Communication and Opinion

Embed Size (px)

Citation preview

  • 7/25/2019 Communication and Opinion

    1/32

    Annu. Rev. Polit. Sci. 1998. 1:16797

    Copyright 1998 by Annual Reviews Inc. All rights reserved

    COMMUNICATION AND OPINIONDonald R. KinderCenter for Political Studies, University of Michigan, Ann Arbor, Michigan 48109-1248;e-mail: [email protected]

    KEY WORDS: mass media, agenda setting, priming, framing, campaigns

    ABSTRACT

    This chapter reviews the vast and sprawling literature that seeeks to illumi-nate and explain the effects of mass communication on American publicopinion. Klappers famous verdict of minimal effects, delivered some 40years ago on this subject, was faithful to the evidence available to him at thetime, but now seems quite mistaken. With sharp improvements in design,measurement, and analysis, and with keener understanding of human infor-mation processing, minimal effects have given way to an entire family of realeffects: agenda-setting, priming, framing, and even, looking in the right

    places, persuasion.

    INTRODUCTIONMy topic is the interplay between mass communication and public opinion incontemporary American society. This terrain was scouted famously by WalterLippmann more than 70 years ago, first in Public Opinion(1922) and then inThe Phantom Public (1925). Persuasively and poetically, Lippmann drovehome the point that in societies the size and complexity of the United States,ordinary citizens must come to depend on others for their news about nationaland world affairs:

    Each of us lives and works on a small part of the earths surface, moves in asmall circle, and of these acquaintances knows only a few intimately. Of any

    public event that has wide effects we see at best only a phase and an aspect....Inevitably our opinions cover a bigger space, a longer reach of time, a greaternumber of things, than we can directly observe. They have, therefore, to be

    pieced together out of what others have reported and what we can imagine.(Lippmann 1922, p. 53)

    To clarify the relationship between what others have reported and what wecan imagine is my primary purpose here.

    Since Lippmanns time, advances in education have dramatically lowered

    cognitive barriers to political enlightenment, and thanks to the invention and167

  • 7/25/2019 Communication and Opinion

    2/32

    rapid diffusion of electronic media, information about politics has becomevastly more accessible. Perhaps not accidentally, these developments have beenaccompanied by a recent inclination in democratic theorizing to place aspects ofcommunicationdeliberation and discussionat the center of politics (e.g.

    Barber 1984; Dahl 1989; Fishkin 1991, 1995; Guttmann & Thompson 1996;Habermas 1989; Mansbridge 1980; Sunstein 1993). If democracy is not entirelygovernment by discussion, it is certainly government by discussion in part.

    With normative concerns lurking in the background, this chapter aspires toorganize a sprawling literature and to suggest productive lines for future re-search. The chapter turns on the notion that the interplay between mass commu-nication and public opinion can be apprehended by imagining that citizens un-dertake three generic tasks: They make sense of the carnival of personalities,issues, and events that, moment to moment, comprise the political realm; theydecide what is important in politics and what is not; and they evaluate politicalalternativescandidates, policy proposals, and morethat are placed beforethem. I treat each of the three separately because it is analytically useful to doso, but without forgetting that in practice they are often as not thoroughly inter-mingled.

    I could refer to these three tasks, alternatively, as memory, attention,and choice. This more explicitly psychological terminology is a sign of thechapters tilt towards the microfoundations of mass communication and opin-ion. No doubt others, given the same assignment, would place the emphasiselsewhere. (For reviews that slice the field differently, see Converse 1996,

    Kinder 1998, and Sniderman 1993.) Some selection is definitely necessary. Inthe preamble to an essay with roughly the same purpose as mine, Lazarsfeld &Merton (1948) complained of their ungrateful task. Why? Because they foundcertified knowledge concerning the political role of mass media impres-sively slight. Fifty years later, the task is perhaps equally ungrateful, but foranother reason entirely. The problem now is avoiding being swept away by theavalanche of research; the trick is keeping on ones feet while attempting to de-termine which parts of the massive flow should be singled out and certified. Ifthe job is impossible, I find consolation in the fact that those who follow will

    be presented with the same opportunity and will no doubt uncover many thingsI missed.

    MAKING SENSE OF POLITICS

    How do citizens go about making sense of the swarming confusion of prob-lems (Lippmann 1925, p. 24) that constitute public life, and what role mightcommunication play in this process? In some ways the single best source on thisquestion remains LippmannsPublic Opinion, brimming with insights into the

    predicament of the citizen in the modern world. Lippmann portrayed citizens as

    168 KINDER

  • 7/25/2019 Communication and Opinion

    3/32

    preoccupied with the trials and tribulations of private life, concerned in publicaffairs, but immersed in private ones, and so likely to think twice beforetaking the trouble to become informed about the machinations of government.

    InAn Economic Theory of Democracy, Downs (1957) turned Lippmanns

    intuitions into a sustained and powerful argument. Downs pointed out thatthere are costs attached to the acquisition and analysis of political informa-

    tionmeasured in time, energy, and opportunityand that rational voters will

    pay such costs only insofar as the information promises a return. But in a large

    society, one persons vote is lost in a sea of votes, and so the instrumental

    benefits from an enlightened vote are infinitesimal. The result, according to

    Downs, is rational ignorance.Rational or not, the depths of ignorance demonstrated by modern mass pub-

    lics can be breathtaking. For example, after more than 40 years of continuous

    Democratic control of the US House of Representatives, only 59% of Americans

    knew in the fall of 1992 that the Democratic Party held the majority of seats in

    the House. In late 1995, more than twice as many Americans held the stupen-

    dously incorrect belief that the federal government spends more money on for-

    eign aid than on Medicare (and on and on; see Delli Carpini & Keeter 1996).1

    Conceding that the average American, most of the time, cares little about

    politics and knows less is not the same as concluding that politics is an incom-

    prehensible mess, that the grand events, issues, and personalities of the time re-

    main always, in Jamess (1890, p. 488) famous formulation, one great,

    blooming, buzzing confusion. Nevertheless, it would be imprudent to forget

    that the fascination political scientists find in politics is seldom shared by the

    general public. Keeping our expectations properly modest, how do citizens

    figure out whats going on?

    Stereotypes, Frames, and Analogies

    At the center of Lippmanns answer to the puzzle of political understandinglies the notion of stereotype. In Lippmanns account, stereotypes are mentalmaps that allow us to navigate successfully the world of politics, a world al-together too big, too complex, and too fleeting for direct acquaintance (Lipp-mann 1922, p. 16).

    This turns out to be a good answer and a prescient one. Lippmann wrote

    Public Opinion before the cognitive revolution that would overtake psychol-

    COMMUNICATION AND OPINION 169

    1 1Things are worse than they seem. Even the most methodologically scrupulous surveys (e.g. theNational Election Study, the General Social Survey) are able to complete interviews with onlyabout 75% of the originally targeted sample. Those who refuse to be questioned, or who are nevercontacted in the first place, and so play no role in estimates of the publics knowledge, are much lesslikely to take an interest in politics than the public as a whole. This means Americans are even lesswell informed than the figures just cited suggestby roughly 25%, in corrections suggested by

    Brehm (1993).

  • 7/25/2019 Communication and Opinion

    4/32

    ogy had appeared on the horizon, yet his approach is in certain respects indis-tinguishable from the central thrust of contemporary cognitive psychology. Inthis line of work, stereotypes are regarded as categories and stereotyping asone ordinary manifestation of the ubiquitous human process of categorization

    (Fiske 1998). Stereotypes are inevitable, an intrinsic and essential aspect ofcognition. Without such categories,

    mental life would be chaotic. If we perceived each entity as unique, wewould be overwhelmed by the sheer diversity of what we experience and un-able to remember more than a minute fraction of what we encounter. And ifeach individual entity needed a distinct name, our language would be stag-geringly complex, and communication virtually impossible. Fortunately,though, we do not perceive, remember, and talk about each object and eventas unique,but rather as an instance of a class or object that we already knowsomething about(Smith & Medin 1981, p. 1; italics added).

    Under this account, citizens understand new developments in politicsemerging candidates, novel proposals, breaking eventsby recognizing suchdevelopments as instances of something they already understand (Graber1984).

    Research on stereotypes in psychology has concentrated in recent years onissues of mental representation, on how social groups are represented in themind. Stereotypes are thought to exist as a network of semantic elements, in-cluding a category label (e.g. blacks), a set of features that represent presump-tive beliefs regarding the most typical characteristics of category members

    (e.g. athletic, dull, violent), and exemplars that illustrate the category (e.g.Jesse Jackson, Michael Jordan). These elements are organized together in atangled hierarchy. Features and exemplars vary in their strength of associationto the category; features are more or less prototypical, just as exemplars aremore or less stereotypical. These can be thought of as vertical links, or associa-tions, within the stereotype. There are also horizontal associations, which con-nect related concepts at a similar level of generality (for example, the categorylabels black and Jew). The features that contribute to a stereotype are widelyassumed to be traits (as in the example above), although they might also in-

    clude visual images, physical attributes, and beliefs and attitudes (e.g. Brewer& Campbell 1976, Fiske & Neuberg 1990, Smith & Zarate 1990, Stangor &Lange 1993).

    Stereotypes, like other knowledge structures, are embedded in a largerinformation processing system. That systems architecture is composed ofa series of independent memories. Most important for present purposes isthe distinction between a short-term or working memory (the capacity ofwhich is severely limited, both as to how much information can be re-

    presented and how much can be successfully transferred), and a long-term

    memory (where stereotypes and other forms of knowledge structures reside),

    170 KINDER

  • 7/25/2019 Communication and Opinion

    5/32

    which can accommodate huge amounts of information stored in a variety offormats (Anderson 1983, Newell 1990, Newell & Simon 1972). This systemis characterized by finite mental energy and limited representational capac-ity. These constraints encourage category-based understanding and imply sat-

    isficing as the operative standard for performance; that is, a search will beterminated when an acceptable, but typically not optimal, understanding hasbeen achieved. A final point is that knowledge structures in long-term mem-ory, like stereotypes, become available through spreading activation. Activa-tion spreads from a currently active location in a knowledge structure (e.g.the category label) to other nearby locations in the tangled hierarchy; thespread is rapid, perhaps automatic (not under conscious attentional control),increases with strength of association, and diminishes sharply with distancefrom the source of activation (Anderson 1983, Collins & Loftus 1975).

    The power of stereotypes in the interpersonal domain is well established.We know that visible features such as skin color or hair texture that are associ-ated distinctively with membership in social groupings are attended to sponta-neously and are used as a basis for initial categorization; that this primitivecategorization rapidly activates personality traits that are associated with thecategory in long-term memory; and that such activated traits influence both

    judgment and behavior in a variety of ways (Brewer 1988, Stangor & Lange1994).

    Stereotypes are but one kind of knowledge structure, and if their mental rep-resentation and dynamic operation are well understood in the domain of inter-

    personal relations, there remains the task of identifying those knowledge struc-tures that dominate political understanding. Schank & Abelson (1977, p. 10)

    put this point forcefully:

    There is a very long theoretical stride, however, from the idea that highlystructured knowledge dominates the understanding process, to the specifica-tion of the details of the most appropriate structures. It does not take one veryfar to say that schemas are important: one must know the content of the sche-mas.

    One possibility is ideology, the proposition that commitment to an ab-stract ideological principle (e.g. liberalism, social Darwinism) would pro-vide citizens with a deep and rich understanding of politics. Should citizensreason in this way, then new political events have more meaning, retentionof political information from the past is more adequate, and political behaviorincreasingly approximates that of sophisticated rational models, which as-sume relatively full information (Converse 1964, p. 391). Perhaps, but nevermind: Only a tiny percentage of citizens appear to be in possession of the ab-stract master principles that could do this work (Converse 1964; Kinder 1983,

    1998).

    COMMUNICATION AND OPINION 171

  • 7/25/2019 Communication and Opinion

    6/32

    Another possibility is what Rahn (1993) calls partisan stereotypes, thepackets of knowledge that citizens develop about prototypic Democratic andRepublican officials, including their likely policy positions, group affinities,and political priorities. The rationale for understanding based on partisan

    stereotypes was well expressed by Stokes (1966, pp. 12627):

    To the average person the affairs of government are remote and complex, andyet the average citizen is asked periodically to formulate opinions aboutthese affairs.... In this dilemma, having the party symbol stamped on certaincandidates, certain issue positions, certain interpretations of political realityis of great psychological convenience.

    Living in a stable two-party system where the party cue is a conspicuousand recurrent feature of political discourse, Americans do indeed seem to relyon partisan stereotypes to fill out their impressions of candidates, to makesense of the choices they confront (e.g. Conover & Feldman 1989, Lodge &Hamill 1986, Hamill et al 1985, Rahn 1993).

    As partisan stereotypes are to political candidates, so frames may be to is-sues. The issues taken up by government are always complex, always subjectto alternative interpretations. What exactly is affirmative action? Is it reversediscrimination? Is it compensation for the injustices of the past? Is it an unde-served handout? Which of these interpretations prevails affects how citizenswill understand affirmative action, and, in the end, what their opinions on af-firmative action turn out to be.

    More precisely, how citizens understand an issuewhich features of it arecentral and which are peripheralis reflected in how the issue is framed. Theanalysis of frames and public opinion is especially indebted to the work ofGamson (1992, Gamson & Modigliani 1987).2 Gamson presumes that the de-fining of issues is central to politics. Every issue is contested; advocates of all

    persuasions are always attempting to define the issue their way, employingmetaphors, catch-phrases, and other condensing symbols that frame the issuein a particular fashion (Gamson & Modigliani 1987, p. 143). Frames seek tocapture the essence of an issue. They define what the problem is and how to

    think about it; often they suggest what, if anything, should be done to remedyit.

    Public opinion does seem to depend in a systematic and intelligible way onhow issues are framed (e.g. Bobo & Kluegel 1993, Chong 1993, Iyengar 1991,

    Nelson & Kinder 1996). More important, framing effects show up reliably in

    172 KINDER

    2 2On other uses of frame, see for example Goffmans (1974) account of how individualsconstruct meaning out of their social experience; Tversky & Kahnemans (1981) celebratedexperimental demonstrations that, contrary to expected utility theory, decision outcomes can besystematically and decisively influenced by altering how the options are framed; and social

    movement frames, which justify, dignify, and animate collective action (Tarrow 1994, p. 22).

  • 7/25/2019 Communication and Opinion

    7/32

    studies that deliberately mimic elite debates and journalistic habits (Iyengar &Kinder 1987; Kinder & Sanders 1990, 1996; Mendelberg 1997; Popkin 1991;Price 1989; Stoker 1997). These demonstrations provide empirical backing to

    Schattschneiders bold contention that [p]olitical conflict is not like an inter-

    collegiate debate in which the opponents agree in advance on a definition ofthe issues. As a matter of fact,the definition of the alternatives is the supreme

    instrument of power; the antagonists can rarely agree on what the issues are be-cause power is involved in the definition. He who determines what politics isabout runs the country (1960, p. 68; italics in original).

    Although the concept of frame provides a promising conceptual tool for ex-

    amining the process by which issues are defined in politics, a number of puz-zles remain. Do frames exist both in political discourse and in cognitive proc-essing? How do citizens choose frames and how do they make them their own

    (Capella & Jamieson 1997, Chong 1996)? And finally, notice that the empiri-cal literature on frames skips over the focal concern of this section: How docitizens come to understand public issues? Invoking alternative frames pro-

    duces alterations in opinion, but it is less clear whether alternative frames leadto systematic alterations in understanding, and if so, what mechanisms mightaccount for such alterations.

    Understanding is the focus of the final possibility taken up here: that citi-

    zens come to understand current events through what Holyoak & Thagard(1995) call analogic reasoning. No doubt for many us, most of the time, thenews of the day flies by; events burst onto the national and international stage

    and expire just as quickly, without citizens feeling any interest or obligation tosort things out. But some developments have more staying power; thankslargely to the focusing function of the news (see below), they remain for some

    time in the forefront of public consciousness. What then?Then, by Holyoak & Thagards account (1995), analogic reasoning comes

    into play. Confronted with a novel and unfamiliar event (the target analog),

    people do their best to understand it in terms of events they know well already

    (the source analog), even when connecting the two requires a mental leap

    (Holyoak & Thagard 1995, p. 2). The burden of Holyoak & Thagards work is

    to show that such mental leaps are not haphazard, that they follow a predictablelogic. When faced with an unfamiliar event, people first select a source analog

    by retrieving information about it from long-term memory (selection); next

    they map the source to the target, considering correspondences of specific at-

    tributes as well as abstract relations, thereby generating inferences about the

    target (mapping); and finally they evaluate and modify these inferences to take

    into account what they happen to know of the target itself (evaluation). Each

    stage in this process is subject to pressures, e.g. parallelism between the roles

    of major protagonists in target and source. Empirical support for this general

    picture of analogic reasoning comes from diverse sources including computer

    COMMUNICATION AND OPINION 173

  • 7/25/2019 Communication and Opinion

    8/32

    simulations, surveys, and experiments (Holyoak & Thagard 1995, Spellman &

    Holyoak 1992, Gentner 1983; for additional arguments and evidence on the

    political importance of reasoning by analogy, see Gilovich 1981, Jervis 1976,

    Neustadt & May 1986).

    Even when events do not capture the publics attention, citizens may findthemselves encouraged to think about new events in analogic ways. That is,

    when coupled with mass communication, analogy can be used as an instru-

    ment of propaganda (Holyoak & Thagard 1995, p. 100). A case in point is

    President Bushs generally effective argument in late 1990 early 1991 that the

    brewing conflict in the Persian Gulf should be understood in terms provided by

    World War II: Iraq was Germany, Saddam Hussein was Hitler, and (therefore)

    appeasement would be disastrous (Spellman & Holyoak 1992).

