Capturing Campaign Effects

Priming and Persuasion in Presidential Campaigns

Priming and Persuasion in Presidential Campaigns Larry M. Bartels IN E V E RY E L E C T I O N C Y C L E , the major parties and their presidential candidates spend vast sums of money and prodigious amounts of energy on the campaign for the White House. Thousands of journalists, campaign operatives, pollsters, and media consultants derive their livings from this yearlong spectacle. Every day, interested citizens can read and watch detailed accounts and analyses of the candidates’ strategies, speeches, and issue stands. As soon as the election is over, star reporters and star campaigners race to produce best-selling inside accounts of the decisions and events that led to victory or defeat. But is it all sound and fury, signifying nothing? After more than half a century of academic election studies, the surprising reality is that political scientists still understand relatively little about how presidential campaigns affect the vote. Even more surprisingly, much of what they do understand seems to provide good grounds for doubting that campaigns signiAcantly affect the vote at all, at least in the most straightforward sense of determining who wins and who loses. Finkel’s (1993, 14) analysis of survey data suggests that “for all the presidential elections in the 1980s, the outcomes were essentially predictable from the electorate’s spring or early summer dispositions.” Moreover, statistical forecasts based upon historical relationships between presidential election outcomes and economic and political “fundamentals” measured before the fall campaign have even begun to match actual election outcomes about as well as opinion polls conducted just before Election Day (Rosenstone 1983; Lewis-Beck and Rice 1992; and many others). If, as Gel78 man and King (1993, 412) assert, summarizing this evidence, “the outcome of recent elections can be predicted within a few percentage points in the popular vote, based on events that have occurred before Labor Day,” we are left to wonder how the millions of dollars and hours expended by candidates , campaign workers, and political journalists in the two months after Labor Day could possibly be well spent. One obvious answer is that even highly effective campaign efforts on both sides may cancel each other out, leaving the election result essentially similar to what it would have been if neither side campaigned at all. This sort of stalemate seems especially plausible in presidential general election campaigns, since both major parties typically draw from roughly similar pools of candidates, spend roughly similar amounts of money, and adopt roughly similar campaign strategies. As I have put it elsewhere, In a world where most campaigners make reasonably effective use of reasonably similar resources and technologies most of the time, much of their effort will necessarily be without visible impact, simply because every campaigner’s efforts are balanced against more or less equally effective efforts to produce the opposite effect. (Bartels 1992, 267) Such countervailing effects are clearly an important part of the story. But they do nothing to account for the fact that campaigns sometimes do produce substantial shifts in the relative standings of the competing candidates, at least as measured in presidential trial heats in commercial opinion polls. Journalists and campaigners point to these substantial shifts to support their assertions that campaign strategies and campaign events are often signiAcant and sometimes decisive. At least in retrospect, the judgments underlying such assertions seem to reBect a good deal of consensus regarding the important turning points of each campaign season. But if these day-to-day campaign events really cause substantial shifts in vote intentions, we are left with the question of how eventual election outcomes could be so predictable in advance. To turn Gelman and King’s (1993) apt question on its head, how are election outcomes so predictable when campaign polls are so variable? A partial resolution of this apparent paradox may be provided by the notion that campaigns matter because they tend to produce congruence between fundamental political conditions and predispositions on the one Priming and Persuasion in Presidential Campaigns 79 hand and vote intentions on the other. This supposition is supported by survey-based Andings on political activation (Berelson, Lazarsfeld, and McPhee 1954; Bartels 1988; Finkel 1993) and by aggregate-level analysis of poll results and election outcomes (Bartels 1992; Gelman and King 1993; Holbrook 1996). Most strikingly, Holbrook (1996) has observed that the net effects of campaign events in the 1984, 1988, and 1992 presidential elections were strongly correlated with the...