Date of Award


Document Type

Open Access Dissertation


Health Services and Policy Management


The Norman J. Arnold School of Public Health

First Advisor

Janice Probst


Racism can be understood as a system of dominance and power designed to uphold the racially privileged of society. One can be exposed to racism through various mechanisms including the Internet and social media. Evidence indicates that exposure to racism is associated with poor mental and physical health outcomes as well as unhealthy behaviors. Current literature studying the effects of racism focus primarily on experienced or perceived racism. Furthermore, the geographic influence of racism on health at the county level and across levels of rurality is currently unknown. The purpose of this study was twofold; 1.) To explore the geographic distribution of observed racism as across the US South and 2.) To determine the influence of rurality on the association between observed racism and black mortality in the US South.

Data for this study were retrospectively combined from a variety of sources. Allcause, age adjusted black mortality rates were derived from the Centers for Disease Control and Prevention Wide-ranging Online Data for Epidemiological Research (CDC WONDER). The independent variables of interest, racism and rurality, were obtained from Twitter and the 2015-2016 Area Health Resource File (AHRF), respectively. Socioeconomic and factors influencing access to care acted as controls in the study and were obtained from the AHRF at the county level. Income inequality ratios for each county were collected from the 2017 Robert Wood Johnson (RWJ) County Health Rankings. Variables measuring social capital were obtained from the publicly available Northeast Regional Center for Rural Development. Twitter data was provided by the Data on Local Life and You (DOLLY) project. Geotagged tweets at the county level were used as the measure of observed racism. A racist tweet was defined as one containing the racial slur “niggers”. Rurality was defined according to the 2013 rural-urban-Continuum-Codes (RUCC). The sample population included 653 counties across seven southern US states; Virginia, North Carolina, South Carolina, Georgia, Alabama, Mississippi, and Louisiana. Counties with no reported black mortality rate were excluded from analysis. Generalized linear modeling was used to test for association. Results were interpreted for significance at 95% confidence intervals (p < 0.05).

Results indicate that rural areas experience lower odds of racist tweeting, in comparison to their urban counterparts. The odds of racist tweeting behavior decreases as a county’s unemployment rate increases. The percent of voters in the 2012 Presidential Election was indirectly associated with racist tweeting behavior, suggesting a protective factor. Hotspots of racist tweeting can be observed in Mid-Georgia near Macon and among Columbia, South Carolina and surrounding counties. Observed racism was found to be significantly associated with black mortality. Counties with racist tweeting in the upper quartiles experienced significantly higher black mortality rates in comparison to counties in the lower quartile of racist tweeting. There was no significant interaction found between rurality and racist tweeting indicating that rurality does not influence the relationship between black mortality and racist tweeting. Rurality and socioeconomic factors like education, income inequality, and unemployment rates continue to be significant predictors of mortality. Furthermore, primary care physician (PCP) supply and hospital bed supply ratios are negatively associated with black mortality as results indicate that the black mortality rate increases as the ratio of PCPs and hospital beds increase.

The findings in this study indicate that rural areas are less likely to experience observed racism, in comparison to their urban peers. There are significant hotspots of racist tweeting occurring mainly in parts of Georgia and South Carolina. Furthermore, when accounting for geographic influence across levels of rurality, higher rates of observed racism is significantly associated with increased black mortality rates. These findings also indicate that even in the presence of racial discrimination, socioeconomic factors still remain significant in predicting black mortality, contributing to health disparities. Future research should explore the relationship between observed racism and mortality as well as other indicators of health further.

Available for download on Wednesday, December 18, 2019