Impacts of the duration of Head Start enrollment on children's academic outcomes: moderation effects of family risk factors and earlier outcomes
Abstract
This study examined effects of the duration of Head Start enrollment on children's academic outcomes. Study questions are as follows: (a) Do Head Start children's baseline characteristics differ among those who entered at age 3 and enrolled in Head Start for 1 year, those who entered at age 4 and stayed for 1 year, and those who entered at age 3 and stayed for 2 years? (b) Do children's academic scores differ among the three groups? (c) Do these effects differ depending on the number of family risk factors and on children's scores at the beginning of the year? Findings indicate that children who entered Head Start at age 3 and stayed for 2 years showed higher academic outcomes than those enrolled later or for less time. Benefits were greater among children with more family risk factors and among those whose academic scores at the beginning of the study year were lower. © 2011 Wiley Periodicals, Inc.
Since its inception in 1965, Head Start has served children from birth to age 5, providing a comprehensive educational program for low-income children with a focus on school readiness. Ideally, participants could begin school at an equal standing with others from higher income households within their age groups and continue to develop positive outcomes later in life (Vinovskis, 2005). Positive impacts of Head Start have been examined repeatedly (Currie & Thomas, 1995; Lee, 2008; Love et al., 2005; McKey et al., 1985; Puma, Bell, Cook, Heid, & Lopez, 2006; Raikes et al., 2006; Westinghouse Learning Corporation, 1969; Zigler & Styfco, 2004). Children who attended Head Start had better outcomes in reading and math and had fewer socioemotional behavioral problems than those who did not attend the program. Despite the well-documented overall positive impact of Head Start, specific impacts of Head Start have been less known. Who receives program benefits and under what conditions? The current study examines whether duration of Head Start enrollment has an impact on Head Start children's academic outcomes. Specific study questions are as follows: (a) Do Head Start children's baseline characteristics differ among those who entered at age 3 and enrolled in Head Start for 1 year, those who entered at age 4 and stayed for 1 year, and those who entered at age 3 and stayed for 2 years? (b) Do children's academic scores differ among the three groups? (c) Do these effects differ depending on the number of family risk factors and on children's scores at the beginning of the year?
Head Start was established as an antipoverty program in the Economic Opportunity Act of 1964 (PL 88–452). Forty-six years later, in 2010, the federal government spent $7.1 billion on Head Start programs. That year, 904,153 children were enrolled in Head Start (U.S. Department of Health and Human Services, Administration for Children and Families [U.S. DHHS, ACF], 2010a), with a national average cost of $7,600 per participant. Head Start, which is a public assistance program, determines the eligibility mainly based on income level. Although it varies, 4-year-old children are given priority over 3-year-old children for admission to the program. Head Start programs also use other criteria to admit children, such as family risk factors. For example, if a child is 3 years old but lives with at-risk factors (single parent, parents with less education, etc.), then that child may be admitted into the program over those without risk factors. Currently, the Head Start bureau is trying to expand the program so that more children can enroll in the program. However, administrators do not have any evident criteria in terms of who should be given the opportunity to enroll in the program. And there is a shortage of funding for the program. Head Start determines eligibility mainly based on income level. However, only one third of eligible low-income children attend Head Start (Currie, 2001).
The demographics of Head Start enrollees have changed, especially in enrollee age composition. In 1980, before the Early Head Start program was implemented, 21% of Head Start children were age five and older, 55% were four years old, and 24% were three years old. Children under the age of three did not attend. By 2009, possibly as a result of Early Head Start, 10% of Head Start children were under three years of age, 36% were three years old, 51% were four years old, while only 3% were five and older (U.S. House of Representatives, Committee on Ways and Means, 2004). Despite the changes in age composition and shortage of funds, most evaluations of Head Start have focused on overall program impact, comparing Head Start participants with children who were not in the program. Evaluations have not focused on specific program efficiencies among Head Start participants. Recently, the Head Start has tried to increase income-eligible guidelines from 100% to 130% of parents so that more children can enroll in the program. There is a need for research-based evidence to determine who should be given the opportunity to enroll in Head Start to maximize the program benefits.
Recently, early childhood educators, social workers, psychologists, and other policy-based professionals have suggested the need for a new research agenda to analyze the optimal timing and duration of Head Start program participation. The question of dosage effects, that is, “Who receives the most benefits and under what conditions?” is of particular importance (Reynolds, 2004). Several nationally conducted Head Start evaluation studies have begun to study dosage effects. The Head Start Impact Study (Puma et al., 2006; Puma, Bell, Cook, & Heid, 2010), the National Early Head Start Research and Evaluation Study (Love et al., 2005; Raikes et al., 2006), and the Head Start Family and Child Experience Survey (FACES, Mathematica Policy Research, Inc. 2007) collected data on outcome variables and examined the impact of children's ages of entry into the program. Puma et al. (2006, 2010) found that Head Start's impact was more significant for children who entered Head Start at age 3 than those who enrolled at age 4. Parents of children who enrolled at age 3 tended to show better parenting practices than those whose children enrolled at age 4. This is particularly important, not only because these studies are among the few randomized experimental studies, but also because they examined Head Start's impact in terms of specific subgroup analysis. Lee (2008) analyzed Head Start's impact over time depending on age of entry and found that children who entered Head Start at age 3 had higher reading and math scores than children who enrolled at age 4. Although recent studies have begun to examine the impact on children of Head Start at entry age, the optimal timing and duration of Head Start enrollment has not been concurrently investigated for its impact on children's developmental academic outcomes.
