Campaigning in Context: A Practical Statewide Study of Correlations between Campaign Contact Methods, Partisanship, Timing, Frequency, Population Density, and Regionalism on Voter Turnout

Abstract

Studies of campaign influence on individual voter turnout typically analyze responses from a single door-to-door, telephonic, or mail interaction involving a non-partisan source contacting individuals in a localized urban area. I argue these investigations are unsatisfying. They do not consider hyperpartisanship, campaign micro-targeting strategy developed from large data collection, regional diversity, and repeated contact attempts. This study examined partisan campaign contact correlation with voter turnout utilizing a “real-world” statewide dataset created from a coordinated partisan get out the vote (GOTV) effort during the 2014 election cycle. Four traditional GOTV methods were investigated: volunteer door-to-door, volunteer telephone call, postal mail, and professional interactions. Treatment and control groups were empirically tested against a dependent variable of whether or not a voter cast a ballot following the attempted partisan contact. This large data set allowed for an analysis of several conditions supporting my argument. These included a voter’s partisan affiliation, when the contact occurred, how often a voter was contacted, the region where the contacted voter lives, and local population density. The results presented many findings distinctive from previous scholarship. Partisan volunteer door-to-door contact was not always the best method to increase voter turnout. Different contact methods show stronger correlations with voting among various partisan groups. Turnout among the treatment groups was higher or lower dependent upon when contact occurred. Any campaign contact closer to Election Day generally improved voting likelihood among aligned partisans, but not with voters registered as unaffiliated or anti-partisan. Additionally, contact frequency resulted in dissimilar turnout levels among treatment groups dependent on contact method and partisan affiliation. The data also showed unique reactions to each contact method contingent on the voter’s congressional district or local population density. These results have implications on our understanding of individual voter behavior, partisanship, contact timing and frequency correlation with turnout, large-district campaign strategy, and regional GOTV efforts.