Abstract
Several of the earliest advocates of proportional representation upheld the “principle of free association”: that electors should be free to choose the constituencies to which they belong. This article analyzes this principle in game-theoretic terms and investigates how far the free association “game” can be simulated through proportional representation. It is assumed that each voter's prime concern is that his most-preferred candidate should be elected. If preferences are single peaked, the outcome of an election based on free association can be predicted; the same outcome would result from a modified form of the single transferable vote.

This publication has 3 references indexed in Scilit: