Racial disparities in health are well-documented and represent a significant public health concern in the US. Racism-related factors contribute to poorer health and higher mortality rates among Blacks compared to other racial groups. However, methods to measure racism and monitor its associations with health at the population-level have remained elusive. In this study, we investigated the utility of a previously developed Internet search-based proxy of area racism as a predictor of Black mortality rates. Area racism was the proportion of Google searches containing the “N-word” in 196 designated market areas (DMAs). Negative binomial regression models were specified taking into account individual age, sex, year of death, and Census region and adjusted to the 2000 US standard population to examine the association between area racism and Black mortality rates, which were derived from death certificates and mid-year population counts collated by the National Center for Health Statistics (2004–2009). DMAs characterized by a one standard deviation greater level of area racism were associated with an 8.2% increase in the all-cause Black mortality rate, equivalent to over 30,000 deaths annually. The magnitude of this effect was attenuated to 5.7% after adjustment for DMA-level demographic and Black socioeconomic covariates. A model controlling for the White mortality rate was used to further adjust for unmeasured confounders that influence mortality overall in a geographic area, and to examine Black-White disparities in the mortality rate. Area racism remained significantly associated with the all-cause Black mortality rate (mortality rate ratio = 1.036; 95% confidence interval = 1.015, 1.057; p = 0.001). Models further examining cause-specific Black mortality rates revealed significant associations with heart disease, cancer, and stroke. These findings are congruent with studies documenting the deleterious impact of racism on health among Blacks. Our study contributes to evidence that racism shapes patterns in mortality and generates racial disparities in health.
Citation: Chae DH, Clouston S, Hatzenbuehler ML, Kramer MR, Cooper HLF, Wilson SM, et al. (2015) Association between an Internet-Based Measure of Area Racism and Black Mortality. PLoS ONE 10(4): e0122963. doi:10.1371/journal.pone.0122963
Academic Editor: Hajo Zeeb, Leibniz Institute for Prevention Research and Epidemiology (BIPS), GERMANY
Received: September 11, 2014; Accepted: February 7, 2015; Published: April 24, 2015
Copyright: © 2015 Chae et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: Interested researchers can apply to access publicly available mortality data from the National Center for Health Statistics at the following website: https://naphsis-web.sharepoint.com/Pages/VitalStatisticsDataResearchRequestProcess.aspx. Data on area-level variables from the American Community Survey can be downloaded through the U.S. Census Bureau at http://www.census.gov/acs/www/data_documentation/data_main/. Internet query data were from a previous study (Stephens-Davidowitz S (2014) The cost of racial animus on a black candidate: Evidence using Google search data. J Public Econ 118: 26–40.) obtained from a third-party commercial entity, Google, Inc. The data are available upon request by contacting email@example.com.
Funding: DHC was supported by the National Institute on Aging of the National Institutes of Health under Award Number K01AG041787. BGL was supported by grants from the Robert Wood Johnson Foundation during the course of the study. SIS-D received funding in the form of salary from Google, Inc. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Robert Wood Johnson Foundation.
Competing interests: During the time of this research, SIS-D was at Harvard University. He was also a quantitative analyst at Google, Inc. SIS-D constructed the area racism measure prior to his affiliation with Google, Inc., and was not with Google, Inc. when the initial version of this manuscript was completed. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials. No other potential competing interests or financial disclosures were reported by the authors of this paper.
A growing body of evidence indicates that the unique constellation of environmental stressors and psychosocial challenges experienced by Blacks in the US contributes to accelerated declines in health and generates racial disparities . Of these stressors, there has been increasing attention to the impact of racism-related factors, including interpersonal experiences of racial discrimination . Racially motivated experiences of discrimination impact health via diminished socioeconomic attainment and by enforcing patterns in racial residential segregation, geographically isolating large segments of the Black population into worse neighborhood conditions . These areas are typically characterized by social anathemas such as poverty and crime, and fewer health-promoting resources, including recreational facilities, parks, supermarkets, and quality healthcare. Such characteristics shape health behaviors such as exercise, diet, and substance use [4, 5]. Racial discrimination in employment can also lead to lower income and greater financial strain, which in turn have been linked to worse mental and physical health outcomes .
