Voting, Voter Suppression, Political Representation, and Constituents' Health

Abstract

Abstract This dissertation seeks to understand how political factors including voter suppression and gerrymandering affect public health. In chapter 1, I investigate the impact of the Voting Rights Act of 1965 on infant death rates in US Jim Crow states. Using mortality data from the National Center for Health Statistics from 1959, the earliest available year, through 1980 and population and demographic data from the 1950, 1960, 1970, and 1980 decennial censuses, I compare trends in infant deaths before and after 1965. I use a difference-in-differences design to obtain causal estimates of the effect of the VRA on infant death rates. Exposure groups are defined based on the criteria specified in Section 4 of the VRA. Within the Jim Crow states, counties that met the criteria, indicating that they had worse records of voter discrimination, were delineated as exposed, and counties that did not meet the criteria were delineated as unexposed. The results of Chapter 1 showed that in VRA-exposed counties Black infant deaths decreased by an average of 11.4 deaths per county beyond the decrease in Black infant deaths seen in unexposed counties between the pre-VRA period (1959-1965) and the post-VRA period (1966-1970). White infant deaths did not display the same statistically significant additional decrease in deaths in covered counties. These results provide evidence that political changes like the VRA can produce measurable changes in health outcomes. In Chapter 2, I look at how gerrymandering can distort public health metrics that are calculated by congressional districts in North Carolina. I combine US census data with the 50StateSimulation dataset created by the Algorithm-Assisted Redistricting Methodology group to look at how 2017-2021 aggregate infant death rates change when calculated using congressional district boundaries that are vs are not gerrymandered. The 50StateSimulation dataset provides a set of 5000 simulated congressional districts that comply with all state regulations but are not gerrymandered. We compare the simulated dataset to three actual enacted districting plans from 2021, 2022, and 2023 to investigate differences in infant death rate as well as between-congressional district variance and within-congressional district variance in rates. We find that under the 2023 plan, which showed strong evidence of gerrymandering, congressional districts that were outliers with regard to voting outcomes plans are also outliers with regard to infant death rates. The results of chapter two provide evidence that gerrymandering can produce districts with infant death rates that are outliers compared to what would have been produced under non-gerrymandered plans. In Chapter 3, I build on the analyses conducted in Chapter 2, to look at how gerrymandering alters understanding of uninsurance need across the US. In this set of analyses, we use the 50StateSimulation dataset as a non-gerrymandered comparison group. We also use data on medical uninsurance from the American Community Survey (2017-2021 five-year estimates) to look at how the percent uninsured, the within-district variance of percent uninsured, and the between-district variance of percent uninsured change when calculated using the enacted state districting plan vs the set of simulated non-gerrymandered redistricting plans. We find that compared to states with weak evidence of gerrymandering, states with strong evidence of partisan gerrymandering were more likely to contain congressional districts with more extreme values of uninsurance rates. We also found that in Republican gerrymandered states, there was a pattern of distortion, such that Republican-leaning congressional districts tended to have lower uninsurance rates and Democrat-leaning districts had higher uninsurance rates than the equivalent values under non-gerrymandered simulated plans. These results provide evidence that partisan gerrymandering can affect determination of CD-level uninsurance rates, distort understanding of public health burdens, and produce an overall more polarized understanding of uninsurance metrics. Across all three aims, I find evidence of some of the ways that political factors including voter suppression and gerrymandering impact public health. These findings underscore the importance of including political factors in public health research and of considering public health impact when changing or enacting new policies.