Table of Contents | Previous | Next |
Chapter 4: Preliminary Information on Child and Family Characteristics; Fall 2002 Data Collection
This final chapter begins with a presentation of preliminary information about the samples of Head Start and non-Head Start children and their families using information from the fall 2002 baseline parent surveys. The overall baseline response rates are presented, along with the response rates for the Head Start and non-Head Start control groups, and a discussion of the implications of these response rates for later data analysis. The chapter ends with a description of the range of preschool and child care arrangements used by children who were randomly assigned to the non-Head Start control group and for whom a parent interview was completed. These non-Head Start children will serve as the basis for calculating the impact of Head Start on child and family outcomes.
Comparison of Head Start and Non-Head Start Groups
Response Rates
Data collection for fall 2002 and spring 2003 has been completed. For the fall 2002 data collection, data entry, initial checking, and statistical weighting have all been completed. Work has also begun to assess the psychometric properties of the direct child assessments. Overall response rates for both child assessments and parent/primary caregiver interviews have been very good, with a combined rate of 80 percent for fall 2002 and a projected rate of 83 percent for spring 2003.25 However, as depicted in Exhibit 7, the fall 2002 response rate for the group of treatment families was higher than that of control families.
| Instrument | Fall Head Start (Treatment) |
2002 Non-Head Start (Control) |
|---|---|---|
| Child Assessment | 85% | 72% |
| Parent Interview | 85% | 74% |
Preliminary response rates for spring 2003 indicate a higher overall response rate than in the fall (i.e., 83% vs. 80%) and a slightly smaller differentiation between treatment and control group response rates (i.e., 88% Treatment vs. 78% Control for child assessments and 86% Treatment vs. 79% Control for the parent interviews).
Respondent Characteristics
It is important to explore whether the differential response rates for the treatment and control groups led to differences in the average characteristics of the two groups. Exhibit 3 showed that at the time of random assignment there were no differences in these two groups on the characteristics measured (child age, gender, race and parent and child language and income eligibility). As shown in Exhibit 8, when looking at fall 2002 respondents only, there is one significant difference on the same variables measured at random assignment. Specifically, families with children whose primary language is English were more likely to respond if they were in the treatment group, rather than the control group. Further data on similarities and differences between the two groups of respondents is also available from the fall 2002 parent/primary caregiver interview and these provide a greater array of characteristics. Exhibits 9 and 10 provide information from the parent/primary caregiver interview on the characteristics of the respondents (both children and their families) assigned to the Head Start and non-Head Start study groups. Data in these tables are weighted (using the final weights) to represent the national population of new Head Start applicants at all Head Start grantees/delegate agencies. (Appendix D provides the same information using unweighted data, data weighted with and without adjustments for non-response, and an analysis of the differences. This technical information is intended to help readers understand the effect of weighting, which will have important implications for later impact analyses.) As shown in Exhibits 9 and 10, the two groups are well matched on all baseline individual-level characteristics tested. Again, as was the case with the data from the random assignment rosters, the only difference found was related to language.
