0 |
| N | 68 | 83 | 53 |
| Child-Adult Ratiosc | |||
| Averagea | 3.2 (2.1) | 3.8 (2.2) | 4.0 (2.1) |
| Range | 0.5 - 10.8 | 0.5 - 11.0 | 0.3 - 9.5 |
| N | 67 | 83 | 54 |
| Source: Based on observations in “eligible” care
arrangements, defined as care that occurs for at least 10 hours
per week outside a child’s home, or by a nonrelative in
the child’s home. Only one arrangement per child was observed.
Note: Individual observations were not conducted for all children at 14 months. Children in the same locations who were scheduled to be observed within three months of each other were assigned the same classroom characteristics. aStandard deviations in parentheses. bThe minimum possible average score on the FDCRS is 1.0 and the maximum possible is 7.0. The minimum possible average for the Arnett is 1.0, and the maximum is 4.0. cChild-adult ratios were recorded six times during each observation. The average presented here is the average of all nonmissing observations. |
[D] |
Source: Based on outside observations of “eligible” care arrangements, defined as care that occurs for at least 10 hours per week outside a child’s home, or by a non-relative in the child’s home. Only one arrangement per child was observed. Note: Individual observations were not conducted for all children at 14 months. Children in the same care setting who were scheduled to be observed within three months of each other were assigned the same family care characteristics. The possible range on each subscale is 1.0 – 7.0. |
2. Quality of Child-Caregiver Interactions (as Measured by the C-COS)
We turn now from measures of global quality to examine child-caregiver interactions as indicators of the quality of the child care settings children were in. Four key C-COS variables are reported in this paper: (1) any caregiver talk to the child (combining responding to the child and initiating verbal interactions with the child)—these include requesting language or communication, requesting action, reading to the child, and other talking or singing; (2) caregiver responding to the child (including requesting language or communication, requesting action, reading to the child, or other talking or singing); (3) caregiver initiating talk with the child (same categories as responding); and (4) incidents of children’s negative behavior (including wandering, upset/crying, and hitting/biting/bothering another child or being hit/bothered by another child). Table III.4 shows average C-COS scores for Early Head Start children in center and family child care at 24 and 36 months of age. At 24 months of age, the frequency of caregiver talk to children was similar (about 30 incidents) for children in center and family child care, while at 36 months of age, caregiver talk was lower in centers. Incidents of caregiver talk to 36-month-old children were, in general, slightly lower than to 24-month-old children, and somewhat higher in family than center child care. Children in center care experienced an average of 26 incidents of any talk, while children in family child care experienced an average of 31.
| C-COS Variable | Center-Based Care | Family Child Care | ||
|---|---|---|---|---|
| 24 Months | 36 Months | 24 Months | 36 Months | |
| Incidents of Any Caregiver Talk to Child |
||||
| Averagea | 29.7 (12.3) | 25.8 (12.5) | 30.4 (13.2) | 31.3 (15.1) |
| Range | 1 - 60 | 0 - 60 | 3 - 60 | 6 - 60 |
| N | 297 | 323 | 90 | 65 |
| Incidents
of Caregiver Responding to Child |
||||
| Averagea | 8.5 (9.0) | 7.6 (7.2) | 8.1 (10.5) | 7.4 (7.7) |
| Range | 0 - 48 | 0 - 38 | 0 - 48 | 0 - 28 |
| N | 297 | 323 | 90 | 65 |
| Incidents
of Caregiver Initiating Talk with Child |
||||
| Averagea | 21.8 (10.8) | 18.8 (11.6) | 22.6 (11.4) | 24.2 (13.5) |
| Range | 0 - 58 | 0 - 57 | 0 - 59 | 1 - 55 |
| N | 297 | 323 | 90 | 65 |
| Incidents
of Child Negative Behavior |
||||
| Averagea | 5.5 (5.8) | 4.0 (4.8) | 5.4 (5.6) | 3.3 (4.4) |
| Range | 0 - 36 | 0 - 36 | 0 - 36 | 0 - 17 |
| N | 297 | 323 | 90 | 65 |
| Source: Based on observations “eligible” care
arrangements, defined as care that occurs for at least 10 hours
per week outside a child’s home, or by a nonrelative in
the child’s home. Only one arrangement per child was observed.
Note: Based on observations at all sites at 24 and 36 months. Possible range of number of incidents is 0 to 60 over a 2-hour observation period. Sample sizes are slightly larger than for the ITERS/ECERS and FDCRS due to fewer missing values. aStandard deviations in parentheses. |
Incidents of the caregiver responding to the child
were much lower than initiations, averaging between seven and eight
incidents per child in the 2-hour observation period. Caregivers
showed slightly more responsiveness to 24- than to 36-month-old
children, in both centers and family child care homes. Children
in family child care at 36 months experienced 24 incidents of caregiver-initiated
talk, while 36-month-old children in center care experienced about
19 incidents, on average.
We might expect more verbal interaction between child and caregiver at 36 months because children have higher verbal ability, but in general children at that age experienced fewer interactions with their caregivers in center settings. This could be due to the rising child-adult ratios or to children having more interactions with peers. Perhaps the ratio indicates the fraction of the caregiver’s attention that each child in her care can receive within a fixed time period. In family child care, however, there were slightly more incidents of any caregiver talk to the child at 36 than at 24 months.
Instances of negative child behavior were observed infrequently. On average, about the same number of incidents of negative behavior were observed for 24-month-old children in centers and family child care (5.5 and 5.4, respectively, out of a possible 60 across all the observation intervals). The range and variance in the number of incidents were also similar in the two settings. Older children displayed fewer incidents of negative behavior: 36-month-old children in child care centers displayed an average of 4.0 incidents of negative behavior, compared with 5.5 incidents for 2-year-olds. Children in family child care experienced 3.3 incidents at age 3, and 5.4 at age 2. The range and variance in negative behaviors were smaller for children in family child care than for children in centers.
Two important indicators of the quality of child-caregiver interactions at age 3 were substantially higher in family child care settings than in center care: (1) incidents of any caregiver talk to child at age 3 were 31.3 and 25.8 in family and center care settings, respectively (a difference of more than one-third of a standard deviation); and (2) incidents of the caregiver initiating talk with the Early Head Start child at age 3 were 24.2 in family care, compared to 18.8 in center care (also a difference of more than a third of a standard deviation). However, since nonresponse was so much higher for family child care than center care, it is likely that the center care we observed is more representative of all center care for Early Head Start children (both community centers and care provided by Early Head Start) than the family child care we observed is for home-based settings more generally. The family child care settings we observed are likely to be of higher quality than the in-home care provided by relatives or nonrelatives used by all Early Head Start children.
3. Quality Experienced by Early Head Start Children in Early Head Start and Community Centers
As reported in an earlier section, Early Head Start children experienced good- or nearly good-quality care averaged across all forms of center care. We further analyzed the average classroom-level quality for children in center care by contrasting the experiences of Early Head Start children in Early Head Start centers and Early Head Start children in community-based centers. The same measures were used—ITERS and ECERS-R, Arnett caregiver ratings, and child-adult ratios—when children were 14, 24, and 36 months old, and the sample includes all children whom we observed in center care across all sites. ITERS and ECERS-R scores were consistently higher, on average, for children in Early Head Start centers (Figure III.4). The average of these global quality ratings was consistently good in Early Head Start centers, ranging between 5.0 and 5.2 across the three time periods.
The largest disparity between Early Head Start and community centers appeared at 14 months, when average ITERS quality in Early Head Start centers was 5.0 and just 3.8 in community centers. The disparity narrowed by 36 months due to the improving scores for community centers—by 36 months of age, only 0.2 points on the ECERS-R separated the quality levels of Early Head Start and community child care centers (Figure III.4).
[D] |
Differences also appeared in the ITERS and ECERS-R subscales. For every subscale at every age (except for Program Structure and Language and Reasoning at 36 months), the Early Head Start-community setting differences were statistically significant (Figure III.5).