The Head Start program was founded based upon ecological theory that examines human development as a complex interrelationship among self, family, and community (Bronfenbrenner, 1979). A number of scholars have studies the impact of Head Start mothers' characteristics. Helen Raikes and associates (2006) show that maternal education is related to gains in Head Start children's reading skills. Maternal verbal test scores, measured by the Armed Forces Qualification Test scores (AFQT), affected Head Start's impact on children (Currie & Thomas, 1999; Levine & Painter, 1999). Lee (2008) also found that maternal education and maternal AFQT scores moderated the effects of early age of entry into Head Start. Head Start children whose mothers had fewer years of education and lower AFQT scores tended to have more short-term gains than those whose mothers had with more years of education.
Family and contextual factors have shown their impacts on other intervention programs. “Compensatory effects,” also known as the “lost resources hypothesis,” have been documented for socioeconomically disadvantaged groups participating in intervention programs (Burchinal, Peisner-Feinberg, Bryant, & Clifford, 2000; Caughy, DiPietro, & Strobino, 1994; Lee, 2008; National Institute of Child Health and Human Development Early Child Care Research Network, 2000; Yoshikawa, 1995). Children raised in environmentally at-risk families—living in poverty, with low maternal education, or with single parents—benefited more from interventions than those who had more advantaged environments (Field, Widmayer, Stringer, & Ignatoff, 1980; Lee, 2003; Olds, & Kitzman, 1993; Ramey & Ramey, 1993; Richardson et al., 1999; Yoshikawq).
Head Start determines eligibility based mainly on family income level. Four-year-old children are usually given priority over 3-year-old children for admission to the program. Head Start programs also use other criteria. However, administrators do not have formal criteria, in terms of who should be given the opportunity to enroll in a program with a long waiting list. Moreover, with the 2007 Head Start Reauthorization Act, a new initiative, “Head Start Roadmap to Excellence and Effectiveness,” emphasized evidence-based research to raise the program quality for children and families. The present study is designed to provide evidence as to which children should enroll in Head Start to enhance program benefit, and for how long.
METHODS
Sample
The study used data collected across 60 local Head Start classrooms. Each year, approximately 1,500 Head Start children enroll in the Head Start program. The current study used data collected during 2007–2008. In 2007–2008, among 1,524 children who participated in Head Start, the 1,260 children who were 3 and 4 years old and had complete data were selected as the final sample. Of these children, 446 had entered Head Start at age 3 and enrolled in Head Start for 1 year (Group 1), 498 had entered at age 4 and stayed for 1 year (Group 2), and 316 had enrolled for 2 years (Group 3, entered at age 3 in a previous year and continued to be in the program).
Children's academic outcomes
Children's academic outcome measures (literacy, math, and science) were collected based on the Head Start and Early Childhood Program Observational Checklist. This Checklist was originally created by the local university team based on the Head Start Outcome Framework (U.S. DHHS, ACF, 2010b). Over the years, Head Start program education managers, supervisors, Head Start teachers, and outcome specialists monitored checklist items similar to other standardized assessments. The team tried to correlate the checklist indicators to those from “Creative Curriculum” (a standardized curriculum) and the Head Start Outcomes Framework (U.S. DHHS, ACF, 2010c). The teachers received initial training when the tool was introduced. Since then, they receive update training when anything was new or different. Teachers were provided a standard manual, which shows how to score the indicators. When new teachers were hired, they received personal training from the supervisor on how to complete the checklist. After training, teacher observed and reported children's developmental outcomes at the beginning of the year and at the end of the year, rating on a 4-point scale, ranging from 1 (not yet) to 4 (excels) for each item. Literacy comprises 23 items: phonological awareness (4 items), book knowledge and appreciation (5 items), print awareness and concepts (6 items), early writing (4 items), and alphabet knowledge (4 items). Mathematics comprises 14 items: numbers and operations (5 items), geometry and spatial sense (6 items), and patterns and measurement (3 items). Science comprises 9 items: scientific skills and methods (4 items) and scientific knowledge (5 items). Cronbaha's alpha from a reliability test for all 46 items was significantly high 0.98 (Literacy =.98, math=.97, and science =.96).
Number of family risk factors
The number of family risk factors was measured using indicators developed by the State Department of Education. It counts 15 items: single parent, unemployed parent, teenage parent, parental loss by divorce or death, incarcerated parent, frequent moving, non-English speaking, low parental school achievement, diagnosed family problems, food insufficiency, parental substance abuse, violent parental temperament, child's special need status, long-term illness, and abuse and neglect. Children with four or more family risk factors were classified as the high-risk group (n=426, 34%).
Baseline variables
Children's gender and ethnicity were coded during family interviews. Program characteristics (morning, afternoon, and full-day programs) was also collected and included in the analyses.