    UncertaintyIt is hard to imagine a purer case of judgment under uncertainty than what is re-quired of citizens in modern, complex societies. On the one hand, publicopinion deals with indirect, unseen, and puzzling facts, and there is nothing ob-vious about them (Lippmann 1922, p. 17). And on the other, as noted above,

    judgments are often based on fragmentary and partial information. As a result,citizens must be at least somewhat uncertain about practically everything.

    Indeed they are. Recent investigations show that however it is measured,

    uncertainty is rampant. Even at the tail end of presidential campaigns, for ex-

    ample, large numbers of voters remain in the dark about the positions taken bythe candidates on the major issues of day (Bartels 1986; see also Alvarez &

    Franklin 1994, Alvarez 1997, Bartels 1988). Such uncertainty is not only per-

    vasive but consequential. Voters appear to be risk averse: They penalize candi-

    dates for being insufficiently clear about their positions (Alvarez 1997, Bartels

    1986).The antecedents of uncertainty are partly to be found in the resources and

    outlooks that citizens bring to politics. Education, cognitive skills, informa-

    tion, attention, and interest all tend to diminish uncertainty (Alvarez 1997,

    Bartels 1988). Uncertainty also varies in ways that directly implicate com-munication processes. Voters are more uncertain about challengers than in-

    cumbents (especially in House elections where challengers are typically un-

    derfinanced); they are more uncertain about minor candidates than major ones;

    and they are more uncertain at the beginning of a campaign than at the

    endparticularly if they have been paying attention to news about the cam-

    paign (Alvarez 1997; Bartels 1986, 1988; Brady & Johnston 1987; Patterson

    1980).Campaigns reduce voter uncertaintyas indeed they are supposed tobut

    equally impressive in these results is how modest the effect of campaigns

    174 KINDER

  • 7/25/2019 Communication and Opinion

    9/32

    seems to be, how much uncertainty remains when they are through. How true

    this is depends partly on the content of the campaign itself. Thus voters uncer-

    tainty about the ideological position of their incumbent senator declines when

    the incumbent runs a campaign emphasizing issues (as against personal quali-

    fications or service to the state), but increases when the opponent attacks theincumbents positions (Franklin 1991). Likewise, in presidential campaigns,

    uncertainty diminishes when candidates stake out distinctive positions on cen-

    tral policy questions (Page & Brody 1972, Page 1978). In short, the mere pres-

    ence of a campaign is insufficient; what candidates choose to say determines

    whether voters go the polls relatively clear-eyed or more confused than ever.

    Franklin takes these results to mean the dominance of politics over institu-

    tions (1991, p. 1210).

    Shortcuts to KnowledgeThat so few Americans know how Boris Yeltsin makes a living or who repre-sents them in Washington is not exactly inspiring, but perhaps not as unsettlingfor democratic sensibilities as it first appears. Granted that encyclopedicknowledge is out of reachand perhaps even irrationalthe public may nev-ertheless muddle through. How? By relying on a variety of sensible and mostlyadaptive shortcuts (Popkin 1991, Sniderman et al 1991).

    All such shortcuts seem to involve taking cues from well-informed or at

    least better-informed sources. The general claim is that although average citi-

    zens do not patrol government looking for problems, they very well may

    pay attention to people who do (Popkin 1993, p. 19). A good illustration is

    provided by Lupias (1994) investigation of California voters contemplating

    several complicated proposals for reform of their states automobile insurance

    industry. Lupia argued that although voters may be impressively ignorant of

    the content of referenda put before them, they may still be able to choose

    wisely if they happen to know the positions taken by well-informed sources

    with clear reputations. In an ingenious analysis, Lupia showed that voters who

    knew little about the details of the various proposals nevertheless made sensi-ble choices, ones that were indistinguishable from those made by well-

    informed voters. The vital piece of information here was the position taken by

    special interest groups. When Californians knew that the insurance industry it-

    self or associations representing trial lawyers supported a proposal, they knew

    enough: They voted against it. So by taking cues from expert sourceselites

    (Brody 1991, 1994), opinion leaders (Katz & Lazarsfeld 1955), interest groups

    (Lupia 1994)citizens may be knowledgeable in their reasoning about po-

    litical choices without necessarily possessing a large body of knowledge about

    politics (Sniderman et al 1991, p. 19).

    COMMUNICATION AND OPINION 175

  • 7/25/2019 Communication and Opinion

    10/32

    Do shortcuts solve the democratic problem of miserably informed citizens?

    No. First, although citizens may be willing to rely on the views expressed by

    experts, we should not presume that experts always get things right, or that citi-

    zens always know what experts think. In the case of automobile insurance re-

    form in California, for example, the insurance industry and the trial lawyersunderstood perfectly well their reputational problem, so that in the advertise-

    ments they purchased, they hid their identity as much as the law would allow

    (Lupia 1994).Second, showing that poorly informed voters can mimic the decisions of

    their better-informed counterparts is not the same as showing that they always

    or often do so. On the contrary, Bartels (1996a) showed that in recent Ameri-

    can presidential elections, poorly informed voters have made decisions in

    ways that differ systematically from those of well-informed voters. Moreover,

    when aggregated, these differences generally do not cancel out: In four of the

    six elections Bartels examined, deviations from hypothetical fully-informed

    election outcomes are both statistically significant and sizable, ranging from

    about three percentage points to nearly six.Third, shortcuts have little to say to the glaring fact of inequalities in infor-

    mation, that some people can recall so little and others so much (Converse

    1990). It would be surprising if inequalities of information as large as this

    turned out to be politically innocuous. They are not. The well informed are

    more likely to express opinions, to use ideological terminology correctly, to

    possess stable opinions, to make use of facts in political discussions, to take an

    active part in politics, and to pick up new information easily and retain it read-

    ily (e.g. Delli Carpini & Keeter 1996, Price & Zaller 1993, Zaller 1992).Finally, we should keep in mind that when we take shortcuts, sometimes we

    end up in the right place and sometimes we get lost. The problem here is not

    just that citizens dont know enough, it is that they know things, or think they

    know things, that are factually incorrect (e.g. that a huge fraction of the na-

    tional treasury is being squandered on foreign aid). In the end, shortcuts to

    knowledge are unlikely to be effective substitutes for the real thing.

    DECIDING WHATS IMPORTANT

    Among the most important decisions in a democratic society are thosethat determine which issues become part of public discussion. In Schatt-schneiders famous and crisp formulation, Some issues are organized into

    politics while others are organized out (1960, p. 71). Put somewhat differ-ently, power is implicated at least as much in the determination of which is-sues are to be decided as in what actions are ultimately decided on (Bachrach

    & Baratz 1970).

    176 KINDER

  • 7/25/2019 Communication and Opinion

    11/32

    My purpose here is to explicate the process of agenda setting, as it is com-

    monly called, at an early stage: agenda setting among the general public, the

    misty swamp (Riker 1993a, p. 2) of everyday thinking and talking about

    politics. How do average citizens decide that crime, to take one example, is

    overwhelming their society? Deciding may be a presumptuous way to putthe question. Somehow citizens come into possession of ideas about what is,

    and what is not, important; which problems are, and which are not, proper sub-

    jects for government action; even what is, and what is not, political. How does

    this happen?Lazarsfeld & Merton (1948) were the first to suggest that citizens beliefs

    about what (and who) is important come from mass media. While generally

    quite skeptical about the power of mass communications in a free society (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 & Merton nevertheless suggested three ways

    that mass media might have measurable effects on society. The first of these is

    agenda setting: The mass media conferstatus on public issues, persons, or-

    ganizations, and social movements (p. 101). The mere fact of recognition, of

    being singled out for attention, is evidently enough to bestow prestige and

    authority.Lazarsfeld & Merton could cite no evidence on behalf of agenda setting,

    and their enunciation of the potential importance of agenda setting had no ap-

    parent impact on empirical research. Klappers (1960) influential review of

    findings on the effects of mass communication was nothing if not thorough,

    yet it managed just two pages on agenda setting, most of that lifted directly

    from Lazarsfeld & Merton (1948) and the remainder stained with such dis-

    heartening phrases as some writers believe (1960, p. 104).But shortly afterward came Bernard Cohens detailed analysis of the role of

    the press in US foreign policy. Based on intensive interviews with leading

    journalists and government officials, Cohen (1963, p. 13) concluded that

    the press is significantly more than a purveyor of information and opinion. It

    may not be successful much of the time in telling people what to think, but itis stunningly successful in telling its readers what to thinkabout. ... The edi-tor may believe he is only printing the things that people want to read, but heis thereby putting a claim on their attention, powerfully determining whatthey will be thinking about, and talking about, until the next wave laps theirshore.

    Cohen had in mind foreign policy, the attentive public, the prestige press,

    and policy experts, so his strong conclusion doesnt speak directly to our ques-

    tion: When it comes to politics, are mass media in fact stunningly successful

    in determining what ordinary Americans think about?

