The U.S. electoral college and spatial biases in voter power

Abstract

The United States does not provide for direct election of its chief executive but utilizes the Electoral College to represent voter choices. Central to this institution is the "winner-take-all" model by which electoral votes in all states but two are awarded to the candidate who garners a plurality of that state's popular vote. As game theorists have long pointed out, this system introduces several biases in voter power that differentially reward or punish voters based on each state's population or electorate. This article offers a historical overview of the Electoral College and the geographic biases in voter power it introduces. It extends the influential binomial model of voter power proposed by Banzhaf in the 1960s to include a multinomial approach sensitive to the presence of more than two parties, the absolute and relative margins of victories, and the number of electoral votes in each state. It then applies this approach to U.S. presidential elections from 1960 to 2004 utilizing a series of cartograms. Next, the article examines voter power differentials between the two major political parties, five ethnic groups, rural and urban areas, and ten religious denominations in the 2000 and 2004 elections. Finally, it links contemporary discussions of voter power to theories of democracy, arguing that the electoral power can only be understood in contingent, temporally fluid, and geographically specific terms.