Abstract
Two-generation human capital programs seek to promote the education of parents and children simultaneously. This study examines relations between family participation in CareerAdvance, which recruits parents of Head Start children into a workforce training program, and children’s Head Start attendance. The sample included 293 children (on average 4 years old) and their parents. After one semester, CareerAdvance children demonstrated higher rates of attendance and lower rates of absence and chronic absence (missing 10% or more of school days) than matched comparison children. These associations were similar across a range of high- and low-risk subgroups at baseline. These findings are discussed in terms of the implications of a family systems approach for improving children’s Head Start attendance.
Original language | English (US) |
---|---|
Pages (from-to) | 1916-1933 |
Number of pages | 18 |
Journal | Child development |
Volume | 91 |
Issue number | 6 |
DOIs | |
State | Published - Nov 1 2020 |
ASJC Scopus subject areas
- Pediatrics, Perinatology, and Child Health
- Education
- Developmental and Educational Psychology
Access to Document
Other files and links
Fingerprint
Dive into the research topics of 'A Two-Generation Education Intervention and Children’s Attendance in Head Start'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
In: Child development, Vol. 91, No. 6, 01.11.2020, p. 1916-1933.
Research output: Contribution to journal › Article › peer-review
}
TY - JOUR
T1 - A Two-Generation Education Intervention and Children’s Attendance in Head Start
AU - Sommer, Teresa Eckrich
AU - Schneider, William
AU - Chor, Elise
AU - Sabol, Terri J.
AU - Chase-Lansdale, P. Lindsay
AU - Brooks-Gunn, Jeanne
AU - Yoshikawa, Hirokazu
AU - Morris, Amanda
AU - King, Christopher
N1 - Funding Information: We are grateful to the many program, policy, and research partners who made this work possible, especially the Community Action Project of Tulsa, Oklahoma (Steven Dow, Executive Director), the Administration for Children and Families (Hilary Bruck, Program Officer), and Ascend at the Aspen Institute (Anne Mosle, Executive Director). We also acknowledge Marisa Sclafani for her assistance in manuscript preparation. This research was supported by the Health Profession Opportunity Grant (HPOG; #90FX00100) and the Health Profession Opportunity Grant‐University Partnership Research Grant (HPOG‐UP; #90PH0020) from the Administration for Children and Families, United States Department of Health and Human Services; the W.K. Kellogg Foundation (Grant #P3020014); and the Foundation for Child Development (Grant #Northwestern 06‐2014). Funding Information: This study draws on data from the CAP Family Life Study, a longitudinal, quasi-experimental study that is designed to compare outcomes of parent–child pairs enrolled in CareerAdvance with a set of matched comparison parents and children over time. Parents learned about CareerAdvance through flyers distributed across CAP Tulsa’s Head Start programs, direct program promotion by Head Start family support staff, and informational meetings led by career coaches at CAP Tulsa Head Start programs throughout Tulsa. Any parent with a child enrolled in a CAP Tulsa Head Start Center was eligible to apply for the family’s enrollment in CareerAdvance. Additional eligibility requirements at application included an interview led by CareerAdvance coaches, background checks, health status, drug testing, absence of a financial or academic hold at the local community college, and English proficiency. Interview criteria included motivation and commitment to school, interest in the health care field, and schedule availability. Based on this process, 317 parents applied to the program, 221 were accepted, and 162 parents enrolled in the program and consented to participate in the CAP Family Life Study. The full CAP Family Life Study included 338 children and their parents (162 CareerAdvance and 176 matched comparison). CareerAdvance parents were enrolled in an education and workforce training program, whereas their children were enrolled in Head Start. The matched comparison group had children enrolled in Head Start only. For our analysis, we required a minimum of 5 months of Head Start data for all sample children during the first 5 months of parents’ enrollment in the career training program. Our sample was reduced by 42 children who lacked valid attendance data and three children who lacked attendance data within the specified time frame. Our final analytic sample included 293 children and their parents (119 CareerAdvance and 174 matched comparison). We compared families in our analytic sample to those who were excluded (see Table S1). First, we estimated the difference in means for a number of sociodemographic indicators between parents and children who were and were not in the analytic sample. Overall, families in our analytic sample (n = 293) have similar characteristics to the full sample (n = 338; the only exception was a slight difference in the proportion of White children between the two samples). Second, we compared differences in means for parents and children who were and were not in the analytic sample by treatment status. Findings from this analysis largely do not reveal evidence of differential attrition. The only exception is children whose parents were in the CareerAdvance group were more likely to leave CAP Tulsa Head Start at age 4, when they became eligible for Tulsa Public Schools, compared to children in the matched comparison group, and thus were not included in the analytic sample. As a result, there is incomplete data during the 5-month window with a greater number of children whose parents were in CareerAdvance compared to the matched comparison. To that end, in all models we include a control for whether the child was age 4 or older and an interaction between child age 4 or older and CareerAdvance participation. We also estimated upper and lower bounds of the magnitude of the association between CareerAdvance participation and child attendance under different assumptions about the pattern of missingness. We were unable to conduct a randomized control trial to test the effect of CareerAdvancegiven that the program was new at the time of the study and parental interest in the program did not exceed supply. About 20–30 parents per cohort enrolled in the program across seven cohorts from 2011 to 2014, and there was no waiting list during this time. Instead, we selected a matched comparison parent for enrollees as each was recruited and enrolled in CareerAdvance. To create the matched comparison group, we drew on administrative Head Start data (i.e., ChildPlus) and selected parents from the full pool of CAP Tulsa Head Start parents (N = 4,985). Parents were included in the matched comparison group based on their observed similarity to CareerAdvance parents, including neighborhood residence, race, gender, age, education, relationship with child, household income, English proficiency, single parent status, and foster parent status, as measured by a propensity score (i.e., the likelihood of participating in CareerAdvance). All parents (in both the CareerAdvance and matched comparison groups) were recruited from CAP Tulsa’s Head Start programs, and thus the difference is that the CareerAdvance group enrolled in an education and training program in addition to Head Start services, whereas parents in the matched comparison group had children enrolled in Head Start only. Propensity