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

This paper examines individual voter turn-out and its putative relationship with voting outcomes at the voting precinct level. Via a GIS-based address matching procedure, we were able to georeference individual voters (registered voters who casted their votes) and non-voters (those registered voters who did not cast their votes) for three recent local referenda in College Station, Texas. We then conducted a scale-sensitive, second-order spatial analysis for the spatial distribution of voter turn-outs, followed by a spatial clustering analysis of the voting results using Getis–Ord’s G_i statistic. We found that the extent of neighborhood effects in local elections is heavily influenced by the voter turn-out. If voter turn-out is clustered at intermediate and large scale, voting results tend to be clustered and also exhibit a sharp polarization between high and low values. If voter turn-out tends to be uniform/regular at intermediate scales but randomly distributed at both small and large scales, there appears to be less clustering in the voting results and thus lack of the neighborhood effect. If the voter turn-out pattern is mixed-uniform/regular at the small scale, random at the intermediate scale, but clustered at the large scale, the voting results show a stronger neighborhood effect.