A GIS-based spatial analysis on neighborhood effects and voter turn-out: a case study in College Station, Texas.

It is typically assumed that the percentage of voter turnout in an area is at least minimally affected by local or neighborhood influence, whether that means local media coverage of a certain issue or varying opinions on what areas of the city or town need to be repaired, using local tax dollars, etcetera.  This study, by Danile Z. Sui and Peter J. Hugill, attempts to spatially illustrate and analyze the trends between voter turnout in an neighborhood and the different issues in each election, called the "neighborhood effect."

Sui and Hugill chose College Station, TX, the home of Texas A&M, for their data set. Because the city has very complex processes to initiate official political moves, such as creating city petitions and demanding recalls, but opperates on a relatively small scale, College Station has a very particular political environment perfect for a study examining local elections in a spatial manner. They chose to explore the results of three referenda on three different years, the first of which addressed the application millions of dollars in tax bonds being set aside to improve the city’s infrastructure (1995), the second a petition to halt the construction of a City Convention Center (1997), and the third an appeal to reopen one of the oldest streets in the city, that was closed by the City Council due to incessant traffic, all compared to the voting turn out in each voting precincts. These precincts were used as if they were neighborhoods, as they coincided with many of the main residential areas and are useful in depicting election results, although the results would not be as effective, or accurate, in larger cities.

In analyzing the data, the researchers matched addresses on voters and non-voters, then applied the voting results for each referenda to a map of the city, in order to compare the spatial distribution in voter turnout and the actual results of each. Sui and Hugill used a fairly complicated method to analyze this data called Ripley’s K-Function to determine “whether a given point pattern departs from randomness toward clustering or regularity” (Sui and Hugill 2002, 163). This is important in determining whether certain neighborhoods had any correlation with the voter turnout, if there was clustering, or whether it was primarily indiscriminate.  

The maps below illustrate the results from this study. In the maps on the left (i, iii, and v), each point represents a registered voter. It is clear that the first issue, the tax bond for repairing city infrastructure was the least controversial, with a low voter turnout and, as seen in figure ii, and passed with a majority in nearly every precinct (with the solid gray sections of the circles representing for the referenda and the white with dots against it in each precinct, seen in map ii). The distribution of these results further illustrates the fact that the areas that would receive these improvements had the most voter turnout, thus the most people who cared about such a relatively mundane issue turned out to vote. In map iii, there is a much higher voter turnout, as well as closer results in many of the precincts, seen in map iv. Evidently, the idea of College Station having a city funded convention center was quite controversial, as it was seen as a pure business expenditure paid for with citizen tax dollars to some, but an effective stimulate for economic growth in the area to others. The measure passed, but not with a large majority.

However, the 1991 issue was the most controversial, illustrated in v and vi. There is a clear concentration of voters in map v, apparently extending south from A&M’s campus. In closing Munson Avenue, citizens needing to go north, presumably mostly students and employees if the college, obviously felt quite strongly that the road needed to be open, probably because it offered more options, thus less traffic, to get where they needed to go.

Sui and Hugill determined that when there is a particularly high voter turnout in the precinct, there tends to be a very clear preference in the area for a particular result, which may be extremely different than the results in other precincts, clearly seen in the first and examples. However,  the difference between these two incidents is the actual voter turnout. There is certainly clustering of voters in the first example, but not to the same scale as the third, where there is a distinct clustering in very large numbers, illustrating the passion the certain neighborhoods felt for repealing the closure. Conversely, when voter turnout is more scattered, there is not obvious polarization of results, as seen in the even spread of voters and  nearly split results overall in the second example. These results are complicated and certainly do not show the full range of components affecting voter turnout, but the use of GIS in this particular geographic area to analyzed political results helps illustrate “the neighborhood effect,” where communities can  have a legitimate impact on voter turnout, and thus influencing the outcomes of particular issues acutely affecting members of these communities. 

Sui, D.Z., & Hugill, P.J. (2002). A GIS-based spatial analysis on neighborhood effects and voter turn-out: a case study in College Station, Texas. Political Geography, 21. Retrieved from https://lms.southwestern.edu/file.php/4373/Literature/Sui-2002-GIS_Voter_Turnout.pdf