Table of Contents | Previous | Next |
Chapter VII: Early Head Start Subgroup Findings
Thus far, we have examined survey data for what they can tell us about the total population of Early Head Start programs. Yet programs differ from one another. For example, a subgroup of programs operating in urban areas may be very different from those providing services in rural areas. In this chapter, we examine how subgroups differ on key characteristics, as comparing subgroups may suggest areas for planning and future research. For example, if subgroups have meaningfully different profiles, providers of training and technical assistance might plan and target support according to the varying needs that must be met. A logical next step would be to conduct research aimed specifically at understanding potential linkages between characteristics and child outcomes. This chapter describes our analytic approach to subgroup analysis, as well as differences across subgroups in their community, program, and family characteristics.
ANALYTIC APPROACH
Using survey data, we categorized programs into mutually exclusive subgroups for comparison. We then compiled variables spanning community, program, and family characteristics of interest and examined them by subgroup. This section describes the subgroups we defined and how we analyzed the data.
We generated 14 subgroups that fall into three broad categories: (1) community (service area, diversity, and change in diversity), (2) program (size, approach, Head Start affiliation, serving pregnant women, and primary caregiver education), and (3) family characteristics (demographic and psychological risks, racial/ethnic and language diversity, teenage mothers, and developmental concerns). See Table VII.1.
Determining Subgroup Differences
First, we must calculate and compare differences, and decide what size difference is large enough to be meaningful. In a study with an experimental design, or a descriptive study in which only a sample of programs are included, we would use inferential statistics to determine the significance of differences, based on the probability of achieving a difference of a given size by chance alone. Because this study covered the universe of Early Head Start programs, differences between subgroups of programs are real differences, rather than estimates, and no test of statistical significance is needed. To standardize differences between subgroups, we calculated effect size units to compare subgroups on key characteristics.
| Category | Subgroups | |
|---|---|---|
| Community Characteristics | Service Area | Urban (N= 290; 45.2 percent) Suburban (N= 61; 9.5 percent) Rural (N=269; 42.0 percent) |
| Community Diversity | High: Rated by Survey Respondents (N = 123; 18.9 percent) Lower: Rated by Survey Respondents (N = 527; 81.1 percent) |
|
| Change in Diversity (Past Five Years) | Increasing: Rated by Survey Respondents (N = 273; 42.3 percent) Stable/Decreasing: Rated by Survey Respondents (N = 372; 57.7 percent) |
|
| Program Characteristics | Program Size | Small: Fewer than 50 Families (N = 209; 31.7 percent) Medium: 51 to 100 Families (N = 271; 41.1 percent) Large: 101 to 150 Families (N = 108; 16.4 percent) Very Large: More than 150 Families (N = 72; 10.9 percent) |
| Program Approach | Home-Based (N = 114; 17.3 percent) Center-Based (N = 152; 23.0 percent) Multiple (N = 334; 50.6 percent) Combination (N = 56; 8.5 percent) |
|
| Preschool Head Start Affiliation | Head Start: (N = 532; 81.6 percent) No Head Start: (N = 120; 18.4 percent) |
|
| Pregnant Women | Serve Pregnant Women (N = 551; 84.5 percent) Do Not Serve Pregnant Women (N = 101; 15.5 percent) |
|
| Primary Caregiver Education | High Education: 50 Percent or More Primary Caregivers Have BA or More (N = 83; 17.6 percent) Lower Education: Less than 50 Percent Primary Caregivers Have BA (N = 388; 82.4 percent) |
|
| Family Characteristics | Demographic Risk | High: More than 25 Percent Enrollment with 3 or More Risks (N = 335; 52.8 percent) Lower: 25 Percent or Less Enrollment with 3 or More Risks (N = 299; 47.2 percent) |
| Psychological Risk | High: More than 25 Percent Enrollment with 2 or More Risks (N = 207; 32.6 percent) Lower: 25 Percent or Less Enrollment with 2 or More Risks (N = 429; 67.5 percent) |
|
| Enrollee Diversity | High: 50 Percent or More Racial/Ethnic Minorities (N = 439; 68.0 percent) |
|
| Language Diversity | High: 25 Percent or More Non-English Speakers (N = 199; 43.0 percent) Lower: Less than 25 Percent Non-English Speakers (N = 264; 57.0 percent) |
|
| Teenage Mothers | High: 10 Percent or More of Enrollment (N = 314; 48.5 percent) Lower: Less than 10 Percent of Enrollment (N = 334; 51.5 percent) |
|
| Developmental Concerns | High: 20 Percent or More of Enrollment (N = 276; 45.3 percent) Lower: Less than 20 Percent of Enrollment (N = 334; 54.8 percent) |
|
An effect size is a statistic that presents differences between means as a standard unit—a fraction of a standard deviation. Effect sizes are calculated as the difference between means divided by the population standard deviation for a given characteristic. In subgroups with only two levels, the effect size is the difference in their means, expressed as a fraction of a standard deviation. For subgroups with more than two levels, we present the maximum effect size.1 However, because large differences in subgroup means can fall between the highest and lowest values, we provide a reference table that shows the point differences needed to reach effect sizes of different magnitudes (Table VII.2).
