Social Proximity and 'Friends-and-Neighbors' Voting in Local Elections

Scholars have long been aware of voters' tendency to prefer candidates from more geographically proximate locations--a tendency V.O. Key (1949) dubs "friends-and-neighbors'' voting. This theory, largely developed and tested in congressional and statewide elections, suggests geographic proximity to a candidate encourages voters to “back the home-town boy” (Key 1949, 41), perhaps by encouraging familiarity or a sense of shared identity. By this conception, friends-and-neighbors voting can have no influence in local elections where all candidates share the same hometown. As a result, little work examines friends-and-neighbors voting in local elections. This inattention presents a problem because local elections seem ripe for geographically-defined voting patterns. In these small-budget, low-information elections, candidate familiarity based on geographic proximity may be a particularly powerful determinant of vote choice. Better understanding how citizens vote in local elections will be valuable because local elections occur in greater number than congressional, statewide, or national elections and locally-elected officials are responsible for implementing policy and allocating a large proportion of the nation’s resources (Trounstine, 2009).

We explore friends-and-neighbors voting using an original dataset derived from historical records that reveal the individual votes of all electors in the 1874 local elections in Newport, Kentucky. Nineteenth century Kentucky employed viva voce election law, requiring all votes to be cast by voice at assigned polling places. Every voter's preference for every office was thus recorded by election clerks. Newport is one of the few cities under this voting law for which the complete poll books have been recovered. We have conducted archival work to pair these individual votes with detailed profiles for all known inhabitants of this mid-sized U.S. city at the time of the elections. Most importantly for our analysis, we have GIS-coded addresses of all known residents, including the candidates for local office. We may thus measure the geographic distance between candidates and voters, permitting us to explore whether a friends-and-neighbors voting pattern occurs at the local level.

To complement the poll books and geographic data, we have worked with a large interdisciplinary team to assemble public records about the cities' social networks at the time of the elections. Using census rolls, tax lists, plat maps, city directories, and other sources, we have identified the familial relations and occupations of each known resident. We have culled church membership lists to link residents to the specific churches they attended. In addition to these social data, the records tell us residents' ethnic background, age, gender, accumulated wealth, and if they own or are in the process of purchasing their home. With these social attributes, we may use a social network analysis approach to measure voters’ social proximity to candidates, allowing us to address a number of previously unexplored questions about how individual-level patterns of friends-and-neighbors voting arise. How strongly does geographic proximity predict social proximity? What types of geography predict candidate support? Does social proximity fully account for the relationship between geography and vote choice? Which forms of social interaction best predict candidate support? By answering these questions, this project can shed new light into the determinants of individual vote choice in local elections while also clarifying when and how friends-and-neighbors voting patterns arise.