    COMMUNICATION AND OPINION 177

  • 7/25/2019 Communication and Opinion

    12/32

    McCombs & Shaw (1972) found that they were. Toward the end of the 1968presidential campaign, McCombs & Shaw interviewed a small sample of un-

    committed voters living in and around Chapel Hill, North Carolina. (Uncom-mitted voters were selected because they were presumed most susceptible to

    the influence of the news medias coverage of the campaign.) Voters wereasked, among other things, to name the problems that government should mosttry to solve. While voters were being interviewed, McCombs & Shaw were

    poring over the Durham Morning Herald, Newsweek, the evening nationalnews broadcasts of CBS and NBC, and more, coding the attention each was

    paying to various national problems. McCombs & Shaw found an almost per-

    fect correlation between the two: Those problems given great prominence inthe news were the same problems voters considered the countrys most seri-ous. They concluded that [i]n choosing and displaying news, editors, news-

    room staff, and broadcasters play an important part in shaping political reality(p. 176).

    McCombs & Shaws wildly successful demonstration, based on a contem-poraneous comparison between the aggregated priorities of voters and the ag-

    gregated priorities of news organizations, inspired numerous replications and,unhappily, considerable confusion. The best contemporaneous analysis foundmodest and mysteriously context-dependent support for agenda setting (Er-

    bring et al 1980): Newspapers emphasis on problems influenced readersjudgments about them only for some problems, only for those readers sensi-tized to the problems by relevant personal experiences, only for those who sel-

    dom discussed politics with friends, and only for those who were also regularviewers of network news programs. Stunningly successful overstates thisevidence considerably.

    But perhaps agenda setting should be investigated dynamically over time,

    as problems emerge, move to the center of the national stage, and then gradu-ally drift back into the wings. Funkhouser (1973) thought so, and was re-warded with strong positive results: close correspondence between the amount

    and timing of attention paid to various problems in the national press duringthe 1960s and the importance accorded those problems by the American public

    over the same period. Public opinion seemed to follow, not lead, the agenda setby the press.

    Funkhouser was on to something. His basic finding has since been repli-cated in more elaborate and demanding time-series analyses. In estimating the

    effect of news agendas on the public, these analyses control for the possibility

    that news organizations are responding to the publics priorities (and not ex-

    clusively the other way around); take into account the independent effects ofreal world conditions (e.g. changes in prices or interest rates); and consider the

    possibility that presidents can alter what the public takes to be important

    through major addresses to the nation. With all these safeguards in place,

    178 KINDER

  • 7/25/2019 Communication and Opinion

    13/32

    strong support for agenda setting emerges. For a wide variety of problems, theAmerican publics concern for political problems closely and rapidly trackschanges over time in the attention paid them by national media (MacKuen1981, 1984; Iyengar & Kinder 1987; McCombs & Zhu 1995).

    These results are neatly complemented by the findings of a series of experi-ments by Iyengar & Kinder (1987). These studies in effect convert the varia-tion in news coverage that occurs naturally over time to contemporaneousvariation across experimental conditions. People shown network broadcastsunobtrusively edited to highlight a particular problem ended up assigninggreater importance to that problemgreater importance than they themselvesdid before the experiment began and greater importance than did people ran-domly assigned to control conditions emphasizing different problems. Theseeffects were apparent immediately after the conclusions of the broadcasts aswell as one week later, were enhanced when the edited stories led off the

    broadcasts, and emerged for a wide array of problems: national defense, pollu-tion, unemployment, inflation, arms control, civil rights, energy, the social se-curity system, education, and drugs.

    News appears to be an important influence on the publics agenda, but it isnot the only one: Americans exposed to the same stream of stories will differ intheir views about what is important for the country as a whole. Such differ-ences can be traced partly to self-interest, partly to political principles, and per-haps most of all to social group membership (Boninger et al 1995, Iyengar &Kinder 1987, Kinder 1998). Nor does news have its way uniformly: Americans

    exposed to the same stream of stories will draw different lessons regardingwhich problems are serious and which are not, with the clearest example againsupplied by social groups. Group members appear especially attuned to newsthat is relevant to their own groups predicament. Coverage of civil rights ismore influential among blacks than whites; stories on unemployment are morecompelling for the unemployed than the employed; coverage of the impending

    bankruptcy of social security makes a deeper impression on the elderly than onthe young (Erbring et al 1980, Iyengar & Kinder 1987). In sum, news appearsto be most effective in telling people what to think about when the stories it

    presents reinforce and ratify the experiences of ordinary life.In Lazarsfeld & Mertons (1948) original formulation, agenda setting

    should apply not only to issues and problems but to persons.3 A terrific venue

    for examining this process is the contemporary American presidential nomina-

    tion system. Because news coverage of the primary season is preoccupied with

    winning and losing (Patterson 1980, 1993; Robinson & Sheehan 1983), even

    COMMUNICATION AND OPINION 179

    3 3Lazarsfeld & Merton also include political organizations and social movements among thoseentities that might be singled out for media attention. Surely theyre right, although there is

    lamentably little research on agenda setting of this variety (Gitlin 1980, McAdam 1996).

  • 7/25/2019 Communication and Opinion

    14/32

    casual observers of the campaign pick up information about how the variouscandidates are doing, about who has just won and whose chances are fading.Voters assessments of which candidates are serious and which arenotwhat Bartels (1988) calls expectationsrespond to such reporting, espe-

    cially for those who are following the campaign carefully (Bartels 1988; seealso Ansolabehere & Iyengar 1994, Brady & Johnston 1987, Popkin 1991).These results are important because, for virtually all voters early in the cam-

    paign and for poorly informed voters throughout the campaign, serious can-didates turn out to be winning candidates. That is, under conditions of high un-certainty, voters are often swayed by signs of a candidates electoral viability(Bartels 1988). A spectacular example is Gary Harts meteoric rise in the earlyspring of 1984. Following his upset victory in the New Hampshire primary,Harts public profile soared. Showered with attention, he was suddenly trans-formed from also-ran to front-runner.

    In psychological terms, agenda setting appears to be one manifestation ofthe availability heuristic at work. Heuristics are strategies of simplificationthat reduce the complexity of judgment tasks to make them tractable for thekind of mind that people happen to have (Kahneman et al 1982, p. xii). Theavailability heuristic comes into play when people assess the frequency of ageneral class of events or judge the plausibility of a particular development. Ina series of clever experimental demonstrations, Tversky & Kahneman (1973)showed that people make such judgments according to the ease with which in-stances or occurrences can be brought to mind. By implication, as editors and

    reporters bring some aspects of politics into prominence while keeping othersin darkness, they thereby influence which bits of political memory can mosteasily be brought to mind.

    The evidence on agenda setting strongly vindicates Lippmanns originalconjecture. For their map of politics, for their picture of the mystery offthere, citizens depend mightily, as Lippmann understood, on journalism.

    EVALUATING ALTERNATIVES

    Citizens react to the political alternatives that are put before them. They decidewhether to support or reject a candidate; they find policy proposals sensible orterrible; they take the consequences of government action to be successful ordisastrous; they judge their presidents performance in office to be inspiring ordeplorable; and so on. Democratic theorists disagree over whether this is allthat citizens should do (e.g. Barber 1984, Dahl 1989, Schattschneider 1960,Schumpeter 1942), and as this chapter shows, it would be foolish to pretendthis is all that citizens do. Still, judgment and choice are prominent aspects of

    political experience, and consequently research on communication and opin-

    ion is occupied largely with describing and explaining how citizens evaluate

    180 KINDER

    casual observers of the campaign pick up information about how the variouscandidates are doing, about who has just won and whose chances are fading.Voters assessments of which candidates are serious and which arenotwhat Bartels (1988) calls expectationsrespond to such reporting, espe-

    cially for those who are following the campaign carefully (Bartels 1988; seealso Ansolabehere & Iyengar 1994, Brady & Johnston 1987, Popkin 1991).These results are important because, for virtually all voters early in the cam-

    paign and for poorly informed voters throughout the campaign, serious can-didates turn out to be winning candidates. That is, under conditions of high un-certainty, voters are often swayed by signs of a candidates electoral viability(Bartels 1988). A spectacular example is Gary Harts meteoric rise in the earlyspring of 1984. Following his upset victory in the New Hampshire primary,Harts public profile soared. Showered with attention, he was suddenly trans-formed from also-ran to front-runner.

    In psychological terms, agenda setting appears to be one manifestation ofthe availability heuristic at work. Heuristics are strategies of simplificationthat reduce the complexity of judgment tasks to make them tractable for thekind of mind that people happen to have (Kahneman et al 1982, p. xii). Theavailability heuristic comes into play when people assess the frequency of ageneral class of events or judge the plausibility of a particular development. Ina series of clever experimental demonstrations, Tversky & Kahneman (1973)showed that people make such judgments according to the ease with which in-stances or occurrences can be brought to mind. By implication, as editors and

    reporters bring some aspects of politics into prominence while keeping othersin darkness, they thereby influence which bits of political memory can mosteasily be brought to mind.

    The evidence on agenda setting strongly vindicates Lippmanns originalconjecture. For their map of politics, for their picture of the mystery offthere, citizens depend mightily, as Lippmann understood, on journalism.

    EVALUATING ALTERNATIVES

    Citizens react to the political alternatives that are put before them. They decidewhether to support or reject a candidate; they find policy proposals sensible orterrible; they take the consequences of government action to be successful ordisastrous; they judge their presidents performance in office to be inspiring ordeplorable; and so on. Democratic theorists disagree over whether this is allthat citizens should do (e.g. Barber 1984, Dahl 1989, Schattschneider 1960,Schumpeter 1942), and as this chapter shows, it would be foolish to pretendthis is all that citizens do. Still, judgment and choice are prominent aspects of

    political experience, and consequently research on communication and opin-

    ion is occupied largely with describing and explaining how citizens evaluate

  • 7/25/2019 Communication and Opinion

    15/32

    prominent aspects of their political world. First I review evidence relevant tothree distinct accounts of how this might happen, then turn in the final sectionto consider campaigns as instruments of persuasion.

    PrimingBy what standards should presidents be evaluated? By their position on hu-

    man rights or abortion or affirmative action? By the vitality of the economy?

    By the quality of their judicial appointments? By their performances at

    press conferences? By the moral example they set? When citizens are asked

    for their evaluation of such complex political entities as presidents, the

    priming model presumes that they do not undertake a comprehensive and ex-

    haustive compilation of everything they know. Instead, their evaluations de-

    pend on a modest sample of what they know, and a sample of convenience at

    that. Put another way, political evaluation is subject to the availability heuris-tic (Tversky & Kahneman 1973). Judgment in politics, as in other domains,

    depends on which pieces of memory happen to come to mind. And which

    pieces come to mind is a matter of circumstance. When political circum-

    stances change, what will come to citizens minds will also change. The cir-

    cumstantial basis for political evaluation has many sources, but among the

    most important is news. The standards citizens employ to evaluate a presi-

    dent (or policy, candidate, or other complex political entity) are determined

    substantially by which stories the news media choose to cover and, conse-

    quently, which standards are made available. The more attention news paysto a particular aspect of political lifethe more frequently that aspect is

    primedthe more citizens will incorporate what they know about it into

    their political evaluations.Defined this way, priming has received strong and consistent support in a

    series of network television news experiments (Iyengar & Kinder 1987).Primed by television news stories that focus on national defense, viewers

    judge the president largely by how well he has provided, as they see it, for thenations defense; primed by stories about inflation, viewers evaluate the presi-

    dent by how he has managed, in their view, to keep prices down; and so on. Inthese experiments priming shows up across a variety of problems, for Demo-cratic and Republican presidents alike, and for good news as well as bad. Prim-ing is also apparent in experiments set in the context of elections. For example,during the race for Congress in Connecticuts third district in 1982, voters whowere shown local news coverage that stressed the state of the economy and the

    presidents economic policies relied heavily on their assessments of economicconditions when deciding which congressional candidate to support. By con-trast, voters who watched local broadcasts devoted to the candidates them-

    selvestheir personal backgrounds, group endorsements, and positions on

    COMMUNICATION AND OPINION 181

  • 7/25/2019 Communication and Opinion

    16/32

    policy questionsassigned great importance to these qualities in theirchoices. Thus congressional elections can be primarily a referendum on the in-cumbent presidents economic performance or primarily a local contest be-tween two distinct candidates, depending on how voters are primed.

    All this evidence comes from experiments, but priming is demonstrated byother methods and in other settings as well. For example, following revelationsin November 1986 that funds from secret arms sales to Iran were being ille-gally channeled to the Nicaraguan Contras, the publics evaluations of Presi-dent Reagan suddenly became preoccupied with matters of foreign policy, justas priming would require (Krosnick & Kinder 1990). Likewise, immediatelyafter public disclosure of Gary Harts sexual escapades in the spring of 1987,moral conservatism, which had been utterly unconnected to the publicsevaluations of Hart prior to the revelations, suddenly became very connected(Stoker 1993). Finally, presidential campaigns evidentally prime voters toconceive of their choice as a judgment on the incumbent partys economic per-formance. In recent campaigns, the effect of citizens assessments of nationaleconomic conditions on their vote has roughly doubled between Labor Dayand Election Day (LM Bartels, unpublished observation; for other demonstra-tions of priming in natural settings, see Krosnick & Brannon 1993, Pollock1994, Mutz 1992, Kinder & Sanders 1996).

    As defined here, priming refers merely to the level of attention news pays toa problem, but news does more than simply draw attention to some problemswhile ignoring others; it also frames problems in particular ways. Especially

    important is the extent to which problems are framed in such a way as to sug-gest that they have political causes or political remedies. Priming is augmentedwhen television coverage implies that the president is responsible either forcausing a problem or for failing to solve it (Iyengar & Kinder 1987). More gen-erally, how news media frame problems influences how citizens come to theiropinions (Cappella & Jamieson 1997; Iyengar 1991; Kinder & Sanders 1990,1996; Mendelberg 1997; Stoker 1997)and even whether they come to anopinion in the first place (Kinder & Nelson 1998). By paying attention to some

    problems while ignoring others, and by framing problems in particular ways,

    media define the terms by which political judgments are rendered and politicalchoices made.

    If this is right, then we might expect that political campaigns would be or-ganized with priming in mind. Various indications are consistent with this ex-

    pectation. For example, Petrocik (1996) has recently shown that candidates de-liberately emphasize (prime) issues that, for historical or structural reasons, fa-vor their side. Rikers (1993b) examination of the public debate over ratifica-tion of the Constitution finds ample support for what he calls the DominancePrinciple: [W]hen one side successfully wins an argument on an issue, the

    other side ceases to discuss it, while the winner continues to exploit it (1993b,

    182 KINDER

    policy questionsassigned great importance to these qualities in theirchoices. Thus congressional elections can be primarily a referendum on the in-cumbent presidents economic performance or primarily a local contest be-tween two distinct candidates, depending on how voters are primed.

    All this evidence comes from experiments, but priming is demonstrated byother methods and in other settings as well. For example, following revelationsin November 1986 that funds from secret arms sales to Iran were being ille-gally channeled to the Nicaraguan Contras, the publics evaluations of Presi-dent Reagan suddenly became preoccupied with matters of foreign policy, justas priming would require (Krosnick & Kinder 1990). Likewise, immediatelyafter public disclosure of Gary Harts sexual escapades in the spring of 1987,moral conservatism, which had been utterly unconnected to the publicsevaluations of Hart prior to the revelations, suddenly became very connected(Stoker 1993). Finally, presidential campaigns evidentally prime voters toconceive of their choice as a judgment on the incumbent partys economic per-formance. In recent campaigns, the effect of citizens assessments of nationaleconomic conditions on their vote has roughly doubled between Labor Dayand Election Day (LM Bartels, unpublished observation; for other demonstra-tions of priming in natural settings, see Krosnick & Brannon 1993, Pollock1994, Mutz 1992, Kinder & Sanders 1996).

    As defined here, priming refers merely to the level of attention news pays toa problem, but news does more than simply draw attention to some problemswhile ignoring others; it also frames problems in particular ways. Especially

    important is the extent to which problems are framed in such a way as to sug-gest that they have political causes or political remedies. Priming is augmentedwhen television coverage implies that the president is responsible either forcausing a problem or for failing to solve it (Iyengar & Kinder 1987). More gen-erally, how news media frame problems influences how citizens come to theiropinions (Cappella & Jamieson 1997; Iyengar 1991; Kinder & Sanders 1990,1996; Mendelberg 1997; Stoker 1997)and even whether they come to anopinion in the first place (Kinder & Nelson 1998). By paying attention to some

    problems while ignoring others, and by framing problems in particular ways,

    media define the terms by which political judgments are rendered and politicalchoices made.

    If this is right, then we might expect that political campaigns would be or-ganized with priming in mind. Various indications are consistent with this ex-

    pectation. For example, Petrocik (1996) has recently shown that candidates de-liberately emphasize (prime) issues that, for historical or structural reasons, fa-vor their side. Rikers (1993b) examination of the public debate over ratifica-tion of the Constitution finds ample support for what he calls the DominancePrinciple: [W]hen one side successfully wins an argument on an issue, the

    other side ceases to discuss it, while the winner continues to exploit it (1993b,

  • 7/25/2019 Communication and Opinion

    17/32

    pp. 8182). Thus campaigns are less debates over a common set of issues than

    they are struggles to define what the election is about. An extensive analysis of

    party platforms since World War II in 23 democracies finds the same pattern

    (Budge 1993). As priming would predict, campaigns are mostly about sali-

    ence, not confrontation (Riker 1993a, p. 4; see also Gelman & King 1993, Ja-cobs & Shapiro 1994, Kinder & Sanders 1996).

