- Shopping Bag ( 0 items )
Campaigns suddenly seem to matter, as do questions about the electoral process in the aftermath of the 2000 presidential election. The authors examine the U.S. electoral process as an integrated event spanning a full year, drawing upon the Annenberg 2000 Election Study. The scale of their fieldwork is such that they have been able to isolate key turning points and that dynamics can be studied within certain segments. Johnston, Hagen and Jamieson have also utilized candidate appearances, news coverage, and ...
Campaigns suddenly seem to matter, as do questions about the electoral process in the aftermath of the 2000 presidential election. The authors examine the U.S. electoral process as an integrated event spanning a full year, drawing upon the Annenberg 2000 Election Study. The scale of their fieldwork is such that they have been able to isolate key turning points and that dynamics can be studied within certain segments. Johnston, Hagen and Jamieson have also utilized candidate appearances, news coverage, and campaign advertising to provide this integrated account of a U.S. campaign. Richard Johnston is Professor and Head of Political Science at the University of British Columbia and an Associate Member of Nuffield College, Oxford. He is co-author of Letting the People Decide (Stanford University Press, 1992) and The Challenge of Direct Democracy (McGill-Queen's University Press, 1996). Michael G. Hagen is Associate Research Professor and Director of the Center for Public Interest Polling at the Eagleton Institute of Politics at Rutgers University. He is co-author of Race and Inequality: A Study in American Values (Chatham House, 1986) and a contributor to Reasoning and Choice: Explorations in Political Psychology (Cambridge, 2003). Kathleen Hall Jamieson is Ware Professor of Communication and Director of the Annenberg Public Policy Center at the University of Pennsylvania. She is author or co-author of twelve books on politics and media including Packaging the Presidency (Oxford University Press, 1988).
This book aims to unite two research traditions that are usually seen as competing. With some noteworthy exceptions on both sides of the disciplinary aisle, one tradition has been articulated mainly by communications scholars and the other mainly by political scientists. To perform the nuptials, we deploy unique bodies of evidence from one of the more compelling presidential elections in living memory, the virtual dead heat of 2000. In the campaign, all the factors that drive political science models were in play at least some of the time - abiding elements of social structure, geography, party identification, and ideology; the economy and other aspects of the record of the previous administration; the perceived fitness of each candidate for executive office; and issues reaching back to the New Deal. But these factors did not operate automatically. They were activated and in some cases critically altered by campaign communication - its overall volume, its partisan direction, the consistency of messages across communications channels, and the rhetorical sophistication of the messages themselves.
To make our case, we focus on three phases in the general campaign and on the critical transitions between them. The first phase was produced by the conventions and lasted for more than a month. In this phase, predictions from econometric forecasting models for a comfortable victory by Al Gore seemed bound for success, as, of course, was Gore himself. This phase came to an abrupt end and the second phase began in late September when perceptions of Al Gore's character - of his honesty in particular - crashed. Overnight he went from being the presumptive victor to fighting for his political life. In the third phase, he called up memories of the New Deal and succeeded in persuading a critical bloc of voters that George W. Bush was a threat to the Social Security system. In the byplay, however, Gore ignored the robust economy, which should have carried him to a comfortable victory. Still, his playing the New Deal card enabled him to win the popular vote.
But he lost the Electoral College and, with it, the election. The divergence between the popular vote and the Electoral vote is another theme of this book. In one sense, this takes us to the foundations of party politics and the origins of the republic. In another sense, it takes us to the frontier of political communication, in particular to the gap between parts of the television broadcast day. Al Gore won the popular vote by, at the end of the campaign, winning the battle for network news. George W. Bush won the Electoral College by, also at the end, winning the battle of the ads. These assertions are possible because by 2000, presidential elections had become a natural experiment on a continental scale. The New Deal-Social Security message that was critical to Gore's recovery was most effective where it did him the least good, in states he could not win and in states he could not lose. In closely fought states, states that were pivotal in the Electoral College, his message was blunted by the sheer weight of pro-Bush advertising.
In short, communication is critical in determining whether and if so how the economy, candidate traits, and issues function in a campaign. Sometimes the communication is directly by a candidate or a closely connected surrogate in intensely covered moments such as conventions or debates. Sometimes the channel is advertising or the news. It matters a lot if ads and news reinforce each other or work at odds. Failure to communicate can be as critical as active attempts at priming or moving opinion.
Saying such things aligns us with research in the tradition of communication studies that emphasizes contingency and the power of rhetoric. But much of that research is supposition, anecdote, or not strictly relevant to an aggregate phenomenon such as an election. The body of research on the other side, attuned more to "necessary" - as opposed to contingent - features of elections seems more robust, more thought through. But much of that research seems oddly antipolitical, ironically so, because most of it is by political scientists. Our view is that many of the propositions about recurring features of elections require communications factors for their proper operation. Explicating those communications factors reveals just where contingency lurks, where strategic choice by candidates is possible or even necessary, and where a candidate can go wrong.
