Racial and Ethnic Disparities in Birthweight in the United States: New Models and New Results

Narayan Sastry, RAND

In this paper we investigate race and ethnic disparities in birthweight using a new modeling approach, namely quantile regression. This technique allows us to fit a model to the median as well as other fixed percentiles of the birthweight distribution using a family of jointly estimated models. Together, these models provide information on how the effects of covariates vary across the entire birthweight distribution and also allow us to focus on the determinants of low birthweight but without resorting to a dichotomous outcome measure. Graphical presentation of the quantile regression results provides valuable additional insights. We use these models to investigate differentials in birthweight by race and ethnicity and the factors that account for these differentials. Data for this study come from the 2001 Natality Data Set, which covers all births in the U.S. and hence allows us to examine finely graded differences in birthweight across the entire distribution.