11 Donald R. Kinder Communication and Politics in the Age of Information Over the latter half of the twentieth century, mass communications have transformed the landscape of American politics, vastly increasing the information about public affairs that is available to ordinary citizens. Through multiple channels—network, cable, and satellite television, radio, newspapers and magazines, regular, overnight and electronic mail, and the Internet and World Wide Web—the volume of information relevant to politics circulating through American society is massive and increasing (Lindbiom, 1977; Mutz, 1998; Pool, 1983). Today we Americans are virtually bombarded with news and propaganda about public affairs: inundated with suggestions about how issues should be understood; instructed on which problems are worth our attention; informed as to how our institutions and officials are performing; told when our opinions are sensible and when they should be altered; and advised what actions, if any, we should take. With what effect? Not much, was Joseph Klapper's (I960) famous and surprising answer. In The Effects of Muss Communication, Klapper presented a careful and thorough review of the available findings. Sifting through the evidence, Klapper concluded that "mass communication functions far more frequently as an agent of reinforcement than as an agent of change" (p. 15). My purpose here is to survey the same terrain of communications and politics that Klapper reviewed 40 years ago. Like Klapper, 1 take my subject to be the effects of mass communication: that is, communication that takes place predominantly one-way, from a small number of professional communicators to a vast number of amateur "receivers" (Pool, 1973). My review concentrates on the contemporary and near contemporary United States, as Klapper's did, and so takes for granted the presence of liberal democratic institutions (fair and frequent elections, the right to vote widely extended, citizens free to express their views; Dahl, 1989) on the one hand and mass communication enterprises driven by commercial imperatives and governed by professional norms of objectivity and balance (Neuman, 1991; Schudson, 1978), on rhe other. Whether different findings might obtain in other so-cieries, arranged in sysrematically different ways, is an open question. Taking up essentially the same subject Klapper did, 1 come to a very different conclusion. Klapper's ruling of "minimal effects" was faithful to the evidence available to him at the time, but now, some four decades and M A SS POLITICAL SEHAVlOR T Communication and Politi « in the Age Qf Info rmatjon a substantial research effort later, the findings lead in quite a different direction, or so I will try to show here. I present this "new look" in communications research in three parts. Each organizes and reviews evidence around a single broad claim to mass communication influence: the first has to do with attention, che second with persuasion, and the third with action. 1 conclude with a few thoughts on how far the science of communication has come, and how far we have yet to go. k. Attention In nations the size and complexity of the United States, the command and control of public attention is accomplished—if it is accomplished at all— primarily through mass communication. Conceivably at stake in this process is influence of three kinds. Mass communication could influence how citizens make sense of politics (what I will call framing); how citizens decide what is important in politics {agenda setting); and how citizens evaluate the alternatives that politics puts before them (priming). In my review, 1 treat framing, agenda setcing, and priming as though they were separate and distinct processes. In the literarure, however, the distinctions are nor so clearly drawn, and particular empirical examples may prove difficult to classify. 1 proceed nevertheless on the idea that it is analytically useful to organize communication findings not around attention in general but around framing, agenda setting, and priming in particular. Framing How might Americans go abour making sense of what Walter Lippmann (1925, p. 24) once called the "mystery off rhere," the "swarming confusion of problems" that populate public life? Lippmann understood that a good answer to this question should begin by recognizing that in modern society, ordinary citizens must rely on orhers for their news of national and world affairs. Such reports inevitably and inescapably privilege and promote particular points of view. Reporters and editors but also presidents, members of Congress, corporate publicists, activists, and policy analysts are all engaged in a more or less continuous conversation over the meaning of current events. In one common vocabulary, this conversation takes place through an exchange of "frames" (Gamson, Croteau, Hoynes, & Sasson, 1992). Frames, it is said, "make the world beyond direct experience look natural" {Gitlin, 1980, p. 6); they "bring order to events by making them something that can be told about; they have power because they make the world make sense" (Manoff, 1986, p. 228); they supply "a central organizing idea or story line that provides meaning to an unfolding strip of events, weaving a connection among them" {Gamson & Modigliani, 1987, p. 143). Frames come in all sizes: "master frames" that coordinate particular accounts of grievance (Snow oc Benford, 1992) or that speil out what politics is about (Capella &C Jamieson, 1997; Weaver, 1972) as well as "issue frames" applied to a single controversy, as in the argument that affirmative action for blacks should be understood as reverse discrimination against whites (Gamson & Modigliani, 1987; Kinder & Sanders, 1990). Frames include both the rhetorical tools fashioned by political elites to advance their ideas and the often unarticulared rules of selection, emphasis, and presentation that govern the work of journalism. Frames take on significance because politics is complex. The issues taken up by government and the events that animate political life are subject to alternative interprerarion; they can always be read in more than one way. Framing arises "whenever there is more than one way to think about a subject" (Popkin, 1993, p. 83). It might be said that in pure form, frames supply no new information. Rather, by offering a particular perspective, frames organize—oi better, reorganize—information that citizens already have in mind.1 Frames suggest how politics should be thought about, encouraging citizens ro understand events and issues in particular ways.- To say that frames supply no new information, that they merely supply a framework for organizing information, is not to say that frames are innocuous or neutral. By defining what the essential issue is and suggesting how to think about it, frames imply what, if anything, should be done. Elites spend as much time and money as they do crafting and disseminating frames on the assumption that the frames that prevail will shape how die public sees politics and therefore influence what the public is prepared to support. Is this assumption correct? So it seems. One empirical test for framing arises out of longstanding concerns about the capacity of ordinary people ro govern themselves. Many perceptive analysts of politics have questioned whether citizens reajly know what they want and need; whether opinions on matters of public policy are actually, in one powerful formulation, "nonattitudes" (Converse, 1970, 2000). Nonattitudes are usually taken as a sign of the average citizens' indifference to politics, but they may also point to the absence of a serious debare among elites. Put the other way around, when elites provide useful frames, citizens may be more likely to see a connection berween what they care about and what politics offers, and so may be more likely to develop real opinions. This hypothesis has been tested in a series of question-wording experiments embedded in national surveys. These experiments compare public opinion elicited in one of two ways: either by questions that refer explicitly to the rival frames that dominate elite discourse (the "framed" condition) or by questions that do not (the "stripped" condition). Here the framed trearmenr is intended to mimic the political situation in which citizens witness a debate between opposing elires, each pushing an alternative definirion of the issue. Meanwhile, the stripped treatment is intended to simulate the situation where there is no elite debate, where citizens are on their own in formulating what the issue is about. As predicted, for a variety of issues, when provided helpful frames, citizens are MASS POLITICAL BEHAVIOR more likely to express opinions, and such opinions are often more stable over time and better anchored in the political considerations that the frames appear to highlight (e.g., Kinder & Nelson, 1990; Kinder & Sanders, 1996; Zaller, 1990). Other