Analyses
To know the difference in baseline variables (research question 1), a one-way analysis of variance test compared baseline characteristics among those who entered at age 3 and enrolled for 1 year (Group 1), those who entered at age 4 and enrolled 1 year (Group 2) and those who entered at age 3 and stayed for 2 years (Group 3). Several steps of regression analyses were conducted to examine whether children's outcomes differed depending on the duration of enrollment. All baseline characteristics and family risk-factor variables were entered into the regression for each of the outcomes measures, followed by the entry age/duration of enrollment (Group 1 and Group 2; Group 3 was the reference group for research question 2) and the interactions between Group status and the family risk factors (research question 3). To examine whether literacy, math, and science scores at the beginning of the year has effects on these associations, children's scores measured at the beginning of the year were entered, followed by interaction effects among group status, family risk factors, and children's academic scores upon enrollment (research question 4).
RESULTS
Findings for Differences in Head Start Children's Baseline Characteristics
Baseline characteristics did not differ significantly depending on group status, except ethnicity. Black children were more likely to enroll Head Start at age 3 and attended for 2 years than children with other ethnicities. More White children tended to enter Head Start at age 4 and stay for just 1 year Table 1.
| Entry age/duration of enrollments | |||||
|---|---|---|---|---|---|
| At age 3 | At age 4 | At age 3 | |||
| Entered | 1 year | 1 year | 2 years | ||
| stadyed for | Group 1 | Group 2 | Group 3 | Total | pc |
| n (%) | 446 | 498 | 316 | 1,260 | |
| Child and family characteristics: | |||||
| Gender (% female) | 0.51 | 0.48 | 0.52 | 0.50 | ns |
| Ethnicity (%) | |||||
| Black | 0.39 | 0.31 | 0.38 | 0.35 | a,c |
| Hispanic | 0.20 | 0.17 | 0.17 | 0.18 | ns |
| White | 0.35 | 0.49 | 0.41 | 0.42 | a,c |
| Othersa | 0.06 | 0.04 | 0.04 | 0.05 | ns |
| ♯ of risk factorsb | 3.16 | 3.08 | 3.01 | 3.09 | ns |
| High risk groups (% 4 or more factors) | 0.36 | 0.34 | 0.31 | 0.34 | ns |
| Program characteristics: | |||||
| AM program | 0.33 | 0.38 | 0.39 | 0.37 | ns |
| PM program | 0.50 | 0.50 | 0.46 | 0.49 | ns |
| Full day program | 0.16 | 0.12 | 0.15 | 0.14 | ns |
| Child's outcome variables: | |||||
| Measured in fall (at the beginning of the program) | |||||
| Literacy | 32.17 | 38.79 | 46.88 | 38.76 | a,b,c |
| Math | 21.61 | 27.18 | 31.2 | 26.4 | a,b,c |
| Science | 13.94 | 16.13 | 18.61 | 16.07 | a,b,c |
| Measured in spring (at the end of the program) | |||||
| Literacy | 50.66 | 63.04 | 66.17 | 59.42 | a,b,c |
| Math | 32.57 | 40.03 | 42.29 | 37.94 | a,b,c |
| Sciencea,b,c | 20.68 | 24.75 | 25.8 | 23.57 | a,b,c |
- a aAmerican Indian, Asian Pacific Island, Asian Indian, Chinese, Cubans, Filipino, Hawaiian, Korean, Pueto Rico, and Vietnamese.
- b bSingle parent, unemployed, teenage parent, parental loss by divorce or death, incarcerated parent, frequent moving, no (limited) English speaking, low school achievement, diagnosed family problems, food insufficiency, substance abuse or addiction, violent temperament, special needs status, long-term illness, physical/sexual abuse and neglect.
- c ca=Group 1 vs Group 2; b=Group 1 vs Group 3; c=Group 2 vs Group 3.
Finding for the Effects of Duration of Enrollments on Academic Outcomes
After controlling baseline variables and family risk factors, regression analyses (Step 1 of Tables 2–4) indicated that children who had entered at age 3 and been enrolled in Head Start for 2 years (Group 3) had higher literacy (β=−.08, p<.05), math (β=−.10, p<.01), and science (β=.07, p<.10) scores than children who entered Head Start at age 4 and stayed for 1 year (Group 2). Group 3 children also had a higher scores on literacy (β=−.41, p<.001), math (β=−.40, p<.001), and science (β=−.35, p<.001) than children who had entered at age 3 and been enrolled in Head Start for 1 year (Group 1). Female children had higher literacy (β=.14, p<.001), math (β=.13, p<.001), and science (β=.09, p<.05) scores than male children. Hispanic children had lower literacy (β=−.06, p<.10), math (β=−.11, p<.01), and science (β=−.09, p<.05) scores than White children. Black children had lower math scores (β=−.09, p<.05) than White children. Children who attended full day programs had lower scores on literacy (β=−.06, p<.10) and science (β=−.06, p<.10) than those who attended half-day morning Head Start programs. No differences were found between morning and afternoon half-day Head Start program for all academic outcomes.
