Observers of U.S. elections have reason to believe that third parties are not relevant political actors since they rarely win many votes or influence which major party wins an election. Researchers should use dependent variables besides vote choice and vote share to find third party effects that are a normal aspect of the American two-party system. A spatial model of elections motivates the hypothesis that a higher likelihood of third party entry induces greater major party candidate divergence. An empirical test that uses candidate positioning data in the 1996 U.S. House elections provides evidence of this third party effect.
political parties, third parties, electoral competition, candidate positioning
Throughout its history, the landscape of American politics has been dominated by two major parties. Only in a handful of years have third parties appeared to play an appreciable role in shaping political outcomes, and even then the discussion is usually limited to presidential elections. 1 One might then conclude that considering only the major parties sufficiently describes the American twoparty system, aside from rare periods of realignment.
This article instead argues that third parties are an integral component of the two-party system, and one should not take their apparent lack of electoral success as an indication of their irrelevance. Their lack of success may be, in fact, evidence that they are influencing political outcomes. This somewhat paradoxical statement is less puzzling once major party strategic anticipation is considered: third parties are not successful, in part, because the major parties take preemptive actions to minimize their success. This influence on electoral strategies is the third party effect that is the focus of this article. The logic of strategic anticipation shows how third parties lend shape to the two-party competition that we observe, even under normal political conditions.
The Search for Third Party Effects
Previous works on third parties tend to focus on their observed effects on final electoral outcomes, such as winning a significant percentage of the vote or stealing enough votes from one major party to flip the outcome of the election. By these standards, third parties are rarely important. Rosenstone, Behr, and Lazarus (1996) argue that their success, in terms of winning votes, is a function of "major party failure," which is a condition that is not regularly met. The two most recent cases of potential spoiler candidates are considered in two articles. Lacy and Burden (1999) find that Perot in 1992 was not a spoiler candidate, contrary to some circles of conventional wisdom, as his candidacy only decreased Clinton's margin of victory over Bush. Herron and Lewis (2007) examine the spoiler effect of Nader in the 2000 presidential election by analyzing ballots cast in Florida. They find that although Nader did steal votes from Gore, he also took nearly as many votes away from Bush. Therefore, the net gain of votes for Bush was modest, and the extreme closeness of the race in Florida was a needed condition to reach the spoiler outcome. Since such tightly contested elections are extraordinary, one would again conclude that third parties rarely influence outcomes.
Differing from the cited works above, Rapoport and Stone (2001, 2005) focus their attention on the actions of the major parties. They outline a dynamic through which third parties can have a lasting influence beyond the election of their initial success. A third party candidate takes advantage of major party failure and wins a significant share of the vote in an initial election. If this candidate attracts a large and identifiable issue constituency, then one or both of the major parties respond by bidding for that group in the following elections. That is, the major party or parties shift their issue commitments to attract third party supporters. Although this dynamic does move the focus onto the actions of the major parties, it still necessitates major party failure, which is not a condition usually satisfied. …