    Reception, Acceptance, Persuasion

    For critics of democracy, the contention that citizens dont know enough to

    participate sensibly in the decisions of government has been a persistent and

    powerful weapon. An especially unsettling version of this general line of argu-

    ment is that electoral change is due largely to the capricious vacillations of the

    least informed voters. As Converse (1962) put it, Not only is the electorate as

    a whole quite uninformed, but it is the least informed members within the elec-torate who seem to hold the critical balance of power, in the sense that alterna-

    tions in governing party depend disproportionately on shifts in their senti-

    ments (p. 578).But a careful look at the stability of partisan attitudes in American national

    political campaigns led Converse to a different conclusion: that it is the moder-

    ately informed who are most likely to be persuaded. Why? Well-informed citi-

    zens are heavily exposed to the campaign but committed in advance, so they

    are hard to persuade. The poorly informed are likewise difficult to convert, not

    because they are adept at defending their precommitments (insofar as theyhave them at all), but because they so rarely encounter messages designed to

    alter their loyalties. This leaves the intermediately informed, who pay suffi-

    cient attention to the campaign to bump into messages that challenge their

    opinions, but who are not so committed to their views as to be immune to per-

    suasion.Meanwhile, working independently, McGuire (1968) was developing a

    general theory of persuasion. McGuires framework depicted opinion change

    as the net outcome of a chain of behavioral steps. As a minimum it requires (a)

    adequate reception (through attention and comprehension) of the persuasivemessage; and (b) yielding to what is comprehended (1968, p. 1139). In this re-

    spect, McGuire was carrying forward a point of view introduced by Hovland

    and his Yale associates (e.g. Hovland et al 1953). McGuire argued that con-

    ventional theories of attitude change tended to run into trouble because they

    overlooked reception, concentrating entirely on yielding. He then developed

    plausible cases wherein individuals most likely to receive a message were least

    likely to accept it (and vice versa). The general result in such cases is a com-

    plex, nonmonotonic relationship between characteristics of individuals and

    overall persuasion.

    COMMUNICATION AND OPINION 183

  • 7/25/2019 Communication and Opinion

    18/32

    Taking off from the insights offered by Converse and McGuire, Zaller

    (1992) devised a general theory of mass communication and opinion change.

    Zallers theory presumes that the essential elements of opinion are considera-

    tions, broadly defined to include any reasons for favoring one side of a dis-

    pute over another, and that most people possess opposing considerations onmost issues. People arrive at their opinions by averaging across the considera-

    tions that happen to be accessible at the moment. Accessibility, in turn, de-

    pends upon a memory search that is probabilistic and incomplete. Considera-

    tions that have been in thought recently are more likely to be sampled (Zaller

    1992, Zaller & Feldman 1992; also see Feldman 1995).To apply this conception of opinion to the dynamics of political persuasion,

    Zaller (1992) introduced two additional assumptions. First, peoples likeli-

    hood of receiving a communication is a direct function of their level of general

    information about politics, where reception involves both exposure to andcomprehension of the given communication (the Reception Axiom). Sec-

    ond, people resist communications that are inconsistent with their political

    predispositions only insofar as they possess sufficient information to notice

    such inconsistency (the Acceptance Axiom).Zallers model has been appliedwith mostly smashing empirical suc-

    cessto a wide variety of cases: the electoral advantages enjoyed by congres-

    sional incumbents (Zaller 1992); shifts in American opinion on school deseg-

    regation and on the Vietnam War (Zaller 1991); variations in popular support

    for Ross Perot during his $73 million presidential adventure of 1992 (Zaller &Hunt 1994, 1995); and the dynamics of presidential primary election cam-

    paigns (Zaller 1989). All these and more can be accommodated within a single

    theoretical vocabulary. On the first page ofThe Nature and Origins of Mass

    Opinion (1992), Zaller wrote that his aim was to integrate as much as possible

    of the dynamics of public opinion within a cohesive theoretical systeman

    audacious ambition and, so far at least, an audaciously successful one.

    Processing On-Line

    Availability (or memory-based) models are not the only way to account forcommunication effects. The strongest alternative model currently in play is

    supplied by Lodge and his Stony Brook associates. Drawing heavily on ideas

    originally developed by Hastie (Hastie & Park 1986; Hastie & Pennington

    1989), Lodge and his colleagues argue that over the course of a campaign, citi-

    zens 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 typically computed at a later date from memory

    traces (Lodge 1995, p. 119). Lodges story runs this way: Campaigns deliver

    Taking off from the insights offered by Converse and McGuire, Zaller

    (1992) devised a general theory of mass communication and opinion change.

    Zallers theory presumes that the essential elements of opinion are considera-

    tions, broadly defined to include any reasons for favoring one side of a dis-

    pute over another, and that most people possess opposing considerations onmost issues. People arrive at their opinions by averaging across the considera-

    tions that happen to be accessible at the moment. Accessibility, in turn, de-

    pends upon a memory search that is probabilistic and incomplete. Considera-

    tions that have been in thought recently are more likely to be sampled (Zaller

    1992, Zaller & Feldman 1992; also see Feldman 1995).To apply this conception of opinion to the dynamics of political persuasion,

    Zaller (1992) introduced two additional assumptions. First, peoples likeli-

    hood of receiving a communication is a direct function of their level of general

    information about politics, where reception involves both exposure to andcomprehension of the given communication (the Reception Axiom). Sec-

    ond, people resist communications that are inconsistent with their political

    predispositions only insofar as they possess sufficient information to notice

    such inconsistency (the Acceptance Axiom).Zallers model has been appliedwith mostly smashing empirical suc-

    cessto a wide variety of cases: the electoral advantages enjoyed by congres-

    sional incumbents (Zaller 1992); shifts in American opinion on school deseg-

    regation and on the Vietnam War (Zaller 1991); variations in popular support

    for Ross Perot during his $73 million presidential adventure of 1992 (Zaller &Hunt 1994, 1995); and the dynamics of presidential primary election cam-

    paigns (Zaller 1989). All these and more can be accommodated within a single

    theoretical vocabulary. On the first page ofThe Nature and Origins of Mass

    Opinion (1992), Zaller wrote that his aim was to integrate as much as possible

    of the dynamics of public opinion within a cohesive theoretical systeman

    audacious ambition and, so far at least, an audaciously successful one.

    Processing On-Line

    Availability (or memory-based) models are not the only way to account forcommunication effects. The strongest alternative model currently in play is

    supplied by Lodge and his Stony Brook associates. Drawing heavily on ideas

    originally developed by Hastie (Hastie & Park 1986; Hastie & Pennington

    1989), Lodge and his colleagues argue that over the course of a campaign, citi-

    zens 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 typically computed at a later date from memory

    traces (Lodge 1995, p. 119). Lodges story runs this way: Campaigns deliver

    184 KINDER

  • 7/25/2019 Communication and Opinion

    19/32

    messages; citizens sometimes notice them; when they do, they detect the im-

    plications of the messages for their evaluation of a candidate; thereupon they

    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 firstplace (for corroborating experimental evidence, see Lodge et al 1989, Lodge &

    Steenbergen 1995, Lodge & Stroh 1993, Rahn et al 1994).One important implication of this model is that citizens may know more,

    perhaps much more, than they can tell. Maybe we shouldnt worry so muchover the repetitious demonstration that ordinary Americans cannot recollect

    basic political facts or remember the names of candidates. Such demonstra-tions do not prove that citizens are wandering around in the dark. From the per-spective of the on-line model, citizens can be responsive to informationin-

    deed, they can be perfectlyresponsive to informationwithout being able torecollect why they feel the way they do. In one experiment, Lodge & Steenber-gen (1995) showed that citizens evaluations of fictitious candidates can be

    predicted from the information that had been presented to them earlier, eventhough they could recall little of it themselves, and what information theycould recall was of trifling value in predicting their overall evaluations.

    Lodge concludes that voters should be judged not by the information theycan recollect but by the kinds of information they entertain moment to momentand by how well they integrate such information into their ongoing politicalevaluations. This argument is well supported by a series of experiments, and itraises a pressing question for future theoretical and empirical work: Assuming

    both memory-based and on-line models are required to account for importantempirical regularities in the relationship between mass communication and

    public opinion, how are the two models to be reconciled?

    Campaigns and Persuasion

    Systematic empirical research on campaigns as instruments of political per-

    suasion began in a context dominated by apprehensions set off by the emer-

    gence of fascism and the rise of commercial advertising. Social scientists ofthe time worried whether ordinary citizens could maintain their democratic

    values and economic autonomy in the new and scary world of mass communi-

    cation. A first and powerful answer was supplied by theThe Peoples Choice,

    Lazarsfeld et als (1948) landmark examination of the 1940 presidential con-

    test between Wendell Willkie and Franklin Roosevelt. In this remarkable

    study, citizens of Erie County, Ohio, were questioned seven times during the

    1940 campaign, beginning in May, as Hitlers armies were sweeping through

    France, and ending just after Election Day. Lazarsfeld et al discovered, to their

    surprise, that relatively few voters changed their minds over this period. By the

    COMMUNICATION AND OPINION 185

  • 7/25/2019 Communication and Opinion

    20/32

    time of the summer conventions, well before the onset of the formal campaign,

    roughly 80% of voters had become permanently committed to one candidate or

    the other. What little change that did occur, moreover, had less to do with the

    campaign than with the personal influence exercised by family and friends.

