The Number of Political Parties: A Reexamination of Duverger's Law

The purpose of this essay is to investigate theory that explains variation in number of candidates for office in single-member constituencies with plurality voting. In other words, I shall reexamine and amplify Maurice Duverger's law-the statement that the simple majority, single ballot system favors two-party system.'1 Many scholars have found this law attractive because it explains structure of institutions as logically derived from rational choice by individuals.2 Macrostructure is thus made to depend on easily explainable and intuitively plausible microbehavior, a neatness of fit common enough in economic theory, but seldom arrived at in political theory. The argument is that supporters of a third party desert to one of two larger parties because they recognize that continued support of their favorite results in wasting their votes, and may even contribute to victory of their most despised. For example, suppose a plurality election in a single-member district at time t1 has produced result: 10,000 votes for A, 9,000 votes for B, and 2,000 votes for C; and suppose that typically supporters of C prefer C to B and B to A. For them to continue at timet to support C appears, ceteris paribus, to guarantee success of A and to be a waste of their second choice. There is considerable indirect empirical support for this theory. The tendency toward two parties is especially strong in United States, where President is elected by a plurality of sorts in what is almost a single-member district. In Great Britain, most elections are by plurality in single-member districts; and, despite persistence of third parties, most constituencies usually have serious candidates from only two parties. Strong third parties in England have typically been one of two major parties in some geographic area, such as Irish