We created tables listing key community, program, and enrollee characteristics, by program subgroups. The tables present means within subgroups on a given characteristic, differences between means, and effect sizes. Because the tables are long, they are placed at the end of the chapter for the reader’s convenience. Table VII.3 presents a summary of differences by key program subgroups. The complete set of tables follows. In the text, we discuss differences in the relative prevalence of characteristics across subgroups, but again, for ease of reading, we do not report means, differences, or effect sizes in the text. Because effect sizes are descriptive statistics without probabilities (p-values) associated with them, we need a guideline for deciding what size differences are practically meaningful. Therefore, for this exploratory work we highlighted differences in the text with an effect size of 0.2 or higher. We note differences of at least 0.1 if they are part of a larger pattern of differences.
Although such an analysis has benefits, it also has some limitations. Many subgroups we defined are highly intercorrelated, and some subgroups and individual characteristics have small sample sizes. In addition, all the analyses presented in this chapter are descriptive; large effect sizes indicate a relationship among variables, but they do not imply causality or the direction of the relationship.
COMMUNITY CONTEXT SUBGROUPS
As detailed in Chapter III, Early Head Start programs function within the context of their communities. Therefore, we expect that community context is related to other program features, such as program auspices, program size, employee education, and population served. In the sections that follow, we describe findings for subgroups according to community characteristics defined by service area, diversity of the community, and recent change in diversity.
Service Area (Table VII.4)
In Chapter III we divided the areas programs serve into three broad categories: urban, rural, and suburban. Programs are about evenly distributed across urban (45 percent) and rural (42 percent) areas. About 10 percent of programs are located in suburban areas, and the rest are in mixed areas. Because cultural populations in the United States vary across urban, rural, and suburban areas, it is perhaps not surprising that programs in each type of service area also differ in their degree of cultural diversity. Although suburban programs make up only a small proportion of all Early Head Start programs, they are the most diverse and are increasing in diversity relative to rural programs. Urban programs are also much more culturally diverse than rural programs and have increased in diversity in recent years.
Urban and suburban programs are more likely than rural ones to operate under the auspices of a community agency. Rural programs are the most likely to have tribal government auspices. Both urban and rural programs are more likely than those in suburban areas to be housed in university settings.
In terms of enrollment, there are fewer differences than might be expected between urban and rural programs. The main differences are between suburban and rural programs, with suburban more likely to be of medium size (enrollment of 51 to 100 children and pregnant women) and rural the most likely to be small.
Suburban programs seem to have greater community resources. They are the most likely to receive a variety of forms of outside funding. In some cases, suburban programs also differ from urban programs (such as in fundraising), but for the most part, the trend is for suburban programs to be the most likely to have various types of funding, followed by urban, and then rural (although urban-rural differences are small). An exception is that rural programs are more likely than the others to receive Part C (early intervention) funds. This could be because rural areas have fewer service providers, so agencies operating Early Head Start programs provide a wider variety of services.
Overall, most programs follow a multiple approach, with only somewhat higher prevalence of this approach for suburban programs. Rural programs are the most likely to follow a home-based model and urban programs a center-based one. This choice of model may be related in part to families’ access to transportation and their geographic distance from the program. In rural areas, families may not be able to take children to an Early Head Start center far from home, especially if public transportation is not available. In contrast, urban programs may be located closer to where families live and may be more accessible by public transportation. Suburban programs are more likely than rural ones to have access to different types of specialists. Specifically, they are likely to have specialists in male involvement, disability, health care, nutrition, mental health, and literacy. Urban programs follow closely behind suburban ones in their access to specialists; again, the trend is for suburban programs to have the highest probability, followed by urban and then rural ones.
Staff education differs by service area, although only for home visitors. Suburban programs have a greater probability than urban and rural programs of employing home visitors with an AA or higher. Rural programs are least likely to have staff with this level of education, although they are more likely to have a home-based program model in which these staff would be employed. Despite having overall better-educated staff, programs in suburban areas have higher rates than urban and rural programs of staff turnover for home visitors. Urban programs have the lowest rates of caregiver turnover compared to other program types.
There are few differences in partnerships by service area, except that rural programs are the most likely to have Part C and health care partnerships. This is consistent with the finding that rural programs are the most likely to receive Part C funding, and may indicate the importance of linking rural families to services.
Enrollee characteristics differ in expected ways by service area. Rural areas have lower minority enrollment than urban and suburban programs. Similarly, rural programs are far less likely than the other two groups to serve families whose primary language is other than English. Enrollee risks tend to be more elevated in urban programs than in rural ones. In general, urban and suburban programs serve families with more demographic risks than do rural programs. Most large differences between urban and rural programs are among single parents, mothers who receive welfare, and teen mothers, all of whom are more prevalent in urban programs. Suburban programs have a higher prevalence than rural ones of families lacking a high school credential and having multiple demographic risks. Psychological risk factors are concentrated in urban programs, which have the highest prevalence of unsafe neighborhoods and multiple risk factors. The exception is that substance abuse is far more common in rural programs (although the overall number of any programs serving many families with substance abuse issues is small).
Community Diversity (Tables VII.5 and VII.6)
We measured community diversity based on survey respondents’ ratings of the diversity of their service areas, and change over the past five years. Eighty percent of programs rate their community diversity as moderate or low; 20 percent rate it as high. However, 42 percent of programs report diversity has increased in the past five years; 56 percent report no change.