score estimation also drew on a measure of motivation and interest in health care training. To measure motivation, CAP Tulsa family support specialists administered a brief survey to all Head Start parents in the fall of each school year to assess parents’ level of interest in pursuing education or job training for the purposes of employment in health care using a 5-point scale. For the CareerAdvance group, we used an equivalent 5-point scale administered during the program interview selection process. This study involves three primary data sources: (a) Head Start administrative data on child attendance and family background characteristics from ChildPlus, part of the federal program’s management information system; (b) a questionnaire for all CAP Tulsa Head Start parents regarding their interest in educational and career advancement in the health care sector (i.e., motivation questionnaire); and (c) in-person parent surveys. Head Start administrative data from ChildPlus were collected by CAP Tulsa Head Start from all new applicants to CAP Tulsa’s Head Start programs at the time ofenrollment. ChildPlus data are a requirement from the Administration for Children and Families and include parent reports of family demographic characteristics and program reports of children’s daily attendance. Once parents consented to the study, they were interviewed by master’s- and doctoral-level research assistants. Structured surveys took place in Head Start centers or in the home, depending on parents’ preference, and lasted approximately 75–90 min. On average, Wave 1 surveys were administered 1.08 months after the first partner meeting (SD = 1.57). The Wave 1 surveys took place in either the fall or winter of the year due to the staggered approach to enrollment, where some cohorts (approximately one-half) began in August and other cohorts began in January. This study draws on data from the CAP Family Life Study, a longitudinal, quasi-experimental study that is designed to compare outcomes of parent–child pairs enrolled in CareerAdvance with a set of matched comparison parents and children over time. Parents learned about CareerAdvance through flyers distributed across CAP Tulsa’s Head Start programs, direct program promotion by Head Start family support staff, and informational meetings led by career coaches at CAP Tulsa Head Start programs throughout Tulsa. Any parent with a child enrolled in a CAP Tulsa Head Start Center was eligible to apply for the family’s enrollment in CareerAdvance. Additional eligibility requirements at application included an interview led by CareerAdvance coaches, background checks, health status, drug testing, absence of a financial or academic hold at the local community college, and English proficiency. Interview criteria included motivation and commitment to school, interest in the health care field, and schedule availability. Based on this process, 317 parents applied to the program, 221 were accepted, and 162 parents enrolled in the program and consented to participate in the CAP Family Life Study. The full CAP Family Life Study included 338 children and their parents (162 CareerAdvance and 176 matched comparison). CareerAdvance parents were enrolled in an education and workforce training program, whereas their children were enrolled in Head Start. The matched comparison group had children enrolled in Head Start only. For our analysis, we required a minimum of 5 months of Head Start data for all sample children during the first 5 months of parents’ enrollment in the career training program. Our sample was reduced by 42 children who lacked valid attendance data and three children who lacked attendance data within the specified time frame. Our final analytic sample included 293 children and their parents (119 CareerAdvance and 174 matched comparison). We compared families in our analytic sample to those who were excluded (see Table S1). First, we estimated the difference in means for a number of sociodemographic indicators between parents and children who were and were not in the analytic sample. Overall, families in our analytic sample (n = 293) have similar characteristics to the full sample (n = 338; the only exception was a slight difference in the proportion of White children between the two samples). Second, we compared differences in means for parents and children who were and were not in the analytic sample by treatment status. Findings from this analysis largely do not reveal evidence of differential attrition. The only exception is children whose parents were in the CareerAdvance group were more likely to leave CAP Tulsa Head Start at age 4, when they became eligible for Tulsa Public Schools, compared to children in the matched comparison group, and thus were not included in the analytic sample. As a result, there is incomplete data during the 5-month window with a greater number of children whose parents were in CareerAdvance compared to the matched comparison. To that end, in all models we include a control for whether the child was age 4 or older and an interaction between child age 4 or older and CareerAdvance participation. We also estimated upper and lower bounds of the magnitude of the association between CareerAdvance participation and child attendance under different assumptions about the pattern of missingness. We were unable to conduct a randomized control trial to test the effect of CareerAdvancegiven that the program was new at the time of the study and parental interest in the program did not exceed supply. About 20–30 parents per cohort enrolled in the program across seven cohorts from 2011 to 2014, and there was no waiting list during this time. Instead, we selected a matched comparison parent for enrollees as each was recruited and enrolled in CareerAdvance. To create the matched comparison group, we drew on administrative Head Start data (i.e., ChildPlus) and selected parents from the full pool of CAP Tulsa Head Start parents (N = 4,985). Parents were included in the matched comparison group based on their observed similarity to CareerAdvance parents, including neighborhood residence, race, gender, age, education, relationship with child, household income, English proficiency, single parent status, and foster parent status, as measured by a propensity score (i.e., the likelihood of participating in CareerAdvance). All parents (in both the CareerAdvance and matched comparison groups) were recruited from CAP Tulsa’s Head Start programs, and thus the difference is that the CareerAdvance group enrolled in an education and training program in addition to Head Start services, whereas parents in the matched comparison group had children enrolled in Head Start only. Propensity score estimation also drew on a measure of motivation and interest in health care training. To measure motivation, CAP Tulsa family support specialists administered a brief survey to all Head Start parents in the fall of each school year to assess parents’ level of interest in pursuing education or job training for the purposes of employment in health care using a 5-point scale. For the CareerAdvance group, we used an equivalent 5-point scale administered during the program interview selection process. This study involves three primary data sources: (a) Head Start administrative data on child attendance and family background characteristics from ChildPlus, part of the federal program’s management information system; (b) a questionnaire for all CAP Tulsa Head Start parents regarding their