Perceptions of racial bias have been linked to poorer circulatory health among Blacks compared with Whites. However, little is known about whether Whites’ actual racial bias contributes to this racial disparity in health. We compiled racial-bias data from 1,391,632 Whites and examined whether racial bias in a given county predicted Black-White disparities in circulatory-disease risk (access to health care, diagnosis of a circulatory disease; Study 1) and circulatory-disease-related death rate (Study 2) in the same county. Results revealed that in counties where Whites reported greater racial bias, Blacks (but not Whites) reported decreased access to health care (Study 1). Furthermore, in counties where Whites reported greater racial bias, both Blacks and Whites showed increased death rates due to circulatory diseases, but this relationship was stronger for Blacks than for Whites (Study 2). These results indicate that racial disparities in risk of circulatory disease and in circulatory-disease-related death rate are more pronounced in communities where Whites harbor more explicit racial bias.

Blacks die at a higher rate than Whites from circulatory-related diseases (e.g., heart disease; U.S. Department of Health and Human Services, 2014). Prominent theories have suggested that one cause of this disparity is that Blacks experience more discrimination, which leads to stress, which in turn has negative health consequences (Clark, Anderson, Clark, & Williams, 1999; Hatzenbuehler, Phelan, & Link, 2013; Major, Mendes, & Dovidio, 2013). Studies supporting this view have found that the perception of discrimination is associated with anxiety, cardiovascular threat response, hypertension, and mortality (Barnes et al., 2008; Mendoza-Denton, Downey, Purdie, Davis, & Pietrzak, 2002; Pascoe & Richman, 2009; Sawyer, Major, Casad, Townsend, & Mendes, 2012; Williams & Mohammed, 2009).

Although this previous work suggests that perceived racial bias contributes to disparities between Blacks’ and Whites’ health, less is known about whether the actual racial bias of Whites contributes to these disparities. A deeper understanding of links between Whites’ racial bias and racial health disparities could help identify communities in greatest need of prejudice-prevention and health-promotion interventions. Therefore, the aim of the current research was to establish whether Black-White disparities in risk of circulatory disease and in circulatory-disease-related death rate are greater in communities where Whites harbor more racial bias.