experiments compare one frame against another. Because alternative frames highlight different features of an issue, they should alter the relative weight given to the interests, group sentiments, and political values that potentially go inco making up an opinion—and so they do (e.g., Ca-pella & Jamieson, 1997; Jacoby, 2000; Kinder & Sanders, 1990, 1996; Nelson, Clawson, & Oxley, 1997; Nelson & Kinder, 1996; Price, 1989). In this respect, frames are like recipes, advice from experts on how citizens should cook up their opinions (Kinder & Sanders, 1996). Shifting the underlying foundations of opinion is one thing, moving opinion itself is another. Thar framing may accomplish the first is no guarantee that it can puil off the second. For to move opinion, a frame must not only be compelling (that is, fit its subject, well), though it must be that, and it must nor only induce large numbers of people to think about the subject in a new way, though it must do that as well- To move opinion, a frame must also induce large numbers of people to think about the subject in a way that pushes them in a new direction. Suppose, to borrow an example from Nelson and his colleagues, the Ku Klux Klan plans to hold a rally in your community. The local press could frame this news to highlight the constitutional protection of assembly and speech or to highlight the government's obligation to preserve order. The two frames are compelling, they induce large numbers of people to think about the subject in a way they otherwise would not, and they push hard in opposite directions. Because Americans believe both in protecting free speech and in preserving social order, when presented with frames that highlight the one as against the other, they end up expressing different positions on whether the rally should be held ar all (Nelson, Clawson, & Oxley, 1997; for other examples and discussions of opinion change through framing, see Chong, 1993; Kinder & Sanders, 1996). Virtually all the framing results 1 have cited so far come from experiments. Experiments have real advantages (Kinder & Palfrey, 1993; Snider-man & Grob, 1996), and these experiments in particular have some very desirable features. For one thing, mosr of them deliberately mimic actual elite debates and everyday journalistic conventions. For another, many are inserted into representative sample surveys, so the common complainr about experiments—that they exploit convenient bur unrepresentative populations (compare Sears, 1986, and Anderson, Lindsay, & Bushman 1999)—does not apply here. Neverrheless, experimental results can always be questioned on their generalizability, ;md framing effects are no exception. The major worry in this respect is that framing experiments—like experiments in mass communication generally—typically erase the distinction between the supply of Communication and Politics in the Age of Information information on rhe one hand and its consumption on the other. That is, experiments are normally carried out in such a way that virtually everyone in rhe audience is hand-delivered the message. The typical experiment thereby avoids a major obsracle standing in the way of communication effects: namely, an inattentive audience, lost in the affairs of private life (more on this point later). By ensuring that frames reach their intended audiences, experiments may exaggerate their power, A more balanced reading of frame effects requires methodological diversification, experiments and studies oriented to the world outside.-* A second problem is that while the experimental literature I've just reviewed makes a good case that frames can affect how (and even whether) people evaluate various matters of politics, it actually skips over the focal concern I began with: how people make sense of these matters in the first place. In all the studies of frames and framing, understanding itself is never directly addressed or measured. This is unfortunate, since democratic institutions often presume rhat ordinary citizens undersrand the matters that come before rhem. Ir is acrually doubly unfortunate, because over the past decade, cognitive psychology has been developing conceptual and methodological tools that are well suited to analyzing the problem of political sense-making. Here I have in mind Pennington's and Hastie's analysis of jurors transforming the jumble of evidence presented to them during a trial into a compelling narrative (Pennington & Hastie, 1992, 1993); Holyoak's and Thagard's (1995) accounr of reasoning by analogy; and Kintsch's (1998) theorizing and research on how people are able to undersrand text". In shorr, the conceptual and methodological tools to undertake a serious examination of how ordinary Americans make sense of the mysteries of polirics now appear to be available (for a first and small step in this direction, see Berinsky & Kinder, 2000). To pin down more precisely what frames are and how they shape undemanding seems a worthwhile project, since, as Gitlin (1980) once put it, frames appear to "organize the world both for journalists who report it and, in some important degree, for us who rely on their reports" (p. 7)." Agenda Setting Among the most important decisions in any society are those that determine which issues become part of politics (Bachrach & Baratz, 1970). In Schattschneider's crisp formulation, "some issues ate organized into politics while others are organized out" (I960, p. 71). The public is somehow implicated in this process: citizens develop ideas about what is and what is not important, which problems are and which are not proper subjects for government action, and these ideas, in turn, shape and constrain what gov-eminent attempts ro do (Bursrein, 1985; Verba & Nie, 1972). How does the public come to believe that crime is an urgent problem and acid rain is not? 362 MASS POIJ1 ICA1. BKHAVIOR Move than 50 years ago, Lazarsfeld and Merton (1948) suggested that the answer co such a question might lie in the agenda setting power of the news media. While expressing considerable skepticism about the influence of mass communications in general—they regarded the automobile as vastly more consequential than the radio and thought that much of what passed for contemporary analysis was infected by "magical" thinking—Lazarsfeld and Merton nevertheless proceeded to sketch out several mechanisms through which mass communications might have real effects. Prominent among these was agenda setting: "mass media confer status on public issues, persons, organizations, and social movements" (p. 101). Lazarsfeld and Merton thought that mere attention was enough, that "enhanced status accrues to those who merely receive attention in the media, quite apart from any editorial support. . . . Recognition by the press or radio or magazines or newsreels testifies that one has arrived, chat one is important enough to have been singled out from the large anonymous masses, that one's behavior and opinions are significant enough to require public notice" (pp. 101 — 102). Good idea—-but an idea only. Lazarsfeld and Merton offered no evidence for their conjecture, and a dozen yeats later, neither could Klapper. Agenda-setting's first and rather oblique brush with evidence did not come until Cohens (1963) perceptive analysis of newspapers and U.S. foreign policy. On the basis of interviews with journalists and government officials, Cohen concluded that the press "may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about" (p. 13). Cohen had in mind policy experts and the attentive public, however, so his strong conclusion doesn't necessarily speak to agenda setting among the general public, what Riket (1993) once called the "misty swamp" (p. 2) of everyday politics. More immediately relevant is the evidence supplied by McCombs and Shaw (1972), who, toward the end of the 1968 presidential campaign, interviewed a small sample or uncommitted voters living in and around Chapel Hill, North Carolina. McCombs and Shaw found an almost perfect correlation between the problems that voters believed were rhe country's most serious and those problems given great prominence in the news they were reading and watching at the time. McCombs and Shaw concluded that "in choosing and displaying news, editors, newsroom staff, and broadcasters play an important part in shaping political reality" (p. 176). McCombs's and Shaw's successful demonstration inspired numerous replications and, for a time, considerable confusion. The best studies turned up modest and often mysteriously contingent support for agenda setting (e.g., Erbring, Goldenberg, & Miller 1979). "Stunningly successful" (Cohen, 1963, p. 13) overstates these results considerably. The problem, looking back on it, was a failure of design. Cross-sectional comparisons miss the real variation in agenda setting, which is temporal Communication and Politics in the Age of Information 363 rather than spatial. Agenda setting is