Search

Issues surrounding race and ethnicity have consistently shaped American political life. Understanding precisely how race has molded American political institutions, changed the debate over representative government, and persistently affected public opinion is the primary goal of this dissertation project. The first project in this dissertation looks at how local political resistance undermined one of the largest expansions of the franchise for minority voters in recent history. Specifically, I use newly digitized data from the Census of Governments to show that entrenched elites in counties where black citizens registered in larger numbers after the passage of the Voting Rights Act in 1965 responded by converting elective offices to appointive ones in order to prevent black voters from electing their most favored candidates. The second essay, co-authored with Rebecca Goldstein (Harvard University), focuses on another historical form of political repression: lynching. This form of political violence discouraged people from assembling to discuss or participate in politics. We use geo-located data on lynching throughout American history to explore the lasting effects of political violence against racial minorities. We connect this information to contemporary survey data and show that people who live close to areas where lynchings took place in the past are considerably more likely to experience voter intimidation or problems with their registration and voter identification. The third paper in this dissertation, co-authored with Jennifer Hochschild (Harvard University), concentrates on public perceptions of the relationship between race, genes, and violent behavior. Our questions in this paper are: how might people incorporate genetic information into their existing views about race, violent behavior, and punishment? Are people eager to adopt essentializing views of genetic underpinnings for violent behavior because they reify existing racial prejudices? Do partisans see things differently? We find that, while learning that the causes of violent behavior might be genetic affects liberals and conservatives very differently, people generally do not take a deterministic view of genes regardless of racial context. The final paper in this dissertation, written with Aaron Kaufman (Harvard University) and Gary King (Harvard University), proposes and validates a new approach to measuring a key feature of one of representative democracy's most important processes: legislative districting. The idea of ``compactness'' for legislative district shapes was introduced as one way of protecting against racial (or partisan) gerrymandering. However, since no precise legal definition of compactness exists, researchers have developed a panoply of proposed measures and left legal and political actors to adjudicate between them - often just by looking. In this paper, we use machine learning to develop a statistical model that measures and accurately predicts what people see as compact.

Coronary artery calcification (CAC), quantified by electron beam computed tomography, is an effective, non-invasive means for detection of extent of coronary atherosclerosis in asymptomatic; and symptomatic individuals. There are several aspects of CAC quantity that make genetic studies of it particularly challenging. These include the non-normality of CAC frequency distributions, the complex nature of the trait and the strong age- and sex-dependence of CAC levels. It is well recognized that coronary artery disease (CAD) and CAC are complex processes that are influenced by genetic, as well as environmental, factors. There has been a great deal of variation in results from studies attempting to identify specific genes for these traits. Methods targeted to detection of small, context-specific effects and which take into account the complex statistical distribution of this quantitative trait need to be implemented in order to allow detection of genetic effects influencing CAC quantity. In order to investigate genetic factors influencing CAC quantity, two candidate gene studies and a genome-wide scan were conducted based on samples from the Epidemiology of Coronary Artery Calcification Study and the Genetic Epidemiology Network of Arteriopathy (GENOA). The first paper, which presents results from a study examining two polymorphisms in the gene for the beta-2 adrenergic receptor, resulted in evidence that this gene may modify the relationship between certain established CAD risk factors (e.g. triglycerides) and CAC. The second paper provides support for a possible epistatic effect between the genes for angiotensinogen and angiotensin-converting enzyme in their influence on CAC. The genome-wide scan based on anonymous markers in a sample of 29 families enriched for hypertension detected regions on chromosomes 6 and 10 that may harbor genes involved in high sex- and age-specific levels of CAC. The results from this dissertation support a genetic component to CAC and demonstrate the need for non-traditional analysis approaches. Confirmatory analyses for all three studies are needed to substantiate and further clarify the associations reported here.