Study 2 Section: Choose Top of page Abstract Whites’ Racial Biases and... Study 1 Study 2 << General Discussion References CITING ARTICLES In Study 2, we examined whether the Black-White disparity in rate of death due to circulatory diseases was more pronounced in counties where Whites harbored more racial bias. Data sources Racial bias Explicit and implicit racial biases were estimated with the methods described for Study 1. Death rate Death records for Blacks and Whites were obtained from the CDC (2014). Specifically, we examined rates (per 100,000) of death from circulatory diseases (e.g., heart disease; Internal Statistical Classification of Diseases and Related Health Problems codes I00–I99) in 2003 through 2013. Circulatory diseases are the leading cause of death in the United States and have shown pervasive racial disparities in prevalence over time (U.S. Department of Health and Human Services, 2014). To account for potential age differences between counties and racial groups and allow for more meaningful comparisons, we used age-adjusted death rates. To derive these age-adjusted rates, we used the default 2000 U.S. standard population detailed in Anderson and Rosenberg (1998). We aggregated the data across males and females, given that sex did not moderate the effects of bias in Study 1, and aggregation minimized missing data (i.e., aggregated data were less likely to be suppressed by the CDC than were separated data for males and females). We were able to obtain death rates for Whites in 3,110 counties and death rates for Blacks in 1,490 counties. Data were unavailable for counties recording fewer than 10 deaths for a given group. Figure 1b shows the location of the 1,149 counties for which we obtained racial-bias data and death-rate data for both Blacks and Whites. To determine whether racial bias predicted Black-White disparities in deaths from causes other than circulatory diseases, we used the CDC (2014) data set to compile county-level data on age-adjusted rates of death due to neoplasm (e.g., cancer) for 2003 through 2013. We focused on death due to neoplasm because neoplasms are the second most prevalent cause of death in the United States (U.S. Department of Health and Human Services, 2014). To isolate effects of interest, we incorporated the same set of county-level covariates used in Study 1, except for sex (because death rates were aggregated across males and females) and age (because death rates were already age adjusted). Results As in Study 1, we used GEEs to estimate all effects. Data from a given county were weighted in the analyses by the number of respondents who completed the racial-bias measure in that county. As in Study 1, race was coded as 0 for Black and 1 for White, and all other predictors were mean-centered. Figure 2c shows the relationship between Whites’ explicit racial bias and Blacks’ and Whites’ death rates due to circulatory diseases before accounting for covariates. To test whether racial disparities in death rate due to circulatory diseases were more pronounced in counties where White respondents showed greater racial bias, even after controlling for a set of county-level covariates, we regressed circulatory-disease-related death rate on race, explicit racial bias, implicit racial bias, the Race × Explicit Racial Bias interaction, the Race × Implicit Racial Bias interaction, all covariates, and the interactions between race and the covariates (Table 3). The effect of explicit racial bias was qualified by the Race × Explicit Racial Bias interaction. Simple-slopes analyses indicated that the relationship between explicit racial bias and rate of death due to circulatory diseases was positive for both Blacks and Whites, but stronger for Blacks, b = 43.200, SE = 12.100, z = 3.559, p = .0004, than for Whites, b = 13.900, SE = 4.970, z = 2.795, p = .0052. As in Study 1, the Race × Explicit Racial Bias interaction was significant over and above the effects of age bias, which suggests that racial bias specifically was related to Black-White disparities in death rate. Neither the main effect for implicit racial bias nor the Race × Implicit Racial Bias interaction was significant. Table 3. Results of the Generalized Estimating Equation Analysis Predicting Death Rate Due to Circulatory Diseases in Study 2 View larger version With all covariates included in the model, in counties with high (1 SD above the mean) explicit racial bias, the difference between Blacks’ and Whites’ death rates was 62 deaths per 100,000. In contrast, in counties with low (1 SD below the mean) explicit bias, the difference was 35 deaths per 100,000. Furthermore, adjusting for all covariates, we estimated how many more Blacks died of circulatory-related diseases annually in counties that were high (1 SD above the mean), rather than low (1 SD below the mean), in explicit racial bias. We made this estimate at the average Black population level in counties for which we had death-rate data for Blacks (average Black population = 28,598); 11 more Blacks per county were predicted to die annually in high-explicit-bias counties (95 deaths) than in low-explicit-bias counties (84 deaths). To determine whether similar effects would emerge for death rate not due to circulatory diseases, we regressed rate of death due to neoplasm on the same predictors as in our analyses of deaths due to circulatory diseases. Neither the main effect of explicit racial bias nor the Race × Explicit Racial Bias interaction was significant, ps > .14. Moreover, when we modeled neoplasm death rate as a covariate in the model predicting death rate due to circulatory diseases, the Race × Explicit Racial Bias interaction remained significant, b = −21.100, SE = 9.240, z = 2.283, p = .0224. These findings suggest that that the relationship between explicit racial bias and death rate was specific to circulatory-related, and not neoplasm-related, disease. (See the Supplemental Material for additional analyses.) Discussion In counties where White respondents harbored more explicit racial bias, the rate of death from circulatory disease was increased for both Whites and Blacks. However, Whites’ explicit racial bias predicted this death rate more strongly for Blacks than for Whites. As in Study 1, explicit racial bias, compared with implicit racial bias, was a stronger predictor of Black-White health disparity.