dynamic—problems emerge, move for a while to the center of the srage, and then gradually drift back ro the wings—and so should be investigated over time. Consistent with this, Funkhouser (1973) found in a simple analysis a close correspondence between the amount of attention paid to various problems in the national press over the 1960s on the one hand and the importance subsequently accorded such problems by the American public, on the other. This basic finding has stood up well to further and more sophisticated testing. Subsequent studies have controlled on the possibility that news organizations are responding to the public's priorities (and not just the other way around); have taken into account rhe independent effects due to real-world conditions (e.g., changes in prices or interest rates); and have simultaneously considered the possibility that presidenrs can alter what the public takes to be important through major addresses to the nation. Under these quite stringent conditions, strong empirical support for agenda setting repeatedly emerges. Rising prices, unemployment, energy shortages, arms control: all these (and more) become high-priority issues for the public after they first become high priority for newspapers and networks. For a wide variety of problems, the American public's concern for political problems closely and rapidly tracks changes over time in the attention paid them by media (e.g., Behr & Iyengar, 1985; Fan, 1992; MacKuen, 1981, 1984; McCombs & Zhu, 1995; Neuman, 1990; Protess et al., 1991). These results are nicely complemented by the findings from experimental research (Iyengar & Kinder, 1987; Miller & Kxosnick, 2000). In effect, these experiments convert the variation in news coverage that occurs naturally over time to contemporaneous variation across experimental conditions. People shown network broadcasts unobtrusively edited to highlight a particular problem subsequently assigned greatet importance ro that problem—-greater importance than they themselves did before the experiment began and greater importance than did people randomly assigned co control conditions that emphasized different problems. These effeccs are apparent immediately after the conclusions of the broadcasts as well as one week later, are enhanced when the edited stories lead off the broadcasts, and emerge for a wide array of problems.'' If agenda-setting effects are robust and powerful, chey also seem to be quite specific. Thus news about energy affects beliefs about the importance of energy and energy alone; news about national defense affects beliefs about national defense and national defense alone; and so on. Stories about one problem don't seem to spill over onto others, even when the problems are related. Such specificity may reflect the way news is typically presented, in tight, self-contained bundles (Weaver, 1972). More important may be the way most Americans approach politics, innocent of broad ideological frameworks that might link one national problem with another (Converse, 1964; Kinder, 1998). 366 M A S S I' O 1. 1 T ICSL BtHA V I O R nick & Kinder 1990). There are exceptions, however, and they show up (so far at least) primarily in a single realm: race. In that realm, communications that have nothing ostensibly to do with blacks and whites nevertheless appear to prime the audience's racial predispositions. Flagrantly racist speech is now out of bounds in the United States, and most white Americans subscribe to facial equality as a matter of principle. At the same time, many whites feel resentful toward blacks and uncomfortable in their presence. These conditions have given rise to euphemistic discourse—appeals to racism that traffic in code words (Kinder 8i Sanders, 1996). Mendelberg (2001) refers co this as "implicit communication." Drawing on a tecent stream of research in psychology (e.g., Banaji & Greenwald, 1995; Bargh & Pietromomaco, 1982; Devine, 1989), Mendelberg argues that carefully calibrated messages can prime racial predispositions automatically, outside the recipient's awareness. A covert appeal to racism is successful precisely because ir evades the self-censorship that whites would exercise if they were to notice that the incoming message violated the norm of equality. A typical if demoralizing example of covert communication is provided by the 1988 American presidential campaign and the carefully crafted and well-coordinated initiative to portray the Democratic nominee, Governor Dukakis, as soft on crime. In some ways 1 988 can be read as a clear example of Petrocik's (1996) issue ownership theory in action: the Republicans stressing their issue (controlling crime) while the Democrats tried to change the subject.1. The 1988 campaign thus appears straightforward—except that the Republicans' effort, though ostensibly about crime, was really about race, featuring the story of one Willie Horton. Hotton, a black man convicted of murder and sentenced to life imprisonment, was granted a weekend leave by the Massachusetts prison furlough program while Dukakis was governor. Horton fled the state and terrorized a white couple in Maryland, beating the man and raping the woman before being recaptured and returned to prison. In the local uproar that followed, Dukakis defended the furlough program and appeared indifferent to Horton's victims. Horton's story became a fixture in Bush's speeches, in Republican campaign fliers, and in a set of memorable television advertisements. This was covert communication about race. On the surface, the message was about crime and crime only, and race words were scrupulously avoided. But Horton's name and story and picture were ever-present. The effort was deliberate (Jamieson, 1992; Kinder & Sanders, 1996; Mendelberg, 2001), and it succeeded: the campaign primed predispositions on race (not crime), moving racially conservative Democrats and independents into the Republican column (Kinder & Sanders, 1996; Mendelberg, 1997, 2001). In the 198S campaign, and in a variety of other cases (Mendelberg, 2001; Valentino, Hurchings, & White, 2002; Winter, 2001), communications that appear to be about one subject—crime, welfare reform, government spending, privatization of Social Security—turn our, in rheir consequences, to be about another, namely, race. " r Communication and Politics in the Age of Information L Persuasion By persuasion I mean "changing people's attitudes [and behavior] through the spoken and written word" (McGuire, 1973, p. 216). Political persuasion entails the supply of arguments and evidence through which people are induced to change rheir minds about some aspect of politics: in light of new information, people come to think that the president is smarter than he first seemed, or that school desegregation is ineffective and should be abandoned, or that more effort and money must now be invested in national defense. On the question of what role mass communication plays in political persuasion, Klapper's answer of course was nor to worry: mass communications seldom persuade, and minimal effects are the rule (others who say this are Hovland, 1959; Key, 1961; McGuire, 1986, 2001; Mueller, 1994; Schudson, 1984). Zalier (1996) disagrees, insisting that the righr answer to the question of persuasion is not minima! effects, but, to paraphrase lightly, massive effects, all the time. The truth, as I will rry to show, lies somewhere in berween. Klapper was influenced heavily by The People's Choice, the landmark examination of the 1940 presidential contest in Erie County, Ohio, that was carried out by a team headed by Paul Lazarsfeld (Lazarsfeld, Berelson, & Gaudet, 1948). To their surprise, Lazarsfeld's team discovered that relatively few voters altered their intentions over the course of the year. Indeed, by the time of the summer conventions, before the formal campaign even began, roughly 80 percent of voters had become permanently committed to one candidate or the other. From spring to fall, only a handful of voters—some 5 percent—actually changed sides. What little change that did occur, moreover, apparently had less to do with the campaign than with the personal influence of family and friends. Lazarsfeld, Berelson, and Gaudet concluded that presidential campaigns are generally ineffective at persuasion. Rather than converting voters from one side to the other, presidential campaigns reinforce the early deciders and activate the latent predispositions of the initially uncommitted. So it was in 1940, and so it was, evidently, in 1948. In The Voter Decides, Berelson, Lazarsfeld, and McPhee (1954) report the results of a similar investigation, situated this time in Elmira, New York, and focusing on the famous Dewey-Truman presidential contest. Once again, or so it seemed, mass communication "crystallizes and reinforces more than it converts" (p. 248). Klapper took the results from these two exemplary studies seriously, and he was wise to do so, for their main conclusions have stood up very well. Despite dramatic changes over the last half-century in politics and society, American presidential campaigns still fail as exercises in political persuasion.7 We know this from panel studies patterned after the original Lazarsfeld design (e.g., Barrels, 1997; Finkel, 1993; Markus, 1982; Patter- MASS POLITICAL BEHAVIOR son, 1980) and from aggregate forecasting models that accurately predict the outcome of American presidential elections from information available before the campaign gets underway (e.g., Banels, 1992; Gelman & King, 1993; Rosensrone, 1983). Klapper was right about presidential campaigns: riot much persuasion, mostly reinforcement and activation.3 Conceding this point.