In addition to these indirect socioeconomic and neighborhood effects, racial discrimination may also directly impact health by engaging psychobiological mechanisms induced in the stress response . Experiences of racially-motivated discrimination are inherently stressful, and may undermine psychological adjustment particularly when viewed as being outside of personal control, resulting in depression, anxiety, and anger . These negative affective and cognitive responses are associated with maladaptive health behaviors and greater risk of chronic disease. Racial discrimination can also affect health by eliciting a cascade of biochemical reactions that over time can damage biological systems . As a source of psychosocial stress, racial discrimination may lead to premature physiologic deterioration or “wear and tear,” ultimately compromising the ability of the body to respond to such challenges and increasing susceptibility to and acceleration of chronic diseases . Research on discrimination has found that it is related to a range of biological markers of stress, including measures of oxidative stress and inflammation [11–13].
Despite this accumulating evidence, a central dilemma in epidemiologic studies is related to the measurement of racism. This research has mostly relied on self-report measures of interpersonal experiences of discrimination, which faces limitations given the motivational ambiguity that characterizes contemporary discrimination . For example, although legally sanctioned forms of criminal profiling may not have an overt racial component, they have been applied inequitably and found to disproportionately impact racial minorities . Racial discrimination in employment and housing markets has also been found to occur covertly despite being expressly prohibited by law [16, 17]. Discrimination can also be experienced more subtly through interpersonal interactions, such as instances of being treated with less respect or courtesy, being followed in stores, or receiving poorer service at restaurants . In contrast to traditional, overtly racially motivated discrimination and prejudice, attributions of intent for these experiences are often less clear. Furthermore, individuals may not disclose experiences of discrimination, or may be unaware “invisible” discrimination and racist events that are not directly experienced . For instance, not all forms of discrimination have an identifiable perpetrator, e.g., when resumes from candidates for employment or housing applicants with “Black names” are systematically rejected. [19, 20]. These measurement challenges may account for inconsistent or null findings on the relationship between health outcomes and self-reported racial discrimination, which is often influenced by subjective appraisal of the motivation for unfair treatment. In fact, results from a systematic review revealed that nearly two-thirds of studies found no statistically significant relationship between self-reported racism experiences and physical health outcomes .
Other studies have examined racial segregation as an indicator of institutionalized racism, which to some extent is perpetuated by discriminatory practices and the racial preferences of Whites . While this line of research has provided important insights into the residential conditions experienced by Blacks, such measures as proxies for racial discrimination and prejudice are limited as they only indirectly capture racial attitudes in a geographic area.
In this study, we examined a previously developed Internet-based measure of “area racism” that did not rely on the provision of responses to survey questions and is less susceptible to social desirability bias; it may also more directly assess racial attitudes in a geographic area . This measure, calculated based on Internet search queries containing the “N-word”, was strongly associated with the differential in 2008 votes for Barack Obama, the Black Democratic presidential candidate, vs. 2004 votes for John Kerry, the White Democratic presidential candidate. Studies have found that Internet searches on other subjects, including religiosity and firearms, reflect socio-demographic characteristics of the underlying population [24, 25]. For example, the percent of a state’s residents believing in God explains 65% of the variation in search volume for the word “God” . Internet queries of health conditions have also been used for disease surveillance, including influenza outbreaks, and have been found to be a stronger predictor than pharmacy records [26, 27]. Socially unacceptable attitudes or actions may also be less likely to be censored on the Internet given perceptions of anonymity, and may in fact serve as an outlet for unpopular beliefs.
An Internet search-based measure of area racism may serve as a more direct indicator of racial attitudes and the extent of discrimination and prejudice towards Blacks in a geographic area, including those experiences of racially motivated bias that are subtle or not observable, and which are not necessarily reported in survey instruments. Accordingly, such measures may improve our ability to investigate racism and help us to examine its consequences for health disparities at the population-level. The purpose of this study was to examine the association between an Internet search-based measure of area racism and Black mortality rates in the US.
The unit of analysis in this study is the media market or designated market area (DMA) as defined by Nielsen Media Research. There are a total of 210 DMAs in the US, which are geographic areas receiving common television broadcasts or radio programming; they may also receive other similar media content, such as newspapers or Internet advertising. Hence, although there is heterogeneity within DMAs, they represent a coherent unit of analysis because residents may be exposed to similar social messages influencing racial attitudes.
The main predictor variable of this study is area racism within DMAs . This previously developed variable was calculated as the proportion of total Internet search queries containing the “N-word” (singular or plural) using Google from 2004–2007. More colloquial forms of the “N-word” (i.e., ending in “-a” or “-as”) were not included given prior analysis of top searches revealing that these versions were used in different contexts compared to searches of the term ending in “-er” or “-ers” . Data on area racism were unavailable for Alaska and small DMAs, resulting in a total of 196 DMAs included in analyses. Area racism was standardized so that a one-unit increase in area racism indicates a one standard deviation higher proportion of Google searches containing the “N-word.” Supporting the validity of this measure were statistically significant correlations found with explicit measures of racial attitudes (i.e., a measure of beliefs about interracial marriage included in the General Social Survey) and area-based measures of racial composition (i.e., the percent of Blacks living in an area) . Google query data for this study are available upon request.