General Discussion Section: Choose Top of page Abstract Whites’ Racial Biases and... Study 1 Study 2 General Discussion << References CITING ARTICLES In counties where White Project Implicit respondents harbored more explicit racial bias, Black-White disparities in access to affordable health care (Study 1) and rate of death due to circulatory diseases (Study 2) were more pronounced. The robustness of these findings was evidenced by the replication of the same pattern across two independent data sets that both included a large number of counties. Although the effects could have been driven by an unmeasured third variable, the fact that racial bias was a significant predictor in models with a large set of covariates supports the notion of a direct relationship between Whites’ racial bias and Black-White health disparities. To our knowledge, this is the first research to show that racial bias from a dominant group (e.g., Whites) predicts negative health outcomes more strongly for the target group (e.g., Blacks) than for the dominant group. These results are consistent with research that has shown that population-level bias (i.e., antigay attitudes), when aggregated across targeted and nontargeted groups, predicts negative health outcomes more strongly for the targeted than the nontargeted group (Hatzenbuehler, Bellatorre, et al., 2014). However, the current results extend this previous work, which did not directly address the issue of whether the effects were driven by the dominant group’s stigmatization of the targeted group or the targeted group’s self-stigmatization. Furthermore, by demonstrating a relationship between Whites’ racial biases and health outcomes of Blacks in the same community, these results support previous findings that Blacks’ subjective perceptions of racism are linked to their own health (Pascoe & Richman, 2009; Williams & Mohammed, 2009). However, because perceptions of racism can be shaped by race-based rejection sensitivity (Mendoza-Denton et al., 2002), we extended this previous work by measuring racial bias directly from Whites. Additionally, to our knowledge, this is the first research to use geo-coded data on implicit bias to examine whether racial health disparities are more pronounced in communities where Whites show more implicit bias. Notably, when implicit bias was modeled without explicit bias, it showed patterns similar to those for explicit bias (see Tables S4 and S6 in the Supplemental Material), but when explicit and implicit biases were modeled together, only explicit bias predicted health outcomes. The null effects of implicit bias are informative, as research is increasingly focusing on the predictive utility of implicit bias in medical contexts (e.g., Hall et al., 2015), and it is important to identify the outcomes that are more strongly related to explicit than to implicit bias. One potential explanation for why Whites’ explicit bias was a stronger predictor than their implicit bias in the current work is that explicit bias has historically exerted stronger effects on the structural factors (e.g., regulation of environmental pollution) and psychological factors (e.g., overtly negative interracial interactions) that ultimately shape health outcomes. One limitation of this research is that the respondents who completed Project Implicit’s racial-bias measures might not have been representative of their counties on all dimensions. For instance, Project Implicit respondents might not reflect racial biases of older community members. Though we employed a poststratification weighting scheme designed to circumvent this limitation, no amount of poststratification weighting can make a sample truly representative on all dimensions. Thus, future research should examine whether the observed effects remain when bias is measured with full probability sampling. Another limitation is that we did not have data on the geographic mobility of specific individuals for whom we assessed health outcomes. It is possible that the deceased in Study 2 had moved to their communities soon before their deaths and had died before the racial bias of their communities had any effect on their health. However, two lines of evidence suggest that this scenario occurs relatively infrequently. First, 95% of the people who died from circulatory-disease-related causes were over age 55 (CDC, 2014), and the median duration of residency for people in this age group is more than 11 years (Mateyka & Marlay, 2011). Second, when people of this age do move, they are more likely to move to another residence within the same county than to a different county (data from the U.S. Census Bureau’s 2009–2013 ACS: report titled Geographic Mobility by Selected Characteristics in the United States). Together, this evidence suggests that many people are geographically stable in the years leading to their death. Nevertheless, future research might further examine the role of geographic mobility in the relationship between racial bias and health. The finding that Whites’ circulatory-disease-related death rate was increased in counties where White respondents harbored greater explicit bias is consistent with research showing that racial bias is linked to negative health outcomes for Whites (Kennedy et al., 1997; Lee et al., 2015; Mendes, Gray, Mendoza-Denton, Major, & Epel, 2007). One explanation for this finding, suggested by recent research (Lee et al., 2015), is that highly biased communities have decreased social capital (i.e., trust and bonding between community members), which in turn predicts negative health outcomes. Though the current research does not establish that White respondents’ racial bias caused the observed health disparities between Whites and Blacks, the relationships between Whites’ racial bias and Black-White health disparities were independent of a large set of county-level socio-demographic characteristics. Thus, our findings raise compelling questions about the mechanisms through which Whites’ racial bias can be related to health outcomes. On the basis of existing theoretical frameworks (Clark et al., 1999; Hatzenbuehler et al., 2013; Major et al., 2013), we posit that multiple causal pathways might account for this relationship. These pathways might include structural (e.g., discrimination in health care), interpersonal (e.g., hostile interactions), emotional (e.g., stress), and behavioral (e.g., maladaptive coping) processes that catalyze biological systems that increase disease risk. We hope that the current work serves to generate future research examining why there is a relationship between racial bias and health.

Acknowledgements We are grateful to Doc Edge for providing insight into our statistical analyses.

Action Editor

Jamin Halberstadt served as action editor for this article. Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article. Funding

This research was supported by the National Science Foundation under Award Numbers 1306709 (to R. Mendoza-Denton and O. Ayduk) and 1514510 (to O. Ayduk and J. B. Leitner), and by a Social Sciences and Humanities Research Council Institutional Grant (to E. Hehman). Supplemental Material

Additional supporting information can be found at http://pss.sagepub.com/content/by/supplemental-data Open Practices

All data and materials have been made publicly available via the Open Science Framework and can be accessed at https://osf.io/5ybhd/. The complete Open Practices Disclosure for this article can be found at http://pss.sagepub.com/content/by/supplemental-data. This article has received badges for Open Data and Open Materials. More information about the Open Practices badges can be found at https://osf.io/tvyxz/wiki/1.%20View%20the%20Badges/ and http://pss.sagepub.com/content/25/1/3.full.

Notes 1.

Unless otherwise noted, all data from the U.S. Census Bureau were downloaded from factfinder.census.gov.