    Such evidence led Lazarsfeld et al to conclude that presidential campaigns aregenerally ineffective at persuasion; rather their functions are to reassure theearly deciders (reinforcement) and to mobilize the latent predispositions of

    the initially uncommitted (activation).Despite dramatic changes over the last half century in the technology of

    campaigning, the rise of television, and the decline of strong partisans in theelectorate, Lazarsfeld et als conclusions have stood up very well. As large-scale, expensive experiments in political persuasion, American presidentialcampaigns are a bust (e.g. Bartels 1997, Berelson et al 1954, Finkel 1993, Mar-

    kus 1982, Miller & Shanks 1982, Patterson 1980). This conclusion is but-tressed from a different angle by the recent development of aggregate forecast-ing models that are capable of predicting the outcome of presidential elections

    based on information that is available before the campaign gets underway(Campbell 1992, Gelman & King 1993, Markus 1988, and especially Rosen-stone 1983).

    Why do presidential campaigns regularly fail as agents of persuasion? Oneanswer is that the campaign of one side is neutralized by the campaign of theother (Lazarsfeld & Merton 1948; this account has been rediscovered in recentyears by Bartels 1992 and by Gelman & King 1993). Under current arrange-

    mentselectoral competition dominated by two well-established parties andpresidential campaigns funded primarily by public sourcesboth sides as-semble teams of roughly equal experience and intelligence, who set aboutspending roughly the same (large) amount of money in roughly the same ways.

    Another reason presidential campaigns enjoy limited success convertingvoters from one side to the other is that they run up against the most basic andenduring of political predispositions: voters attachment to party. For manystrong partisans, the details brought forward by any particular campaign are

    beside the point. For them, partisanship is a standing decision, largely im-

    mune to whatever surprises the campaign might bring (Campbell et al 1960,Converse 1966). [Alternative conceptions of partisan identification differ onthis point of immunity to the campaign (e.g. Fiorina 1981, Green & Palmquist1990, MacKuen et al 1989).]

    Even those less frozen in their partisan loyalties do not come to the cam-

    paign without prejudice. They tend to encounter information that dispropor-tionately favors their own candidate, and more important, they selectively in-terpret whatever information comes their way in a fashion that supports their

    political inclinations (Lazarsfeld et al 1948, Sears & Whitney 1973). This

    point has been made in scores of studies of presidential debates, which show

    186 KINDER

  • 7/25/2019 Communication and Opinion

    21/32

    how sharply evaluations of debate performance polarize along partisan lines.Debates dont persuade; they reinforce and strengthen viewers prior commit-ments (Katz & Feldman 1962, Sears & Chaffee 1979).

    Finally, if strong partisans are essentially unmovable because they are so

    deeply committed, other potential voters are difficult to persuade becausetheyre not paying attention. Analyzing the flood of political propaganda un-leashed by the 1940 presidential campaign, Lazarsfeld et al (1948, p. 121) no-ticed that far from being drowned in information, most voters did not even get

    their feet wet. Perhaps the principal obstacle to persuasion is indifference:The problem for most political propagandists is not that they fail to reach theirenemies but that they fail to reach anyone at all (Sears & Whitney 1973, p. 7;Hyman & Sheatsley 1947).

    Having said all that, we should keep in mind that if only a little persuasiontakes place, the collective consequences can be very great if the race is close.

    Moreover, even the most successful forecasting models leave some room forcampaign effects. A fine example was furnished, quite inadvertently, byGeorge McGovern, who in 1972 offered the United States the worst man-aged campaign in this century, most notably by selecting and then abandoning

    a vice-presidential nominee who had been treated for depression with shocktherapy (Bartels 1992, p. 266; on McGoverns troubles in 1972, see Popkin etal 1976).

    Granting that little persuasion ordinarily takes place from Labor Day toElection Day, it would be a mistake to conclude that the presidential cam-

    paign is generally ineffective. Activation and reinforcement may not be asglamorous as persuasion, but they are important processes nonetheless.Campaigns activate voters by arousing their interest and providing them in-formation, thereby allowing them to choose wiselyor in any case, morewisely than they could have in the absence of a campaign. Campaigns rein-

    force voters by providing good reasons and reminding them why they areDemocrats or Republicans, thereby keeping partisans in line and defections toa minimum.

    Nor, finally, should we conclude that because presidential campaigns

    rarely succeed in persuading significant numbers of voters to switch from oneside to the other, other kinds of campaigns will fare no better. One clear exam-

    ple to the contrary comes from the systematic study of House campaigns. Aprincipal finding of such studies is the importance of money. In particular, themore money challengers raise, the better the campaign they can mount; the

    better the campaign, the more voters they can reach, and the better impressionthey can establish (e.g. Jacobson 1980, Green & Krasno 1988). Candidateswho challenge House incumbents generally have a difficult time, but withoutadequate financing, without a real campaign, they have virtually no chance at

    all.

    COMMUNICATION AND OPINION 187

  • 7/25/2019 Communication and Opinion

    22/32

    A second case of campaigns as effective instruments of persuasion is pro-vided by the sequence of primaries and open caucuses through which Ameri-

    can presidential candidates are selected. The current system is interesting fromthe perspective of persuasion because, first, the competition naturally takes

    place within the party. This means that voters party attachments are largelynullified as a basis for choice, so voters are less anchored in the primary sea-son than they will be later in the fall. Second, candidates vary in the recogni-

    tion and resources they command at the start of the process. Finally, and per-haps most significantly, the selection process takes place over time, in atightly bunched sequence of contests, which would seem to invite dynamic

    forces.This complex system is splendidly analyzed inPresidential Primaries and

    the Dynamics of Public Choice(Bartels 1988). Bartels first develops a theoryof choice based on the voters expectations, uncertainty, and political predis-

    positions, and then applies the theory to the primary campaigns of 1976, 1980,

    and 1984. Bartels demonstrates that variation in the outcome of primary elec-tions can be explained by the interplay among differences in voters political

    predispositions from one state to the next; by differences in voters expecta-tions, which are shaped by fluctuations in electoral fortunes; and by differ-ences in voters uncertainty, which generally diminish over the course of the

    campaign.To estimate the impact of the primary campaign, Bartels compares what ac-

    tually happened in particular years with what would have happened under a

    one-day national primary. This comparison yields some stunning results. Forexample, in 1976, Carter would have been a far weaker candidate; GeorgeWallace would have been the likely beneficiary of Carters decline; and theDemocratic convention of that year probably would have had real work to do.

    Such results suggest that the primary campaign can matter, and in a major way.Especially consequential are early and unanticipated outcomessurprisingvictories or shocking defeatsthat change voters expectations and, in turn,

    alter their choices.And finally, if we ask not about the short term but about the longer haul, not

    about individuals but about the public as a whole, and not about campaigns perse but about the general flow of information and debate carried by mass media,

    then persuasion turns out to be the rule, not the exception. Consider as just one

    example the findings from research devoted to explaining fluctuations over

    time in public support for the president. From Truman to Clinton, these per-

    formance ratings move up and down, sometimes dramatically. Such twists and

    turns may seem enigmatic, but they are are in fact quite predictable. Unem-

    ployment, inflation, economic growth, flagrant violations of public trust, the

    human toll of war, international crises, dramatic displays of presidential

    authorityall these influence the presidents standing in the general public

    188 KINDER

  • 7/25/2019 Communication and Opinion

    23/32

    (e.g. Brody 1991, Hibbs et al 1982, Kernell 1978, MacKuen 1983, Mueller

    1970, Ostrom & Simon 1985). Thus assessments of presidential performance

    in the aggregate and over the long haul seem very well informed, highly re-

    sponsive to developments that most citizens learn about primarily through the

    mass media.In these investigations as in others (e.g. Fan 1988, MacKuen et al 1989,

    Page & Shapiro 1992, Stimson 1991), public opinion in the aggregate emerges

    as responsive to social, economic, and political changeoften exquisitely so.

    If citizens on average know so little, how can the public as a whole choose so

    wisely? One powerful answer comes from the sheer mechanical process of sta-

    tistical aggregation, the law of large numbers applied to public opinion. Given

    sufficient cases, even faint signals can emerge from a sea of noise (Converse

    1990, p. 382; for a review of the evidence on the dynamics of collective public

    opinion, see Kinder 1998). The role of the media in these dynamics is not al-

    ways explicit, but it is always implied. Looked at in a particular way, political

    persuasion is ubiquitous.

    PAST AND FUTURE

    Klappers (1960) famous conclusion of minimal effects was faithful to the

    evidence available at the time. Nearly 40 years later, it seems not so much

    wrong as incomplete. With more incisive research designs, greater care in

    measurement, more adept statistical analysis, greater attention to long-termchange, and deeper appreciation for the human information processing system,

    minimal effects have given way to a family of real effects: agenda setting,

    priming, framing, and even, looking in the right places, ample persuasion. If,

    as Bartels has recently maintained, the minimal effects verdict should be re-

    garded as one of the most notable embarrassments of modern social science

    (Bartels 1993, p. 267), we are now relieved of that burden.4

    Of course we still have some distance to go. One limitation of the current

    literature is its preoccupation with individuals. Politics, as Huckfeldt & Spra-

    gue (1993) remind us, involve individuals and groups tied together in com-plex relations. And yet, when we address citizen politics in the mass, the

    temptation appears overwhelming to shift the level of understanding and

    analysis to that of independent individualsindividuals abstracted from time,

    place, and setting (p. 281). This temptation is indeed powerful, though the in-

    clination to investigate public opinion within its social context has a long and

    honorable history (e.g. Katz & Lazarsfeld 1955, Lazarsfeld et al 1948) and

    COMMUNICATION AND OPINION 189

    4 4Not everyone thinks so. Skeptics in the Klapper tradition include McGuire (1986), Mueller(1994), and Schudson (1984). The strongest argument on the other side is advanced by Zaller

    (1996).

