Are racial disparities in health conditional on socioeconomic status?.

This research examined two basic questions: (1) Is there a racial gap in morbidity and self-rated health over 20 years? (2) If yes, are the racial differences conditional on socioeconomic status (SES)? Regarding the first question, the results show that black adults began the study in poorer health: they were much more likely to suffer from serious illness and to rate their health poorly. In the examination of incident chronic and serious illness, the rate of change in morbidity over the 20 years was similar for white and black adults. This nonsignificant effect of race on changes in morbidity shows that over the 20 years black adults did not experience a higher incidence of morbidity than white adults—the racial disparity in health did not widen over the study period. The nonsignificant effect of race also means, however, that the racial disparity in health did not diminish over the 20 years. Black adults began the study period in poorer health, and the disadvantage endured. Whereas they did not experience an improvement in serious illness either, they ended the study period in poorer health than white adults. Overall, the findings point to stable racial differences or persistent health inequality (Ferraro & Farmer, 1996b; Kelly Moore, & Ferraro, 2004).

The results regarding the racial gap in morbidity should be interpreted in light of two elements of the study design. First, the measures of morbidity were based on doctor diagnosis and therefore dependent on accessing health care. As black Americans are less likely to access formal health care through primary care facilities (Davis, Liu, & Gibbons, 2003; Escarce & Puffer, 1997; Shi, 1999), they may be less likely to have illnesses diagnosed by a doctor. This could lead to an underestimation of the true incidence of the illnesses in the black American population and more conservative estimates. Second, these models directly test for selective attrition in the sample. In the models for serious illness and chronic illness, the selection effect for death was positive and highly significant which indicates that those who did not complete the study were likely to have had more illnesses. In the preliminary stages of this research, all models were estimated without the selection variables for death or refusals/lost. Without the selection effect in the model, the effect for race was modestly significant for changes in serious illness. Thus, the inclusion of the attrition terms may account for the difference; black and white persons with incident morbidity did not survive the 20-year study.

The second research question focused on whether the racial disparity in health was conditional on SES. To this end, interactions between race and the measures of SES were tested in all models to examine if racial disparities were consistent across levels of SES. In the cross-sectional analyses, black adults of all employment groups, including the unemployed, tended to report more serious illness and poorer self-rated health than white workers. For self-rated health, however, the interactions between race and income and between race and occupational prestige showed the racial gap to be largest at the highest levels of income/occupational prestige. These findings provide evidence for the diminishing returns hypothesis because black adults did not enjoy similar ‘returns’ in self-rated health as white adults with increased income and occupational prestige.

For self-rated health, the longitudinal and cross-sectional results showed that racial differences in changing self-rated health remained and persisted despite the significant effects of SES; the racial gap in self-rated health was largest at the highest level of education. Black adults at the lowest educational levels actually reported slightly better self-rated health than their white counterparts. As education increased, there was substantial improvement in the self-rated health for white adults but there was no improvement for black adults. Therefore, black adults at the highest education levels reported significantly poorer self-rated health than their white counterparts. In short, the racial gap was the largest at the higher levels of education, offering further support for the diminishing returns hypothesis.

In research on educational returns and occupational mobility, minorities have long been shown to receive fewer payoffs for education than their white counterparts (Bowles & Gintis, 1976). In other words, white adults receive a greater return than black adults in terms of high-paying jobs for their level of educational attainment. Other research shows that although education is associated with higher income for both black and white persons, the relationship is stronger for white males ( Farley, 1984). The findings from the current research can be seen as differential educational returns on self-rated health. As educational levels increased, black Americans did not experience the same improvement in self-rated health as white Americans.

Willie (1989a) argued that the kinds of institutional changes needed to help black people are different than those for white people. The findings in the current research indicate that educational institutions help the self-rated health of white adults, but do not provide the same boost in self-rated health for black adults. Although it is unclear why black Americans did not enjoy the same health benefit from education, it is possible that increased awareness of racial oppression and discrimination may be a factor. One could posit on the basis of equity theory that unfairness leads to feelings of distress ( Walster, Walster, & Berscheid, 1978), and psychological distress brought on by racial discrimination may adversely affect physical health ( Krieger & Sidney, 1996; Kessler, Mickelson, & Williams, 1999; Williams & Collins, 1995). Perhaps, as education levels increase, black adults become more aware of the social injustices and racial discrimination that influence their life-chances. This social awareness of hardships faced in everyday life could be internalized and manifest itself in self-rated health. Those who sense they are not in control of their life-chances could be transferring their feelings of frustration into their interpretation of health. Of course, further research is needed to examine such a mechanism.

The rising expectations hypothesis (Geschwender, 1964) and status incongruence hypothesis ( Dressler, 1996) may also provide insight into these findings. The original rising expectations hypothesis outlined that increasing expectations will lead to dissatisfaction and attempts to change the social order. It is frequently the case that increased education, income, and occupational prestige bring higher expectations for a certain standard of living. If their expectations were not realized in life, the inconsistency would increase levels of distress. As outlined by Dressler (1996), it is the discrepancy between one's lifestyle and economic status, or the "status incongruence" which can lead to conflict and stress. High levels of distress could, in turn, have a negative impact on self-rated health ( Farmer & Ferraro, 1997) and ultimately the incongruence could lead to health problems such as hypertension ( Appel et al., 2002; Dressler, 1996).

Wilson's (1980) controversial research on the declining significance of race in determining life chances also drew attention to the unique hardships faced by poor minority persons. Willie (1989b) argued, however, that the significance of race is actually increasing, especially for middle-class black adults. He noted that even racial minorities who are educated experience important disadvantages in society. Findings from the present research show that all people with lower levels of education, income, or occupation suffered health consequences, but that black people did not reap the same health benefit from higher SES. Perhaps, individuals with lower levels of SES are faced with so many other difficulties in employment, adequate housing, and budgeting for meals that their economic situation defines their health status ( Winkleby & Cubbin, 2003). Once issues dealing with the basic standard of living are met with increasing levels of SES, racial inequalities in health care access, health promotion and occupational segregation play a more prominent role in health status. Whatever the case, race remains an important risk marker for a host of health outcomes ( Appel et al., 2002).

The interpretation of these findings should be considered in the context of a two study limitations. First, there was a substantial amount of missing data on occupational prestige. Supplemental analyses showed that the respondents missing on occupational prestige were mostly women and many reported working status to be "home maker." Although we attempted to account for the missing data in the models, it is still possible that women who did not work outside the home were not adequately represented in the analysis. The second limitation is that the data allow for research on white and black Americans only. The sample size is not adequate to examine other racial/ethnic groups nor is there any information on ethnic origins of the black and white population. Future research is needed to examine these hypotheses beyond the confines of the black–white dichotomy.

The present research adds evidence to the emerging body of research showing that the effect of race on health is conditional on SES. Stated differently, the SES gradient on health is distinct for black and white people. Moreover, this research identified two alternative forms of the potential interaction of race and SES on health outcomes—minority poverty and diminishing returns—and tested for these hypotheses. The results provide some evidence for the diminishing returns hypothesis on self-rated health and document a substantial and enduring racial gap in health status over 20 years of observation.