| Characteristic | Head Start (Treatment) Group |
Non-Head Start(Control) Group |
Difference: (Head Start) - (Non-Head Start) |
|---|---|---|---|
| Child Age Cohort: | |||
| 3-year old | 53.3% | 53.1% | 0.2% |
| 4-year-old | 46.7% | 46.8% | -0.2% |
| Child Gender: | |||
| Boys | 49.0% | 47.9% | 1.1% |
| Girls | 51.0% | 52.1% | -1.1% |
| Child Race/Ethnicity: | |||
| Hispanic | 42.1% | 44.1% | -2.0% |
| Black | 27.3% | 25.0% | 2.5% |
| White | 28.2% | 29.4% | -0.2% |
| Other | 2.3% | 2.6% | -0.3% |
| Child Language: | |||
| English | 66.3% | 62.3% | 4.0%** |
| Spanish | 31.4% | 34.2% | -2.8% |
| Other | 2.3% | 3.5% | -1.2% |
| Parent Language: | |||
| English | 68.6% | 66.7% | 1.9% |
| Spanish | 30.5% | 32.5% | -2.0% |
| Other | 1.0% | 1.0% | 0.1% |
| Head Start Income Eligibility: | |||
| Yes | 93.3% | 91.7% | 1.6% |
| No | 6.7% | 8.3% | -1.6% |
| **= p=0.05 Note: Data source: Roster information used at time of random assignment |
| Child Characteristics | Head Start (Treatment) Group |
Non-Head Start (Control) Group |
Difference (Treatment- Control) |
|---|---|---|---|
| Age on 9/1/02: | |||
| 3 years old | 60.2% | 60.8% | -0.6% |
| 4 years old | 39.8% | 39.2% | -0.6% |
| Gender: | |||
| Boys | 49.5% | 49.4% | 0.1% |
| Girls | 50.5% | 50.6% | -0.1% |
| Race/Ethnicity: | |||
| White, Non-Hispanic | 28.8% | 28.3% | 0.5% |
| Black, Non-Hispanic | 24.9% | 23.7% | 1.2% |
| Hispanic | 40.3% | 40.5% | -0.2% |
| Other | 6.1% | 7.6% | -1.5% |
| Home Language: | |||
| English | 67.7% | 64.8% | 2.9%(**) |
| Spanish | 28.9% | 30.5% | -1.6% |
| Other | 3.3% | 4.6% | -1.3% |
| Disability: | |||
| Yes | 13.8% | 12.4% | 1.3% |
| No | 86.2% | 87.6% | -1.3% |
| **= p=0.05 Note: Data source: Fall 2002 Parent Interview. |
| Maternal Characteristics | Head Start (Treatment) Group |
Non-Head Start (Control) Group |
Difference (Treatment- Control) |
|---|---|---|---|
| Age: | |||
| Under 18 years | 0.1% | 0.0% | 0.1% |
| 18-21 years | 9.2% | 9.8% | -0.6% |
| 22-25 years | 28.8% | 28.0% | 0.8% |
| 26-30 years | 28.7% | 32.6% | -3.9% |
| Over 30 years | 33.1% | 29.6% | 3.5% |
| Marital Status: | |||
| Married | 45.3% | 48.3% | -3.0% |
| Separated | 7.7% | 7.9% | -0.2% |
| Divorced | 8.5% | 7.5% | 1.0% |
| Widowed | 0.7% | 0.3% | 0.4% |
| Never Married | 37.8% | 36.0% | 1.8% |
| Race/Ethnicity: | |||
| White, Non-Hispanic | 31.0% | 32.5% | -1.4% |
| Black, Non-Hispanic | 24.6% | 24.0% | 0.6% |
| Hispanic | 38.8% | 38.6% | 0.1% |
| Other | 5.5% | 4.9% | 0.6% |
| Education Level: | |||
| No High School/GED | 34.8% | 35.6% | -0.9% |
| GED | 5.5% | 6.5% | -1.0% |
| High School Diploma | 28.8% | 27.4% | 1.4% |
| Some Postsecondary | 22.6% | 22.8% | -0.2% |
| Associate Degree | 3.7% | 4.0% | -0.3% |
| Bachelor's Degree | 3.4% | 2.7% | 0.8% |
| Graduate Degree | 1.1% | 1.0% | 0.1% |
| Employment Status: | |||
| Full-time | 32.7% | 32.7% | 0.1% |
| Part-time | 16.1% | 18.1% | -2.1% |
| Other | 51.2% | 49.2% | 2.0% |
| Monthly Household Income: | |||
| $250 or Less | 3.4% | 2.1% | 1.3% |
| $251-$500 | 8.3% | 9.0% | -0.7% |
| $501-$1,000 | 25.6% | 25.0% | 0.7% |
| $1,001-$1,500 | 24.2% | 25.2% | -0.9% |
| $1,501-$2,000 | 17.3% | 18.6% | -1.3% |
| $2,001-$2,500 | 9.9% | 7.5% | 2.4% |
| Over $2,500 | 11.2% | 12.6% | -1.4% |
| Receipt of Public Assistance: | |||
| Yes | 53.8% | 52.1% | 1.8% |
| No | 46.2% | 47.9% | -1.8% |
| Notes: (1) Data source: Fall 2002 Parent Interview; (2) There were no statistically significant differences. |
Early Care Experiences of Children Assigned to the Control Group
This section describes the preschool and child care arrangements used by children who were randomly assigned to the non-Head Start control group. It is important to emphasize that this preliminary descriptive information does not reflect the characteristics, or the preschool and child care arrangements, used by all low-income families. Because there is strong reason to believe that families applying to Head Start may be different from the overall population of low-income families, at least in terms of motivation to enroll their children in a preschool program prior to kindergarten, the data cannot speak to the arrangements used by low-income families overall. In addition, the data do not provide information about the overall availability of preschool and child care arrangements nationally or for all parts of the communities involved in the study.
