"A GIS-Based Spatial Analysis on Neighborhood Effects and Voter Turn-out: A Case Study in College Station, Texas"
The phrase "All politics is local" was coined by the Speaker of the House of Representatives, Tip O'Neill, in 1994. In this article, Peter Hugill and Danile Sui attempt to examine the "neighborhood effect" in local elections, in addition to using a GIS-based address matching to map the voting results at precinct level and the spatial distribution of voter turn-out. This allows the authors to analyze individual voter data in each of three city-wide referendums held in College Station, Texas between 1995-1999. In turn, the authors were able to "gauge the nature of spatial clustering of actual voter turn-out and how the spatial distribution of voters influences the neighborhood effect of voting results at the precinct level" (170).
The authors found that a myriad of factors influence an individual voters' understanding of local politics and referendums, including personal observation of events and situations, informal interaction with inter-personal communications, organizationally-based interactions, and mass media all potential factors for an individual's voting behavior. The authors conclude that while the "neighborhood effect" (spatial dependence) is prevalent, the reasons for neighborhoods voting the way they do may be affected by the plethora of factors, such as those listed above, opening future avenues for political scientists to study spatial heterogeneity.
Sui and Hugill obtained actual voter lists for their study for three referendums from the City Secretary's office. They first established a baseline comparison, geocoding registered voters who did not cast their ballots in the city-wide election (or "non-voters"). The authors used ArcView GIS Address Matching Module to geocode each voter who did cast their ballots during the three referenda. Next, the authors mapped individual voter distribution and voting results at the precinct-level. To examine the spatial patterns of actual voter distribution, the authors forged a "second-order voter distribution analysis" using Ripley's K-function (163). Ripley's K-function is used for analyzing the spatial patterns of any phenomenon whose spatial locations are measured as points (in this article's case, a circle of radius d centered on each point, where each point represents a registered voter). To explore the neighborhood effects, voting results at the precinct-level were analyzed using Getis-Ord's G-statistic, which reveals local clustering of specific phenomenon (used in this case to explore the relationship between the distribution of actual voter turn-out and the neighborhood effect of voting in local politics). Implementing GIS "greatly facilitates the implementation of these spatial statistical tendencies and stimulates further development of individual-based analytical techniques" (165).
The results from the case study of College Station portend that neighborhood effects in local politics are heavily influenced by the distribution of actual voters. For the specific findings from College Station, the authors summarize possible linkages between the spatial distribution of voter-turn out and the neighborhood effect in the Table below:
The results from the case study of College Station suggest that neighborhood effects in local politics are heavily influenced by the distribution of actual voters. However, the overarching message of this article is that the role of GIS has shifted from being a data storage tool to a a program ripe for the field of electoral geography, as it handles both ecological and individual voter data.