Political scientists have long noted that congressional elections are often uncompetitive, often extremely so. Many scholars argue that the cause lies in the partisan redistricting of congressional districts, or “gerrymandering”. Other scholars emphasize polarization created by a fragmented news media, or the candidate choices made by a more ideological primary electorate. All these explanations identify the cause of party-safe elections in institutions of various kinds. This dissertation, by contrast, presents a structural explanation of uncompetitive elections. My theory is that population composition and patterns of migration are significant causes and predictors of election results in Florida. I test this theory empirically by comparing the predictions from four hypotheses against aggregate data, using the county as the unit of analysis. The first hypothesis is that Florida can be divided into clearly distinguishable, persistent partisan sections. This hypothesis is confirmed. The second hypothesis is that Florida voters have become increasingly partisan over time. This hypothesis is confirmed. The third hypothesis is that the degree of migration into a county predicts how that county will vote. This hypothesis finds some confirmation. The last hypothesis is that the degree of religiosity of a county predicts how that county will vote. This hypothesis is confirmed. By identifying the structural causes of party-safe elections, this study not only contributes to our understanding of elections in Florida, but also sheds light on the current polarization in American politics.

This research explains the variance in voter turnout and ballot roll-off in county and circuit nonpartisan judicial elections in Florida from 2014 and 2016. Based on theory, a collection of constituent (demographic and socioeconomic), candidate, competition, and contextual variables is gathered to construct four regression models. Two full regression models were constructed for turnout and roll-off and analyzed using SPSS software, in addition to two best regression models analyzing five statistically significant variables found within each full model. Presidential year elections and higher populations age 65 and up had positive impacts on voter turnout, while primary elections, campaign expenditures, and populations of minor (other) party registered voters had statistically significant negative effects on turnout. Increases in ballot roll-off were associated with presidential year elections, and populations with more college degrees, higher median household income, and higher percentages of voters registered with no party affiliation or minor political parties. Roll-off decreased in primary elections. While various contextual, competition, and constituent variables had significant impact on both turnout and roll-off in Florida judicial elections, candidate characteristic variables had no significant impact on differences in voter turnout and ballot roll-off.