versely related to employment opportunities for Black males (Wilson, 1987). Lack of meaningful and adequately compensated employment can create frustration and hostility, which have emerged as important risk factors for coronary heart disease, obesity, and other health conditions. Levels of hostility, however, are patterned by SES; and residence in poor living conditions predicts the higher likelihood of hostile responses (Harburg et al., 1973).

As the findings on race and mental health status suggest, research is also critically needed to identify the reason why Blacks have lower rates than Whites on indicators of subjective well-being but comparable or better rates than Whites on other indicators of mental health. More attention needs to be focused on the resources and cultural strengths within minority communities. High levels of religious involvement, family and kin support systems, psychological resources, such as John Henryism and racial identity, and processes of attribution have all been identified as potential adaptive resources within minority populations (James, 1993, 1994; Neighbors et al., 1996; Williams, 1998).

Future research must give greater attention to comprehensively assessing racial minority status and including identifiers for ethnic variation within each of the five OMB categories. The availability of adequate data for Native Americans or Alaska Natives, Hispanics, and Asians or Pacific Islanders is still a major problem. Because of the relatively small sizes of some of these groups and their geographic distribution, standard sampling strategies for national populations do not yield adequate sample sizes to provide reliable estimates for the distribution of diseases in these groups or to explore heterogeneity within a given racial group. Surveys focused on a particular geographic area with a high concentration of a racial subgroup, as opposed to national ones, are necessary to provide data for these groups. Combining multiple years of data in ongoing surveys is another useful strategy for obtaining health information for small population groups.

Researchers also need to be more self-critical about the collection, analysis, and interpretation of racial data. Greater consideration must be given to why race/ethnicity identification is being collected, the limitations of racial data, and how the findings should be interpreted. Data on racial differences should routinely be stratified by SES within racial groups. Failure to do this may misspecify complex health risks and even lead to harmful social stereotypes and consequences. Whenever feasible, additional information that captures these characteristics should be collected. This will include the assessment of SES, acculturation, and economic and noneconomic aspects of discrimination (Williams, 1997). There are limited opportunities to collect additional information in vital-statistics systems and in record-based surveys. Even in these contexts, how-