Our primary outcome was mortality rates within DMAs for Blacks age 25 and over. Data included information from death certificates and mid-year population counts from 2004–2009 collated by the National Center for Health Statistics (NCHS) . Data are publicly available. Interested researchers can apply to the National Association for Public Health Statistics and Information Systems to access the data at the following website: https://naphsis-web.sharepoint.com/Pages/VitalStatisticsDataResearchRequestProcess.aspx. We examined deaths among those 25 years of age and older, which represent the largest burden of mortality from chronic diseases that are likely to be influenced by social stressors stemming from racism. Across this period, among Blacks there was an average of 23.1 million person-years and approximately 270,000 deaths from all causes each year. For each year, DMA-specific age- and sex-adjusted Black mortality rates weighted using the US 2000 standard population were calculated per 100,000 person years.
We analyzed data on all-cause mortality and the four leading specific causes of death among Blacks identified using International Classification of Disease, Version 10 codes: heart disease (I00-I09, I11, I13, I20-I51); cancer (C00-C97); stroke (I60-I69); and diabetes (E11-E14) . To adjust for relevant area-level covariates, we used data from the American Community Survey (2004–2009), which were merged with and aggregated within DMA-level group identifiers . These data are publicly accessible through the U.S. Census Bureau and can be downloaded at http://www.census.gov/acs/www/data_documentation/data_main/. We examined the following DMA-level covariates: percent of the population that is Black; urbanicity, defined as the percent of people within a DMA living in a city with 50,000 or more people; the percent of Blacks with up to a high school education; and the percent of Black households living in poverty. We also adjusted for all-cause and cause-specific mortality rates among Whites 25 years of age and older, which were calculated using 2004–2009 NCHS data. Using similar methods, age and sex-adjusted mortality rates weighted using the US 2000 standard population were calculated among Whites. White mortality rates per 100,000 person-years were calculated each year at the DMA-level.
Our primary outcomes were derived based on the number of deaths within DMAs and are thus best modeled using analytic methods appropriate for examining count data. Negative binomial regression models were used to examine the association between area racism and Black mortality rates. This was used in lieu of Poisson regression because data are over-dispersed (α>0), which occurs when the variance is larger than the mean . Over-dispersion was evident in all of our models; however, when data are not over-dispersed, results derived from negative binomial and Poisson regression are equivalent. Mid-year population counts were used to model the population at risk of death. Because data include repeated observations within geographic regions, analyses were adjusted at the DMA-level using a Huber-White clustered standard error calculation . For ease of interpretation, mortality rate ratios [MRR], 95% confidence intervals 95 [CI], and p-values are provided. Model fit was assessed using Pseudo-R2 and accompanying p-values were calculated using an F-test. To compare models, we further provide the Akaike information criteria (AIC) and calculate the relative likelihood (RL), which estimates the probability that a nested model is better than its more parsimonious counterpart . We interpreted an RL<0.05 to indicate that the second model is preferable to its comparison. Statistical analyses were done using Stata 13/IC.
We specified four nested models: Model 1 examined the association between area racism and the Black mortality rate; Model 2 further adjusted for overall DMA-level covariates, namely, urbanicity and % Black; Model 3 incorporated race-specific education and poverty among Blacks at the DMA-level; and finally, Model 4 adjusted for the White mortality rate at the DMA-level to further adjust for unmeasured confounders that influence mortality overall in a geographic area, and to examine Black-White disparities in the mortality rate. In models predicting Black mortality rates from heart disease, cancer, stroke, and diabetes, we adjusted for corresponding White cause-specific mortality rates. All models accounted for the decedent’s age (in ten-year groups), sex, year of death, and US Census region of residence (Northeast, Midwest, South, and West), which were recorded from death certificates.
This study involved secondary analysis of existing data and therefore is exempt from human subjects review. Mortality and population data were collated by the National Center for Health Statistics and the US Census Bureau, which are anonymized and publicly available. Google query data were in aggregate at the DMA-level and contain no individual identifiers.
The distribution of area racism by DMAs is illustrated in Fig 1. Notably, this shows that greater proportions of Google search queries containing the “N-word” were concentrated in the rural Northeast and South of the US.