| Head Start children's literacy scores | |||||
|---|---|---|---|---|---|
| Step 1 | Step 2 | Step 3 | Step 4 | Step 5 | |
| Variables (Constant) | 64.91*** | 66.83*** | 31.29*** | 30.58*** | 30.57*** |
| Gender (female) | 4.79*** | 4.89*** | 0.95 | 0.94 | 0.95 |
| (0.98) | (0.98) | (0.70) | (0.69) | (0.69) | |
| [.14] | [.15] | [.03] | [.03] | [.03] | |
| Ethnicity | |||||
| Black | −1.69 | −1.57 | −0.06 | −0.23 | −0.33 |
| (1.15) | (1.15) | (0.80) | (0.80) | (0.80) | |
| [−.05] | [−.04] | [.00] | [.00] | [−.01] | |
| Hispanic | −2.75+ | −2.65+ | −1.17 | −1.20 | −1.18 |
| (1.37) | (1.37) | (0.96) | (0.95) | (0.95) | |
| [−.06] | [−.06] | [−.03] | [−.03] | [−.03] | |
| Others | 3.93+ | 4.05+ | 0.16 | 0.05 | −0.06 |
| (2.30) | (2.30) | (1.61) | (1.60) | (1.60) | |
| [.05] | [.05] | [.00] | [.00] | [.00] | |
| White | Reference group | ||||
| Program characteristics | |||||
| PM program | −0.74 | −0.75 | −1.47+ | −1.46+ | −1.43+ |
| (1.06) | (1.06) | (0.74) | (0.74) | (0.74) | |
| [−.02] | [−.02] | [−.04] | [−.04] | [−.04] | |
| Full-day program | −2.94+ | −2.91+ | −7.01*** | −7.09*** | −6.96*** |
| (1.60) | (1.60) | (1.13) | (1.12) | (1.12) | |
| [−.06] | [−.06] | [−.14] | [−.14] | [−.14] | |
| AM program | Reference group | ||||
| High risk factors+A4a | −0.69* | −1.36** | −0.48 | −0.47 | −0.46 |
| (0.29) | (0.47) | (0.33) | (0.33) | (0.33) | |
| [−.07] | [−.14] | [−.05] | [−.05] | [−.05] | |
| Duration of enrollment | −14.27*** | −14.93*** | −0.91 | −12.20*** | −10.21* |
| Entered at age 3/stayed 1 year (Group 1) | (1.25) | (2.55) | (1.84) | (3.42) | (4.38) |
| [−.41] | [−.42] | [−.03] | [−.35] | [−.29] | |
| Entered at age 4/stayed 1 year (Group 2) | −2.79* | 0.04 | 5.43** | −1.70 | 10.63* |
| (1.22) | (2.54) | (1.78) | (3.41) | (4.73) | |
| [−.08] | [.00] | [.16] | [−.05] | [.31] | |
| Entered at age 3/stayed 2 years (Group 3) | Reference group | ||||
| Interactions between | 0.20 | −0.06 | 0.48 | −0.62 | |
| Group 1×High risk factors | (0.73) | (0.51) | (0.51) | (1.06) | |
| [.02] | [−.01] | [.01] | [−.07] | ||
| Group 2×High risk factors | −0.93 | −0.38 | −0.29 | −4.04*** | |
| (0.74) | (0.52) | (0.51) | (1.12) | ||
| [−.10] | [−.04] | [−.03] | [−.44] | ||
| Scores upon enrollment (Measured in fall) | 0.91*** | 0.78*** | 0.78*** | ||
| (0.03) | (0.05) | (0.05) | |||
| [.74] | [.63] | [.63] | |||
| Interactions between | 0.23*** | 0.23+ | |||
| Group 1×Scores upon enrollment | (0.07) | (0.12) | |||
| [.28] | [.21] | ||||
| Group 2×Scores upon enrollment | 0.15* | −0.16 | |||
| (0.06) | (0.10) | ||||
| [.18] | [−.19] | ||||
| Interactions among | 0.02 | ||||
| Group 1×Risk factors×scores in fall | (0.03) | ||||
| [.08] | |||||
| Group 2×Risk factors×scores in fall | 0.10*** | ||||
| (0.03) | |||||
| [.42] | |||||
| Adjusted R | 0.17 | 0.17 | 0.60 | 0.60 | 0.61 |
- a a1=those who had 4 or more family risk factors, 0=else.
- + +p<.10; *p<.05; **p<.01; ***p<.001.
