The Relative Efficiency of Approval and Condorcet Voting Procedures

Abstract

This article compares empirically the average utility for the electorate of the candidates selected by approval voting and by ranking-based voting with a search for a Condorcet winner (a Condorcet-completion procedure). “Feeling thermometer” scores from the presidential election surveys for 1972 through 1984 of the Institute for Survey Research are used for distributions of utilities of candidates for voters. Several hypotheses about voting behavior under approval voting are explored, while voters using a Condorcet completion procedure are assumed to rank the candidates according to the utility they offer. The average social utilities of the candidates chosen by these voting systems are remarkably similar, while that for single-vote plurality (under sincere voting) is significantly lower.