  • 7/25/2019 Communication and Opinion

    24/32

    now appears to be making a comeback, perhaps as design and analysis becomemore adequate to the task (e.g. Achen & Shively 1994, Huckfeldt & Sprague1995, King 1997). These results remind us that public opinion is not only an

    achievement of individual psychology but the product of a complex and multi-

    layered social experience as well.A second defect afflicting contemporary research on communication andopinion has to do with emotion, conspicuous by its absence. Current studiesemphasize cognition and the management of information, as in the tussle be-

    tween on-line and memory-based models of information processing. Aboutthe emotional requirements of citizenship, or the emotional foundations of po-litical opinion, little is heard. A generation or two ago things were different. In-spired by psychodynamic theory and impressed with the possibilities of in-

    depth interviewing, analysts of public opinion sought to reveal how, inLasswells (1930) famous formulation, private affects were displaced onto

    political objects (e.g. Adorno et al 1950, Lane 1962, Smith et al 1956). But thescientific reputation of psychodynamic theory soon came under ferocious at-

    tack; intensive interviewing as a mode of inquiry went out of fashion; and so-cial psychology, which had contributed so much to the study of public opinion,was overtaken by a cognitive revolution (Tompkins 1981). Now things havetaken a different turn. Within psychology, theories of emotion are proliferating

    and empirical work is booming (Ekman & Davidson 1994). Provoked in partby all this commotion, a modest resurgence of interest has occurred in the po-litical analysis of emotion (e.g. Abelson et al 1982, Conover & Feldman 1986,

    Esses et al 1993, Kinder 1993, Marcus 1988, Marcus & MacKuen 1993, Sulli-van & Masters 1988). These demonstrations suggest that if we are to under-stand communication and opinion, we must attend to the whole person, to rea-son and emotion alike.

    Citizens not only think and feel, they also (at least on occasion) act. Butfaithful to the fields inclinations, I have so far ignored the role of mass com-munication in political action. An example of research running against this

    tide, one well worth imitating, was prompted by what some observers saw as anew and troubling development in American politics: the emergence of nega-

    tive campaigns. With the advent of television, campaigns at all levels havesupposedly turned increasingly hostile and ugly. More often than not, candi-

    dates criticize, discredit, or belittle their opponents rather than promoting theirown ideas and programs (Ansolabehere et al 1994, Jamieson 1992, West1993). Is negative campaigning corrosive? A good if not final answer comesfrom the research of Ansolabehere & Iyengar (1995, Ansolabehere et al 1994),

    based primarily on a series of well-crafted and ambitious field experiments inwhich negative advertisements are embedded, more or less seamlessly, withinactual ongoing campaigns. The principal result of these experiments is that

    negative campaigning is demobilizing. Negative advertisements strengthen

    190 KINDER

  • 7/25/2019 Communication and Opinion

    25/32

    citizens impression that government is unresponsive; induced to witness cam-paigns that seldom rise above exchanges of snide and petty accusations, citi-zens withdraw.5

    My focus here has been to clarify the effects of mass communication on

    public opinion in contemporary American society. Complicated and difficultas that task is, it represents just one piece of the story of communication in thebroader workings of politics. The larger story involves the dynamic interplayamong public officials, interest groups, issue entrepreneurs, and social move-ments, in addition to the media and the public, all played out against a land-scape of turbulent events and changing political opportunities (see Bartels1991, 1996b; Burstein1985; Carmines & Stimson 1989; Gamson & Modigli-ani 1987; Hallin 1984; Hilgartner & Bosk 1988; Kingdon 1984; Jones 1994;Lindblom 1977; Page 1996; Protess et al 1991; Riker 1982; Schudson 1978;Stimson et al 1995; Verba & Nie 1972; Verba et al 1995).

    LippmannsPublic Opinionemphasized the heavy responsibility borne bythe press in a free society, since the press alone describes and interprets theevents of public life that few citizens experience directly. The press handlesthis responsibility poorly, in Lippmanns view, for it is capricious in what itchooses to disclose, like the beam of a searchlight that moves restlessly about,

    bringing one episode and than another out of the darkness into vision. Maybeso. Placing the effects documented here in a broader conception of the political

    process will enable a deeper analysisand critiqueof communication andopinion than Lippmann could manage in his time.

    ACKNOWLEDGMENTS

    I am grateful to Adam Berinksy, Kathy Cramer, Janet Weiss, and NicholasWinter for their timely and helpful advice on an earlier draft of this chapter.

    Visit theAnnual Reviews home page at

    http://www.AnnualReviews.org.

    COMMUNICATION AND OPINION 191

    5 5It seems a mistake to tie negative campaigning to the development of television, however;Riker (1996) discovered plenty of it in the storied and often romanticized campaign for the

    ratification of the US Constitution.

  • 7/25/2019 Communication and Opinion

    26/32

    192 KINDER

    Literature Cited

    Abelson RP, Kinder DR, Peters MD, Fiske ST.1982. Affective and semantic componentsin political person perception.J. Pers. Soc.Psychol. 42:61930

    Achen CH, Shively WP. 1995. Cross-Level In-ference. Chicago: Univ. Chicago Press.248 pp.

    Adorno TW, Frenkel-Brunswick E, LevinsonDJ, Sanford RN. 1950. The AuthoritarianPersonality. New York: Harper & Row.990 pp.

    Alvarez RM. 1997.Information and Elections.Ann Arbor: Univ. Mich. Press. 257 pp.

    Alvarez RM, Franklin CH. 1994. Uncertaintyand political perceptions. J. Polit. 56:67188

    Anderson JR. 1983.The Architecture of Cog-

    nition. Cambridge, MA: Harvard Univ.Press. 345 pp.Ansolabehere S, Iyengar S. 1994. Of horse-

    shoes and horse races: experimental stud-ies of the impact of poll results on electoral

    behavior.Polit. Commun.11:41330Ansolabehere S, Iyengar S. 1995. Going

    Negative: How Political AdvertisementsShrink and Polarize the Electorate. NewYork: Free. 243 pp.

    Ansolabehere S, Iyengar S, Simon A, Valen-tino N. 1994. Does attack advertising de-mobilize the electorate? Am. Polit. Sci.Rev. 88:82938

    Bachrach P, Baratz MS. 1970.Power and Pov-erty. New York: Oxford Univ. Press. 220

    pp.Barber B. 1984.Strong Democracy. Berkeley:

    Univ. Calif. Press. 320 pp.Bartels LM. 1986. Issue voting under uncer-

    tainty: an empirical test.Am. J. Polit. Sci.30:70928

    Bartels LM. 1988.Presidential Primaries andthe Dynamics of Public Choice. Princeton,

    NJ: Princeton Univ. Press. 374 pp.Bartels LM. 1991. Constituency opinion and

    congressional policy making: the Reagan

    defense build-up.Am. Polit. Sci. Rev. 85:45774Bartels LM. 1992. The impact of electioneer-

    ing in the United States. InElectioneering,ed. D Butler, A Ranney, pp. 24477.NewYork: Oxford Univ. Press

    Bartels LM. 1993. Messages received: the po-litical impact of media exposure.Am. Po-lit. Sci. Rev.87:26785

    Bartels LM. 1996a.Politicians and the press:Who leads, who follows? Presented atAnnu. Meet. Am. Polit. Sci. Assoc., SanFrancisco

    Bartels LM. 1996b. Uninformed votes: infor-

    mation effects in presidential elections.Am. J. Polit. Sci. 40:194230

    Berelson B, Lazarsfeld P, McPhee WN. 1954.Voting: A Study of Opinion Formation in a

    Presidential Election. Chicago: Univ. Chi-cago Press. 395 pp.

    Bobo L, Kleugel JR. 1993. Opposition to racetargeting: self-interest, stratification ideol-ogy, or racial attitudes?Am. Sociol. Rev.58:44364

    Boninger DS, Krosnick JA, Berent MK. 1995.Origins of attitude importance: self-interest, social identification, and valuerelevance.J. Pers. Soc. Psychol. 68:6180

    Brady HE, Johnston R. 1987. Whats the pri-mary message: horse race or issue journal-ism. In Media and Momentum: The New

    Hampshire Primary and Nomination Poli-tics, ed. GR Orren, NW Polsby, pp.12786. Chatham, NJ: Chatham House

    Brehm, J. 1993. The Phantom Respondents:Opinion Surveys and Political Representa-tion. Ann Arbor: Univ. Mich. Press. 266

    pp.Brewer MB. 1988. A dual process model of

    impression formation. InAdvances in So-cial Cognition, ed. TK Srull, RS Wyer,Vol. 1, pp. 1337. Hillsdale, NJ: