Eli McKown-Dawson: How Geography and Social Networks Impact Voting Behavior and Policy Preferences

My name is Eli McKown-Dawson, and I am a junior majoring in Political Science with minors in Statistics and Philosophy. During my sophomore year, I enrolled in the Political Science Department’s Research Intensive Bachelor’s Certification Program and began working with Dr. Lonna Atkeson to study election science and voting behavior. I am interested in partisan polarization and how Americans’ election-administrative environments (i.e., the laws that govern the conduct of elections and the quality of election administration) impact their confidence in election outcomes and decisions at the ballot box. Beyond authoring academic research papers with Dr. Atkeson, I have published elections research with the MIT Election Data Science Lab and am a contributor at YouGov. After graduating from FSU, I plan to pursue a Ph.D. in political science and continue studying American political behavior.

For my IDEA Grant project, I will assess how Americans’ political beliefs are affected by the partisanship of those with whom they live and interact. For example, do Democrats who live in “bluer” areas of the country hold more liberal views than Democrats in “redder” areas and vice versa for Republicans? Testing this question is an important contribution to the political science literature, as scholars consistently find that national political figures exert heightened influence over local political contests (Carsey & Wright 1998; Sievert & McKee 2018; Caughey, Dunham, & Warshaw 2018), and Democrats and Republicans are increasingly likely to distrust and dislike each other (Iyengar et al. 2019; Abramowitz & Saunders 2008). I am interested in whether living and interacting with those who support a particular political party will nudge Americans to support candidates and issues supported by that party (Huckfeldt & Sprague 1991; Beck et al. 2002).

Dr. Atkeson and I developed this project during the fall of 2022 while designing a 2022 post-election survey of Florida voters. We were struck by the inaccuracy of some public opinion surveys and hypothesized that these errors occurred because the surveys failed to capture a certain subset of Americans: those in their local political minority. Specifically, we theorized that those in red areas are more likely to live, work, and interact with Republicans, while those in blue areas are more likely to interact with Democrats (Gimpel & Karnes 2006; Jacobs & Munis 2020; Conover 1984). These social networks should make voters in red areas more likely to support Republican candidates, approve of Republican elected officials, and hold more conservative policy positions, while the inverse should occur in bluer areas. 

I plan to investigate this question using data from the Cooperative Election Study (CES), an online political science survey conducted yearly in November by Harvard University and YouGov between 2006 and 2022. The pooled CES dataset has over 400,000 respondents. I have already collected these data and will combine them with historical presidential vote returns to measure local party support and partisan social influence. I will then employ hierarchical item response models to assess how—controlling for party affiliation and demographic characteristics—local party support influences Americans’ propensity to approve of and vote for presidents and presidential candidates from a given party. I will also examine whether these social networks impact policy preferences. For example, I will test whether living in more Democratic (Republican) counties makes respondents more (less) likely to support abortion access, environmental regulations, and gun control. 

I will also analyze the 2010-2014 CES panel survey, where the same individuals were surveyed multiple times over four years. Using these data, I can test whether moving to a county with a different political leaning changes Americans’ policy preferences. In other words, if a Republican moves to a Democratic county, do their policy opinions become more liberal? Using these panel data to supplement my cross-sectional analysis will help me assess whether my hypothesized relationship is causal. The figure below shows the distribution of county partisanship for CES respondents from 2006 to 2020. Positive values indicate Democratic counties and negative values indicate Republican counties. The figure demonstrates that there are a substantial number of Republicans living in blue areas and Democrats living in red areas.

I expect my statistical tests to confirm that place-based social networks influence how Americans think about politics and policy. My study will provide insight into how place impacts within-party variations in political behavior and how accounting for this relationship can improve the accuracy of public opinion surveys. Over the summer, I plan to complete these analyses and author a research paper based on my findings, with the eventual goal of submitting this paper to a peer-reviewed journal. I will also solicit feedback from the Political Science Department’s U.S. Politics Workshop and incorporate those changes into my project. Overall, I am very excited to continue my research with the support of an IDEA Grant.