Descriptive characteristics of variables under investigation are shown in Table 1. On average, these indicate that around 11% of the population aged 25 and over in included DMAs was Black. More than two-thirds of the population lived in or around urban areas. Around 50% were aged 45 and older. Approximately 80% of Blacks had up to a high school degree or equivalent, and almost 30% lived below the poverty line. Across DMAs, Blacks had higher all-cause and cause-specific mortality rates (heart disease, stroke, cancer, and diabetes) compared to Whites. Rates per 100,000 person-years within each of these racial groups are shown in Table 1.
Results from negative binomial regression predicting the all-cause Black mortality rate per 100,000 person years adjusting for age, sex, year of death, and region are shown in Table 2 (Model 1). We found a significant positive relationship between area racism and the Black all-cause mortality rate (MRR = 1.082; 95% CI = 1.056, 1.108; p<0.001). Adding urbanicity and the percent of Blacks significantly improved model fit (Model 2), and attenuated the effect of area racism on the all-cause Black mortality rate (MRR = 1.076; 95% CI = 1.052, 1.10; p<0.001). Adding socioeconomic variables (education and poverty among Blacks) also resulted in a better fitting model (Model 3) and diminished the effect of area racism, but it remained a significant predictor (MRR = 1.057; 95% CI = 1.034, 1.080; p<0.001). Finally, adding the White all-cause mortality rate significantly improved the fit of the model (Model 4). Area racism continued to be significantly associated with the Black all-cause mortality rate (MRR = 1.036; 95% CI = 1.015, 1.057; p = 0.001).
Also in this final model, increasing age was associated with greater Black mortality rates; for example, with those aged 65–74 (MRR = 17.594, 95% CI = 17.031, 18.176; p<0.001) having greater morality rates compared to those aged 25–34 (the reference). Men also had greater morality rates compared to women (MRR = 1.362; 95% CI = 1.284, 1.444; p<0.001). There was a decrease in the Black mortality rate over time (MRR = 0.980; 95% CI = 0.977, 0.983; p<0.001). Compared to the Northeast, residence in the Midwest (MRR = 1.135; 95% CI = 1.050, 1.227; p = 0.001) and South (MRR = 1.120; 95% CI = 1.040, 1.205; p = 0.003) were associated with a greater Black mortality rate; there was no significant difference in the Black mortality rate associated with living in the West compared to the Northeast (MRR = 1.036; 95% CI = 0.947, 1.149; p = 0.387).
Results from models examining cause-specific mortality rates for leading causes of death among Blacks are summarized in Table 3. These show estimates from models equivalent to final models in Table 2, but adjusting for corresponding White cause-specific mortality rates. Area racism was significantly associated with Black mortality rates for heart disease (MRR = 1.043; 95% CI = 1.015, 1.072; p = .003), cancer (MRR = 1.026; 95% CI = 1.002, 1.050; p = 0037), and stroke (MRR = 1.033; 95% CI = 1.002, 1.065; p = .038). Area racism was not significantly associated with diabetes mortality rates (MRR = 0.946; 95% CI = 0.878, 1.019; p = 0.141).
Sensitivity analyses were conducted examining alternative years of mortality data: 2004–2007, 2005–2008, and 2006–2009. These were conducted in order to examine whether considering shorter four-year consecutive periods of mortality data that were concurrent with the years for Internet query data (2004–2007) or occurred in future years would change our results. The magnitude of associations with area racism for each model when using alternative years of mortality data did not change the MRR by more than 10%. Further, the statistical significance of area racism was the same for Black all-cause mortality in all models, as well as for the same specific causes of death when using the mortality data from 2004–2009.
Results from our study indicate that living in an area characterized by a one standard deviation greater proportion of racist Google searches is associated with an 8.2% increase in the all-cause mortality rate among Blacks. This effect estimate amounts to over 30,000 deaths among Blacks annually nationwide. Although this magnitude was attenuated to 5.7% after adjusting for area-level demographic and socioeconomic covariates, particularly with the inclusion of education and poverty among Blacks, these factors are also influenced by racial prejudice and discrimination and therefore could be on the causal pathway. Area racism was statistically significant even after controlling for White mortality rates (3.6% greater mortality rate). These findings indicate that area racism, as indexed by the proportion of Google searches containing the “N-word”, is significantly associated with not only the all-cause Black mortality rate, but also Black-White disparities in mortality.
