The conventional wisdom among students of elections is that the choices of voters are largely driven by powerful forces that have lasting effects from one election to the nextenduring political orientations and retrospective judgments about economic performance. Despite the media's focus on personalities and the individual traits of the candidates, the scholarly consensus is that "campaigns don't matter"--the choices of voters are largely beyond the influence of candidates when weighed against enduring electoral forces.
Our article argues, however, that campaigns and voters' evaluations of the personality traits of candidates (such as their morality, knowledge, and leadership) do matter. Previous research has tested the impact on vote choice (often in presidential elections) of two sets of influences--voter evaluation of candidate personality traits, which is treated as measuring the effects of campaigns, and such enduring electoral forces as party identification, ideological orientation, and retrospective judgments. What is particularly telling is that this tradition of research has focused on two-party competition and has asked if candidate traits or enduring political forces have the greatest impact on the choice of voters between the Republican and Democratic candidates.
We argue that this previous research tradition of parsing out the relative effects of personality traits and party identification on the choice among the two main political parties misses a fundamental aspect of contemporary elections, especially presidential elections: the substantial and decisive votes won by third party candidates. No president in the past three elections has won a majority of votes cast; third party candidates are drawing enough votes to swing the popular vote of some recent elections. Put simply, scholars have been looking in the wrong place for campaign effects.
Our article synthesizes contradictory findings about the effects of campaigns. It largely acknowledges the consistent finding that enduring electoral forces severely limit the impact of Democratic and Republican campaigns while taking seriously the direct evidence from campaigns that they can have some influence. In particular, our analysis examines the effects of candidate trait ratings on the decision to vote for third party candidates in the 1992, 1996, and 2000 presidential elections. The results challenge the scholarly consensus that candidate traits do not matter and suggest that campaign effects are sufficient to determine who wins contemporary presidential elections.
Debating the Effects of Campaigns
A long line of research stretching over half a century has provided strong support for the "enduring effects" interpretation of elections and vote choice. The enduring effects school has identified three sets of factors as severely limiting the effects of campaigns. First, research since the beginning of behavioralism has found that the votes of many Americans were "locked up" before the fall general election campaign even began. For instance, Lazarsfeld, Berelson, and Gaudet (1944) found that voters in the 1940 election expressed a vote intention in the spring, with very few changing their choice by election day in November. From this evidence, Lazarsfeld, Berelson, and Gaudet concluded that campaigns only serve to reinforce predispositions, rather than change the choices of voters about the candidates. Second, the hold of political party loyalty has also been consistently identified as a powerful check on campaigns. Campbell et al. (1960) reported in The American Voter that party identification was a lasting political orientation and that it was the single most important influence on vote choice. Campaigns introduced short-term effects, but these were dwarfed by the overwhelming influence of party identification. Third, V. O. Key (1966) reported that retrospective evaluations of the national economy introduced another significant longer term effect on elections that severely restricted the impact of campaigns (see also Frankovic 1993; Fiorina 1981; Kinder, Adams, and Gronke 1989; Markus 1988; Pomper 1993). …