Explicating communications factors also forces us to question the role of campaign communications in enlightening voters. At one level, the 2000 campaign clearly did enlighten the electorate. The incidence of basically correct perception of candidates' positions on issues was greater at the end than at the beginning. But not every effect of the campaign lay in the domain of interests and issues. The campaign also processed highly manipulated images of candidates' character. And within the domain of issues, much depended on what was said - and not said - and on the resources each side could command to get its message out.
Our claims rest on analyses of three bodies of data. Most important is a massive "rolling cross section" survey of the 2000 electorate, the National Annenberg Election Survey. Fieldwork began in November 1999 and finished in January 2001. This book focuses on the over thirty-seven thousand respondents interviewed between Independence Day and Election Day. Alongside the survey and sometimes joined to it are bodies of advertising data, organized by day and by media market. The spatiotemporal pattern in ad buys, when combined with the rolling cross-section survey data, enables us to estimate the impact of ads with considerable efficiency. Finally, we tie the rolling cross-section data to a detailed analysis of campaign coverage in major newspapers but, more importantly, on the national TV networks.
The next part of this chapter lays out the book's analytic stakes in some detail. Then we further describe the survey, advertising, and news data. Finally, we describe the order of argument and the plan of the book.
Forecasting Models and the Record of the Previous Administration
According to all forecasting models, Al Gore was supposed to win handily. Not only was the economy robust but also ratings of Bill Clinton's handling of his job were very high. Even discounting for the fact that Gore was not Clinton and that the Democrats were shooting for a third consecutive term, Gore should not have lost. Zaller (1998) reinforced this expectation by claiming that Bill Clinton's popularity in the face of the Monica Lewinsky scandal confirmed the importance of "fundamentals," most importantly the economy. The predictions to this effect and the body of research from which they derive are ably captured by a symposium in PS: Political Science and Politics in 2001. The seeds of a difficulty already lurked in the research, however, especially in Gelman and King (1993) and Campbell (2000). Both argue that one function of a campaign is to prime the economy and so ensure that this most "fundamental" of considerations operates as forecasting models say it should. Both argue that the economy will always be primed in fact, as it would always be in one side's interest to do so, just by different sides in different contexts.
In the first transition of the 2000 campaign, exactly this happened when Bill Clinton told the Democrats in convention to be more optimistic about and to take credit for the economy over which he had presided. Democratic identifiers did as they were told. This did not increase partisan bias in economic perceptions, it just removed a partisan perversity in perception that reflects the ongoing class basis of U.S. politics. The convention thus fulfilled the preconditions for the presidential election forecasting models. But when Gore failed to champion the message of Democratic prosperity, the effect faded. Despite both robust economic indicators and general public satisfaction, the economy could not burnish Gore's prospects on its own. The rhetoric of the convention got the preconditions right; Gore's silence stilled its potential effect.
Gore's refusal to prime the economy may have reflected anxiety about associating himself too closely with his predecessor. This anxiety underscores a contingency taken for granted in the forecasting models. The standard claim is that popular approval of an administration mainly reflects its management of the economy. Other elements in popularity are similarly policy-driven or reflect ongoing partisan bias that carries no net predictive significance. Judgment on the incumbent's personal life is just not a factor. What forecasters did not forecast is that a president could be regarded as a competent chief executive but a bad human being, someone whose moral failings undercut his successor's ability to embrace prosperity. We surmise that Gore calculated that a tie to Bill Clinton the man would diminish his prospects more than a tie to the Clinton economy would enhance them. We suspect he was wrong. Whatever we think, the link between Al Gore and Bill Clinton created a contingency, a strategic choice and a challenge to Gore's rhetorical skill. The unfolding of the story reversed the logic outlined in Zaller (1998).
The Persuasibility of the Electorate
A pivotal event was the collapse in Al Gore's reputation as a man of character. There are many things to say about this but the first is that the collapse was quintessentially a media phenomenon, involving ads and news working in concert. A claim that media effects matter to the electoral bottom line is still controversial, notwithstanding the emergence of serious research with data from the field. For decades, the standing position in political science scholarship was the "minimal effects" model. It is useful to think of this model in terms of two mediating factors in any attempt at persuasive communication:
• How likely is the message to be received by the target audience?
• How likely are receivers, once they get the message, to yield to its persuasive content?
Among political scientists, these questions are associated with Zaller (1992). This sequence was first identified by the Yale studies of attitude change and social influence, typified by Hovland and Janis (1959) and brilliantly synthesized by McGuire (1968, 1969). It was independently identified by Converse (1962), although he seems never to have connected his insight to the Yale school one. Converse's 1962 idea lay mainly dormant, however, until Zaller (1990, 1991, 1992) resurrected it and explicitly linked it to McGuire's synthesis.1 Early communications studies in sociology and social psychology drew skeptical conclusions about each mediating factor. The standard view was that persuasive messages are unlikely to reach their target audience, at least not in an unmediated way. Audience members who do get the message resist it. Those susceptible to the message never get it. Katz and Lazarsfeld's (1955) classic study of the two-step flow of social communication argued that the audience never gets the message, not in its original form. Most hear it only second hand, after it has been reinterpreted by opinion leaders to blunt its dynamic intent. Besides, most persons are well armed to resist messages. Berelson, Lazarsfeld, and McPhee (1954) documented cognitive bias in perception of candidates' issue positions, bias motivated by voters' own prior partisan commitments. Partisan stereotyping is another mechanism by which a persuasive message can be frustrated (Conover and Feldman, 1989).2 By 1960, the various mutually reinforcing elements in the pattern had come to be seen as the "minimal effects" model (Klapper, 1960). The continuing grip of the model is nicely captured by Finkel (1993).
By the 1970s, however, skepticism about the minimal effects model could already by heard. Steven Chaffee, for instance, pointedly claimed that "the limited effects model is simply not believed" by the contributors to his edited volume (Chaffee, 1975: p. 19). It is telling, however, that this cri de coeur came from the field of communications research - with its professional stake in finding effects from the very thing it studies - not from political science. But Chaffee was not merely whistling in the wind. McCombs and Shaw (1972) had already staked out an empirical case that the media could at least set the agenda for political discourse and by 1981 their perspective had become commonplace.3 By the 1980s political scientists were willing to pick up the thread. Erbring, Goldenberg, and Miller (1980) were the first to acknowledge agenda setting, and Iyengar and Kinder (1987) documented the phenomenon on a national scale. Outright persuasion was still not on the screen, however.4 This changed with Page, Shapiro, and Dempsey (1987), who showed how the news moved opinion, although not on the time line of campaigns. Jamieson (1992) raised the specter of persuasion inside campaigns, as did Johnston, Blais, Brady, and Crête (1992), who found that news and ads affected judgment on party leaders and vote intention.5
It is one thing for persuasion to occur, it is another for its effect to be permanent. The electorate, once moved, may typically return to its original position. In a time-series sense, the electorate may be basically "stationary." It is moveable by an external shock, but without continuing pressure from whatever administered the shock (a stock market collapse in mid-campaign, for example), quasi-autonomic forces undo the initial movement. If such a shock occurs right at the end as an accident of timing, it may turn the electoral tide. But provided shocks occur early enough in the campaign, their effects will be undone. Such an aggregate pattern should prevail if the dominant mode of political cognition among individuals is "memory-based" (Lodge, McGraw, and Stroh, 1989; Lodge, Steenbergen, and Brau, 1995).6 But Lodge and his colleagues argue that the dominant mode of political cognition is not memory-based but "on-line." On the on-line view, when a shock causes voters to shift their evaluation of a political object, they quickly forget the reason for the reevaluation, they just update and move on. They may shift again but only under the pressure of another shock. If this is indeed the dominant mode of political cognition, then campaign persuasion starts to look very consequential. The Lodge et al. claim is not universally accepted, however, and evidence for it is mainly experimental.7
This book provides a direct test. The fact that we identify phases in the 2000 campaign testifies, we argue, to the power of on-line cognition. Particularly impressive is the shift that this section started with: the collapse of perceptions of Al Gore's character. We show that this is the pivotal event for the entire campaign and that it was induced by an intense, but very short burst of bad news whose effects were permanent. Gore was never able to undo the particular damage, and his recovery came about only because he was able to shift the agenda to another question. But other shocks - some of them remarkable in their initial impact - saw their effects dissipate and so were less consequential. Evidently, alternative modes of political cognition coexist. Some voters may have better memories than others. Different modes may be triggered by different events or by different media channels, news versus ads, for instance.
News and Ads
Most of the evidence on the "minimal effects" record involves lack of impact from news. When audience research was in its infancy, so was political advertising as we now understand the term. As well, the minimal effects model could be said to turn on a communications stream that is geographically localized and mediated by personal influence. As such, its relevance may be more historical than current. Personal influence networks are less binding than they once were (Putnam, 2000) and so citizens may have no choice but to look to impersonal sources. Newspaper ownership and market share have consolidated (Mutz, 1998) even as the broadcast media have become the central news source. The media have become more intrusive and citizens may have become less resistant. Accordingly, as Mutz (1998) argues, we have witnessed increasingly pervasive impersonal influence, reflecting growth in the collective consciousness of society.8 Mutz's own work still concentrates on news, however, and for her the problematic thing is whether the news gets it right. But the changes she documents for mass media also apply to their role as carriers for ads.