[D] |
Source: Based on outside observations of “eligible” care arrangements, defined as care that occurs for at least 10 hours per week outside a child’s home, or by a non-relative in the child’s home. Only one arrangement per child was observed. Note: Individual observations were not conducted for all children at 14 months. Children in the same care setting who were scheduled to be observed within three months of each other were assigned the same classroom characteristics. The possible range on each subscale is 1.0 – 7.0. **Difference is significant at the .05 level. |
The differences between Early Head Start and community settings on the Arnett scale were less dramatic but followed the same pattern, and only the difference at 14 months was significant (Figure III.6). Average Arnett ratings at Early Head Start centers across all three time periods were between 3.4 and 3.5. Between 14 and 36 months of age, average quality in community centers by this measure increased from 3.1 to 3.4, so that by the time children were 2 and 3 years old there was little difference between Early Head Start and community centers.
In another indication of the good quality provided by Early Head Start programs, child-adult ratios were consistently lower (that is, fewer children per adult) in Early Head Start than community child care centers (Figure III.7). Average child-adult ratios were 2.6 to 1 for Early Head Start centers enrolling 14-month-old children and 3.0 to 1 for Early Head Start 24-month-old children. Average ratios increased from 2.6 to 1 to 4.5 to 1 in Early Head Start centers between 14 and 36 months of age. In community centers, ratios increased from 3.9 to 1 when children were 14 months old to 6.1 children per adult at 36 months. As noted earlier, by 36 months, many Early Head Start children had left the program and were likely to have been in preschool classrooms where higher ratios are acceptable.
Considering specific caregiver and child behaviors and interactions as quality indicators, Figure III.8 shows the average number of incidents of three caregiver interaction behaviors and one child behavior coded on the C-COS when children were 24 and 36 months old. The patterns of better quality in Early Head Start than community centers seen with the global measures appear here also, but the differences are less pronounced. Children in Early Head Start centers experienced higher levels of any caregiver talk than children at community centers at 36 months, 28 versus 25 incidents of talk (Figure III.8a). This difference is driven by the difference in talk initiated by the caregiver (Figure III.8b), with Early Head Start children experiencing an average of about 22 incidents and children in community centers just under 18. When they were 2 years old, children across the two settings did not experience differential amounts of caregiver talk. The small differences between the number of incidents of children’s negative behavior in Early Head Start and community centers were not significant at either 24 or 36 months of age (Figures III.8c and III.8d).
In this section, we have seen multiple measures of program quality that almost unanimously demonstrate that Early Head Start children experienced higher levels of quality in centers operated by Early Head Start programs when compared with community-based centers, and higher quality in centers than in family child care settings, although we compare centers and family care cautiously because of small samples of the latter and the potential for bias in which family care settings were observed. C-COS data suggested that Early Head Start children in family child care experienced somewhat more caregiver talk than children in center care, in contrast to the global quality differences between the two modes of care. Quality in community settings, however, apparently improved somewhat as children got older. The older children became, the less difference there was in the quality of care received in Early Head Start-operated and community-based child care centers. This may be due to the efforts that Early Head Start programs expended to improve quality among community partners, or to the fact that environmental ratings are generally higher for preschoolers than for infants (Cost, Quality, and Child Outcomes Study Team 1995).
[D] |
 [D] |
[D] |
1 Appendix Tables A.1, A.2, and A.3 provide response rates by type of care and by site for the three data collection periods. (back)
2 In Section C of this chapter, we report the quality of care children experienced in Early Head Start centers. (back)
3 It should be noted, however, that because the measure of quality changed also (from the ITERS at 14 and 24 months to the ECERS-R at 36 months of age), the change in the observational instrument may have contributed to the apparent age difference found. (back)
4 We know of no standard convention in the literature to indicate the rating on the Arnett that is accepted as "good quality," in the sense that 5.0 on the ITERS, ECERS-R, and FDCRS is. However, a rating of 3 out of a possible 4 indicates the statements (such as, "speaks warmly to children") are "quite a bit" characteristic of the primary caregiver. (back)
5 The Head Start Program Performance Standards specify ratios of 4:1 for children under 3 years of age. At the 36-month age point in the Early Head Start evaluation, some children were observed when they were slightly younger than 36 months, but most were 36 months or older. Therefore, some might be subject to the performance standards for 3-year-olds, which allow 13 to 15 children for 2 adults. (back)
| Table of Contents | Previous | Next |