Disproportionality and bias in US Presidential Elections: How geography helped Bush defeat Gore but couldn't help Kerry beat Bush

The 2000 and 2004 US Presidential elections were closely fought contests, with in the first case victory in the Electoral College being denied to the candidate with the largest share of the popular vote. Disproportionality in the translation of votes into seats (in this case, from popular votes to votes in the Electoral College) is common to contests using a winner-takes-all electoral system. So is bias, whereby that disproportionality does not apply equally to each candidate. Analysis of the bias at those two elections shows that Bush was favoured at the first but not at the second. Identification of the bias components shows that Bush was advantaged by variations in the number of popular votes per Electoral College voter across the states, and also by variation in turnout. In 2000, his popular votes were also more efficiently distributed than Gore's; in 2004 they were less efficiently distributed than Kerry's, largely because of increased turnout - producing larger numbers of surplus votes - in states that were already safe for Bush.