| Head Start children's math scores | |||||
|---|---|---|---|---|---|
| Step 1 | Step 2 | Step 3 | Step 4 | Step 5 | |
| Variables (Constant) | 41.69*** | 42.99*** | 18.58*** | 17.68*** | 17.67*** |
| Gender (female) | 2.65*** | 2.73*** | 0.39 | 0.35 | 0.35 |
| (0.63) | (0.63) | (0.43) | (0.43) | (0.43) | |
| [.13] | [.13] | [.02] | [.02] | [.02] | |
| Ethnicity | |||||
| Black | −1.99* | −1.92* | −0.15 | −0.20 | −0.25 |
| (0.74) | (0.74) | (0.50) | (0.50) | (0.50) | |
| [−.09] | [−.09] | [−.01] | [−.01] | [−.01] | |
| Hispanic | −2.87** | −2.85** | −0.21 | −0.22 | −0.19 |
| (0.88) | (0.88) | (0.60) | (0.60) | (0.60) | |
| [−.11] | [−.10] | [−.01] | [−.01] | [−.01] | |
| Others | 1.52 | 1.55 | 0.76 | 0.85 | 0.80 |
| (1.49) | (1.48) | (1.00) | (1.00) | (1.00) | |
| [.03] | [.03] | [.02] | [.02] | [.02] | |
| White | Reference group | ||||
| Program characteristics | |||||
| PM program | −0.31 | 0.34 | −0.78+ | −0.79+ | −0.79+ |
| (0.68) | (0.68) | (0.46) | (0.46) | (0.46) | |
| [−.01] | [−.02] | [−.04] | [−.04] | [−.04] | |
| Full-day program | −0.53 | −0.57 | −2.43*** | −2.44*** | −2.34** |
| (1.02) | (1.03) | (0.70) | (0.69) | (0.69) | |
| [−.02] | [−.02] | [−.08] | [−.08] | [−.07] | |
| AM program | Reference group | ||||
| High risk factorsa | −0.49* | −0.93** | −0.43* | −0.40+ | −0.40+ |
| (0.19) | (0.30) | (0.21) | (0.21) | (0.21) | |
| [−.08] | [−.15] | [−.07] | [−.07] | [−.07] | |
| Duration of enrollment | |||||
| Entered at age 3/stayed 1 year (Group 1) | −9.04*** | −8.34*** | 0.29 | −5.85** | −5.28+ |
| (0.80) | (1.64) | (1.14) | (2.16) | (2.68) | |
| [−.40] | [−.37] | [.01] | [−.26] | [−.24] | |
| Entered at age 4/stayed 1 year (Group 2) | −2.20** | 0.34 | 3.50** | −1.60 | 6.56** |
| (0.78) | (1.63) | (1.10) | (2.17) | (2.98) | |
| [−.10] | [.02] | [.16] | [−.07] | [.30] | |
| Entered at age 3/stayed 2 years (Group 3) | Reference group | ||||
| Interactions between | |||||
| Group 1×High risk factors | −0.24 | −0.34 | −0.23 | −0.42 | |
| 0.47 | (0.32) | (0.32) | (0.62) | ||
| [−.04] | [−.06] | [−.04] | [−.07] | ||
| Group 2×High risk factors | −0.84+ | −0.70* | −0.60+ | −2.99*** | |
| 0.47 | (0.32) | (0.32) | (0.68) | ||
| [−.14] | [−.12] | [−.10] | [−.51] | ||
| Scores upon enrollment (Measured in fall) | 0.87*** | 0.75*** | 0.75*** | ||
| (0.03) | (0.05) | (0.05) | |||
| [.76] | [.65] | [.65] | |||
| Interactions between | |||||
| Group 1×Scores upon enrollment | 0.21** | 0.19+ | |||
| 0.07 | (0.10) | ||||
| [.22] | [.19] | ||||
| Group 2×Scores upon enrollment | 0.16** | −0.13 | |||
| (0.06) | (0.10) | ||||
| [.21] | [−.18] | ||||
| Interactions among | |||||
| Group 1×Risk factors×scores in fall | 0.01 | ||||
| (0.03) | |||||
| [.04] | |||||
| Group 2×Risk factors×scores in fall | 0.09*** | ||||
| (0.02) | |||||
| [.42] | |||||
| Adjusted R | 0.16 | 0.16 | 0.62 | 0.62 | 0.63 |
- a a1=those who had 4 or more family risk factors, 0=else.
- + +p<.10; *p<.05; **p<.01; ***p<.001.
| Head Start children's science scores | |||||
|---|---|---|---|---|---|
| Step 1 | Step 2 | Step 3 | Step 4 | Step 5 | |
| Variables (Constant) | 26.03*** | 26.73*** | 13.26*** | 12.90*** | 12.89*** |
| Gender (female) | 1.15* | 1.19** | 0.44 | 0.45 | 0.45 |
| (0.38) | (0.38) | (0.28) | (0.28) | (0.28) | |
| [.09] | [.09] | [.04] | [.04] | [.04] | |
| Ethnicity | |||||
| Black | −0.53 | −0.49 | 0.36 | 0.29 | 0.27 |
| (0.45) | (0.45) | (0.33) | (0.33) | (0.33) | |
| [−.04] | [−.04] | [−.03] | [.02] | [.02] | |
| Hispanic | −1.43* | −1.41* | −0.06 | −0.11 | −0.09 |
| (0.53) | (0.53) | (0.40) | (0.40) | (0.40) | |
| [−.09] | [−.09] | [.00] | [−.01] | [−.01] | |
| Others | 0.71 | 0.72 | 0.90 | 0.83 | 0.81 |
| (0.90) | (0.90) | (0.67) | (0.66) | (0.66) | |
| [.02] | [.03] | [.03] | [.03] | [.03] | |
| White | Reference group | ||||
| Program c+A56 characteristics | |||||
| PM program | −0.57 | −0.59 | −0.73* | −0.73* | −0.71* |
| (0.41) | (0.41) | (0.31) | (0.30) | (0.30) | |
| [−.05] | [−.05] | [−.06] | [−.06] | [−.06] | |
| Full-day program | −1.04+ | −1.06+ | −2.64*** | −2.69*** | −2.66*** |
| (0.62) | (0.62) | (0.46) | (0.46) | (0.46) | |
| [−.06] | [−.06] | [−.14] | [−.15] | [−.14] | |
| AM program | Reference group | ||||
| High risk factors+A67a | −0.26* | −0.50* | −0.25+ | −0.25+ | −0.25+ |
| (0.11) | (0.18) | (0.14) | (0.14) | (0.14) | |
| [−.07] | [−.14] | [−.07] | [−.07] | [−.07] | |
| Duration of enrollment | |||||
| Entered at age 3/Stayed 1 year (Group 1) | −4.60*** | −4.36*** | −0.50 | −4.97*** | −4.26* |
