Panel Paper: How Does Head Start Compare? Evidence from Three Contemporary Datasets

Thursday, November 6, 2014 : 9:30 AM
Jemez (Convention Center)

*Names in bold indicate Presenter

RaeHyuck Lee1, Jeanne Brooks-Gunn1, Wen-Jui Han2, Jane Waldfogel1 and Fuhua Zhai3, (1)Columbia University, (2)New York University, (3)Fordham University
Head Start was introduced in 1965 as a compensatory early childhood education program for low income children, at a time when few other preschool options for them existed. Today, however, numerous other child care arrangements are available, and so to understand the effects of Head Start, it is essential to know how it works in comparison to available alternatives. Nonetheless, most Head Start studies have overlooked this issue, which is how to define the control or comparison group to which Head Start is compared. Therefore, the question we focus on in this study is not whether Head Start works, but how Head Start compares to other available child care and preschool arrangements. To do so, using three contemporary datasets, we compare comprehensive domains of school readiness at the age of five years between children in Head Start programs and those in each of the four comparison groups.

Our analyses draw on three datasets—the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B), the ECLS-Kindergarten Class (ECLS-K) of 2010-2011, and the Fragile Families and Child Wellbeing Study (FFCWS). We define five child care and preschool arrangements at the age of four years—Head Start; prekindergarten (i.e., state-funded prekindergartens or preschools); other center-based care (i.e., day care centers, nursery schools, or other preschool programs); home-based care (i.e., care from relatives or non-relatives in the child’s home or another person’s home); and exclusive parental care. We employ a wide set of school readiness outcome measures—academic achievement (i.e., early reading and math scores (ECLS-B, ECLS-K), PPVT and WJ-Word Identification scores in the FFCWS); socio-emotional development (i.e., externalizing, internalizing, and attention problems and pro-social skills); health and nutrition (i.e., BMI, overweight and obesity, healthy and unhealthy eating patterns, and medical and dental checkups); and maltreatment (i.e., mothers’ use of spanking). To address selection bias, we use regressions with an extensive set of control variables and propensity score weighted regressions.

We find that Head Start is associated with better reading skills compared to parental care or home-based care, but also associated with worse reading skills compared to prekindergarten. Regarding behavior outcomes, Head Start is associated with higher levels of externalizing behavior problems compared to parental care. With respect to health and nutrition outcomes, Head Start tends to be associated with lower BMI scores and lower probability of being overweight or obese compared to home-based care. Head Start is also associated with healthier eating patterns compared to parental care. Finally, Head Start is associated with more dental checkups, regardless of what the comparison group is, but the associations are much more pronounced when compared to parental care or home-based care.

The findings of this study make an important contribution to the Head Start research by showing the importance of clearly defining a comparison group. Also, this study provide clear evidence on Head Start’s short-term effects on diverse dimensions of school readiness among children from the three largest contemporary national datasets. Finally, some findings of this study provide some indication of the need to improve the quality of Head Start services.