There are few differences between programs that operate in areas rated high in cultural diversity, or as increasing in diversity, except in expected ways. For example, programs with high or increasing community diversity are more likely to be in urban or suburban areas and to have higher prevalence of some risks associated with urban settings, such as many families receiving welfare and living in unsafe neighborhoods. As would be expected, these programs serve high proportions of racial/ethnic and language minority groups. Programs with high community diversity also have more management turnover than programs in communities that are less diverse.
PROGRAM CHARACTERISTICS
In this section, we describe subgroups based on program characteristics, including size, affiliation with a Head Start program, serving pregnant women, staff education, and approach to service delivery.
Program Size (Table VII.7)
We characterized programs by the size of their enrollment of children and pregnant women, creating four groups of small (32 percent), medium (41 percent), large (16 percent), and very large (11 percent) programs. In terms of community characteristics, very large programs are most likely to be in urban areas, and small programs are most likely to be in rural ones. Suburban programs fall between these extremes and are most likely to be of medium size. Similarly, larger programs tend to be in areas of at least moderate and increasing community diversity. Small programs tend to be in low-diversity areas that have not changed in recent years.
Program characteristics vary by size in that larger programs are most likely to operate out of community agencies and universities and least likely out of schools. Small programs are the most likely to have tribal government auspices. School auspices are most prevalent in small and medium-sized programs. Larger programs tend to operate in multiple sites, whereas small programs in a single site, as would be expected. Very large programs are most likely to be affiliated with a preschool Head Start. The picture is mixed when we look at types of funding programs receive; however, large and very large programs tend to be more likely than small and medium-sized programs to receive almost every type of funding. Large and very large programs are also the most likely to have a management information system (MIS).
In terms of program approach, small and medium-sized programs are most likely to have a home-based approach. Both the small and the very largest programs are most likely to be center-based. Although combination models are relatively rare, small programs are more likely to use a combination approach and least likely to use a multiple option. Small programs are most likely to choose a single program approach (center, home, or combination), while larger programs, with more staff and perhaps greater diversity of family needs, are more likely to choose more than one option (multiple approach). Larger programs tend to have greater access to specialists than medium-sized or small programs. There are especially large differences between larger and smaller programs in having male involvement specialists, nurses, and dietitians; again, these differences are to be expected given the greater resources of larger programs overall. Large programs are likely to be more complex than smaller ones, perhaps because of economies of scale that make specialists affordable. Small and medium-sized programs have the highest levels of staff education for primary caregivers and home visitors, as well as the highest rates of employee turnover (although with smaller numbers of staff, having just a few employees leave could result in a large rate of turnover). Small programs are least likely to have had turnover of the director in the previous year. Very large programs have the greatest likelihood of formal partnership with each type of provider—Part C, child care, mental and health care providers—relative to all other groups. However, the smallest programs have the highest average number of Head Start program partnerships for transitions—more than twice as many as very large programs.
Very large programs have the highest concentration of racial/ethnic minorities and people who speak languages other than English. This association may be related to the somewhat higher prevalence of very large programs in urban areas, which also have more minority residents. There is a mixed pattern of specific risks; in general, however, these are most prevalent in larger programs. Among demographic risks, large programs have the most prevalence of low enrollee education, and very large programs encounter the highest unemployment. Although the profile of individual demographic risk factors varies by program size, the difference in percentage of enrolled population at the highest risk levels (more than risk factors) does not differ by program size. All psychological risk factors, including unsafe neighborhoods, family violence, and substance use, are most prevalent in very large programs, except for mental health problems, which are most prevalent in large programs. Similarly, very large programs, on average, have the highest percentage of families with more than two psychological risk factors.
Program Service Approach (Table VII.8)
Chapter IV describes the way we characterize program models (center-based, home-based, multiple, and combination), based on services provided, with the multiple approach most prevalent. As noted earlier, center-based programs are most likely to be in urban areas; conversely, home-based programs are most likely to be in rural ones. Multiple and combination approach programs are evenly distributed across service areas. Home-based programs are most likely to be in areas of low diversity and least likely to be in those of moderate cultural diversity.
In terms of program characteristics, combination programs are the most likely to be in community agencies, and multiple programs least likely—note, however, that there are not many combination programs. Although few programs overall do so, center-based programs are most likely to have government agency auspices and multiple programs to have school auspices. Home-based programs are by far the most likely to operate in a single site, while combination programs are most likely to have multiple sites, followed closely by center-based and multiple programs. As we describe under the program size subgroups, small programs are the most likely to be combination models and the least likely to be multiple. Multiple programs are the most likely to be large or very large. Combination programs are also the most likely to operate a Head Start program; home-based are the least likely.
In general, home-based programs have lower receipt of outside funding (perhaps related to their overall smaller size). Center-based programs are the most likely to have state child care subsidies, fee-for-service, and funds from individual fundraising contributions. Programs with center-based services may use these additional funding sources to supplement Early Head Start funds to cover the cost of full-day center-based child care. Combination programs are the most likely to have other, rarer, funding sources, including contracts, grants from businesses, and other sources.
Multiple approach and center-based programs are most likely to have an MIS, in part perhaps because they also tend to be larger. Center-based programs are most likely to use these systems for producing reports on services or on individual progress.
