Re-Assessing the Impacts of Head Start on Children's Cognitive and Health Outcomes

Kumiko Imai, Cornell University

Positive evidence on the benefits of Head Start using large, nationally representative data comes from a handful of studies that used family fixed models, which assume that unobserved family characteristics determine participation in Head Start. Using the National Longitudinal Survey of Youth 1979 (NLSY79) and NLSY79 Children data, I present evidence from difference-in-differences estimators that use paired matching for control selection. Specifically, I match each Head Start child with a set of control children who are similar in terms of demographic and socioeconomic variables, and compare pre- and post-Head Start outcomes with comparable outcomes for control children. My matched-pairs difference-in-differences estimators suggest that, contrary to previous findings, Head Start has little effects on child cognitive outcomes, but positive short-run effects on health measures. My study demonstrates the importance of selecting appropriate controls and controlling for unobserved child-specific characteristics in evaluating early childhood interventions.