It is well known that to win a US Presidential election it is not sufficient win the popular vote. Rather, because of the electoral college, the geographical distribution of votes is crucial. Data from past elections shows that a useful statistic describing the distribution of votes is its "dimension". I will explain the topological definition for the "dimension" of a subset of Euclidean space, estimate it for the Kerry and Bush votes in the 2004 US Presidential election, and recall the consequences for the outcome of that election.