interest in educational and career advancement in the health care sector (i.e., motivation questionnaire); and (c) in-person parent surveys. Head Start administrative data from ChildPlus were collected by CAP Tulsa Head Start from all new applicants to CAP Tulsa’s Head Start programs at the time ofenrollment. ChildPlus data are a requirement from the Administration for Children and Families and include parent reports of family demographic characteristics and program reports of children’s daily attendance. Once parents consented to the study, they were interviewed by master’s- and doctoral-level research assistants. Structured surveys took place in Head Start centers or in the home, depending on parents’ preference, and lasted approximately 75–90 min. On average, Wave 1 surveys were administered 1.08 months after the first partner meeting (SD = 1.57). The Wave 1 surveys took place in either the fall or winter of the year due to the staggered approach to enrollment, where some cohorts (approximately one-half) began in August and other cohorts began in January. Table 1 presents the baseline characteristics of our full analytic sample (n = 293) and the CareerAdvance and matched comparison groups separately. All families in the study were low-income due to Head Start requirements. The average annual household income at baseline was $14,342.69 (SD = 12,079.62). Forty-eight percent of children were female, and the full sample of children had a mean age of 46.21 months (3.85 years; SD = 12.27). The child sample was racially and ethnically diverse: 41% Black, 18% White, 10% Hispanic, and 31% non-Hispanic other. Ninety percent of parents spoke English as their primary language. Nearly one-third of children (32%) had a single parent and the average household size was approximately four individuals (SD = 1.43). Fifty-two percent of parents in the sample had a high school degree, general education diploma (GED), or less, and 49% of sample parents were employed at baseline. Table 1 also indicates that that there are no statistically significant differences between the CareerAdvance and matched comparison groups after our two-stage propensity score matching. Full sample M (SD) (n = 293) CareerAdvance M (SD) (n = 119) Matched comparison M (SD) (n = 174) GED = general education diploma; HSIS = Head Start Impact Study. *p <.10. Federal Head Start guidelines require that centers track daily attendance for each child. We capitalize on these data and examine individual child attendance rates. Attendance and absence rates were calculated based on the number of days a child attended or missed CAP Tulsa Head Start, divided by the number of days the child was enrolled. The staggered nature of enrollment in CareerAdvance, with some parents entering the career training program in the fall and others in the spring, meant that families were differentially exposed to CareerAdvance across the Head Start year between September and June. Families who entered CareerAdvance in the fall could be enrolled for 10 months prior to the end of the child’s academic year, whereas families with parents who entered the career training program in the spring could be enrolled for 5 months. To equalize CareerAdvance exposure across cohorts, we examined children’s attendance, absence, and chronic absence over the initial 5 months of CareerAdvance enrollment, or approximately one semester, for both fall and spring CareerAdvance enrollees and matched comparison families. Due to the fact that child attendance rates were strongly right-skewed (where most are in the upper end of the distribution and with a ceiling effect at 100%; see Figure 1), we logged attendance rates as our outcome measure. Relying on log transformation reduces the influence of skew in regression analysis and better represents the shape of the distribution. We logged absence rates for similar reasons. We created an indicator variable denoting chronic absenteeism if a child was absent for 10% or more of the offered school days during the time of Head Start enrollment. The 10% benchmark for chronic absence is standard practice in Head Start (United States Department of Health and Human Services, Office of Head Start, 2016). In addition, we drew on a number of family, parent, and child measures of risk and protection from the full sample to examine potential differential associations with CareerAdvance. We created a family demographic risk scale modeled after the “household risk index” used in the 2002 Head Start Impact Study (HSIS; United States Department of Health and Human Services, Administration for Children & Families, Office of Planning, Research & Evaluation, 2018). We adapted the HSIS household risk index by focusing only on the primary parent and excluding employment status due to the concern that the parent’s employment status would be endogenous to program participation. This study’s family demographic risk scale includes the following household attributes (all dummy coded 1 = yes; 0 = no): (a) household receives public assistance; (b) parent’s education is a high school degree, GED, or lower; (c) parent was a teenage parent; and (d) parent is a single parent. Each of the elements of the above scale has been shown to be associated with a variety of adverse outcomes (Klebanov, Brooks-Gunn, & Duncan, 1994). We then summed the four items and coded families as high risk if they scored above the median (3 or 4) and low risk if they scored below the median (1 or 2). We created a scale measuring a parent’s level of readiness for college based on four parent and household attributes (all dummy coded 1 = yes; 0 = no): (a) parent is above the median age in the sample (i.e., 28 years old); (b) household income is above the sample median (i.e., $20,000–$24,999); (c) there is more than one adult in the household; (d) parent has a high school degree or higher. The above indicators individually and collectively may represent a readiness to engage in the rigors of college education. Older students with higher household income and family resources may be more likely to succeed in college (Jacobs & King, 2002). We then summed the four items and dichotomized the scale at the median so that parents were coded as high college-ready if they possessed all four attributes, and low college-ready if they possessed three or fewer. We used the Bracken School Readiness assessment, administered by CAP Tulsa at baseline, to measure a child’s level of school readiness. We coded children as low school-ready if they scored below the national mean of 100 points on the Bracken School Readiness assessment, and high school-ready if they scored above 100 points. Parents who lived more than 4.14 miles from Head Start (above the median distance from home to Head Start) were identified as high risk, whereas parents who lived less than 4.14 miles were labeled as low risk. Analyses of child attendance and chronic absence by distance from home to Head Start included five cohorts of study participants (not the usual seven for other subgroup analyses) due to missing home address data for two cohorts. We used a dichotomous indicator of whether parents reported that they were currently married or single. The five factors did not correlate with one another with the following exceptions: family demographic risk (−0.26) was negatively associated with parent college readiness, and single parent status was associated with family demographic risk (0.41) and parent college readiness (−0.34; Table S2). In the absence of an experiment, we