—that little persuasion takes place from Labor Day to Election Day—should not, however, lead us to conclude thac presidential campaigns are therefore dispensable. To the contrary, activation and reinforcement are vital political processes, as Lazarsfeld and his colleagues knew. Campaigns activate voters by arousing their interest and providing them with information, thereby allowing them to choose wisely—or in any case, more wisely than they would have in the absence of a campaign (Barrels, 1988; Gelman & King, 1993; Johnston, Blais, Brady, & Crete, 1992; Shanks & Miller, 1992). And campaigns reinforce voters by providing good reasons for their choices, reminding them why they are Democrats or Republicans, thereby keeping partisans in line and defections to a minimum (Barrels, 1993; Katz & Fddman, 1962; Sears & Chaffee, 1979). Acknowledging the importance of activation and reinforcement, the main and arresting lesson of more than a half-century of empirical research on presidential campaigns is the failure of persuasion. In the next section ! sketch out three reasons why this might be so—^three principle obstacles to persuasion: neutralization, resistance, and indifference. Neutralization One reason why presidential campaigns are ineffective as instruments of persuasion is that the campaign mounted by one side is "neutralized'' by the campaign mounted by the other (Lazarsfeld & Merton, 1948; Battels, 1992; Gelman & King, 1993). Under current arrangements—electoral competition dominated by two well-established parties and presidential campaigns funded primarily by public sources—both sides assemble roughly equally capable teams, of roughly equal experience and intelligence, who set about spending roughly the same (large) amount of money in roughly the same ways." Whatever persuasive eflect is accomplished by the one is offset by the other. Imagine a different world. What would we see if one of the political parties managed somehow to monopolize mass communications? European experience with totalitarian control of communications suggests that under some conditions the opposition may be whittled down until only the firmly convinced die-hards remain. In many parts of this country, there are probably relatively few people who would tenaciously maintain their political views in the face of a continuous flow of hostile arguments. Most people want—-and need—to be told that they arc right and to know that other people agree with them. Thus, die parries could Communication and Politics in the Age of Information forego their propagandizing only ar considerable risk, and never on a unilateral basis. (Lazarsfeld, Berelson, & Gaudet 1948, p. 87) It seems unimaginable that either party will unilaterally disarm, for the reasons that Lazarsfeld, Berelson, and Gaudet suggest, and so neutralization—a stalemate of roughly equal and opposing forces—is likely to be with us indefinirely in presidential campaigns. But campaigns need not be as two-sided as presidential contests normally are. Consider, as a concrarv and real example, campaigns for the U.S. House, which these days are famously lopsided. House incumbents typically raise vast sums of money, much more than their challengers, and this enables them to run virtually-continuous campaigns for reelection fjacobson, 1980, 2001). As a consequence, incumbents entice significant numbers of voters who ought to be voting for their challengers to vote for them instead, and these "defections'* turn out to be concentrated among moderately attentive voters; just those who are most likely to notice the incumbent's campaign but miss the weaker campaign mounted by the challenger (Zaller, 1992). A similar point is suggested by the remarkable success enjoyed by Ross Perot in 1992. Third-party candidates arc a regular though usually invisible-feature of American presidential elections, but in 1992 Perot won nearly 20 percent of the popular vote, surpassing rhe record of any third-party candidate since Teddy Rooseveh. Perot did so well at least in part because of his ability to mount a serious campaign. Perot outspent both Clinton and Bush in 1992, shelling out some 73 million dollars, nearly nil of it drawn against his own personal fortune. And he apparently spent it well, building and maintaining a national and professional organization and mounting a memorable advertising campaign and spending 45 million dollars on television alone. Most third-party campaigns are underfinanced, ragged affairs; Perot's quite diffetence experience shows the potential persuasive power of campaigns (Alvarez & Nagler, 1995; Rosenstone, Behr, & Lazarus 1996; Zaller & Hunt, 1994, 1995). Resistance Klapper (1960) thought that a principal cause of stability in voters' partisan preferences over the course of presidential campaigns was selective exposure: "by and large, people tend to expose themselves to those mass communications which are in accord with their existing attitudes and interests. Consciously or unconsciously, they avoid communications of opposite hue" {p. 19). Klapper was sure that this was so, and he was not alone (Behavioral Sciences Subpanel, 1962; Berelson & Steiner, 1964; Festinger, 1957, 1 964; Hyman & Sheatsley, 1947; Lazarsfeld, Berelson, & Gaudet, 1948). Despite all the early confidence, the evidence for selective exposure turns out to be thin. We now know chat people do not, for the most pan, seek out mass communications that reinforce their political predispositions, 370 MASS POI.U 1CAL BEHAVIOR nor do they choose to seal themselves off from communications that threaten them (e.g., Eagly & Chaiken, 1998; Freedman & Sears, 1965; Iyengar et al., 2001; Milburn, 1979; Mutt &; Martin, 2001; Sears & Freedman, 1967). All things considered, this may be a good thing, since selective exposure assumes that people are "living autistically in a fool's paradise, endeavoring to remain blissfully ignorant of belief-threatening material, even though in actual environments it is often adaptive to acquaint oneself with the opposition arguments (McGuire, 1985, p. 275). Maladaptive or- not, selective exposure cannot account for persuasion's failure."1 People are adept at defending their views, not by hiding from threatening messages but by rejecting them (Abelson, 1959, 1968; Sears & Freedman, 1967). Thar is, persuasion fails not through selectivity in exposure but through selectivity in acceptance. In scores of studies of presidential debates, for example, voters' evaluations of debate performance polarize sharply along partisan lines (Katz & Feldman, 1962; Sears & Chaffee, 3979). More generally, people assess communications that challenge their attitudes as weakly argued, unconvincing, and laced with error (e.g., Lord, Ross, & Lepper, 1979; Pomerantz, Chaiken, & Tordesillas, 1995), and they subject such communications to more active counterarguing and greater scrutiny than they do communications that confirm their attitudes (e.g., Cacioppo & Petty, 1979; Edwards & Smith, 1996; Feather, 1963). From this point of view, presidential campaigns enjoy such limited success in converting voters from one side to the other partly because they run up against heavy resistance. Resistance is especially heavy because presidential campaigns play directly on the voters' primary political predisposition, their attachment to party. For many partisans, the details brought forward by any particular campaign are simply further corroboration of their party's comparative virtuosity (Barrels, 1993; Campbell et al., 1960; Converse, S966; on the decline and revival of strong partisanship in the American electorate, see Bartels, 2000)." Resistance is a serious obstacle to political persuasion, but if it cannot be avoided altogether, it can be reduced. For example, persuasion should be more common during the presidential nomination process, where the strong predisposition of partisanship is less relevant (Barrels, 1988). Like-wise, emerging issues or topics that elicit relatively weak attitudes should be opportunities for persuasion as well (Bassili, 1993, 1996; Krosnick & Abel-son, 1992). More generally, resistance will diminish when people lack the contextual information that would otherwise enable them to notice that an