Prenatal Care and Infant Health: Elementary, Dear Watson? Accounting for Selection Bias in Nonexperimental Data

Satvika Chalasani, Pennsylvania State University

This paper reexamines the relationship between prenatal care and infant health using cross-sectional data from India. Given the dearth of experimental evidence on the nature and extent of the effects of prenatal care on birthweight and infant mortality, we have mostly drawn conclusions from observational data. These results often suffer from the nagging problem of selection bias: women that seek a certain level of prenatal care may be systematically different in a way that also affects their birth outcomes. I employ (i)a sibling-difference model, and (ii)a propensity score model to address this unobserved heterogeneity. Neither of these approaches has been used before in the literature on prenatal care. The use of data from a non-Western setting is also a first for the prenatal care-selection literature. The huge investments made in prenatal care worldwide as part of the effort to improve maternal and child health make this a question worth revisiting and answering right.