Ethnic gaps in mortality persist in the United States but the specific
causes remain elusive. We propose a broader mortality framework that
includes neighborhood characteristics that We test using data from a f
ile that links the National Health Interview Survey (NHIS) from 1986 t
hrough 1990 with information from death certificates from the National
Death Index (NDI), and additional census tract-level data from the 19
90 Census STF-3A files. Cox proportional hazards models, which include
measures of minority concentration and median income at the neighborh
ood level, for all-cause mortality during the follow-up, are estimated
for men and women separately The concentration of African Americans i
n the neighborhood of residence, in addition to individual socioeconom
ic status, fully account for differential mortality between African Am
erican and non-Hispanic white men and women. For Mexican Americans, th
e concentration of Hispanics in the neighborhood slightly enhances the
ir significant mortality advantage. From additional analyses, it appea
rs that the pathway between residential segregation and mortality is r
outed through poorer neighborhood economic conditions for men and high
levels of female headship in segregated neighborhoods for women. The
final analysis conducted for men by age at death shows that both young
and middle-aged African American men are affected by the concentratio
n of African Americans in the community.