incoming message is inconsistent with their predispositions (Zaller, 1992). Indifference H strong partisans are essentially unmovable because they are so deeply committed, other potential voters are difficult to persuade because they're not paying attention. In the midst of analyzing the Hood ol political prop- Communicarion and Politics in (he Age of Information aganda unleashed by the 1940 presidential campaign, Lazarsfeld, Berelson, and Gaudet (1948) noticed that far from being drowned in information, most voters "did not even get their feet wet" (p. 121). Lazarsfeld and his associates were teferring to presidential campaigns, but their point is a very general one. "Perhaps the principal incontestable moral of the data about politics and mass communications," V. O. Key once w'rote, "is that many of the political messages catried by the communications networks do not reach many people. The limits of the audience fix the area of direct influence of the mass media: a message unheard is a message unheeded" (1961, p. 345). A major obstacle in the way of persuasion is a habitually inattentive and often distracted audience (Converse, 1975; Delii Carpini & Keeter, 1996; Hyman & Sheatsley, 1947).12 That said, there remain two general openings for persuasion. First is what I will call low-information persuasion. The major claim here is that generally uninterested citizens might still be susceptible to persuasion whets elite, sources supply simple and decisive cues; simple so that the cue can be easily communicated and decisive so that the cue can make a difference. An excellent demonstration of this poinr is provided by Lupin (1994), who showed that California voters who knew little about the details of various complicated proposals to reform the automobile insurance industry in rheir state nevertheless made choices that were indistinguishable from those made by well-informed voters. All the generally ill-informed voters needed to know was which proposals were backed by which interest groups. When Californians knew, for example, that the insurance industry itself was behind a particular proposal, they knew enough to vote against it. In terms of Petty and Cacioppo's elaboration likelihood model of attitude change, Lupia is tracing out a peripheral route to persuasion (Petty & Cacioppo, 1986; for a related fotmulation, see Chaiken, Liberman, & Eagly, 1989).13 By taking cues from those who are well informed (e.g., Brody, 1991; Katz & Lazarsfeld, 1955; Huckfeldt & Sprague, 1995; Lupia & McCubbins, 1998), common citizens can "be knowledgeable in their reasoning about political choices without necessarily possessing a large body oi knowledge about politics" (Sniderman, Brody, & Tetlock, 1991, p. 19).'1 The second opportunity is persuasion over the long haul. An excellent case in point is supplied by Page and Shapiro (1992), who find in a wide-ranging study that in those instances when American public opinion on matters of policy has changed, "it has not done so wildly or capriciously or randomly; it has generally shifted in comprehensible ways, in response to new information and changing condirions" (p. 321). The changing conditions that Page and Shapiro have in mind are for the most part historic events—World War II, riots that raced through American cities in the 1960s, the skyrocketing inflation of the late 1970s, Chernobyl—and the new information that they say moves opinion comes notably from presidents, commentators, and experts, supplied to the general public through mass communication. 372 MASS. POLITICAL B E 11 A V 1 O R Page and Shapiro's result turns out to be quite representative of the empirical returns from a wide range of recent inquiry into the dynamics of public opinion (e.g., Fan, 1988; Haller & Norpoth, 1994; Haynes & Jacobs, 1994; Hibbs, Rivers, & Vasilatos, 1982; Kellstadt, 2000; MacKuen, 1983; MacKuen, Erikson. & Stimson, 1989, 1992; Mueller, 1971, 199L 1994; Ostrom & Simon, 1985; Stimson, 1991/1998). In these various investigations, public opinion in the aggregate seems quite responsive to social, economic, and political change, indeed sometimes exquisitely so. The role • of the media in these dynamics is not always explicit, but it is always at least implied. In short, if we ask not about the short term but the longer haul, and not about individuals but about the public as a whole, then persuasion seems to be the rule, not the exception. How can this be? How, especially, can this result be reconciled with the supposedly imposing obstacle to persuasion posed by the low exposure and low interest of the average citizen? One answer is that the engine of opinion change pointed to in these studies is history in capital letters—not humdrum politics but historic turning points, like World War II or the civil rights movement, that are exceptional in the attention they command and the interest they arouse. A second answer is that the movement in collective opinion that these studies uncover, may be due primarily to change among the most attentive members of the public. In aggregating to the public as a whole, the random error contributed by individual ignorance, indifference, and confusion may be canceled out (though not always and never completely; see, for example, Althaus, 1998; Battels, 1996a; Berinsky, 1999). Given a sufficient number of cases, clear signals can emerge from even a "sea of noise" (Converse, 1990, p. 382). The implication here is that the systematic and sensible movement of public opinion in the aggregate may be determined dispro-portionately by a mere handful of citizens who are paying close attention (Converse, 1990; Stimson, 1991/1998; Zaller, 1992).,s To sum up my discussion so far: political persuasion is contingent on circumstance. Persuasion grows more likely when campaigns face little opposition, when resistance is diminished, when well-placed sources provide simple and decisive cues, and when history intrudes on attentive citizens. Neutralization, resistance, and indifference are formidable but not insuperable obstacles to persuasion. If KJapper was right to say that presidential campaigns in particular seldom persuade, we seem nevertheless to have come quite far from his general conclusion of "minimal effects." Zaller (1996) wants to take us fatther still, arguing that the correct conclusion is massive persuasion, almost all the time. We've failed to see this, Zaller says, for three reasons. First, we have measured exposure to mass communication poorly (here echoing Bartels's [1993] complaint). Second, we have looked for evidence of opinion change where there is no change in mass communication content."' And third (the heart of the matter), we have failed to model persuasion properly. In particular, Zaller contends that Comi- tion and Politics in the A«e of Inforn 373 standard assessments of persuasion overlook the crucial fact that citizens are often exposed to countervailing messages and so are pushed simultaneously in opposite directions. What appear in standard analysis to be minimal effects are actually, in Zaller's telling, massive but offsetting effects. Trying to do better, Zaller begins by drawing on the same set of psychological ideas that inform much recent research on communications generally. Zaller presumes that people arrive at their opinions by averaging across the considerations that happen to be accessible at the moment. Accessibility, in turn, depends on memory retrieval that is probabilistic and incomplete. Considerations that have been "in thought'' recently are more likely to be sampled. To make this model of opinion come alive for understanding political persuasion, Zaller (1992) introduces rwo additional assumprions, building on insights offered by Hovland (1959; Hovland. Lumsdaine, & Sheffield, 1949), Converse (1962), and McGuire (1968). The first is that people will be more likely to receive a communication as a direct function of their level of general information about politics, where reception involves both exposure to and comprehension of the given communication. Second, people will resist communications that are inconsistent with their political predispositions only insofar as they possess sufficient information to detect such inconsistency. Zaller's model recognizes that citizens differ sharply from one another in their partisan and ideological predispositions and that they differ enormously from one another in the care and attention they invest in politics. So specified, Zaller's model can account for a variety of empirical cases; the electoral advantages enjoyed by congressional incumbents, shifts in American opinion on school desegregation, variations in popular support for Ross Perot during his 73 million-dollar presidential campaign adventure of 1992, and changes in American opinion oil the Vietnam War, among others (Zaller, 1989, 1991, 1992; Zaller & Hunt, 1994, 1995). In all these instances, public opinion appears to move in response to