Instead, the descriptive information specifically portrays the alternatives used by families that were eligible for, and applied to enroll their children in, Head Start but did not gain access to Head Start services. The alternatives ultimately chosen by these families depend on both parents’ preferences and the availability and accessibility of various options. However, this preliminary analysis does not attempt to disentangle the extent to which the observed arrangements are a function of supply versus demand, nor does it address the quality of preschool or child care arrangements used by children in the Head Start and non-Head Start groups. These considerations will be addressed in future analyses.
Data and Methods
The information reported in this final section use data collected from parents in fall 2002 regarding where their child regularly spends time Monday through Friday and who is responsible for their child during that time. This information, as described earlier, is used to explore children’s weekday arrangements in two ways: (1) the main daytime arrangement, defined as the setting where a child spends the majority of his or her time between 9am and 3pm, Monday through Friday and (2) all daytime weekday arrangements, defined as one or more non-parental arrangements used regularly between the hours of 8am and 6pm, Monday through Friday, for at least 5 hours per week (this includes the main preschool or child care arrangement).
The data are weighted to represent the population of newly entering children in communities that are not saturated with Head Start services and adjusted for survey non-response (see Chapter 2). Data are presented separately for children in the 3- and 4-year-old sample cohorts26 because (1) families of younger preschool-age children may use different preschool and child care arrangements than families of older preschool-age children, and (2) impacts will be assessed separately for 1-year versus 2-year participants in Head Start.
Children’s Care Settings: Non-Head Start Group
As shown in the first section of Exhibit 11, parent care serves as the main arrangement in fall of 2002 for about half of the children assigned to the non-Head Start group (45 to 48 percent depending on the cohort). Center-based care is the principal alternative among children who do not have parent care as their main arrangement.27 Overall, center-based care serves as the main arrangement for 39 to 40 percent of children assigned to the non-Head Start group in fall 2002. All other types of non-parental care, which include care in someone else’s home and in the child’s own home by someone other than the parent, make up the main arrangement for a total of 13 to 14 percent of children in the non-Head Start group.
