347 The Direct and Indirect Effects of Metropolitan Area Inequality on Mortality A Hierarchical Analysis NORMAN J. WAITZMAN, a,b KEN R. SMITH, c AND ANTOINETTE STROUP c b Department of Economics, University of Utah, Salt Lake City, Utah 84112, USA c Department of Family and Consumer Studies, University of Utah, Salt Lake City, Utah 84112USA BACKGROUND In previous research, we delineated several of the conceptual issues regarding the pathways by which socioeconomic characteristics of areas might affect individual health 1 and constructed first-time empirical analyses of the effects of poverty-area residence 2,3 and of residential economic segregation in metropolitan areas 1 on indi- vidual mortality using national samples. These analyses showed that area character- istics independently influenced mortality rather than serving as mere proxies for socioeconomic characteristics of individual residents. They also suggested that both the level and spatial distribution of resources in relatively large metropolitan areas significantly affected the life chances of residents. In this analysis, we extended our earlier research by considering (1) the extent to which the health effects associated with spatial economic segregation might be attributable to the statistical distribution of area income, and also by providing (2) a preliminary assessment as to hierarchical structure, that is, the extent to which area effects indirectly affect mortality by mod- ifying the effects on mortality of individual socioeconomic status. METHODS The data for our analysis, described in detail elsewhere, 1 comprised nine succes- sive annual National Health Interview Surveys (NHIS) from 1986 to 1994 matched to death certificates through 1995. Metropolitan statistical area (MSA) identifiers were appended to the data for respondents residing in thirty-four of the largest met- ropolitan statistical areas (MSAs) in the United States. The sample was restricted to 136,956 respondents aged 35 to 65 years residing in one of those MSAs, of which 3,715 had died by 1996. Economic segregation, or spatial inequality , in each area was gauged by the so- called “p index” of poverty, roughly the average across tracts of the probability of within-tract encounters between residents above and below the poverty line in each MSA in 1990. Statistical inequality was measured by the 1990 Gini coefficient on family income in each MSA. Incorporation of both measures provided a test of the a Address for correspondence: Norman J. Waitzman, Ph.D., Department of Economics, University of Utah, 1645 E. Central Campus Dr. Front, Salt Lake City, UT 84112, USA. 801- 581-7600 (voice); 801-585-5649 (fax). e-mail: waitzman@econ.sbs.utah.edu