As a source of stress experienced across the life-course, racism contributes to “weathering” or accelerated declines in health across biological systems caused by the mobilization of resources to meet environmental demands, thus potentially impacting a range of health outcomes [7, 9]. For example, experimentally manipulated experiences of discrimination have been associated with cardiovascular reactivity, particularly in response to scenarios that were motivationally ambiguous . We found robust associations between area racism and heart disease, cancer, and stroke, leading causes of death among Blacks. Racial disparities in mortality from these diseases may be influenced by racism through biobehavioral channels engaged in the threat response . As a source of chronic psychosocial stress, repeated racism may result in a heighted pro-inflammatory state that can have particularly detrimental consequences for the etiology and progression of cardiovascular and other immune disorders . Studies on discrimination have found evidence for adverse consequences for hypertension, atherosclerosis, and their inflammatory mediators [11–13, 37–40]. A recent study found that racism-related factors may also be associated with accelerated aging at the cellular level .
We posit that an Internet search-based measure of racism at the DMA-level reflects the extent of racism in a geographic area, which has been shown to negatively impact health outcomes. Self-reports of discrimination may not capture the major and everyday social insults experienced by Blacks, particularly those without an identifiable perpetrator, or because the perceived reasons for negative treatment may be uncertain . Furthermore, explicit measures of racial attitudes are susceptible to the provision of socially desirable responses or self-censorship in reporting publicly unacceptable beliefs . Area-based measures of segregation and isolation also do not directly assess racial attitudes in a geographic area. Along these lines, Internet-search based proxies of underlying population attitudes may be particularly useful in examining beliefs and actions that are not socially sanctioned, such as those related to racism. Accordingly, such measures may be utilized to examine racism that is systematic, institutionalized, subtle, or otherwise hidden, which may not necessarily be captured in self-report instruments. This measure does not entail that all Internet queries containing the “N-word” are motivated by racist attitudes, or that everyone in a region with a racial bias conducts such searches. Using “big data” and aggregating millions of Internet searches yields a high signal-to-noise ratio despite potential sources of measurement error .
There are several caveats to our study to be noted. Given its ecological design, our results are limited in making inferences regarding causality. We cannot conclude that racism at either the individual or area-level necessarily causes deaths among Blacks or contributes to an increased risk of mortality among Blacks in a geographic area. Nevertheless, while these results cannot make such causal conclusions, they do question why the extent of Internet search queries containing the “N-word” is associated with Black mortality rates.
Because of the correlational nature of the data, there are also challenges in interpreting the direction of the associations that we found, i.e., the temporal sequence of our primary exposure and outcomes could not be determined. The mortality data we examined overlaps with the search timeframe of the Internet measure. However, sensitivity analyses revealed minor differences when considering alternative periods of mortality data. Although the effects of racism on health have been shown to accumulate over time, racism can also have more immediate effects on mortality (e.g., discrimination in health care settings) . Racism-related factors impact survival from chronic diseases and increase the risk of acute fatal events such as myocardial infarction [44, 45]. Furthermore, recent analyses suggest that current measures of racial bias are reflections of past negative attitudes and are relatively stable. Data on racial attitudes from 1975–2012 in the General Social Survey indicate that while the absolute level of racial bias changes across states, the relative position of states to each other does not substantially differ (K. Keyes, PhD, analysis of publicly available data, November 2013). Correlations were consistently greater than 0.60 (p<0.01 for all), indicating a high level of stability in the relative rank of a state for racial bias across the last 30+ years. Thus, there is suggestive evidence that the Internet search-based measure of area racism is also likely to be relatively invariant from earlier periods.
There are also limitations inherent in studying cause-specific mortality using death certificates, as was done here . In particular, causes of death are likely to be best identified when the cause is unique and obvious, while the potential for bias increases when comorbidities exist or when the cause of death is occluded by more proximal sources. Specifically, previous research indicates that sensitivity in detecting diabetes-related mortality through death certificates is sub-optimal, which may contribute to the lack of association found between area racism and diabetes mortality rates . Furthermore, post-hoc analyses suggested that this null finding may have been due to competing causes of death among Blacks which may mask an association. However, in order for substantive conclusions on other cause-specific outcomes to be artifacts of measurement error, death certificates would need to be systematically biased towards over-diagnosis of heart disease, cancer, and stroke among Blacks compared to Whites in areas with higher levels of area racism. Although our inability to explore this is a limitation of our study, examining all-cause mortality provides some protection against this potential bias.