addressed parents’ possible nonrandom selection into the program, which would yield biased estimates of treatment effects, by applying inverse propensity score weights in our regression analyses. We first used one-to-one propensity score matching to identify a group of Head Start parents similar to CareerAdvance parents to recruit to the study. A small amount of imbalance on observable characteristics remained after this first round of matching, so we conducted a second round of matching to compare children of CareerAdvance enrollees to parents of nonenrollees with similar observable characteristics, drawing on higher quality CAP Family Life Study data and allowing for matching across cohorts. However, matching strategies do not account for differences between the treatment and comparison groups along unobservable characteristics (Murnane & Willett, 2010; Rubin, 2001). We do not have concrete data about why the matched comparison group did not participate in the program, but our hypothesis is a lack of informational resources (e.g., matched comparison parents just did not know about the program). We do not believe that employment is a reason that the matched comparison group did not participate in the program because both groups had similar rates of baseline employment (Table 1). To employ inverse propensity score weighting in our second round of matching, we drew upon Wave 1 parent survey data, which was not available at the time of sample recruitment. Our matching variables included: gender, English spoken in the home, single parent, age, race (White, Black, Hispanic, and Other), number of adults in the household, number of children in the household, parent education (less than high school, high school diploma or GED, or higher), neighborhood (one of four in the Tulsa area), fall or spring semester program or study entry, year of program or study entry, and household income. We also addressed possible differences in motivation for school and career between CareerAdvance and matched comparison parents based on a “motivation score.” With data available on the motivation for CareerAdvance from parent interview scores, we created a comparable score based on a short motivation questionnaire administered to all other Head Start parents. The measures were equivalent and each uses a 5-point Likert scale. We drew all variables from parent survey data except for the neighborhood of residence and household income, which came from CAP Tulsa administrative data. As shown in Table 1, inverse propensity score weighting led to a balanced sample between the CareerAdvance and matched comparison groups based on observable baseline characteristics. We investigated the association between families’ participation in CareerAdvance and children’s rates of attendance, rates of absence, and chronic absence (yes/no) in Head Start during the first semester of enrollment (approximately 5 months), compared to participation in typical Head Start services. We also examined whether relations between participation in CareerAdvance and children’s attendance, absence, and chronic absence were greater or smaller for families, parents, and children with a range of baseline characteristics. We expected greater gains in children’s attendance (and reductions in absence and chronic absence) among families with high levels of household risk compared to families with low levels of household risk, parents with high college readiness compared to parents with low college readiness, children with high school readiness compared to children with low school readiness, families who lived far from Head Start compared to families who lived near Head Start, and children whose parents were single compared to children whose parents were married. Our hypotheses largely center on greater gains among at-risk populations, with the exception that parents with high college readiness may experience greater gains than parents with low college readiness. As shown in Table S2, high parent college readiness and family demographic risk are not highly correlated, indicating the possibility that both parents who experience high risk and parents who are college ready are likely to benefit from CareerAdvance. To examine associations among CareerAdvance participation and children’s (logged) attendance rates, (logged) absence rates, and chronic absence, we estimated coefficients from ordinary least squares regressions (OLS) with inverse propensity score weights, controlling for child gender, age, and race, as well as whether the parent entered the career training program during the fall or spring semester. We did not include parent employment in our models because of concern that it would be endogenous to program participation. We used linear probability models to measure associations with the likelihood of a child being chronically absent with the same set of controls and weights. To address differences by family demographic risk, parent college readiness, child school readiness, distance from home to Head Start, and parent marital status, we again estimated coefficients from OLS models, stratifying by high and low (and near and far distance from home to Head Start). We conducted Chow tests to determine whether estimates from the stratified models were significantly different from each other. We found that the treatment and matched comparison groups were well-balanced across risk and protective factors. We also estimated upper and lower bounds of the association between CareerAdvance participation and children’s attendance under different assumptions about the pattern of missing data. We ran several follow-up tests in which we included the 45 children who were excluded in our study due to having < 5 months of attendance data as well as the 293 children who had complete attendance data (for a total N of 338). Using our analytic sample (n = 293), we examined the distribution and used the top 5% to indicate high values on the given outcome and the bottom 5% to indicate low values on the given outcome. For example, the top fifth percentile for attendance was 99% and the bottom fifth percentile for attendance was 66%. To construct an upper bound estimate, we assumed that the nature of the missing data was such that we would estimate the largest possible treatment effect, with the greatest treatment versus comparison group differences. For an upper bound, we assumed that treatment group children had high attendance rates, low absence rates, and were not chronically absent (set at the top/bottom 5% of the distribution), and matched comparison group children had low attendance rates, high absence rates, and were chronically absent (also set at the top/bottom 5% of the distribution). We conducted a similar exercise to construct a lower bound, assuming that the nature of missingness was such that we would estimate the smallest possible treatment effect. Table S3 shows similar outcomes under both sets of assumptions, suggesting that differential attrition by treatment status does not appear to be an issue of concern given that even lower bound estimates retain the signs and nearly always the significance of the estimated associations. Regular daily attendance is necessary for children to achieve gains from the developmental programming Head Start provides, and even small increases in children’s attendance can lead to advances in child development (Ramey, Ramey, & Stokes, 2009; Ready, 2010; Reynolds et al., 2014). One of the few studies that have used a nationally representative sample to examine chronic absenteeism (Early Childhood Longitudinal Study’s Kindergarten Cohort; Gottfried, 2014) found negative effects of chronic absence (missing more than 14 days per year) on children’s educational and social engagement and school achievement in math and reading. For these reasons, Head Start has made child attendance a high priority and has set guidelines for achieving center-based attendance goals (United States Department of Health and Human Services, Office of Head Start, 2016). However, the Administration for Children and Families (through the Office of Head Start) is not prescriptive about center-based strategies to improve attendance as there is little evidence as to whether or why various strategies are effective. This article is about understanding new ways to improve children’s attendance in Head Start. In particular, we study an innovative program that draws lessons from developmental science about the “linked lives” between parents and children (Bronfenbrenner & Morris, 1998; Elder, 1998), suggesting the potential for a program that offers direct services to parents to improve the attendance of children. More specifically, we argue that a family systems approach that simultaneously targets parents’ and children’s well-being together will be associated with higher attendance compared to Head Start alone. Two additional perspectives—ecological and work support theories—provide useful frameworks for considering the mechanisms by which two-generation programs may affect children’s attendance in Head Start. Ecological theory suggests that families are embedded in a larger macrosystem of economic opportunities and constraints (Bronfenbrenner, 1977; Bronfenbrenner & Morris, 2006). Promoting the human capital (e.g., education and employment) of low-income parents helps to give them access to full participation in the broader economy, and may improve material and psychological resources. Likewise, both parents and children are embedded in the family and extrafamilial microsystems such as Head Start. Head Start was originally designed to meet the developmental and family needs of parents and children together (Zigler & Muenchow, 1992). A work support perspective provides nuance to an ecological approach, linking multiple environments and contexts to how families operate internally (Cox & Paley, 2003). For parents, two-generation programs offer a key work support that makes employment or schooling possible. Low-income mothers’ participation in the labor force has risen dramatically in recent decades due in part to welfare reform and associated work requirements and the expansion of the Earned Income Tax Credit (Haskins, 2006). The availability of affordable child care and public schooling has also been associated with increases in mothers’ employment (Gelbach, 2002). Higher child-care prices may interfere with mothers’ employment among both single and married mothers (Connelly & Kimmel, 2003; Han & Waldfogel, 2001). Importantly, costs of child care have led some women, especially those with the least skills, to forgo entering the labor market or to seek informal child-care arrangements (Henly & Lyons, 2000). Head Start, the largest publicly funded early childhood education program for low-income families, may provide the key work support to allow mothers to enter or remain in school or the workforce (Sabol & Chase-Lansdale, 2015). Because two-generation programs explicitly address the interconnectedness of the daily lives of parents and children, we expect that children's attendance will improve as a result of family participation in CareerAdvance. Two-generation programs intentionally support the family system, offering services to promote the education of parents and child at the same time, and helping parents see the connection between their own educational goals and their children’s. For instance, these programs coordinate parent–child school schedules, making attendance logistics easier for both generations. Using Head Start as a platform, two-generation programs also provide social networks and peer connections as well as financial and supportive services to families that will likely reduce attendance barriers. Early childhood education, including Head Start, has become a central policy lever for expanding opportunity, especially among low-income children in the United States (Barnett, 2001; Heckman, 2006; Lee, Brooks-Gunn, Schnur, & Liaw, 1990; Magnuson & Waldfogel, 2005; Yoshikawa et al., 2013). Children must be present if they are to reap the benefits of early childhood education programs, yet attendance is a concern for most programs. The Administration for Children and Families stipulates that Head Start children should attend 85% of the program days offered in any given month, but the national average attendance rate in Head Start is 75% (United States Department of Health and Human Services, Office of Head Start, 2016). Another area of increasing concern is the subgroup of children who miss many attendance days, that is, chronically absent children. For instance, the proportion of children that is chronically absent from publicly funded early childhood education programs is 26% in Baltimore, 36% in Chicago, and 49% in New York City (Balfanz & Byrnes, 2013; Connolly & Olson, 2012; Ehrlich, Gwynne, Stitziel Pareja, & Allensworth, 2014). From a developmental perspective, when children are absent for many school days, they miss the potential gains afforded by cognitive, socioemotional, language, and literacy services of Head Start (Ready, 2010; Reynolds et al., 2014). From a policy perspective, high chronic absenteeism in Head Start has the added effect of reducing average center attendance rates, a key federal program performance measure, with consequences for continued funding. This article examines a novel approach to improving children’s Head Start attendance that takes a family systems perspective by addressing the interdependence of parents and children and the micro and macrosystems influencing their lives. More specifically, the study aims to examine the following: (a) differences in attendance and absence rates between children whose families participate in a two-generation program and children who participate in Head Start only; (b) differences in chronic absence between children whose families participate in a two-generation program and children who participate in Head Start only; and (c) whether certain types of families (i.e., by baseline parent, child, and family characteristics) are more likely to benefit from CareerAdvance than others. We expect that a number of family, parent, and child risk and protective factors may lead to potential differential associations between family CareerAdvance participation and children’s Head Start attendance. For example, families with high or low family demographic risk, parent college readiness, and child school readiness may have stronger or weaker child Head Start attendance as a function of these characteristics. Families who live a far or near distance from home to Head Start or parents who are married or single may also experience differential associations between CareerAdvance participation and children’s Head Start attendance. Head Start is the nation’s oldest federally funded preschool program. This program has maintained an explicit whole family focus since its inception in the 1960s by combining early education services for children with a variety of services for parents to promote children’s development and foster family