| (0.49) | (1.00) | (0.75) | (1.41) | (1.82) | |
| [−.35] | [−.33] | [−.04] | [−.37] | [−.32] | |
| Entered at age 4/Stayed 1 year (Group 2) | −0.91+ | 0.37 | 1.67* | −1.45 | 5.10** |
| (0.47) | (0.99) | (0.73) | (1.40) | (1.91) | |
| [−.07] | [.03] | [.13] | [−.11] | [.40] | |
| Entered at age 3/Stayed 2 years (Group 3) | Reference group | ||||
| Interactions between | −0.84 | −0.15 | −0.13 | −0.36 | |
| Group 1×High risk factors | (0.29) | (0.21) | (0.21) | (0.43) | |
| [−.02] | [.−05] | [−.04] | [−.10] | ||
| Group 2×High risk factors | −0.42 | −0.23 | −0.20 | −2.23*** | |
| 0.29 | (0.21) | (0.21) | (0.46) | ||
| [−.12] | [−.07] | [−.06] | [−.65] | ||
| Scores upon enrollment (Measured in fall) | 0.80*** | 0.66*** | 0.65*** | ||
| (0.03) | (0.05) | (0.05) | |||
| [.68] | [.56] | [.56] | |||
| Interactions between | 0.27*** | 0.22+ | |||
| Group 1×Scores upon enrollment | (0.07) | (0.11) | |||
| [.30] | [.25] | ||||
| Group 2×Scores upon enrollment | 0.17* | −0.23* | |||
| (0.07) | (0.10) | ||||
| [.23] | [−.30] | ||||
| Interactions among | 0.02 | ||||
| Group 1×Risk factors×scores in fall | (0.03) | ||||
| [.07] | |||||
| Group 2×Risk factors×scores in fall | 0.12*** | ||||
| (0.03) | |||||
| [.60] | |||||
| Adjusted R | 0.12 | 0.12 | 0.52 | 0.53 | 0.54 |
- a a1= those who had 4 or more family risk factors, 0=else.
- + +p<.10; *p<.05; **p<.01; ***p<.001.
Findings for Effects of Risk Factors on the Associations Between Duration of Enrollments and Academic Outcomes
Children who had more than four risk factors (the higher risk group) had significantly lower literacy (β=−.07, p<.05), math (β=−.08, p<.05), and science (β=−.07, p<.05) scores than those who had three or fewer risk factors (the lower risk group). As Step 2, Table 3 indicates that interaction effects between family risk factors and group status were found on math scores (β=−.14, p<.10). As shown in Figure 1, among children with more family risk factors (4 or more), those who attended Head Start for 2 years (Group 3, 41.3) had higher math scores than those who attended for 1 year (Group 1 [34.7] and Group 2 [37.4]). However, among children who had fewer family risk factors, differences were not apparent between Group 2 (40.7) and Group 3 (40.3), although Group 2 children had higher math scores than Group 1 (36.2).
Interactions between risk factors and duration of enrollment on math scores.
Findings for the Effects of Scores Upon Enrollment on Associations Between Duration of Enrollment and Academic Developmental Outcomes
As regression analyses of Step 3 (Tables 2–4) indicated, children who had higher scores at the beginning of the year had higher literacy (β=.74, p<.001), math (β=.76, p<.001), and science (β=.68, p<.001) scores at the end of the year. Further, interaction effects between duration of enrollment and scores upon enrollment were also found (Step 4, Tables 2–4). Children with lower academic scores at the beginning of the year had higher literacy, math, and science scores at the end of the year, comparing those who had been in the program for 2 years with those in Head Start for just 1 year. Figure 2 indicates that among children with lower initial science scores (less than 16.1) at the beginning of the year, those who stayed in the program for 2 years (Group 3) had higher science scores (23.3) than those who entered at age 3 and stayed for only 1 year (Group 1, 19.9) and those who entered at age 4 and stayed for 1 year (Group 2, 22.4). In contrast, among children whose initial science scores were higher at the beginning of the year, science scores at the end of the year were similar across three groups of children.
Interactions between scores at the beginning of the year and duration of enrollment on science scores.
Similar patterns were revealed for the math scores. Among children who had lower math scores at the beginning of the year (less than 26.4), math scores measured at the end of the year were higher for children who had been enrolled for 2 years than for 1 year (31.1, 37.0, 33.7). Among children who had higher math scores upon enrollments, Group 2 and Group 3 children had similar math scores (39.8, 44.4, 44.6; Group 1, 2, and 3).
Findings for Interaction Effects Among Groups Status, Risk Factors, and Scores Upon Enrollments
As shown in regression analyses of Step 5 (Tables 2–4), the positive effects of longer enrollment duration in Head Start were greater among children with more family risk factors and lower academic scores at the beginning of the year.
Figure 3 indicates that among children with lower literacy scores at the beginning of the year (≤38.76), when they had more family risk factors, the children in Group 3 had higher literacy scores than those in Group 2 and Group 1 (47.1, 54.4, 56.7, Group 1, 2, & 3). When children had lower literacy scores upon enrollment and a lower number of family risk factors, children in Group 3 had lower literacy scores than those in Group 2 (49.2, 59.7, 55.1, Group 1, 2, and 3). Comparatively, among children with higher literacy scores upon enrollment, as shown in Figure 4, children in Group 3 had higher literacy scores than those in Group 2 and Group 1 both for children with lower (63.2, 69.7, 70.5) and higher (61.6, 68.6, 69.4, Group 1, 2, and 3) family risk factors.
Those who had lower literacy scores at the beginning of the year.
Those who had higher literacy scores at the beginning of the year.
Similarly, children's math scores in Group 3 were higher when they had been enrolled for 2 years rather than for 1 year (30.4, 30.7, 38.0) for those who had high family risk factors and lower math scores upon enrollment. The differences were much less, comparing those who had attended for 2 years and for 1 year, among children with lower math scores upon enrollment and low numbers of family risk factors (31.9, 36.8, 36.0). Math scores between those who enrolled for 1 year and 2 years were similar for children who had higher math scores at the beginning of the year regardless of their number of family risk factors (approximately 45 for all groups). The same patterns were found for science scores. Children's science scores at the end of the program were higher for children in Group 3 than those in Group 2, when they had higher number of family risk factors and when they had lower science scores at the beginning of the year.
The study did not find significant differences in family characteristics between children who had been enrolled in Head Start for 1 year and for 2 years. Children enrolled in Head Start for 2 years showed higher literacy, math, and science scores than those attending Head Start for 1 year, regardless of entry age at 3 or 4. Children with more family risk factors had lower academic developmental outcomes. However, among children with more family risk factors, when they attended Head Start for 2 years, they had higher math scores than those who had enrolled for only 1 year. Moreover, the positive effects of longer enrollment duration were greater for children with more family risk factors and for those with lower academic scores upon enrollment.
CONCLUSION
Children who enrolled in Head Start for 2 years had higher scores in literacy, math, and science than those who stayed for only 1 year. This suggests that children who enrolled Head Start for a longer duration benefited from the program directly. Building on previous studies that indicate positive effects of early entry into Head Start (Lee, 2008; Puma et al., 2006, 2010), the current study suggests that children receive greater benefits when they enter Head Start at an earlier age and stay for a longer duration. Several studies (Puma et al., 2006; Zigler, 2010; Zigler & Styfco, 2010) examined the duration of enrollment on Head Start children's outcomes and found positive effects from longer intervention (Skibbe, Connor, Morrison, & Jewkes, 2011; Spieker, Nelson, Petras, Jolley, & Banard, 2003). Lee (2008) also analyzed the effects age of entry into Head Start by controlling the duration of enrollment. No study, however, considered the “inoculation” effects of Head Start by concurrently considering duration and entry age as the main analysis variables, comparing children who enrolled Head Start at age 3 and stayed for 2 years with children who entered at age 3 and stayed for 1 year and with those who entered at age 4 and stayed for 1 year. The positive effects found for children who stayed for 2 years, compared with children who stayed for 1 year regardless of entry age (at age 3 or at age 4), suggests that both components, entry age and duration of Head Start enrollments, should be considered in determining program effects.
Family Risk Factors
Positive effects of longer Head Start enrollment were found among children who had more family risk factors. Children's math scores increased among children with higher numbers of family risk factors, if they stayed enrolled in Head Start for 2 years. Consistent with the previous study findings (Lee, 2008; Zigler, 2010; Zigler & Styfco, 2004), this may be because of the “compensate dosage effects” of Head Start for children with many family risk factors, such as having a single parent, unemployed parents, teenage parent, parental loss by divorce or death, incarcerated parent, frequent moving, and others. Families with more risk factors have more difficulty providing nurturing home environments for their children. This, in turn, negatively affects their children's development (Bradley et al., 1989; Brooks-Gunn, Klebanov, Smith, Duncan, & Lee, 2003; Lee, 2009). Thus, when children facing more family risk factors receive a longer “dosage” of Head Start, their marginal benefits may be greater than those with less family risk. However, these dosage effects were not found for literacy and science scores. As indicated in the previous study (Lee, 2008), the question of whether “dosage” effects are possibly associated with the each component of quality of Head Start's curriculum should be examined further (Raver & Zigler, 1997; Zigler & Bishop-Josef, 2006).
Baseline Academic Scores
Compensatory dosage effects were greater for the children with lower academic scores at the beginning of the year. Children who had lower reading, math, and science scores at the beginning of the year had higher scores at the end of the year if the children had attended Head Start for 2 years rather than for 1 year. This difference was greater for children with more family risk factors. As suggested by Reynolds, Mavrogenes, Bezruczko, and Hagemann (1996), cognitive gains obtained by Head Start participants by the time they leave the program should continuously influence children's outcomes later in life (Consortium for Longitudinal Studies, 1983; Reynolds, 2004; Schweinhart, Barnes, & Weikart, 1993). Despite the importance of cognitive gains, there has been little discussion of how children obtain solid “cognitive gains” that are sustainable through life. Just participating in Head Start for 1 year might not be sufficient intervention for low-income children who face many cumulative risk factors.
