Household Composition , Race / Ethnicity and Maternal Reports of Child Health in the United

* Authors are listed in alphabetical order, denoting equal contributions. Abstract The primary objective of this paper is to model the association between household composition and maternal reports of child health for a large representative sample of young children in the United States. Using the 1991 National Maternal and Infant Health Survey—Longitudinal Follow-up, we examine these relationships first for all children and then separately by race. Our results suggest that children in nuclear families and children living with their mother and a partner experience the most positive health outcomes; on the other hand, children in extended households (presence of other adults) and female-headed households experience the worst physical health outcomes. A small portion of this relationship is attributable to racial differences that are associated with differential household arrangements and a larger portion of this relationship is due to SES. Maternal depression is strongly correlated with child health and well-being, and accounts for virtually all of the household compositional differences (female-headed households notwithstanding) in the physical health of children. Finally, the negative health effects of extended and female-headed households are more pronounced for children of non-Hispanic Black mothers, net of potentially confounding variables such as SES and depression.