Strategic Voting in Multicandidate Elections under Uncertainty and under Risk

Abstract

A number of voting systems for multicandidate elections stipulate that the voter may, under certain constraints, cast v _i ≥ 0 votes for the i-th candidate, i = 1,..., K (where K is the number of candidates), and that the candidate receiving the greatest total vote wins. We will refer to these methods as numerical voting systems. Each can be characterized by the constraints imposed: e.g., simple plurality voting permits one of the v_i to be 1 but the others must be 0; approval voting permits each of the v _i to be either 0 or 1; the Borda system requires that the set |v _i : i = 1,..., K} consist of the numbers K-1, K-2, ..., 0; cardinal measure voting requires that 0 ≤ v _i ≤ M for some fixed positive number M. For background on these and other multicandidate voting systems, see Black [1958], Brams [1978], Merrill [1980], and Weber [1978].