alterations in the supply of information provided by elites. The story is complicated because it takes into account differences in the motivation and skill that citizens bring to politics, and because it recognizes that overall shifts in public opinion typically conceal underlying combinations of changes that move in opposite and partially offsetting directions. Zaller offered bis model as an attempt to provide a general account of public opinion. On the first page of The Nature and Origins of Mass Opinion (1992), Zaller announced his aim to "to integrare as much as possible ot the dynamics of public opinion within a cohesive theoretical system." In fact, the model does vety well, even judged against that audacious aspiration. But Zaller's is not the only way to represent and understand political persuasion. The strongest alternative psychological model currently in play is supplied by Milton Lodge (1995) and his associates, who hope to specify "the main architectural and procedural features of a psychologically realistic model of the candidate-evaluation process" (p. 111). In this pursuit. Lodge 1 AS S POLITIC A L BtHAVIOR takes inspiration from the information-processing approach to human cognition generally, drawing heavily on the "on-line" model of information processing developed by Hastie in particular (Hastie & Park, 1986; Hastie & Pennington, 1989). Lodge argues that over the course of a campaign, most citizens most of the time develop their impressions of the candidates on-line: that is, "each piece of campaign information is immediately evaluated and linked to the candidate node in working memory at the time of exposure, when the information is in the senses, so to speak, and not typ--ically computed at a later date fcom memory traces" (p. 119). Campaigns deliver messages; citizens sometimes notice them; when they do, they (somehow) detect the implications of the messages for their evaluations of a candidate; thereupon they immediately integrate these implications into their summary evaluation (or "running tally") of the candidate; transfer their now updated overall evaluation to long-term memory; and quickly forget the details that prompted the updating in the first place (for corroborating evidence for this account, all supplied by experiments, see Lodge, McGraw, & Stroh, 1989; Lodge & Steenbergen, 1995; Lodge & Stroh, 1993; Rahn, Aldrich, & Borgida, 1994). We need both on-line and accessibility (or memory-based) models. How can we reconcile their differences? One possibility is to dissolve or at least soften the differences between the two, by specifying intermediate or hybrid modes of opinion formation and change, partly on-line and partly memory-based (Hastie & Pennington, 1989). Anothet possibility is that people differ systematically and persistently in their reliance on one process or the other (Jarvis &: Petty, 1996). Still another is to say rhat the two models cover different domains. Perhaps on-line processing applies primarily to the evaluation of candidates. Citizens understand that when the campaign comes to a close they will be asked to make a decision, which should encourage on-line processing, in contrast, perhaps memory-based models apply when people are surprised by a request for a judgment on mattets that they had previously regarded as insignificant (Hastie & Park, 1986). Just such a surprise occurs, Zaller (1992) suggests, when people are accosted by an interviewer wanting to know what they think about health care or the war in Afghanistan. k. Action Up to now, 1 have been concerned with mass communication as an instrument for directing attention and changing attitudes. I will now turn briefly to the role that communication might play in influencing action. Participation in politics can take a variety of forms, but the most common is turning out to vote (Rosenstone & Hansen, 1993; Tilly, 1978; Verba, Schlozman, & Brady, 1995; Verba & Nie, 1972). And on the question of mobilizing turnout, the evidence suggests that mass communication Communication and Politics in the Age of Information ■ is less important than personal influence. Both surveys and experiments show that people who are contacted by a parry representative during the campaign are more likely to vote (e.g., Caldeira, Clausen, & Patterson, 1990; Eldersveld, 1956; Gosnell, 1927; Kramer, 1970; Rosenstone & Hansen, 1993; Wielhouwer & Lockerbie, 1994). In a field experiment carried our recently in New Haven, Connecticut, for example, Gerber and Green (2000) found that turnout was increased substantially by personal canvassing but only slightly by direct mail and not at all by appeals over telephone. These results suggest an explanation for the disappearing American voter. That turnout in American national elections is down sharply from the high-water marks of the early part of the twentieth century (Burnham, 1970; Converse, 1972; McDonald & Popkin, 2001) may have to do in part with the transformations that have come to political campaigns over this period. Where mobilization once relied on face-to-face contact between prospective voters and party activists (Gosnell, 1937; Wolfinger, 1974), mobilization belongs now to "professional campaign consultants, direct mail vendors, and commercial phone banks" (Gerber & Green, 2000, p. 653). As the political parties substituted telephone calls and direct mailings for personal canvassing, more and more potential voters remained home (Rosenstone & Hansen, 1993; Gerber & Green, 2000). While candidates and parties have been spending less of their resources' on personal canvassing, they have been spending more on mass communications, especially television. And as campaigns are waged increasingly over the airwaves, they are rurning more and more negative. So at least is the contention of Ansolobehere and Iyengar. In the television age, they say, campaigns have become "hostile and ugly. More often than not, candidates criticize, discredit, or belittle their opponents rather than promoting their own ideas and programs" (Ansolobehere, Iyengar, Simon, & Valentino, 1994). This development is not just unseemly, again according to Ansolobehere and Iyengar, but is real trouble for the American political system. For negative campaigns are demobilizing: they sour citizens on politics and drive them away from the polls. Ansolobehere and Iyengar reach this conclusion primarily on the basis of a series of well-crafted experiments in which negative advertisements are embedded within actual campaigns. They find essentially the same result—negative campaigns turn voters off—when they examine voting rates in Senate elections that vary in campaign negativity, and when they analyze turnout among Americans questioned in a pair of National Election Study surveys, some of whom were witness to negative advertisements and some of whom were not (Ansolobehere et al., 1994; Ansolobehere & Iyengar, 1995; Ansolobehere, Iyengar, & Simon, 1999). This is an impressive line of research, but it seems to me that settled conclusions on the effects of negative campaigns are not yet possible. Negative campaigns do not always lead to demobilization (Lau, Sigelman, Held-man, & Babbitt, 1999), and some forms of negative advertising, by clarifying that important differences separate the candidates, may actually 376 MASS POLITICAL B E H A V I O R enhance turnout (Kahn & Kenney, 1999; West, 1993). Nor, finally, is [t obvious that recent campaigns are all that negative (Jamieson, 1992; West, 1993). To Herbert Hoover, the New Deal was "Fascism," "despotism," and "the poisoning of Americanism" (Sundquisc, 1983, p. 301). Bryce characterized American campaigns of the late nineteenth century as "rhick with charges, defences, recriminations, till rhe voter knows not what to believe" (1888/1995, p. 879, quoted in Lau et al., 1999, p. 851). And Raker (1996) turned up plenty of criticism and hostility in the campaign for ratification or the U.S. Constitution.'