Despite these limitations, our results have important implications for methods to measure and monitor racism, and to investigate its impact on mortality. Our findings resonate with those of other research documenting the impact of broader macro-social forces on health outcomes. Previous studies have suggested that political representation, social policies, and high coverage race-related events have an influence on health, mortality, as well as indicators of biological dysregulation among Blacks [48–52]. This emerging body of evidence lends additional credibility to our interpretation of findings. Our study is largely consistent with prior research documenting the detrimental health consequences of discrimination and prejudice, and advances directions for population-level health disparities research through the application of an Internet search-based measure of racism to examine associations with Black mortality.
Conceived and designed the experiments: DHC SC MLH. Analyzed the data: SC. Contributed reagents/materials/analysis tools: SC SIS-D BGL. Wrote the paper: DHC SC MLH MRK HLFC SMW SIS-D RSG BGL.
- 1. Chae DH, Nuru-Jeter AM, Francis D, Lincoln KD. Conceptualizing racial disparities in health: advancement of a socio-psychobiological approach. Du Bois Rev. 2011;8: 63–77. doi: 10.1017/s1742058x11000166
- 2. Williams DR, Neighbors HW, Jackson JS. Racial/ethnic discrimination and health: findings from community studies. Am J Public Health. 2008;98: S29–37. pmid:18687616 doi: 10.2105/ajph.98.supplement_1.s29
- 3. Williams DR, Collins C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep. 2001;116: 404–416. pmid:12042604 doi: 10.1016/s0033-3549(04)50068-7
- 4. Dinwiddie GY, Gaskin DJ, Chan KS, Norrington J, McCleary R. Residential segregation, geographic proximity and type of services used: evidence for racial/ethnic disparities in mental health. Soc Sci Med. 2013;80: 67–75. doi: 10.1016/j.socscimed.2012.11.024. pmid:23312305
- 5. Duncan DT, Kawachi I, White K, Williams DR. The geography of recreational open space: influence of neighborhood racial composition and neighborhood poverty. J Urban Health. 2013;90: 618–631. doi: 10.1007/s11524-012-9770-y. pmid:23099625
- 6. Darity WA Jr. Employment discrimination, segregation, and health. Am J Public Health. 2003;93: 226–231. pmid:12554574 doi: 10.2105/ajph.93.2.226
- 7. Clark R, Anderson NB, Clark VR, Williams DR. Racism as a stressor for African Americans. A biopsychosocial model. Am Psychol. 1999;54: 805–816. pmid:10540593 doi: 10.1037//0003-066x.54.10.805
- 8. Pittman CT. Getting mad but ending up sad: the mental health consequences for African Americans using anger to cope with racism. J Black Stud. 2011;42: 1106–1124. pmid:22165423 doi: 10.1177/0021934711401737
- 9. Geronimus AT, Hicken M, Keene D, Bound J. "Weathering" and age patterns of allostatic load scores among blacks and whites in the United States. Am J Public Health. 2006;96: 826–833. pmid:16380565 doi: 10.2105/ajph.2004.060749
- 10. McEwen BS. Stress, adaptation, and disease. Allostasis and allostatic load. Ann N Y Acad Sci. 1998;840: 33–44. pmid:9629234 doi: 10.1111/j.1749-6632.1998.tb09546.x
- 11. Cooper DC, Mills PJ, Bardwell WA, Ziegler MG, Dimsdale JE. The effects of ethnic discrimination and socioeconomic status on endothelin-1 among blacks and whites. Am J Hypertens. 2009;22: 698–704. doi: 10.1038/ajh.2009.72. pmid:19390511
- 12. Szanton SL, Rifkind JM, Mohanty JG, Miller ER 3rd, Thorpe RJ, Nagababu E, et al. Racial discrimination is associated with a measure of red blood cell oxidative stress: a potential pathway for racial health disparities. Int J Behav Med. 2012;19: 489–495. doi: 10.1007/s12529-011-9188-z. pmid:21913047
- 13. Lewis TT, Aiello AE, Leurgans S, Kelly J, Barnes LL. Self-reported experiences of everyday discrimination are associated with elevated C-reactive protein levels in older African-American adults. Brain Behav Immun. 2010;24: 438–443. doi: 10.1016/j.bbi.2009.11.011. pmid:19944144
- 14. Krieger N, Waterman PD, Kosheleva A, Chen JT, Carney DR, Smith KW, et al. Exposing racial discrimination: implicit & explicit measures—the My Body, My Story study of 1005 US-born black & white community health center members. PLoS One. 2011;6: e27636. doi: 10.1371/journal.pone.0027636. pmid:22125618
- 15. Gelman A, Fagan J, Kiss A. An analysis of the New York City Police Department's "Stop-and-Frisk" policy in the context of claims of racial bias. J Am Stat Assoc. 2007;102: 813–823. doi: 10.1198/016214506000001040
- 16. Kau J, Keenan D, Munneke H. Racial Discrimination and Mortgage Lending. J Real Estate Financ. 2012;45: 289–304. doi: 10.1007/s11146-011-9330-3
- 17. US Equal Employment Opportunity Commission. EEOC Files Suit Against Two Employers for Use of Criminal Background Checks. 11 Jun 2013. Available: http://www.eeoc.gov/eeoc/newsroom/release/6-11-13.cfm. Accessed 2013 Oct 24.