Substantial numbers of children with low academic scores upon enrollment and with four or more family risk factors had been in the Head Start in the previous year. For literacy scores, 9% of the 316 children with both low literacy scores and high numbers of risk factors were returning students. These children's academic outcomes improved after they completed a second year in Head Start, compared with children who had only 1 year of Head Start. The 1-year enrollment (in a half day program of 3 and 1/2 hours per day for 4 days a week) was not sufficient, particularly for children experiencing many risk factors. Head Start's impact may be cumulative and dormant, with some sleeper effects appearing later. Although the quality of Head Start varies, studies indicate that Head Start is higher quality relative to typical childcare programs. Head Start provides comprehensive educational, nutritional, and health interventions for children and families. Children with many family risk factors who were not enrolled in Head Start would not receive comprehensive care in other childcare programs or by staying at home. Comparatively, Head Start children's learning experiences during the year when the child was 3 years old may be necessary to acquire program benefits for the 4-year-old child in Head Start. Learning environments can have cumulative effects over time. A targeted program with a longer duration (at least 2 years) is necessary to achieve the full program benefits for more disadvantaged families. With a longer program, disadvantaged participants can obtain solid and stable developmental outcomes.
Despite the positive effects of a longer duration of Head Start, children who had better academic skills at the beginning of the year gained fewer additional benefits. Most Head Start classrooms consist of multiage groups and mixed groups of beginners and children returning to Head Start. The Head Start curriculum follows the National Association of the Education of Young Children (NAEYC) guidelines, with a strong emphasis on developmentally appropriate practice. Within this framework, the curriculum should also include more challenging tasks for children who have already achieved the performance standard. A more individualized curriculum could be provided to children who need next level of knowledge (Bloom, 1956). Further, the annual Head Start evaluation should not focus too much on evaluating the compliance or the noncompliance of minimum performance standard, laws, and regulations. Even the 2007 Head Start reauthorization act (Public Law 110–134), “Improving Head Start for School Readiness Act of 2007,” expands the targeted enrollment population and the increases the qualifications of staff, but does not specifically allocate the necessary funding to improve curriculum quality. Should Head Start set curriculum as a compensatory educational programs focusing on risk factors? Or should it provide a high-quality intervention program, developing children's resilience?
Other Baseline Characteristic Variables
Other baseline variables also affected Head Start children's academic developmental outcomes. Consistent with other Head Start impact studies (Lee, 2008; Love et al., 2005; Puma et al., 2006, 2010), girls scored higher on reading and math than did boys. How the gender differences in academic outcomes develop over time should be examined further. Hispanic Head Start children had lower scores on reading and math than non-Hispanic White Head Start children. Although Head Start programs included bilingual Spanish-speaking staff, the entire curriculum was not in Spanish. Possibly, some Spanish-speaking children faced language benefits or other cultural barriers. Children who attended full-day Head Start programs had lower literacy, math, and science scores than those who attended half-day morning Head Start programs. Most Head Start classrooms offered half-day programs. Those who need a full-day care were not able to stay one program but have to move to other schools. Not all programs offered full time programs. Children whose parents could not meet TANF work requirements were pulled out of the program due to lack of state financial support. These children returned to program when their parents went back to work, but missed continuity. Thus, most children who attended full-day Head Start programs did not have a stable childcare setting. The stability of childcare arrangements affects children's cognitive development (Morrissey, 2009; Moss & Brannen, 1987; Tran & Weinraub, 2006). Children of working families or who need a longer care should be able to enroll in a full-day, year-round Head Start program.
Limitations
The current study examined only children who participated in Head Start programs. It did not have a matched control group, eligible but not participating in Head Start. Although several baseline variables were included in the analyses, as well as initial literacy, math, and science scores, the causality of Head Start enrollment should not be inferred. The Behavior Observation Checklist was measured by teacher's reports and based on teachers' observation. Children's outcomes measured by other standardized measures would provide more comparable analyses. The current study examined a 1-year cohort of children and used only three indicators of academic outcome. A study that includes more cohort years and examines more comprehensive outcome measures, including health and nutrition, would be important as a next step.
Contributions/Implications
These results support the recommendation that Head Start be more targeted, providing a longer duration of enrollment and accepting younger children. Without sufficient funding, the recent increase in income eligibility can leave the most vulnerable at-risk children out of the program. Target efficiency—who should be enrolled in the Head Start program to maximize Head Start program benefits—is important. The Early Head Start program admits children under the age of 3, but has limited spaces. Based on this research, Early Head Start programs should be expanded and coordinated with regular Head Start programs so children can participate in the program from a younger age and can stay longer. When funding is limited, the children who need Head Start most should have priority. Head Start should provide more full-day programs, providing stable childcare arrangements for families facing TANF work requirements. Most important, this study supports that Head Start should accept younger children and have them stay longer in the program. Enrollment criteria should consider family risk factors in addition to the child's age, even among low-income families. Then, Head start provides enough program duration for those who needs the program the most. Head Start continues to make unique contributions to enhance children's developmental outcomes, especially for children with many at-risk factors, by providing comprehensive early intervention programs.
