Vote-Maximizing Parties

Typically, spatial models of voting and elections are based on the behavioral assumption that political parties competing in elections seek to garner the highest number of ballots possible and thereby obtain governmental power. In other words, they are vote-maximizing parties.

The Downsian proximity spatial model, and many derivatives thereof, arrive at the median voter convergence result in part by making the vote-maximizing party assumption. Parties, in these basic models, are simply groups of individuals who promote candidates to seek governmental power. Note that, in contrast to more sociological accounts, parties are assumed to propose policy in order to win office, rather than winning office to implement particular policies. Consider Figure 1: here the (approximately symmetric, unimodel) distribution of voters is bisected by the broken line that represents a median (middle) voter's ideal point. The x-axis is some unidimensional, say “left—right,” space. L represents the electoral position of a party on the left, and R the position of a party on the right. If voters cast their ballots according to a proximity rule—that is, they vote for the party closest to them in space—then party R will clearly win more votes than party L. The outcome, by assumption, is that party R wins governmental office. Notice that vote maximization drives both parties to place themselves as close as possible to the median voter. The model, and the assumption of vote maximization, was originally designed to fit U.S. presidential elections—in which the entire country is the constituency and “winner takes all”—especially for the mid- to late-20th century. In these models, vote maximization is simply the means to a governmental end; to wit, there is an uncertainty as to whether, independently of actually winning office, more votes are considered a good in and of themselves for parties.

Tinkering with the Downsian model in relatively small ways can lead to nonconvergence, even while maintaining vote maximization. Suppose, for example, that voters in the tails of the distribution in Figure 1 disproportionately abstain as the parties move away from them toward the median. In this case, a party may maximize votes by locating away from the median to shore up its base support. Alternatively, theorists have suggested that parties are more complicated organizations than the “black box” logic above suggests. That is, they have “grassroots” members, usually more ideologically extreme and committed than the median voter in the population. If these members have power over the electoral profile of their party, or the party leaders have some other reason to care more about policy than votes, vote maximization may, again, be an overly crude assumption.

Both theoretically and empirically, the assumptions of vote-maximizing parties have been deemed appropriate for plurality or majoritarian constituencies (be they districts or nations). Outside of the United States, commentators have discussed vote maximization, median voter convergence, and the proximity model with respect to the elections of Tony Blair in the United Kingdom and Gerhard Schröder in Germany.

Figure 1 Vote-maximizing parties and median voter convergence

[Page 694]Directional spatial models posit that voters perceive themselves as on a particular “side” of an ideological scale and consider voting only for candidates on the same side. Importantly, within their side, voters favor the candidate or party most removed (i.e., farther) from their own location. A vote-maximizing party, then, will move away from voters in ideological space and toward extremes.

...