An empirical study of some election systems

Uses presidential election data of the American Psychological Association for 5 yrs to compare a number of election systems in frequency of selecting Condorcet winners, consistency over subslates, and the effect of relevant and irrelevant candidates. The results from each system could be scaled in 1 dimension for each of the 5 elections, using the Kemeny metric. These scales were closely related to the weighting given to the variance of the candidates' rankings. An interpretation is given in terms of the polarizing effect of a candidate on an electorate. Election systems are discussed as strategies for the accumulation of subelectorates to reach a decisive set of voters. (20 ref)