~ k. Conclusions and Implications If, as Barrels maintains (perhaps mischievously), the minimal effects verdict should be regarded as "one of the most nocable embarrassments of modern social science" (1993, p. 267), I am happy to announce that we are now relieved of at least that burden. Better yet, we have replaced a sweeping generalization with a set of reasonably refined and contingent conclusions. When Klapper was writing his review, television was not yet a settled part of political life; the digital revolution, personal computers, the Internet, and the Web were still decades off. Dramatic changes have certainly come to the technology of mass communications. This system is evolving rapidly, and ir would be foolish to claim that these developments will prove irrelevant to politics (Mutz & Martin, 2001; Neuman, 1991). Nevertheless, change in our thinking about mass communications and politics over the last 40 years has had to do with alterations in social science more than in communications technology. Such alterations are partly methodological: stronger designs, more sensitive instruments, and better analysis. Srill more important, or so it seems lo me, are theoretical and conceptual advances, an infusion of good ideas. While Klapper was putting the finishing touches on The Effects of Mass Communication, a revolution was brewing in psychology, one that placed cognition at the center of study. Much of the "new look" in communication research is indebted to and inspired by this turn to cognition (e.g., Anderson, 1983; Collins & Lottus, 1975; Simon, 1955; Tversky & Kahneman, 1973, 1981). In possession of a more sophisticated understanding of human judgment and choice, research has begun to develop a dependable, complex, and detailed account of how mass communications influence the politics of everyday citizens. Such influence proceeds in a variety of ways, as I have tried to show here. Mass communications frame and organize information and facts, thereby encouraging citizens to see and understand politics in cerrain ways. By singling out some aspects of politics for special attention, media influence what problems citizens see as urgenr (and which candidates they take as serious). By dwelling on some problems and issues and neglecting others, media prime certain memories and not others, thereby altering the standards Communication and Politics in die Age of Information 377 ■ citizens apply when they evaluate the wisdom of a policy, the virtue of a candidacy, or the performance of their government. By providing compelling arguments and decisive cues, mass communications persuade citizens to replace one opinion with another. And by trafficking in certain kinds of negative advertising, media (perhaps) discourage patticipation in politics. All in all, in these various ways, the media's power seems quite impressive—but ir is not without limits. For one rhing, framing, agenda setting, priming, and persuasion are all constrained by the anticipated reaction of the audience or, to use a differenr language, by what the American political culture finds permissible. If communications wander too far afield, they will be rejected (e.g., Gamson & Modigliani, 1987; Sherif & Hovland, 1961; Petty, Cacioppo, & Haugtvedt, 1992). Likewise, agenda setting seems limited to altering the priorities that cirizens artach to such plausible problems as unemployment or drug abuse, problems that are reasonably regarded as relevant to the national interest and widely understood as potentially affecting the lives of millions of Americans. In the same way, priming appears to be restricted to elevating the importance of reasonable standards: to privilege national security or health care above other pressing problems. Media determine which accounts, problems, and standards predominate in public discourse, but these selections are taken from a short list of predominantly serious possibilities contending for attention. And or course not everyone is influenced. The standard currency of average effects that I have adopted here overlooks the near certainty that Americans differ from one another in ways thar are relevant to their susceptibility to media influence. The one variable that has drawn the most research attention so far in this regard is engagement in politics. That some Americans are deeply interested in politics while others couldn't care less is widely appreciated, but how this difference is implicated in citizen's vulnerability to media influence is unclear {e.g., Iyengar & Kinder, 1987; Kinder & Sanders, 1990; Krosnick & Brannon, 1993; Krosnick & Kinder, 1990;, Miller & Krosnick, 2000; MacKuen, 1981, 1984; Weaver, Graber, McCombs, & Eyal, 1981).Much clearer are the results on credibility. Those who mistrust media organizarions are less subject to their influence: less raken by their frames (Druckman, 2001); less shaped by their agendas (Iyengar & Kinder, 1985); less thoroughly primed (Miller & Krosnick. 2000); and less persuaded by their arguments (Lupia &C McCubbins, 1998). This result is corroborated by scores of attitude change experiments in social psychology, which show that the credibility of the source is a vita! ingredient in persuasion (McGuire, 1969, 1985; Petty & Duane 1998). Taken all around, we now seem quite a way further along toward the "science of communication" that Klapper (I960), Hoviand (1954) and other founders of the field originally hoped for. Of course, rhere is still quire a bit left to do. For one thing, given all the evidence about how mass communications matter, we need to get smarter about how information i.s creared and disseminated. For this we need theories of campaigns and jour- MASS POLITI CAL BEHAVIOR w nalism, and we need systematic empirical work that connects the "information system" on the one hand with the judgments made and actions taken by individual citizens on the other (for promising steps in this direction, see Armstrong, Carpenter, & Hojnacki, 2000; Barrels, 1996b; Bovitz, Druckman, &i Lupia, 2000; Freedman, 1999; Gamson & Modigliani, 1987, 1989; Glaser, 1996; Hilgartner & Bosk, 1988; Jones, 5994; Page, 1996; Petrocik, 1996; Schudson, 1978). For another, we need clarification of the psychological mechanisms that mediate mass communication influence. The standard interpretation for framing, agenda setting, and priming builds directly on basic research in cognition and gives center stage to automatic processing (Cappella & )a-mieson, 1997; Price & Tewksbury, 1997). The story runs roughly this way: communications highlight some aspects of politics at the expense of others; when citizens notice such communications, relevant parts ot their memory are automatically activated; those bits and pieces of activated memory are thereby rendered accessible; and accessible constructs and information exercise disproportionate influence over the opinions and evaluations that citizens express. Under this account, mental processes are set into motion by environmental provocations and operate outside ot conscious awareness. Whether framing, agenda setting, and priming effects can actually be explained in this way is a matter of current contention. The evidence for accessibility and automatic processing is either mixed (e.g., Valentino, Hutchings, & White, 2002, find evidence that priming is mediated by accessibility, while Miller & Krosnick, 2000, do not) or indirect (e.g., Men-delberg, 2001 )■ 1 he major alternative to accessibility entails a more thoughtful, self-conscious, and effortful process. In the case of framing, for example, the argument is that by singling out certain features of an issue or event, frames imply which considerations to take into account, but the final arbiter is the citizen, who chaoses which of the available considerations are relevant and who decides how important each consideration should be (Gross, 200!; Nelson, Clawson, & Oxley, 1997; Nelson, Oxley, & Clawson, 1997; Chong, 1996). This debate about the psychological underpinnings of framing recapitulates a broader conversation in psychology over the extent to which everyday judgments, decisions, and behavior are under conscious control (see, for example, Posner & Snyder, 1975; Bargh & Ferguson, 2000). My point here is simply that a deep understanding of communication and politics in our age of information requires specification of psychological mechanism. Finally, there remains the task of explicating the downstream consequences of mass communication influence, of spelling out the political differences that framing and persuading and such actually make. The obligation here is to integrate communication results into the larger story of politics. It might be thought that this work is someone else's business, that it belongs to mainstream political Science, not to political psychology. Per- Communication and Politics in the Age of Information haps. But it seems to me that a truly successful science of mass communications cannot be just a wing of cognitive psychology, as important as psychology has been to the development of the field. The full story musr ar least be informed by the ways—partial, incomplete, intermittent and halting as they may be—that the worries and aspirations and sentiments of ordinary citizens give shape and direction to government.19 k. /Votes 1 thank the editors for their patience, and Skip Lupia, David Sears, and Nick Valentino for their valuable comments on an earlier version of the chapter. !. In this regard, framing effects in political communication resemble Tversky and Kihneman's (1981) famous experiments! demonstrations of the difference that alternative frames make to decision-making. 