This dissertation uses spatial analyses to examine several contemporary issues within the field of political science. Overall, the bulk of information used within the field of political science to draw conclusions about the electorate has primarily come from survey data. This dissertation not only uses survey data but also relies on election data. Using precinct-level election data as opposed to county data, allows for a higher level of detail than is typically found in previous research. Conclusions from this dissertation rely on a spatial analysis of precincts in the fifteen counties of the Cincinnati Metropolitan Statistical Area.
Within the field of political science, survey data provide the means by which voting patterns and beliefs are most commonly analyzed, particularly among and between the groups in society. The data collected from surveys continues to provide critical information and trends in the electorate, however it has not been able to provide conclusive results about polarization and there remains much disagreement over the level of polarization within the electorate, whether there is a link between ideology and party and the degree of that link, and whether people who vote and think alike politically are living in close proximity to one another or dispersed amongst each other. Academics and pundits on opposite sides of the issue have produced empirical evidence from survey data to support their points of view on the matters. These same camps use state and county level data with inconclusive results. However, with the additions of spatial statistics in Geographic Information Systems, or GIS, there is an opportunity empirically analyze and provide new and statistically sound findings to lend new conclusions to old arguments.
This dissertation, using spatial analysis, examines the phenomena of geographic clustering of like-minded people, polarization, and ideological sorting within the American electorate using several sources. First, an analysis of presidential voting by residential location between the years of 1952-2000 as well the 2008 presidential election via the American National Election Studies (ANES) and between the years of 1968-2008 via the General Social Survey is provided. Second, a case study of voting results within the Cincinnati Metropolitan Statistical Area is analyzed using spatial statistics available in ArcMap GIS. This dissertation tests the following hypotheses:
1. Voters are polarized;
2. Polarized voters are the most active;
3. Ideology and party are closely related; and
4. Those who are like-minded politically are sorted geographically.
Through the use of precinct-level election data, this dissertation provide answers to three contemporary questions: (1) whether the electorate is more polarized, (2) whether politically like-minded people cluster with others like themselves, and (3) whether partisans are sorting into parties based on ideology. This dissertation advances the field of political science by demonstrating the ability of spatial analysis to provide answers to contemporary questions within the field and by exploring the application of GIS methods in election analysis.