Abstract
OBJECTIVES: As data for multiple race groups have only recently been collected and tabulated, the current understanding of the health of multiple race groups is not well developed. In the National Health Interview Survey (NHIS), survey respondents who report more than one race are asked to identify a primary race. This report compares selected health and demographic measures by the response to the primary race question and compares estimates for specific primary race groups to corresponding estimates for single race groups. METHODS: Using 1997-2003 NHIS data, several demographic and health measures were compared by reported primary race within the four largest multiple race groups. Then estimates by primary race for these four multiple race groups were compared to those for their single race counterparts. RESULTS: There were few statistically significant associations between reported primary race and health or demographic variables within the four multiple race groups. This lack of association may be due to the small number of multiple race respondents (which results in large standard errors) rather than similarity of point estimates among the subgroups. A greater number of differences between estimates for single race groups and for multiple race respondents who reported the same single race as their primary race were identified. CONCLUSIONS: The apparent lack of association between primary race within a multiple race group and health outcomes suggests that tabulating multiple race responses by primary race is unnecessary for valid health estimates for multiple race groups, at least with available statistical power. However, differences between single race and primary race estimates within a multiple race group suggest that inferences for multiple race respondents using single race estimates may be imprecise for some outcomes.