stability. CareerAdvance was designed and is delivered by the Community Action Project of Tulsa County (CAP Tulsa), a large anti-poverty organization and the Head Start delegate agency for Tulsa County, Oklahoma, and builds on Head Start’s family-centered approach. CareerAdvance pairs high-quality Head Start, which has proven benefits for children (Gormley, Phillips, & Gayer, 2008), with tuition-free education training and other support services at no charge to parents, including coaching support, peer group meetings, and additional child care outside of normal Head Start hours. Other early education programs, such as public prekindergarten programs, may be less effective platforms for promoting two-generation strategies and are not a focus of this article (Sommer, Sabol, Chase-Lansdale, & Brooks-Gunn, 2016). Regular daily attendance is necessary for children to achieve gains from the developmental programming Head Start provides, and even small increases in children’s attendance can lead to advances in child development (Ramey, Ramey, & Stokes, 2009; Ready, 2010; Reynolds et al., 2014). One of the few studies that have used a nationally representative sample to examine chronic absenteeism (Early Childhood Longitudinal Study’s Kindergarten Cohort; Gottfried, 2014) found negative effects of chronic absence (missing more than 14 days per year) on children’s educational and social engagement and school achievement in math and reading. For these reasons, Head Start has made child attendance a high priority and has set guidelines for achieving center-based attendance goals (United States Department of Health and Human Services, Office of Head Start, 2016). However, the Administration for Children and Families (through the Office of Head Start) is not prescriptive about center-based strategies to improve attendance as there is little evidence as to whether or why various strategies are effective. This article is about understanding new ways to improve children’s attendance in Head Start. In particular, we study an innovative program that draws lessons from developmental science about the “linked lives” between parents and children (Bronfenbrenner & Morris, 1998; Elder, 1998), suggesting the potential for a program that offers direct services to parents to improve the attendance of children. More specifically, we argue that a family systems approach that simultaneously targets parents’ and children’s well-being together will be associated with higher attendance compared to Head Start alone. Two additional perspectives—ecological and work support theories—provide useful frameworks for considering the mechanisms by which two-generation programs may affect children’s attendance in Head Start. Ecological theory suggests that families are embedded in a larger macrosystem of economic opportunities and constraints (Bronfenbrenner, 1977; Bronfenbrenner & Morris, 2006). Promoting the human capital (e.g., education and employment) of low-income parents helps to give them access to full participation in the broader economy, and may improve material and psychological resources. Likewise, both parents and children are embedded in the family and extrafamilial microsystems such as Head Start. Head Start was originally designed to meet the developmental and family needs of parents and children together (Zigler & Muenchow, 1992). A work support perspective provides nuance to an ecological approach, linking multiple environments and contexts to how families operate internally (Cox & Paley, 2003). For parents, two-generation programs offer a key work support that makes employment or schooling possible. Low-income mothers’ participation in the labor force has risen dramatically in recent decades due in part to welfare reform and associated work requirements and the expansion of the Earned Income Tax Credit (Haskins, 2006). The availability of affordable child care and public schooling has also been associated with increases in mothers’ employment (Gelbach, 2002). Higher child-care prices may interfere with mothers’ employment among both single and married mothers (Connelly & Kimmel, 2003; Han & Waldfogel, 2001). Importantly, costs of child care have led some women, especially those with the least skills, to forgo entering the labor market or to seek informal child-care arrangements (Henly & Lyons, 2000). Head Start, the largest publicly funded early childhood education program for low-income families, may provide the key work support to allow mothers to enter or remain in school or the workforce (Sabol & Chase-Lansdale, 2015). Because two-generation programs explicitly address the interconnectedness of the daily lives of parents and children, we expect that children's attendance will improve as a result of family participation in CareerAdvance. Two-generation programs intentionally support the family system, offering services to promote the education of parents and child at the same time, and helping parents see the connection between their own educational goals and their children’s. For instance, these programs coordinate parent–child school schedules, making attendance logistics easier for both generations. Using Head Start as a platform, two-generation programs also provide social networks and peer connections as well as financial and supportive services to families that will likely reduce attendance barriers. Past efforts to improve children’s attendance in early childhood education programs and elementary school have primarily targeted communication and interactions among parents, teachers, and staff. These types of communications (e.g., over the phone, in-person at home or school, or via text messages and letters to parents) tend to rely on the assumption that parents may actively choose not to take their children to school, may lack informational resources about when and how often their children are supposed to attend, or they simply forget, so reminders may help. This strategy may be reasonable given that parents are central to the school attendance of young children. Yet the approach misses the larger system of ecological influences that often play a key role in whether families are able to realize the goals that they have for their children, including regular school attendance. Some low-cost, light-touch interventions have been modestly effective in Head Start and elementary school. For example, behavioral nudges have had some success in the short term in boosting early childhood attendance. In a randomized experiment of 26,000 K-5 students, parents of students in the bottom 60% of attendance across 10 school districts received mailers that reported the number of days their children had been absent from school and offered general information about the importance of attendance in early grades (Robinson, Lee, Dearing, & Rogers, 2017). Students in families who received the attendance mailings were absent 0.53 fewer days over the course of the entire school year, on average, than students in households who did not receive them. Rates of chronic absence (missing 10% or more of days offered) were 4.64% in the treatment group and 5.45% in the control group, a 14.9% reduction in chronic absenteeism. In early childhood education, there are very few experimental studies of intervention strategies to improve children’s attendance. One exception is a randomized trial of preschool children and their parents (N = 780 families) which involved a series of behaviorally informed text messages over 18 weeks (Kalil & Mayer, 2017). The study