- 18. Williams DR, Yan Y, Jackson JS, Anderson NB. Racial Differences in Physical and Mental Health: Socio-economic Status, Stress and Discrimination. J Health Psychol. 1997;2: 335–351. doi: 10.1177/135910539700200305. pmid:22013026
- 19. Bertrand M, Mullainathan S. Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. National Bureau of Economic Research. 2003. Available: http://www.nber.org/papers/w9873.pdf.
- 20. Dovidio JF, Gaertner SL. Aversive racism and selection decisions: 1989 and 1999. Psychol Sci. 2000;11: 315–319. pmid:11273391 doi: 10.1111/1467-9280.00262
- 21. Paradies Y. A systematic review of empirical research on self-reported racism and health. Int J Epidemiol. 2006;35: 888–901. pmid:16585055 doi: 10.1093/ije/dyl056
- 22. Bobo L, Zubrinsky CL. Attitudes on residential integration: Perceived status differences, mere in-group preference, or racial prejudice? Soc Forces. 1996;74(3): 883–909. doi: 10.1093/sf/74.3.883
- 23. Stephens-Davidowitz S. The cost of racial animus on a black candidate: Evidence using Google search data. J Public Econ. 2014;118: 26–40. doi: 10.1016/j.jpubeco.2014.04.010
- 24. Choi H, Varian H. Predicting the Present with Google Trends. Econ Rec. 2012;88: 2–9. doi: 10.1111/j.1475-4932.2012.00809.x
- 25. Scheitle CP. Google's Insights for Search: A Note Evaluating the Use of Search Engine Data in Social Research. Soc Sci Q. 2011;92: 285–295. doi: 10.1111/j.1540-6237.2011.00768.x
- 26. Carneiro HA, Mylonakis E. Google trends: a web-based tool for real-time surveillance of disease outbreaks. Clin Infect Dis. 2009;49: 1557–1564. doi: 10.1086/630200. pmid:19845471
- 27. Ginsberg J, Mohebbi MH, Patel RS, Brammer L, Smolinski MS, Brilliant L. Detecting influenza epidemics using search engine query data. Nature. 2009;457: 1012–1014. doi: 10.1038/nature07634. pmid:19020500
- 28. National Center for Health Statistics. Compressed Mortality File, 1968–2009 (machine readable data file and documentation, CD-ROM Series 20, Version 2A, 2E, 2N) as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. Hyattsville, MD; 2012.
- 29. Murphy SL, Xu JQ, Kochanek KD. Deaths: Final data for 2010. National Vital Statistics Reports. Hyattsville, MD: National Center for Health Statistics; 2013. pmid:24979975
- 30. US Census Bureau. Summary File. 2004–2009 American Community Survey. US Census Bureau’s American Community Survey Office; 2012. Accessed: http://www.census.gov/acs/www/data_documentation/data_main/. pmid:2004
- 31. Gardner W, Mulvey EP, Shaw EC. Regression analyses of counts and rates: Poisson, overdispersed Poisson, and negative binomial models. Psychol Bull. 1995;118: 392–404. pmid:7501743 doi: 10.1037//0033-2909.118.3.392
- 32. Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics. 2000; 56: 645–646. pmid:10877330 doi: 10.1111/j.0006-341x.2000.00645.x
- 33. Wagenmakers EJ, Farrell S. AIC model selection using Akaike weights. Psychon B Rev. 2004;11: 192–196. pmid:15117008 doi: 10.3758/bf03206482
- 34. Merritt MM, Bennett GG Jr., Williams RB, Edwards CL, Sollers JJ 3rd. Perceived racism and cardiovascular reactivity and recovery to personally relevant stress. Health Psychol. 2006;25: 364–369. pmid:16719608 doi: 10.1037/0278-6188.8.131.524
- 35. Duru OK, Harawa NT, Kermah D, Norris KC. Allostatic load burden and racial disparities in mortality. J Natl Med Assoc. 2012;104: 89–95. pmid:22708252
- 36. Hansel A, Hong S, Camara RJ, von Kanel R. Inflammation as a psychophysiological biomarker in chronic psychosocial stress. Neurosci Biobehav Rev. 2010;35: 115–121. doi: 10.1016/j.neubiorev.2009.12.012. pmid:20026349