Combination programs have access to the most specialists, including mental health, disability, literacy, speech, and health care specialists. Multiple programs have the most likelihood of a male involvement specialist, and they follow closely behind combination programs in access to mental health, disability, and health care specialists relative to home- or center-based programs. Home-based programs are least likely to have access to specialists. Combination programs and those offering a wider range of service options may need more specialists to implement these services.
Multiple approach programs have the highest likelihood of only having primary caregivers and home visitors with an AA. However, staff education among other program models is difficult to interpret, because a few programs that offered only home services also reported on education and turnover of primary caregivers; similarly, a few programs offering only center-based services reported on home visitors. Therefore, home visitor education is highest in center-based programs and primary caregiver highest in home-based programs. Home-based programs have highest rates of turnover among home visitors, although this is also a function of program approach. Multiple programs have the highest levels of management turnover and combination programs the lowest.
Center- and home-based programs are most likely to have formal partnerships with child care partners, depending on which definition of program model we use (direct services only or direct and partner-provided services). Center-based programs are most likely to have partnerships with health and mental health providers. All program models have a high likelihood of having formal Part C partnerships. Otherwise, home-based programs have the least likelihood of formal partnerships with health or mental health agencies.
Among enrollee characteristics, center-based programs have the highest prevalence of minority enrollment and home-based the lowest. This is consistent with the tendency of center-based programs to be in areas with high and increasing diversity. Combination programs are the most likely to serve families that speak a language other than English. Enrollee turnover is highest among home-based programs and lowest among center-based ones. Combination programs are more likely than the other models to have children enter the program at older ages (2 to 3 years).
Patterns of enrollee risk vary in expected ways by program approach, particularly when keeping in mind the needs these risk factors represent. For demographic risks, home-based programs are the most likely to serve families receiving welfare and those with an unemployed primary caregiver and the least likely to serve teens. Center-based programs tend to have more single and employed parents, who likely need child care. Combination programs tend to serve families without high school credentials and teen mothers. Multiple programs have the highest levels of multiple demographic risks. They also have the highest levels of many individual psychological risk factors as well as multiple psychological risks among program approaches. Multiple approach programs have higher prevalence of unsafe neighborhoods, family violence, and multiple psychological risks than the other models. Mental health problems have highest prevalence in combination and home-based programs.
Operates Preschool Head Start (Table VII.9)
Most Early Head Start programs also have an affiliation with a preschool Head Start program (82 percent). Programs that are part of agencies that also operate a preschool Head Start do not differ in community characteristics from programs that do not have such an affiliation. They are just as likely to be in urban or rural areas and areas of higher or lower diversity. Programs with a preschool Head Start are much more likely than programs without one to operate under community agency auspices. Conversely, programs without a Head Start affiliation are more likely than those with one to operate under the auspices of a university. Programs with a Head Start affiliation are more likely to have multiple sites and to have an MIS. However, they are less likely to receive some types of outside funding. In their service approaches, programs with a preschool Head Start are less likely to be home-based, but, differences are small.
In general, Head Start affiliated programs have access to more specialists overall. They particularly tend to have mental health, literacy, and speech/language specialists more so than other programs. Programs with a Head Start affiliation have lower staff turnover (both management and primary caregivers) and a higher likelihood of community partnerships with health and mental health providers. Not surprisingly, they have formal agreements with more preschool Head Start programs than do programs without such an affiliation. There are few differences in enrollment characteristics, although enrollment turnover is lower in affiliated programs. Head Start-affiliated programs serve fewer families with psychological and demographic risks, especially family violence than other programs. However, they serve more families with substance abuse issues than do those without Head Start affiliation.
Serving Pregnant Women (Table VII.10)
As we describe in Chapter III, 84 percent of programs serve pregnant women. Here we find that programs serving pregnant women do not differ in community characteristics from programs that do not serve them, although programs report increasing diversity in recent years. There are a few differences among program characteristics in that programs that serve pregnant women tend to be large or very large, but they are less likely than programs that do not serve pregnant women to have an affiliation with a preschool Head Start. Perhaps because programs serve pregnant women in their homes, they are more likely to use a home-based or a multiple approach. Programs that do not serve pregnant women are more likely than programs that do to use a center-based or combination model. Programs that serve pregnant women have greater access to all specialists, but in particular to specialists in male involvement, disability specialists, and nurses. Programs that serve pregnant women have higher primary caregiver education, but lower home visitor education than other programs. They are most likely however, to employ only primary caregivers and home visitors with at least an AA. They are also more likely to serve families that speak languages other than English. Finally, families in programs that serve pregnant women are twice as likely to have high numbers of demographic and psychological risks.
Caregiver Education (Table VII.11)
The performance standards require that at least half of primary caregivers hold an AA. Here we applied a higher standard to measure very high staff education—programs in which at least half their primary caregivers hold a BA (a relatively small group of 83 programs). We find that “high staff education” programs do not differ in community characteristics from programs with lower levels of staff education. In terms of program characteristics, high staff education programs are more likely to have a single site, be affiliated with a university, and be smaller than programs with lower staff education. Programs high in staff education are less likely to have an MIS, and less likely to use it for information on services when they do have one. These programs also tend to pursue a multiple approach and are less likely to use center-based or combination approaches. Obviously, these programs have higher overall staff education in the various ways we define it. They tend to have lower management and frontline staff turnover than programs with lower staff education. All high staff education programs have Part C partnerships, but they are less likely than programs with lower levels of staff education to have formal partnerships with health or mental health care providers. Minority enrollment tends to be lower in programs with higher staff education. Demographic risk factors are multiple, with single parents lower in prevalence among high staff education programs but higher in welfare receipt and unemployment. There are no differences in terms of psychological risk factors.
