some role in the widening health inequality (Makenbach et al., 1989), increases in income and wealth inequality in both the United States and Western Europe (Danziger and Gottschalk, 1993) appear to be the driving force behind the widening health disparities (Williams and Collins, 1995).

A high degree of inequality in a given location (e.g., country, state, county, district, city) may itself be a health hazard. The countries with the smallest spread of incomes and the smallest proportion of the population in relative poverty have the longest life expectancies (Wilkinson, 1994). Evidence from multiple sources suggests that the greater the concentration of income at the upper end of the income distribution, the higher the mortality and morbidity rates (Wilkinson, 1994; Kaplan et al., 1996; Lynch et al., 1998). Socioeconomic inequality also affects health in more complicated ways. It is widely recognized that at the aggregate level average health is negatively correlated to the degree of income inequality. However, if health status depends not on absolute income but on income relative to that of some reference group, then the relationship between income and health is determined by the relative size of within-group and between-group inequality (Deaton, 1999). When the ratio of between-group to within-group inequality changes, the mix of high- and low-income status in any particular group changes. This change alters the measure of the relationship between health and income. Existing community-level evidence about socioeconomic inequality and its relationship to health implies the need for more detailed inquiries into appropriate measures of inequality.

There is considerable variation in health outcomes at all levels of socioeconomic hierarchies. Of particular importance, health outcome variance is greater at the bottom of these hierarchies—for low levels of education and income—than at the upper end. The fact that some persons low in these hierarchies have unexpectedly positive health outcomes (compared to the norm for their level) calls for future inquiry regarding the psychological, social, behavioral, and biological factors that confer protection for some individuals at the low end of the socioeconomic hierarchy.

The provision of defensible explanations for associations between position in socioeconomic hierarchies and morbidity and mortality requires integrated investigation of psychosocial and physiological interrelationships over the life course. Most studies attempting to identify stressful life experiences and other features that predict later disease and mortality have incorporated neither long-term processes nor cumulative experience. There is a singular lack of studies integrating community-level and individuallevel behavioral influences with parallel physiological dynamics to identify multilevel pathways to diverse health outcomes. Evidence does suggest that