- 37. Friedman EM, Williams DR, Singer BH, Ryff CD. Chronic discrimination predicts higher circulating levels of E-selectin in a national sample: the MIDUS study. Brain Behav Immun. 2009;23: 684–692. doi: 10.1016/j.bbi.2009.01.002. pmid:19171188
- 38. Cunningham TJ, Seeman TE, Kawachi I, Gortmaker SL, Jacobs DR, Kiefe CI, et al. Racial/ethnic and gender differences in the association between self-reported experiences of racial/ethnic discrimination and inflammation in the CARDIA cohort of 4 US communities. Soc Sci Med. 2012;75: 922–931. doi: 10.1016/j.socscimed.2012.04.027. pmid:22682683
- 39. Chae DH, Nuru-Jeter AM, Adler NE. Implicit racial bias as a moderator of the association between racial discrimination and hypertension: a study of Midlife African American men. Psychosom Med. 2012;74: 961–964. doi: 10.1097/PSY.0b013e3182733665. pmid:23107842
- 40. Lewis TT, Everson-Rose SA, Powell LH, Matthews KA, Brown C, Karavolos K, et al. Chronic exposure to everyday discrimination and coronary artery calcification in African-American women: the SWAN Heart Study. Psychosom Med. 2006;68: 362–368. pmid:16738065 doi: 10.1097/01.psy.0000221360.94700.16
- 41. Chae DH, Nuru-Jeter AM, Adler NE, Brody GH, Lin J, Blackburn EH, et al. Discrimination, racial bias, and telomere length in African-American men. Am J Prev Med. 2014;46: 103–111. doi: 10.1016/j.amepre.2013.10.020. pmid:24439343
- 42. National Research Council. Measuring Racial Discrimination. Washington, DC: National Academies Press; 2004.
- 43. National Research Council. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington, DC: National Academies Press; 2003.
- 44. Hayanga AJ, Zeliadt SB, Backhus LM. Residential segregation and lung cancer mortality in the United States. JAMA Surg. 2013;148: 37–42. doi: 10.1001/jamasurgery.2013.408. pmid:23324839
- 45. Chae DH, Lincoln KD, Adler NE, Syme SL. Do experiences of racial discrimination predict cardiovascular disease among African American men? The moderating role of internalized negative racial group attitudes. Soc Sci Med. 2010;71: 1182–1188. doi: 10.1016/j.socscimed.2010.05.045. pmid:20659782
- 46. Halanych JH, Shuaib F, Parmar G, Tanikella R, Howard VJ, Roth DL, et al. Agreement on cause of death between proxies, death certificates, and clinician adjudicators in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study. Am J Epidemiol. 2011;173: 1319–1326. doi: 10.1093/aje/kwr033. pmid:21540327
- 47. Cheng WS, Wingard DL, Kritz-Silverstein D, Barrett-Connor E. Sensitivity and specificity of death certificates for diabetes: as good as it gets? Diabetes Care. 2008;31: 279–284. pmid:17959866 doi: 10.2337/dc07-1327
- 48. Richman LS, Jonassaint C. The effects of race-related stress on cortisol reactivity in the laboratory: implications of the Duke lacrosse scandal. Ann Behav Med. 2008;35: 105–110. doi: 10.1007/s12160-007-9013-8. pmid:18347910
- 49. Laveist TA. Segregation, poverty, and empowerment: health consequences for African Americans. Milbank Q. 1993;71: 41–64. pmid:8450822 doi: 10.2307/3350274
- 50. Malat J, Timberlake JM, Williams DR. The effects of Obama's political success on the self-rated health of blacks, Hispanics, and whites. Ethn Dis. 2011;21: 349–355. pmid:21942169
- 51. Williams DR, Collins C. US socioeconomic and racial differences in health: patterns and explanations. Ann Rev Sociol. 1995;21: 349–386. doi: 10.1146/annurev.so.21.080195.002025
- 52. Cooper R, Steinhauer M, Schatzkin A, Miller W. Improved mortality among U.S. Blacks, 1968–1978: the role of antiracist struggle. Int J Health Serv. 1981;11: 511–522. pmid:7333722 doi: 10.2190/wjep-vb83-gaut-xejr