FAMILY CHARACTERISTICS
As we have stressed throughout this report, programs must, as a first step to being effective, adapt their services to meet family needs. Clearly, family characteristics are highly correlated with community characteristics. Although these analyses do not allow conclusions about the direction of influence among these factors, it is logical to think of programs as operating in and reacting to the context of families and the communities they serve. Head Start programs target those most in need and serve many families with numerous risk factors. Next we describe programs with high concentrations of families with two different types of multiple risk factors (demographic and psychological), then programs with many teenage mothers, and then those that serve many children with suspected or diagnosed developmental concerns. Finally, we describe programs with high concentrations of minority families.
Demographic and Psychological Risks (Tables VII.12 and VII.13)
Programs rated their enrollment population on the proportions they serve with each of five demographic risks: (1) single parents, (2) receiving welfare, (3) primary caregivers without a high school credential, (4) primary caregivers unemployed/not in school, and (5) teenage mothers. Programs in the high-risk group are those that serve a majority of enrollees with three or more of these risks. We refer to these programs as “serving high-risk enrollees” and to the rest as “serving lower-risk enrollees.” Programs serving enrollees with high demographic risks do not differ in community characteristics except that they are in areas of increasing diversity and are less likely to be in rural settings. They are more likely to be associated with school systems and universities and less likely to be in community agencies. Programs with high-risk enrollees have greater access to several types of specialists, including male involvement, disability, literacy speech/language, and health care specialists. They are less likely to have an MIS or to use it for reports on service or individual progress.
Programs serving high demographic risk families are more likely to pursue a multiple approach, and least likely to have only center-based services. They are more likely to have had turnover of management-level staff in the previous year. In terms of family characteristics, on average, they have a higher percentage of minority families, and, not surprisingly, they tend to have higher levels of each individual demographic, as well as psychological, risks (described below).
The profile of programs serving many enrollees with high levels of psychological risks is similar to that of programs serving high levels of families with demographic risks. As in our approach to demographic risks, we classified programs as serving enrollees at high psychological risk if at least half their enrollment had two of the following: (1) unsafe neighborhood, (2) family violence, (3) mental health problems, and (4) substance abuse. Programs serving enrollees with high psychological risk differ in a few ways from those serving lower-risk families. They are more likely to be in urban settings (although diversity does not differ). Program characteristics include lower likelihood of community agency auspices and higher of school auspices, between high- and lower-risk programs, respectively. Programs with high-risk enrollees are much more likely than those with lower-risk enrollees to have very large enrollments (and less likely to be small). They are more likely to pursue a multiple service approach, and less so to have a single service option (home- or center-based). Programs with high risk enrollees are less likely to be affiliated with a preschool Head Start. They have higher primary caregiver education, although not home visitor and they tend to have mental health specialists but are less likely to have speech/language specialists. Enrollee characteristics differ in expected ways, with programs serving high-risk enrollees having higher minority enrollment, as well much higher prevalence of all types of demographic and individual psychological risks.
Teenage Mothers (Table VII.14)
Teen mothers are not present in high concentrations across program enrollment; in about half of programs, they make up at least 10 percent of enrollment; in the other half, the figure is less than 10 percent. Programs with a higher prevalence of teen mothers differ very little from programs with a lower one. There are no community-level differences. Programs with many teenage mothers receive more government grants and Part C funding, but there are few other programmatic differences. They are less likely to be home-based, and tend to have somewhat lower staff education. As would be expected, programs with many teenage mothers also have higher levels of certain risks, including single parents and parents without a high school credential. They also have higher levels of family violence and mental health problems.
Developmental Concerns (Table VII.15)
As we describe in Chapter III, programs must set aside 10 percent of their enrollment slots for children with disabilities. Here, we examine programs that report 20 percent or more of their enrollment as having disabilities (38 percent) relative to programs with a lower prevalence. We find very few differences in programs that enroll a substantial proportion of children with developmental concerns relative to programs with lower such enrollment. Among programs with high levels of developmental concerns, we find that although they are equally likely to have a formal Part C partnership, they are somewhat more likely to have staff meetings as a part of that agreement. Staff have higher levels of education both among primary caregivers and home visitors. Enrollee characteristics of programs with high levels of children with disabilities include lower prevalence of blacks and minorities; enrollment turnover is also higher in these programs.
Program Diversity (Tables VII.16 and VII.17)
We considered several family characteristics that we describe under the rubric of diversity. These include racial/ethnic minority enrollment and speaking a primary language other than English. Race/ethnicity and primary language are highly correlated, and the profiles of programs serving many families that are racial/ethnic minorities or that speak languages other than English are similar. Programs with racial/ethnic minorities making up at least half their enrollment are “high-minority,” whereas those with fewer are “lower minority.”