| Type of Arrangement | Main Daytime Arrangement (9am-3pm) |
All Daytime Weekday Arrangement(s)* (8am-6pm) |
||
|---|---|---|---|---|
| 3-year- old cohort |
4-year-old cohort |
3-year-old cohort |
4-year-old cohort |
|
| Parent Care | 45 | 48 | N/A | N/A |
| Non-Parental Care: | 55 | 52 | 63 | 62 |
| Center | 40 | 39 | 42 | 45 |
| Non-Relative's Home | 6 | 5 | 9 | 7 |
| Relative's Home | 4 | 4 | 9 | 8 |
| Child's Home w/Relative | 4 | 4 | 7 | 7 |
| Child's Home w/Non-relative | <1 | <1 | <1 | 1 |
| * Children could be in more than one arrangement |
These preliminary descriptive analyses of the fall 2002 baseline data also indicate that the main arrangements for children not admitted to Head Start do not differ substantially by age. This is a somewhat surprising finding given that research about the use of center-based programs by 3- and 4-year-olds in population-based samples tends to show that 4-year-olds are more likely than younger children to be enrolled in center-based programs.29 Furthermore, public funding for pre-kindergarten and preschool programs is often targeted at 4-year-olds which would suggest that low-income parents of 4-year-olds might find it easier to access center-based services than parents of 3-year-olds. As a result, it was hypothesized that Head Start applicants in the 4-year-old cohort would be at least slightly more likely than the 3-year-old cohort to use a center-based alternative when they could not gain access to Head Start. However, parents of children in the 3-year-old cohort who want, but cannot access, Head Start tend to use other types of center-based care at the same rate as parents of children in the 4-year-old cohort, at least initially. Some of the differences between the pattern of preschool and child care use for the 3- and 4-year-olds in the present study, as compared to the findings from other studies, could be due to the close proximity between notification of not being able to enroll in Head Start and the fall 2002 data collection. That is, parents may not have had time to make alternative arrangements. The spring 2003 and subsequent data collection waves may show greater use of non-parental care among the non-Head Start group families.
In addition to their main arrangement, some children are enrolled in other preschool or child care arrangements during daytime hours. Moving to a consideration of all non-parental preschool or child care arrangements used between the hours of 8am and 6pm, Monday through Friday (for at least five hours per week), the second panel of Exhibit 11 shows that almost two-thirds of children in the non-Head Start group (62 to 63 percent) regularly use at least one non-parental child care or preschool arrangement. Again, the use of any center-based care is the most prevalent, with 42 to 45 percent of children using any center-based care during these hours. Use of other arrangements (i.e., care in a relative’s home, a non-relative’s home, or by a relative or non-relative in the child’s home) is lower, with seven to nine percent of children in most of these settings for at least five hours per week (only 1 percent or fewer children spend at least five hours per week cared for by a non-relative in their own home). Consequently, even when considering the use of secondary arrangements, over one-third (37 to 38 percent) of children in the control group are exclusively in parent care during daytime hours.
Geographic Variation in Children’s Settings
To determine whether the national pattern in Exhibit 11 typifies the patterns of arrangements in particular communities, the data were analyzed separately for each of the 25 geographically distinct clusters from which grantees/delegate agencies were selected for the Head Start Impact Study (see previous discussion of sample selection). These analyses show an extraordinary amount of variation in the patterns of preschool and child care arrangements from one geographic area to another. For example, the proportion of children in the non-Head Start group with parent care as their main arrangement ranges from a low of 4 percent to a high of 72 percent. The proportion of children in center-based care ranges from a low of 10 percent to a high of 93 percent.
Similar variation was observed when looking at differences in the use of any regular daytime arrangement across communities. For example, between 4 and 67 percent of children in the control group are exclusively cared for by their parents during daytime hours, and between 18 and 93 percent of children regularly spend at least 5 hours per week in a center-based program.
These findings, that national averages for the Head Start Impact Study’s non-Head Start group generally do not typify the child care and preschool patterns in particular communities, are consistent with other research that shows a great deal of variation in the use of different types of arrangements across different communities.30 As a result, the impacts of Head Start that will be reported in later publications will include an analysis of the extent to which impacts vary across this wide range of alternatives being used in different communities.
Comparison to Arrangements Used by a National Sample of Low-income Families
One of the key motivations for the Head Start Impact Study’s reliance on equivalent Head Start and non-Head Start control groups is to ensure that measured effects of Head Start are not confounded by other family characteristics that could affect children’s outcomes. For example, parents who apply to Head Start may, in general, be more motivated than other low-income parents to help their children prepare for kindergarten and more likely to seek formal out-of-home education and socialization opportunities for their children. This hypothesis, that the population of families applying to Head Start chooses different preschool and child care arrangements than the overall population of low-income families, is supported when the arrangements used by the non-Head Start group are compared to the arrangements used by all children living in poverty.