2. No one should imagine that every analyst who makes use of the idea of frame means just the same thing by it (Enrman, 1993). The conceptual diveisity on display in this area of scholarship is impressive and perhaps regrettable, bur probably not fatal. Consider as a conceivably comparable and instructive case the idea of "paradigm'. By Mastcrman's (1970) count, Thomas Kuhn's book The Structure of Scientific Revolution (1964) used "patadigm" in 21 distinct ways. We might wish that Kuhn had been more precise, but his idea, we should remember, changed tile way everyone think'; about science. 3. One might expect to find such demonstrations in the literature on social movements. To explain the emergence and occasional successes of social movements, researchers (by now a bit ot a movement themselves) point to various contributing conditions: precipitating grievances, material resources, political constraints and opportunities, pie-existing organizational structures (e.g., McAdam, 1982; McCarthy & Zald, 1977; Tar-row, 1994; Tilly, 1978), and, increasingly in recent years, compelling frames (e.g., Gam-son, 1992; McAdam, McCarthy, & Zald, 1996; Snow & Benford, 1992). But while there now exists considerable evidence that movement leaders spend a fair amount of time formulating frames and strategizing about theit dissemination (e.g., Branch. J988; Ellingson, 1995; Freedman, 1999; Garrow, 1978; Gerhards & Rucht, 1992), this is not the same as demonstrating that such frame work matters, and empirical studies in natural settings with this object in mind are in short supply (McAdam, 1996). 4. For a thoughtful and fascinating essay on the normative difficulties that darning makes for democratic politics, see Barrels, 2002. 5. In Lazarsfeld and Merton's (1948) original formulation, agenda setting should apply not only to issues and problems but to persons as well. A splendid venue for examining this process is provided by the contempoiary American presidential nomination system, which features multiple challengers competing in a sequence of closely bunched contests. Under this system, an unexpectedly strong showing can bring a barrage of media attention. Picked out of the crowd of ptesidential hopefuls, a lucky candidate can become, almost overnight, a serious contender, as Gary Hart showed in 1984 and as John McCain demonstrated again in 2000. What most voters learn in such circumstances, at least in the short run, has very little to do with the candidate's policy ptoposals or personal background or general philosophy. The lesson is more rudimentary: 380 M A S S POLITICAL BEHAVIOR it is that this candidate has arrived, chat tins candidate is to be taken seriously. Showered with attention, such candidates ate handed an opportunity to make their case to the public (Barrels, 1988). 6. Indeed at the time, it was seen this way by the national press (Mendelberg, 2001). 7. As do presidential debates, which might be thought of as campaigns in miniature: here too, there is little persuasion and Iocs of activation and reinforcement (Katz & Feldman, 1962; Sears & Chaffee, 1979). 8. In reaching his conclusion of minimal effects, Kkpper also drew quite heavily on the results from a scries of studies carried out by Carl Hovland and his colleagues for rhe Information and Education Division of the "war Department during World War 11, presented in Experiments on Mass Communication (1949). Holland's team carried out randomized experiments on large and representative samples of American recruits to assess the effects of War Department propaganda films. They discovered that the films were generally quite successful in conveying salient iacts bur almost entirely ineffective in changing attitudes (Hovland, Lumsdaine, Si Sheffield, 1949, p. 65). This result— information gain without corresponding attitude change (see also Robinson, 1976)— would seem to be a sign of resist/wee, discussed later in this section. 9. Nor to mention rhe candidates themselves, also roughly equally matched and battle tested. Were nominees selected by lottery, we no doubt would see more evidence of campaign persuasion, as rhe candidates staggered horn one disaster to rhe next. 10. Without deliberately choosing to do so, people typically find themselves in informational environments that are on balance reinforcing, more so in the selection of friends and mares than in rhe selection of communication sources (Murr. cc" Martin, 2001). Sears and Frcednian (1967) refer to rhis as "de facto selectivity." 11. Conceptions of partisan identification differ over this point of immunity to the campaign (see, for instance, Fiorina, 1981; Jennings & Markus, 1984; Markus &; Converse. 1979), 12. The best predictor ol exposure to new information is a peison's preexisting fund of information (Delli Carpini & Kceter, 1996; Gilens, 2001; Graber, 1984; Price & Zaller, 1993). This gives rise ro a kind oi virtuous evele, where information breeds interest and interesr breeds an appetite for more information and where the costs of assimilating new information decline steeply among those already in the know. This means that rhe audience for political communications consists disproportionately of the relatively well informed, and this fact has a crisscrossing implicarion for persuasion. On the one hand the audience that actually shows up for any media campaign is comparatively well equipped to comprehend, absorb, and retain rhe message; but on rhe other the audience is also comparatively well equipped to defend against and ultimately reject messages that challenge preexisting views. 13. In some ways this example is reminiscent of the claim of The American Voter (Campbell et al., 1960) rhar citizens much preler simple constructions of group interest to ideological abstraction when it comes to ev.iluaring parties and candidates ("ideology hy proxy1'). 14. Research on cue-taking is flourishing, as it should, but it is unlikely in rhe end, to provide a complcrely satisfying solution to the problem of ignorance, for reasons 1 spell our elsewhere (Kinder 1998). 15- Whether rhe public as a whole really deserves the attribution of rationality that some analysts wish to bestow on it is therefore open ro serious question (Converse, 1975, 1990; Kinder, 1998). Communication and Politics in the Age of Information 381 16. Zailer (1996) claims that "rhe flow of political communication in the United States on many (and perhaps most) important matters is relatively srabie over rime— locked into fixed parterns that reflect underlying divisions of power, partisanship, and social inertia" (p, 19). 17. On the possibility that voters might-also be rurned off by news coverage that emphasizes rhe strategic side of campaigns at the expense of issues, see Capella and jamieson (1997) and Valentino, Beckmann, and Buhr (2001). 18. Parr of the problem is that engagement is likely to have offsetting consequences at different stages of the influence process. Part of the problem is that different aspects of engagement have distinctive or even opposite implications for influence (Krosnick & Brannon, 1993). And, finally, whether engagement in politics makes people more or less susceptible to media influence may also depend on the magnitude ol the "story." The terrorist attack on New York and Washington on September 11, 2001, was such big news that it probably washed away much of rhe normal variation in exposure. More often, news commands far less attencion, and in this routine, world, many less-engaged citizens may miss the message altogether. Bold claims notwithstanding (Miller & Krosnick, 2000), it seems premature at this stage to say anything conclusive about the relationship between engagement and influence. 19. As in, for example. Key (1961); Barrels (1991); and Stimson, MjcKuen, and Erickson (1995). k. References Abelson, R. P. (1959). Modes of resolution in belief dilemmas, Journal of Corf in AV--olution, 3, 343-352. Abelson, R. P. (1968). Psychological implication, in R. P Abeíson, E. Aronson. W.J, McGuire, T. M. Newcomb, M.J. Rosenberg, & R, H. Tannenbaum (FJs.i, Theories of cognitive consistency: A sourcebook (pp. 112-139). Chicago: Rami McNally. AJthaus, S. L. (1998). Information effects in collective preferences. American Political Science Review, 92. 545-558. Alvarez, R. M„ & Naglcr, J. (1995). Economics, issues and the Perot candidacy: Voter choice in the 1992 presidential election. American Journal of Political Scieiue. 39, 714-744. Anderson, C. A., Lindsay, J..i.. &c Bushman, B. (1999). Research in the psychological laboratory: Truth or triviality? Current Directions in Psychological Science, 8. 3-9. Anderson, J. R. (1983). The Architecture of cognition. Cambridge, MA: Harvard Universit)' Press. Ansolabehere, S. D., & Iyengar, S. (1995). Going negative: How political advertising shrinks and polarizes the electorate. New York: Flee Press. Ansolabehere, S. D., Iyengar, S,, Simon, A., & Valentino, N. (1994). Does attack advertising demobilize the electorate? American Political Science Review, 88, 829-838. Ansolabehere, S. D., Iyengar, S., & Simon, A. (19991. Replicating experimenrs using aggregate and survey data: The case of negative advertising and turnout. American Political Science Review, 93, 901 -909.