High-minority programs differ from lower-minority programs in community context. Specifically, they are more likely to be in urban communities, and in neighborhoods with moderate to high and increasing cultural diversity.
High- and lower-minority programs also differ in auspices, in that high-minority programs are less likely to be in community agencies and more likely to be tribal government settings. However, they are also more likely to have very large program enrollment. High minority programs are less likely to have an MIS but are more likely to use one for tracking individual progress. These programs are less likely than lower-minority ones to use home-based or multiple approaches and more likely to use center-based or combination ones. In terms of staff turnover, high-minority programs have lower primary caregiver turnover, but higher management-level turnover. They also have less access to literacy specialists and have lower staff education. Partnerships with Part C providers are lower, but partnerships with health care providers (important for this population) are higher in high-minority programs. As expected, enrollee characteristics differ in that high-minority programs serve more non-English speakers and have increased prevalence of several demographic and psychological risk factors, such as single parent, lack of a high school credential, teen mother, unsafe neighborhood, family violence, and multiple psychological risks. Conversely, high-minority programs are lower on several risk factors as well, including employment and mental health problems.
Characteristics of programs that serve a substantial proportion of non-English-speaking enrollees (defined here as at least 25 percent of enrollment) have only a few notable differences from those serving fewer, mostly concentrated within community and family characteristics. These programs are more likely to be in urban and suburban communities than in rural ones, and to have a greater number of preschool Head Start programs in the community (also related to service area, as reported earlier). Compared with programs that have fewer non-English-speaking enrollees, programs with many non-English speakers are less likely to have an MIS, more likely to use a combination approach, and less likely to use a center-based one. Programs with many non-English speakers have lower primary caregiver turnover, and are mixed in staff education (more likely to have highly educated primary caregivers but less likely to have similarly educated home visitors) relative to their counterparts. Part C and mental health provider partnerships have lower prevalence in non-English programs; other partnerships do not differ between groups. As with race/ethnicity, programs with greater proportions of non-English speakers are also more likely to have high levels of racial/ethnic minorities. Patterns of risks are mixed, with lower levels of single parents but increased prevalence of primary caregivers without a high school credential, family violence, and unsafe neighborhoods.
SUMMARY OF KEY FINDINGS
Here, we present as a quick reference a synopsis of program characteristics within key subgroups we studied (Table VII.3). By and large, we find that programs differ in mostly expected ways when we examine subgroups of community, program, and family characteristics, and we view this consistency as a validity check on survey responses. The subgroups with the most marked differences are those related to service area, program size, and program approach, and these are summarized below. For example, urban and suburban programs differ in similar ways as larger and small programs. Among other subgroups we studied, there are few other differences, except in expected directions, that are based on the remaining program and enrollee characteristics.
| Effect Size | ||||||
|---|---|---|---|---|---|---|
| 0.1 | 0.2 | 0.3 | 0.4 | |||
| Community Characteristics | Service Area | Mainly urban | 5.0 | 10.0 | 15.0 | 20.0 |
| Mainly rural | 4.9 | 9.8 | 14.8 | 19.7 | ||
| Mainly suburban | 3.0 | 5.9 | 8.9 | 11.8 | ||
| Mixed | 1.5 | 3.0 | 4.6 | 6.1 | ||
| Other | 0.9 | 1.9 | 2.8 | 3.8 | ||
| Community Diversity | High | 3.9 | 7.8 | 11.8 | 15.7 | |
| Moderate | 4.9 | 9.9 | 14.8 | 19.7 | ||
| Low | 4.9 | 9.8 | 14.7 | 19.6 | ||
| Diversity Past Five Years | Increased | 4.9 | 9.9 | 14.8 | 19.8 | |
| Stayed the same | 5.0 | 9.9 | 14.9 | 19.9 | ||
| Decreased | 1.2 | 2.3 | 3.5 | 4.7 | ||
| Program Characteristics | Program Auspice | Community agency | 4.6 | 9.2 | 13.8 | 18.4 |
| Government agency | 2.3 | 4.7 | 7.0 | 9.3 | ||
| Tribal government | 2.0 | 4.1 | 6.1 | 8.1 | ||
| School | 3.0 | 6.0 | 8.9 | 11.9 | ||
| University | 1.8 | 3.7 | 5.5 | 7.4 | ||
| Hospital or health care provider | 1.8 | 3.6 | 5.4 | 7.2 | ||
| Other | 1.8 | 3.6 | 5.5 | 7.3 | ||
| Number of Sites | Single | 4.8 | 9.6 | 14.4 | 19.2 | |
| Multiple | 4.8 | 9.6 | 14.4 | 19.2 | ||
| Enrollment (Number of Children and Pregnant Women) | 50 or fewer | 4.7 | 9.3 | 14.0 | 18.7 | |
| 51 to 100 | 4.9 | 9.8 | 14.7 | 19.6 | ||
| 101 to 150 | 3.7 | 7.4 | 11.1 | 14.9 | ||
| 151 or more | 3.2 | 6.3 | 9.5 | 12.6 | ||
| Operates Own Preschool Head Start | 3.9 | 7.7 | 11.6 | 15.5 | ||
| Outside Funding Sources | State child care subsidies/block grant | 4.8 | 9.5 | 14.3 | 19.0 | |
| State government grant | 3.8 | 7.6 | 11.4 | 15.2 | ||
| Private foundation grants | 3.6 | 7.1 | 10.7 | 14.3 | ||
| Fundraising activities | 3.4 | 6.7 | 10.1 | 13.5 | ||
| Fee-for-service reimbursements | 2.8 | 5.6 | 8.4 | 11.2 | ||
| County/municipal government grant | 2.7 | 5.5 | 8.2 | 11.0 | ||
| Part C funds | 2.4 | 4.8 | 7.3 | 9.7 | ||
| Contracts | 2.3 | 4.6 | 6.9 | 9.3 | ||
| Grants provided by businesses | 2.2 | 4.5 | 6.7 | 9.0 | ||
| Other source | 2.5 | 4.9 | 7.4 | 9.9 | ||
| Has an MIS | 3.2 | 6.4 | 9.6 | 12.8 | ||
| Uses MIS for Reports on Services(Among Programs Using an MIS) | 3.7 | 7.4 | 11.2 | 14.9 | ||
| Uses MIS for Individual Progress Reports (Among Programs Using an MIS) | 4.8 | 9.6 | 14.4 | 19.2 | ||
| Overall Program Approach | Home-based | 3.8 | 7.5 | 11.3 | 15.1 | |
| Center-based | 4.2 | 8.4 | 12.7 | 16.9 | ||
| Multiple | 5.0 | 10.0 | 15.0 | 20.0 | ||
| Combination | 2.8 | 5.7 | 8.5 | 11.3 | ||
| Staff Characteristics | Program Employs or Has Access to: | Male involvement specialist | 5.0 | 9.9 | 14.9 | 19.8 |
| Mental health specialist | 2.1 | 4.2 | 6.2 | 8.3 | ||
| Disability specialist | 2.7 | 5.4 | 8.2 | 10.9 | ||
| Literacy specialist | 5.0 | 9.9 | 14.9 | 19.8 | ||
| Speech or language specialist | 4.8 | 9.6 | 14.3 | 19.1 | ||
| Heath care professional or nurse | 2.9 | 5.7 | 8.6 | 11.5 | ||
| Other specialist | 4.5 | 9.0 | 13.5 | 18.0 | ||
| Dietitian or nutritionist | 3.9 | 7.8 | 11.7 | 15.6 | ||
| Employs Only Primary Caregivers with at Least an AA | 3.4 | 6.8 | 10.1 | 13.5 | ||
| Employees Only Home Visitors with at Least an AA | 5.0 | 10.0 | 15.0 | 20.0 | ||
| Employs Only Primary Caregivers and Home Visitors with at Least an AA | 3.0 | 6.1 | 9.1 | 12.2 | ||
| Lost Director or Manager in Past 12 Months | 4.8 | 9.6 | 14.4 | 19.2 | ||
| Rate of Employee Turnover | Caregivers employed by program | 2.3 | 4.5 | 6.8 | 9.0 | |
| Home visitors | 2.9 | 5.7 | 8.6 | 11.4 | ||
| Program Partnerships | Has Formal Agreement with Part C Partner | 1.8 | 3.6 | 5.4 | 7.2 | |
| Part C Partnership Features | Referrals to Part C | 0.9 | 1.9 | 2.8 | 3.7 | |
| Referrals to Early Head Start | 1.3 | 2.6 | 4.0 | 5.3 | ||
| Share assessments | 1.6 | 3.3 | 4.9 | 6.5 | ||
| Staff meetings | 3.1 | 6.3 | 9.4 | 12.6 | ||
| Has Formal Agreement with Child Care Partner | 4.9 | 9.9 | 14.8 | 19.8 | ||
| Has Formal Agreement with Health Care Provider | 4.2 | 8.4 | 12.5 | 16.7 | ||
| Has Formal Agreement with Mental Health Care Provider | 3.7 | 7.5 | 11.2 | 15.0 | ||
| Number of Preschool Head Start Programs with Formal Agreement to Coordinate Transition Services | 0.7 | 1.4 | 2.1 | 2.8 | ||
| Family Characteristics | Average Enrollment Turnover | 3.0 | 6.0 | 8.9 | 11.9 | |
| Program Enrollment | Mostly white | 4.6 | 9.2 | 13.8 | 18.4 | |
| Mostly black | 4.2 | 8.5 | 12.7 | 17.0 | ||
| Mostly Hispanic | 3.9 | 7.7 | 11.6 | 15.5 | ||
| Mostly minority | 4.6 | 9.3 | 13.9 | 18.6 | ||
| Serves Any Families Speaking Primary Language Other than English | 4.5 | 9.0 | 13.5 | 18.0 | ||
| Age at Program Entry | Prenatal | 1.5 | 3.1 | 4.6 | 6.2 | |
| 0 to 2 years old | 2.0 | 3.9 | 5.9 | 7.9 | ||
| 2 to 3 years old | 1.6 | 3.2 | 4.8 | 6.4 | ||
| Primarily Serves Families with the Following Demographic Risk Factors | Single parent | 5.0 | 10.0 | 15.0 | 20.0 | |
| Receiving welfare payments | 4.5 | 8.9 | 13.4 | 17.8 | ||
