In polities in which governments are formed by coalitions of parties or in which the government is frequently divided between an executive and a legislature of different parties, there is in general no single winning party or candidate. Rather, the outcome is a composite of the interests of several parties or their representatives. Accordingly, a rational voter may be expected to use his vote to try to achieve what that voter perceives as the best policy outcome rather than expressively voting for the most attractive party or candidate. Kedar (2002) develops a model that is intended to implement this objective. She notes, “In this model, voters reward parties not only for presenting a platform proximate to their own positions, but also for pulling policy outcomes in their direction” (2002: 2). Thus outcome-oriented behavior is menu-dependent. When policy is formed by compromise among multiple players, voters understand that their support for a single party will be diluted by other parties in the system. Hence they may have an incentive to “overpull” by voting for a party more extreme than their own policy position.

For example, suppose that four parties that are expected to form a governing coalition are of about equal strength and arranged from left to right in the following order: A, B, C, and D. The compromise policy is likely to be near the midpoint between parties B and C. A voter whose ideal point is near that of party B may prefer policy to the left of this midpoint and would thus like to move the coalition policy leftward.