found a small, but statistically significant (i.e., 1.8 percentage point) difference in attendance rates between treatment and control groups and a somewhat larger (i.e., 7.4 percentage point) difference in rates of chronic absenteeism between the two groups. Similarly, in an experimental study of the Child Attendance and Social Capital Project, parents of children enrolled in Head Start were given the opportunity to form voluntary partnerships with one another (e.g., a “buddy system”) in support of their children’s attendance and their children were assigned to classrooms based on neighborhood of residence (Sommer et al., 2017). The children of parents who received both treatments had higher rates of Head Start attendance (during winter when average center attendance was the lowest) compared to parents and children who received business-as-usual Head Start services. The Head Start program’s primary approach to the problem of low center-based child attendance is home visits by teachers or other school staff (including nurses), which are to occur when children experience two or more consecutive, unexcused absences or are frequently absent across the school year (United States Department of Health and Human Services, Office of Head Start, 2016). The idea behind home visits is to identify the root causes of chronic absenteeism, such as transportation concerns, and to devise individualized solutions to address them (Balfanz & Byrnes, 2012). Home visits in general are associated with significant improvements in Head Start children’s academic and socioemotional outcomes (Cook, Dodge, Gifford, & Schulting, 2015), and they have the potential to generate significant benefits to family systems. However, home visits in Head Start have not been shown to be particularly effective in increasing children’s attendance despite the prevalence of the practice (Jones-Harden, Chazan-Cohen, Raikes, & Vogel, 2012). For instance, a 2-year study of a program involving family nurse practitioners with specialized expertise in addressing health-related issues (N = 130), believed to be a main source of children’s absence, found no improvements in Head Start attendance in the first or second year, and only modest improvements in elementary school attendance in the first year (Kerr et al., 2012). The lack of proven attendance strategies in early childhood education is especially concerning given that young children with attendance issues often continue patterns of absence throughout elementary school. For example, one-third of chronically absent 4-year-olds are also chronically absent in kindergarten, and one-third of this group remains chronically absent in second grade (Ehrlich, Gwynne, & Allensworth, 2018). Attendance interventions that focus on the child alone may miss the larger ecosystems in which low-income families operate, including significant obstacles like low levels of parent education and family income which may interfere with children’s regular school attendance in the short and longterm. New, more intensive interventions aimed at addressing the macro and microecological systems may hold promise for improving children’s Head Start attendance. CareerAdvance is designed to promote parents’ career identity and human and social capital in ways that benefit parents’ own economic lives as well as their children’s cognitive and socioemotional development. CareerAdvance is an optional add-on to Head Start, and parents are recruited from Head Start centers to participate. In this study, we hypothesize that family participation in this two-generation program may improve children’s attendance in the shortterm because the most common barrier is addressed—namely, a low supply of quality, affordable child care. In addition, CareerAdvance focuses on other key logistical, financial, psychological, and social barriers to children’s school attendance (Gardner, Brooks-Gunn, & Chase-Lansdale, 2017). Below we describe the program elements of CareerAdvance and how they may be related to improvements in children’s Head Start attendance. As a full-day program, Head Start provides a reliable, low-cost work support for parents pursuing education and job training. Moreover, CAP Tulsa’s Head Start program for 3- and 4-year-olds has been documented as high-quality and effective for children in the short and longterm (Gormley et al., 2008; Phillips, Gormley, & Anderson, 2016). Like most Head Start programs, CAP Tulsa also provides additional services to parents, including parenting literacy programs and case management. CareerAdvance goes above and beyond business-as-usual in Head Start by also offering comprehensive, intensive education and training with additional supportive services to parents of children enrolled in Head Start. A key element of the design of the CareerAdvance program is stackable training, which allows individuals to exit (temporarily or permanently) at multiple points along a pathway with an industry-valid credential. For example, it is expected to take participants approximately 4 years to complete the nursing career ladder to become a registered nurse, balancing the needs of family, work, and school along the way. Moreover, the program offers coordinated parent and child school schedules and provides early education services before and after school to ensure that children are fully cared for while parents attend classes and training. The intervention also helps parents find quality, community-based before- and after-school care for older children and infants, fully covering these childcare costs. Comprehensive, high-quality, and flexible care for children gives parents the opportunity to enroll in school or go to work, which is also likely to improve the odds that their children show up to Head Start. In fact, parents may need their children to attend school regularly if they are to participate in their own schooling or employment regularly too. Parents are also likely to face a number of financial and psychological barriers to children’s regular school attendance. CareerAdvance provides financial incentives to parents for achieving education and participation goals in addition to free tuition, books, and materials (such as scrubs and stethoscopes). Parents are eligible to receive financial compensation for achieving certain milestones (e.g., $300 each time they receive a certificate or become employed), meeting certain grade and credit requirements (e.g., maintaining a 3.0 average) or attending peer meetings or courses each month (in total up to $3,000 per year). When CareerAdvance parents receive financial incentives to go to school, they and their children may be more likely to participate. Other individual factors that may impede children’s Head Start attendance include maternal depression and psychological distress which, among other personal and family challenges, may limit parents’ bandwidth for solving the logistical, scheduling, and transportation problems that arise in getting themselves and their children to school (Gennetian & Shafir, 2015; Mullainathan & Shafir, 2013). By offering CareerAdvance from a platform of Head Start services, the program incorporates a range of intensive supports for parents, including Head Start family advocates who provide case management, emergency funding, and connections to other resources (e.g., counseling services), as well as coaches hired by CAP Tulsa to provide academic and career advice. These services offer motivational, psychological, and practical support to parents that may