| Primary caregiver does not have diploma/GED | 3.8 | 7.6 | 11.3 | 15.1 | ||
| Primary caregiver unemployed or not in school | 3.6 | 7.3 | 10.9 | 14.5 | ||
| Teen mother | 2.3 | 4.6 | 7.0 | 9.3 | ||
| More than three risk factors (above) | 4.1 | 8.1 | 12.2 | 16.3 | ||
| Primarily Serves Families with the Following Psychological Risk Factors: | Unsafe neighborhood | 4.2 | 8.4 | 12.7 | 16.9 | |
| Experience family violence | 2.6 | 5.2 | 7.8 | 10.3 | ||
| Mental health problems | 2.5 | 5.0 | 7.4 | 9.9 | ||
| Substance abuse | 1.7 | 3.5 | 5.2 | 7.0 | ||
| More than two risk factors (above) | 3.3 | 6.6 | 9.9 | 13.2 | ||
|
Source: Survey of Early Head Start Programs. a We calculated point differences needed for each effect size with the same equation we used to calculate actual effect size, this time solving for the point difference needed to achieve particular effect sizes: ESi = X/σx, where ESi is an effect size of value i, X is the point difference between subgroup means, and σx is the population standard deviation for a given characteristic, x. |
| Characteristics | Service Area | Program Size | Program Approach | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Urban | Suburban | Rural | Small | Large | Home-Based | Center-Based | Multiple | Combination | ||
| Community Characteristics | Urban | - | - | + | + | + | ||||
| Rural | + | + | - | + | + | |||||
| High/Increasing Diversity | + | + | - | - | + | |||||
| Program Characteristics | Community Agency Auspice | + | + | - | - | - | + | |||
| School Auspice | - | + | + | + | + | + | - | |||
| Multiple Sites | - | + | - | + | + | + | ||||
| Large Enrollment | - | + | + | |||||||
| Operates Preschool Head Start | - | + | + | + | ||||||
| Several Funding Sources | - | + | - | - | + | - | + | + | + | |
| Home-Based Approach | - | + | + | |||||||
| Center-Based Approach | + | + | - | + | - | |||||
| Multiple Approach | - | + | - | + | ||||||
| Combination Approach | + | |||||||||
| Access To Specialists | + | + | - | - | + | - | + | + | + | |
| High Staff Education | + | + | - | + | - | + | + | + | - | |
| High Primary Caregiver Turnover | - | + | + | - | ||||||
| High Home Visitor Turnover | - | + | + | - | + | + | ||||
| Family Characteristics | High Demographic Risk | + | - | + | - | |||||
| High Psychological Risk | + | - | - | - | - | + | ||||
| High Minority | + | + | - | - | - | + | + | + | ||
| Serve Non-English Speakers | + | + | - | - | + | - | + | |||
|
Source: Survey of Early Head Start programs. Note: For brevity, we have collapsed levels of program size subgroups and of some key characteristics. In the case of program size subgroups, +/-- indicates a difference with effect size at least 0.2 relative to the category with the highest or lowest prevalence, whether or not that category is shown (for example, “large” programs may differ at this level on a given characteristic from “medium” programs although medium is a level not shown.)
+ = Subgroup is more likely to have a characteristic (0.2 effect size or higher). – = Subgroup is less likely to have a characteristic (0.2 effect size or higher). Blank cells indicate no difference (effect size less than 0.2). |
| Characteristics | Subgroup Levels | ||||||
|---|---|---|---|---|---|---|---|
| Percentage of Programs | |||||||
| Urban (N = 290) |
Suburban (N = 61) | Rural (N = 269) |
Difference | Effect Size b | |||
| Community Characteristics | Service Area | Mainly urban | -- | -- | -- | -- | -- |
| Mainly rural | -- | -- | -- | -- | -- | ||
| Mainly suburban | -- | -- | -- | -- | -- | ||
| Mixed | -- | -- | -- | -- | -- | ||
| Community Diversity | High | 29.4 | 32.3 | 5.5 | 26.8 | 0.7 | |
| Moderate | 45.9 | 48.2 | 33.3 | 14.9 | 0.3 | ||
| Low | 24.7 | 19.5 | 61.2 | 41.7 | 0.9 | ||
| Diversity Past Five Years | Increased | 48.9 | 57.7 | 30.9 | 26.8 | 0.5 | |
| Stayed the same | 49.2 | 40.6 | 68.4 | 27.8 | 0.6 | ||
| Decreased | 2.0 | 1.7 | 0.8 | 1.2 | 0.1 | ||
| Program Characteristics | Program Auspice | Community agency | 72.3 | 79.0 | 66.1 | 12.9 | 0.3 |
| Government agency | 6.0 | 4.6 | 5.6 | 1.4 | 0.1 | ||
| Tribal government | 0.7 | 0.0 | 7.9 | 7.9 | 0.4 | ||
| School | 8.1 | 12.8 | 10.9 | 4.7 | 0.2 | ||
| University | 4.9 | 0.0 | 3.3 | 4.9 | 0.3 | ||
| Hospital or health care provider | 4.2 | 1.7 | 2.9 | 2.5 | 0.1 | ||
| Other | 3.9 | 2.0 | 3.4 | 1.9 | 0.1 | ||
| Number of Sites | |||||||