The relatively slow rise in emphasis on ads reflects the historical record. The first presidential television ads were played only in 1952. The most controversial early ads - the Johnson campaign's "Daisy" ad, for instance - postdate most of the "minimal effects" classics in audience research. TV advertising has only recently acquired its current scale and scope, so it should be little surprise that academic research has only begun to catch up. Much of the work is devoted to characterizing the content of ads (Kern, 1989; Jamieson, 1996; West, 1997). As ads are even more ephemeral than broadcast news (at least we know when to look for news), merely gauging the volume of plays, much less assessing their impact, is difficult.9
The earliest ad impact study with dynamic evidence was of the 1988 Canadian election (Johnston et al., 1992), and that analysis required heroic assumptions. The earliest U.S.-based work was in the laboratory, most tellingly by Ansolabehere and Iyengar (1995).10 Ansolabehere and Iyengar also managed to take their laboratory insights to the field. Their contribution has been partly obscured by controversy over their claim that negative ads depress turnout.11 Ansolabehere and Iyengar also made claims about ads' directional impact, however, and their findings appear to leave the core claim of the older literature strangely intact. They show that although citizens' vote intentions can be shifted, ads are most effective when they work with, rather than against, predisposition.12 But a major crack in the edifice appeared with Shaw (1999b), who argues with data organized by state and week of campaign that ad volumes make a net difference at the margin. Romer, Jamieson, and Cappella (2000) question Shaw's claim about the magnitude of ad effects but not the fact of their existence.
Perceptions of Candidate Traits
Shaw and Romer, Jamieson, and Cappella aside, most work on ad impact has worked with ad content and has generally focused on messages that clarify means-ends relationships in the domain of issues. The domain in question is inherently positional. But campaigns also process valence information. In most political science accounts the valence consideration in question is the economy. But another valence factor also commonly pervades ads: personality traits of the candidates. Honesty (unlike, say, abortion) is something everybody agrees is a good thing. Of course, perception will be biased, as Democratic identifiers typically see the Democratic candidate in a good light, and so on. But this is just bias, not a position on an issue that actually divides the parties. Presumably, not all trait perception is projective in this sense. Many citizens have no partisan reason to project in the first place and even those who do project may yield to new evidence furnished by the campaign.
Exactly this happened in 2000. The transition between the first and the second phase of the campaign - the undoing of Al Gore - shows how perceptions of candidates' traits are shaped by communication. The most critical shift was in perceptions of Al Gore's honesty. The shift was induced by a rough coincidence of ad and news messages, where Republican ads basically handed a message to TV news. The news in turn undermined perceptions of Gore, a process that was only accelerated by the first debate, which was treated in TV news as a further example of the problems first identified in Republican ads. The fact that ads and news worked together at this point magnified the overall effect.13 Gore's predicament was somewhat mitigated by the fact that the campaign also worsened perceptions of George W. Bush's basic competence. In the end, voters saw a tradeoff between Bush and Gore. Had the election taken place six weeks earlier, the choice before voters would have seemed simpler.
The factor that produced the shift, perception of Al Gore's personal character, is not commonly seen as a major electoral consideration. The literature on voting and elections takes due notice of candidate traits, to be sure. A multitrait battery is a regular feature of U.S. National Election Study (NES) instrumentation, going back to Kinder, Abelson, and Fiske (1979). (Indeed, a version of the Kinder battery in the Annenberg survey is the basis of our own claims about trait perceptions and effects.) Candidate assessment is a stage in the Miller-Shanks (1996) multistage model that is now the industry standard. But few would argue that personality perceptions are the key to distinguishing elections from each other. Bartels's (2002) recent review suggests that candidate perceptions were, if anything, a smaller factor in 2000 than in other years, although potentially important because of the closeness of the result. We confirm Bartels in detail - as far as he goes.
What his analysis misses, however, is that perceptions of Gore were the dynamic key to the campaign. What made them dynamically critical is that movement in trait perception tends to be unidirectional. Bias persists, indeed it increases, but no group in the electorate actually grew fonder of Gore as all others were growing less fond. Increase in bias came about only as some groups reevaluated Gore more quickly and more totally than others.
Acknowledgements; 1. Introduction; 2. The evolution of vote intentions; 3. The landscape; 4. Ads and news: the campaign as a natural experiment; 5. The economy, Clinton and the first phase; 6. Candidate traits and the second phase; 7. Social security and the third phase; 8. Conclusions; Appendix; References; Index.