To carry out this comparison, data from the 1999 National Survey of America’s Families (NSAF) were used to estimate the use of any non-parental care, and of different types of non-parental care, for at least 5 hours per week by 3- and 4-year-olds living in poverty nationally. NSAF data differ from Head Start Impact Study data in a few key ways that likely account for some, but probably not all, of the observed differences in the arrangements used by families in the Head Start Impact Study. First, NSAF data reflect arrangements used at any time during the entire week while data for the Head Start Impact Study only reflect arrangements used, at least in part, during weekday hours. Second, the methods used to sort children by age differ across the two studies, i.e., NSAF age groupings are based on the age of the child at the time of the parent/primary caregiver interview while Head Start Impact Study age groupings are based on the number of years children were expected to participate in Head Start before entering kindergarten.
Third, data for the Head Start Impact Study were collected in 2002, while NSAF data were collected in 1999. Although the effect of the time difference on the results is difficult to predict, there is evidence that rates of use of different child care arrangements have shifted over time31 which may account for some differences in the two populations. Finally, NSAF data are representative of families living in poverty nationwide, while Head Start Impact Study data are nationally representative of the population of children applying to enroll in Head Start for the first time in communities where there are more Head Start applicants than available spaces.
Despite these differences, a comparison of the arrangements used by families in the Head Start Impact Study’s control group against the arrangements used by all low-income families not reporting Head Start use shows two important contrasts, especially for the 3-year-olds (the patterns appear largely the same among 4-year-olds in the two populations). Compared to the overall population of children living in poverty, a higher percentage of 3-year-olds in the control group are in center-based arrangements. In addition, children in the Head Start Impact Study’s 3-year-old cohort are less likely than low-income children overall to be in any relative care.
Although preliminary, these findings support the idea that families that apply to Head Start, especially when their children are 3 years old, are specifically seeking, and are more successful at finding, a center-based preschool or child care arrangement than low-income families as a whole. In other words, we expected parents who apply to Head Start to be different from other low-income parents, and these preliminary analyses support this hypothesis. This will, as discussed below, have implications for the later estimates of the impact of Head Start, and serves as a further reminder that data from the Impact Study cannot be generalized to describe all low-income children and their families.
Implications
These preliminary descriptive analyses show that most families tend to turn to one of two care alternatives in the initial few months immediately after notification that their child will not be able to enroll in Head Start. Nationally, parent care is the alternative for about half of the children whose families wanted Head Start but could not gain access to the program, with center-based programs providing the usual alternative to Head Start for children who are not in parent care. The proportion of children cared for in non-parental home-based settings by a relative or non-relative is generally low. Within and across communities, the results also show substantial variation in the proportion of control group families that use each type of arrangement.
These results make it clear that the impacts of Head Start will not be evaluated against a pure “no-services” alternative. Rather, they will be assessed against a mixture of alternatives ranging from parent care to center-based programs, of which some may look very much like Head Start and some may look very different from Head Start. Although all types of alternatives, including parent care, may offer an environment that effectively supports children’s development, parent care and center-based programs may be thought of as falling on opposite ends of a continuum in terms of the likelihood that the environment delivers a set of services and experiences that is similar to Head Start.
Based on the fall 2002 baseline data collection, children in the control group tend to be concentrated at the two ends of this continuum. Furthermore, the proportion of children using each type of arrangement varies dramatically from one community to another, with children in some communities evenly split between parent care and center-based arrangements and children in other communities more heavily concentrated in either parent care or center-based programs. This reality will complicate the interpretation of Head Start impact estimates that are based on comparing the average outcomes for children in Head Start to the average outcomes of control group children who spend their time in a wide variety of different arrangements.
Among control group children who are mainly in parental care, the analyses also show that approximately one in five children are enrolled in another arrangement for at least 5 hours per week. Because these other arrangements may offer children some of the same out-of-home enrichment opportunities as Head Start, it will be important to consider whether exposure to any group setting is related to outcomes of children in the control group. Consequently, impact analyses that attempt to understand how different alternatives to Head Start are related to variations in the impact of Head Start from one community to another will need to consider main arrangements as well as use of any type of non-parental preschool or child care. The impact analyses will also need to account for potential changes in the types of care settings individual children use, especially as other alternative care settings may become more available over time.
