Using Administrative Data for Impact Evaluation: A Case Study for Family Connects Postpartum Home Visiting Program

Anna Rybinska , Duke University
Kenneth A. Dodge, Duke University

Experiences of child abuse and neglect are linked to poor developmental, behavioral, and health outcomes throughout the life course, including outcomes as diverse as depression and substance abuse, type 2 diabetes and cardiovascular disease, incarceration, and teenage pregnancy. While long-term targeted home-visiting services were shown to reduce child maltreatment prevalence, there is a need to evaluate short-term and population-level interventions in order to reduce cost and expand program reach. In this project, we investigate whether random assignment to a population-level short postpartum home-visiting intervention (Family Connects) can reduce risk of child maltreatment. We also examine whether the effects of the intervention spill-over to reduce the risk of child maltreatment not only for the target infant but also for all children within the family and whether this effect is moderated by family characteristics such as family structure. We use birth records data for North Carolina to identify children who in the Family Connects randomized control trial in 2009 and 2010. Multivariate logistic and Poisson regressions are employed. The examination of child-level and family-level effects of a home-visiting intervention offers new insights about how intervention effects radiate across family members and furthers knowledge about the variation in intervention impact across different types of families. In addition, the findings from the study will demonstrate the feasibility of identifying intention-to-treat effects for maternal, infant, and early childhood public health interventions using administrative records.

See extended abstract

 Presented in Session 18. The impact of policies on demographic outcomes