By Chance or Design? On the Locations of Controversial Billboards in Urban Ohio during the 2012 Election

Abstract

The 2012 American Presidential Election was preceded by widespread efforts to promote stronger voting requirements in states across the nation. In one case, a private, anonymous foundation purchased advertising spots on billboards in urban Ohio to communicate that voter fraud is a felony punishable by fines and imprisonment. This action drew criticism from civil rights groups, who argued that a majority of the billboards were located in minority census tracts, and that the advertisements utilized intimidating language and imagery. This article applies probability theory to the debate in order to determine the likelihood that the observed patterns of billboards in two cities—Cleveland and Columbus—could have occurred by chance. First, simulation is employed to compare the observed allocations of billboards to white and non-white census tracts to patterns generated under Complete Spatial Randomness (CSR). Second, simulation draws from probit regression models, in which the dependent variable is the presence of a billboard within a census tract, are used to generate distributions of first differences between1) the expected value of the dependent variable given that the census tract is majority white and2) the expected value of the dependent variable given that the census tract is majority non-white. The results suggest that the billboard locations in Cleveland were not significantly different from patterns produced under CSR, and that the probability of a given census tract containing a billboard is not significantly different for white and non-white majority tracts. The opposite inferences are made for Columbus.