These preliminary descriptive data on the preschool and child care arrangements for children in the non-Head Start group also indicate that families that apply for Head Start have different patterns of preschool and child care use than the overall population of families in poverty, at least in the few months immediately after not being able to enroll their child in Head Start. In particular, it appears that families that apply to Head Start when their children are 3 years old are more likely to access center-based preschool or child care than low-income families as a whole, even when they cannot gain access to Head Start. This finding underscores the importance of conducting research on the impacts of Head Start using a rigorous random-assignment methodology.
Head Start services are delivered on the premise that a comprehensive, high quality early childhood program can make a difference in child and family outcomes. As a result, basic information about the preschool and child care arrangements of children assigned to the non-Head Start group provides important background for the impact findings to be reported later. It is likely that greater access to good quality, comprehensive early childhood programs among children in the control group will be related to smaller differences in outcomes between these children and children assigned to the Head Start treatment group. Although the preliminary results reported here do not speak to the type or quality of services received in each type of arrangement, future reports will build on these findings to understand the extent to which children in the control group are in arrangements that are similar to Head Start, in terms of both the type and the quality of the services received.
25The spring 2003 response rates have not yet been finalized.(back)
26Children were assigned to cohorts at the time they applied to Head Start based on how local program staff classified them for the purpose of making enrollment decisions. Therefore, these cohorts may not correspond exactly to the age of children at the time they would have begun Head Start.(back)
27A small number of parents of children in the non-Head Start group reported that their children were enrolled in Head Start. Preliminary data about these arrangements indicate that some of these children were enrolled by Head Start grantees selected for the study, that some of these children gained access to Head Start through grantees not selected for the study, and that some of these children were not enrolled in federally-funded Head Start program but were in a different program that parents referred to as “Head Start.” Our estimate is that about 11 percent of children assigned to the control group found their way into Head Start. For this analysis, all of these children were categorized as being in a center-based child care or preschool arrangement.(back)
28“Main Daytime Weekday Arrangement” is the one arrangement where a child spends the most time between 9am and 3pm, Monday-Friday; “All Daytime Weekday Arrangement” includes any non-parental settings regularly used between the hours of 8am and 6pm Monday-Friday for at least 5 hours per week (multiple arrangements are allowed). For children for whom parent care is not the main arrangement, data were not collected to determine how many children regularly spend at least 5 hours per week in parent care.(back)
29See: U.S. Department of Education, National Center for Education Statistics, (2002). The Condition of Education 2002.
Washington, D.C.: U.S. Government Printing Office (NCES Publication No. 2002-025). The Urban Institute, (2003), Percentage of Three- and Four-Year-Olds in Poverty in Different Types of Child Care Arrangements. Unpublished calculations based on data from the 1999 National Survey of America’s Families.(back)
30Sonenstein, Freya, Gary Gates, Stefanie Schmidt, and Natalya Bolshun.. (2003), Primary Child Care Arrangements of Employed Parents: Findings from the 1999 National Survey of America’s Families. Washington, D.C.: The Urban Institute. (Assessing the New Federalism, Occasional Paper No. 55.) O’Neil, Grace and Martin O’Connell (2001). State Estimates of Child Care Establishments: 1977 – 1997. Washington, D.C.: U.S. Census Bureau Population Division. (Working Paper Series No. 55).(back)
31See: U.S. Department of Education, National Center for Education Statistics, (2002), The Condition of Education 2002.
Washington, D.C.: U.S. Government Printing Office (NCES Publication No. 2002-025). Sonenstein, Freya, Gary Gates, Stefanie Schmidt, and Natalya Bolshun., (2003), Primary Child Care Arrangements of Employed Parents: Findings from the 1999 National Survey of America’s Families. Washington, D.C.: The Urban Institute. (Assessing the New Federalism, Occasional Paper No. 55.)(back)
| Table of Contents | Previous | Next |