translate into improved attendance among parents and children alike. Social isolation can be another barrier to attendance among low-income families who may have low institutional trust and limited social capital, that is, social relations among individuals that allow for an exchange of social, emotional, and instrumental resources (Coleman, 1988; Lin, Cook, & Burt, 2001). The CareerAdvance program explicitly fosters social capital through two key strategies: a cohort model for program enrollment and regular peer meetings with other participating parents. Parents enroll in CareerAdvance with small groups (30 participants on average) and take certification classes in health care (typically 4 days per week for 5–6 hr per day while children attend Head Start) together. Parents also participate in weekly meetings with other parents (usually 2–3 hr per week) led by trained career coaches employed by CAP Tulsa. Peer meetings involve a range of topics, including skill building, time management, and children’s development. Career coaches serve as important informational resources and connect parents to an array of services (Scrivener & Coghlan, 2011). CareerAdvance offers microsystems of social support through coaches placed in Head Start centers and course instruction and meetings with groups of parents who are likely to engage with one another to support children’s attendance. Child attendance issues, especially chronic absence, tend to be highest among families who face a range of household risk factors, including single and teenage parenthood, low levels of parent education and household income, and public assistance receipt (Romero & Lee, 2008). We hypothesize that certain types of families (i.e., families facing multiple household risk factors) may bring their children to school more often because of their involvement with the CareerAdvance program compared to other low-income families (i.e., families with fewer household risk factors) who were not involved. These families may be extra motivated to overcome typical obstacles to children’s regular Head Start attendance (e.g., transportation issues) so that they too can attend classes or work. Conversely, parents who are more prepared to participate in postsecondary education (i.e., more “college-ready”), compared to less prepared parents (i.e., less “college-ready”), may be in the best position to take advantage of education and career training, and thus to recognize the benefits of early childhood education to their own school success. Parents whose children are more prepared to enter kindergarten, compared to parents whose children are less prepared, may also be especially motivated to bring their children to school and continue to see them thrive. Structural barriers such as the distance from a family’s home to Head Start may also influence children’s school attendance (Gottfried, 2017; Romero & Lee, 2008; Susman-Stillman, Englund, Storm, & Bailey, 2018). The benefits of parents’ participation in career training to children’s attendance may be greater for families who live farther away from rather than closer to their children’s Head Start center. The longer travel distance is likely to increase the need for schedule coordination and reliable transportation options that may not be regularly available to low-income families. Single parents may also experience greater benefits from the logistical coordination of parent and child school schedules in CareerAdvance compared to married parents who have another adult to support bringing a child to Head Start. Our main hypothesis is that family enrollment in CareerAdvance would promote higher levels of child school attendance and lower rates of absence and chronic absence compared to Head Start children from families that did not participate in the program. We also test whether the benefits of CareerAdvance to child attendance, absence, and chronic absence were correspondingly higher among subgroups within our sample based on family, parent, and child characteristics at baseline. Specifically, we examine differential patterns in attendance, absence, and chronic absence among five subgroups based on family demographic risk, parent college readiness, child school readiness, distance from home to Head Start, and parent marital status. Thus, this study asks three central questions: What is the difference in attendance rates between children whose families participate in a two-generation program, CareerAdvance, which combines education and training for parents with Head Start for children, and children who only participate in Head Start (i.e., the matched comparison group)? What is the difference in absence and chronic absence rates between children whose families participate in CareerAdvance and children who only participate in Head Start? Are certain types of families (i.e., by baseline parent, child, and family characteristics) more likely to benefit from CareerAdvance than others? Answers to these questions suggest a new strategy to promote children’s development in Head Start that draws from family systems theory and links between parent and child outcomes. The innovation is to structure a program so that when parents improve their education while their children participate in Head Start, children's outcomes also improve through regular, daily attendance. We hypothesize that as parents regularly attend work and school, by necessity their children will show up more often to Head Start. Publisher Copyright: © 2020 Society for Research in Child Development
PY - 2020/11/1
Y1 - 2020/11/1
N2 - Two-generation human capital programs seek to promote the education of parents and children simultaneously. This study examines relations between family participation in CareerAdvance, which recruits parents of Head Start children into a workforce training program, and children’s Head Start attendance. The sample included 293 children (on average 4 years old) and their parents. After one semester, CareerAdvance children demonstrated higher rates of attendance and lower rates of absence and chronic absence (missing 10% or more of school days) than matched comparison children. These associations were similar across a range of high- and low-risk subgroups at baseline. These findings are discussed in terms of the implications of a family systems approach for improving children’s Head Start attendance.
AB - Two-generation human capital programs seek to promote the education of parents and children simultaneously. This study examines relations between family participation in CareerAdvance, which recruits parents of Head Start children into a workforce training program, and children’s Head Start attendance. The sample included 293 children (on average 4 years old) and their parents. After one semester, CareerAdvance children demonstrated higher rates of attendance and lower rates of absence and chronic absence (missing 10% or more of school days) than matched comparison children. These associations were similar across a range of high- and low-risk subgroups at baseline. These findings are discussed in terms of the implications of a family systems approach for improving children’s Head Start attendance.
UR - http://www.scopus.com/inward/record.url?scp=85089154049&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85089154049&partnerID=8YFLogxK
U2 - 10.1111/cdev.13397
DO - 10.1111/cdev.13397
M3 - Article
C2 - 33460082
AN - SCOPUS:85089154049
SN - 0009-3920
VL - 91
SP - 1916
EP - 1933
JO - Child development
JF - Child development
IS - 6
ER -