all the officially reported morbidity and mortality rates for Black males in these age groups are 11 to 13 percent too high. For the 1990 census, in addition to demographic analysis, the Bureau of the Census conducted a postcensus enumeration survey (PES) in which the undercount was estimated on a case-by-case matching of census records with those obtained by PES of 165,000 households. According to PES, the undercount rates for Hispanics and for American Indians residing on reservations were even higher than the undercount for Blacks (Hogan, 1993). Future research must give greater attention to the magnitude and size of the undercount and of its impact on the quality of health data.

Health is socially embedded in the larger conditions in which individuals and groups live and work (Amick et al., 1995; Engels, 1984). Biological differences do exist between human population groups, but the existing racial categories do not capture those differences (American Association of Physical Anthropology, 1996; Montagu, 1965). About 75 percent of known genetic factors are innate and identical in all humans; about 95 percent of human genetic variation exists within racial groups (Lewontin, 1974, 1982). Thus, there is more genetic variation within races than between them, and, at best, genetic and biological differences play a minor role in accounting for the observed racial disparities in health (Kaufman and Cooper, 1995).

A prominent hypothesis in the health literature is that racial differences in socioeconomic status (SES) account for the racial variations in health. A robust inverse association persists between SES and health across a broad range of health outcomes in both the industrialized and the developing worlds (Antonovsky, 1967; Bunker et al., 1989; Williams, 1990). Moreover, some research suggests there is a stepwise progression of diminished risk with each higher level of SES (Adler et al., 1993; Marmot et al., 1991). Race is strongly associated with SES, and adjusting Black-White disparities in health for SES sometimes eliminates, but always substantially reduces, these differences (Williams and Collins, 1995; Lillie-Blanton et al., 1996). It is frequently found, however, that even when education and income level are held constant, Blacks have higher levels of ill health than Whites (Williams, 1996b). Some studies find that Black-White differences in health status actually increase with rising SES (Schoendorf et al., 1992; Singh and Yu, 1995).

Greater attention to the construct of racism can serve to inform and structure our understanding of racial inequalities in health (Cooper et al., 1981b; Hummer, 1996; Krieger et al., 1993; LaVeist, 1996; Williams, 1996c,