Facilitating Communication across Lines of Political Difference: The Role of Mass Media Author(s): Diana C.

Mutz and Paul S. Martin Source: The American Political Science Review, Vol. 95, No. 1 (Mar., 2001), pp. 97-114 Published by: American Political Science Association Stable URL: http://www.jstor.org/stable/3117631 Accessed: 02/09/2009 11:13
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=apsa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

American Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to The American Political Science Review.

http://www.jstor.org

AmericanPoliticalScience Review

Vol. 95, No. 1 March2001

Communication across Lines Facilitating Role of Mass The Media DIANA C. MUTZ The Ohio State University PAUL S. MARTIN UniversityOklahoma of

of

Political

Difference:

that and expose peopleto dissimilar politicalviews.Wehypothesize the individual's ability desire to exerciseselectiveexposure a keyfactor in determining a is whether given sourceproduces is exposureto dissimilarviews.Althougha lack of diverseperspectives a common complaintagainst are Americannewsmedia,wefind thatindividuals exposedto far moredissimilar politicalviewsvia news media than throughinterpersonal discussants.The media advantageis rooted in the relative political as of exposing oneselfto thosesourcesof information, wellas the lesserdesireto do so, difficulty selectively natureof mass media. given the impersonal

We

the various usenational data sources survey to examine extentto which information ofpolitical

rationalesfor views with.whichthey disagree,even an cutting political viewpoints has become of inoutcome they do not like acquiresgreaterlegitimacy. concernto observersof Americanpolcreasing believe such For example, the literature on political tolerance itics.Advocatesof deliberative democracy argues that educationis importantbecause it "putsa exposure is essential in order for alternativesto be contrastedeffectively(Fishkin 1991). Others consider person in touch withpeople whoseideasand valuesare exposureto dissimilarviews indispensablein forming diferent from one's own"(Stouffer1955, 127, emphasis in valid opinions and in learningto appreciatethe per- in original).Likewise,differences tolerancebetween spectivesof others (Arendt1968;Benhabib1992).Still men and women and between urban and rural resiothers point to the value of exposureto cross-cutting dents have been attributed to the more parochial views for purposesof establishingpoliticallegitimacy. contactsof womenand ruraldwellers(Nunn,Crockett, views ensures "thatno one and Williams 1978; Sullivan, Piereson, and Marcus Exposureto cross-cutting could see the end result as arbitraryrather than 1982). Along similar lines, authoritarianism negais reasonableand justifiable,even if not what he or she tively related to diversityof experience (Altemeyer 1996; Marcuset al. 1995), and politicallydiverseper(Fearon 1998,62, happenedto see as mostjustifiable" sonal networks increase awareness of oppositional emphasisin original). Exposure to conflictingviews is deemed a central viewpointsand politicaltolerance(Mutz 1999). element-if not the sine qua non-of the kind of In short,both in politicaltheoryand empirical work, needed to maintain a democratic there is near unanimousagreementthat exposureto political dialogue citizenry(e.g., Barber1984;Bellah et al. 1985;Haber- diverse political views is good for democracy and mas 1989). In contrast,political talk that centers on should be encouraged.Most social scientists concur a reinforcing sharedviewpointdoes little to encourage that political attitudes and opinions are formed deliberation on multiple perspectivesor promote a through social interaction, political discussion, and 1988, 220; Schudson 1995). personalreflection,and these processesare of a higher public sphere (Calhoun Accordingto the most often cited proponentof com- qualitywhen people are exposedto dissimilar perspecmunicationacrosslines of difference, John StuartMill, tives. Nonetheless,there is little empiricalworkon the "if the opinion is right, [people] are deprivedof the contextsin whichsuch exposureoccurs.Moreover,the of exchangingerror for truth; if wrong, recent trendtowardresidentialbalkanization based on opportunity they lose what is almost as great a benefit,the clearer sharedlifestylesheightensconcernsaboutcommunicaof perceptionandlivelierimpression truthproducedby tion across lines of political differencein the United its collision with error"(Mill [1859] 1956, 21). Mill's States. To the extent that people live among homogestatementpoints to two potentialbenefitsof exposure neous othersin self-selectedenclaves,theirexposureto to oppositionalviews, the opportunity changeone's dissimilarviews may be limited. to mind and adopt a normativelybetter viewpoint,and Some theoristsproposethat the futureof communithe deeper understandingof one's own position ac- cation across lines of political differencelies in techquired through confrontingdifferentperspectives.A nologies that transcendgeographicspace. As Calhoun third benefit is legitimationof an undesiredoutcome. (1988, 225) argues, "in modern societies, most of the informationwe have about people differentfrom ourselves comes not throughanydirectrelationships, even Diana C. Mutzis Professorof PoliticalScienceand Journalism and in Ohio State University, OH Communication, Columbus, 43210-1373. the casualones formedconstantly urbanstreets and PaulS. Martinis AssistantProfessorof PoliticalScience,University shops. Rather it comes throughprint and electronic of Oklahoma, Norman,OK 73019. media."Yet, much of what is known about the strucThis research supported a grantto Mutzfrom the Spencer was by ture and news gatheringpracticesof Americanmedia Foundation by the Centerfor Advanced and Studyin the Behavioral Sciences.We are gratefulto the editorand anonymous reviewers for suggeststhattheyare unlikelyto playa veryusefulrole. this The goal of this studyis to evaluateconflictingclaims improving manuscript. 97

he extent to whichpeopleareexposed to cross- To the extent that people are at least exposed to

acrossLines of PoliticalDifference FacilitatingCommunication

March2001

regarding the media's contribution to cross-cutting ongoingobstacleto the formationof a publicsphereis the persistenttrade-offbetween amountof interaction politicalexposure. We begin by reviewingrelevant researchon inter- and heterogeneityof interaction.Strongties and freand personalcommunication the massmedia.We then quentcontacttend to characterize homogeneousinteruse two nationalsurveys test the proposition to thatthe actions,which do not bringnew views to one's attenmedia make a greatercontribution than interpersonal tion (Granovetter 1973). Quantitative evidence of networksto Americans' exposureto dissimilar political selective exposurein interpersonal politicalcommuniviews. Drawingon surveydata across differentmedia cation is buttressed by qualitative accounts of the we environments, examinethree independenttests of courage requiredto speak up among heterogeneous whether selective exposure explains our finding of others (e.g., Mansbridge 1980;Schudson1984),as well views throughnews as the lengthsto whichpeople sometimesgo in orderto greater exposureto cross-cutting media.The resultssuggestthat the structure Amer- avoid discussingpolitics (Eliasoph 1998). Selectivity of icans' informationenvironmentsplaces an extraordi- appears to play a significant role in the kinds of nary burden on the mass media to bring diverse conversationspeople choose to have and, thus, the perspectivesto public attention, a burden the news kinds of political networksthey form. If residential media may be increasingly choices increasingly facilitate de facto selective expoill-equippedto shoulder. sure, and if people activelydodge any politicalconflict that enters their lives, then the prospects for crossINTERPERSONAL EXPOSURETO cutting interpersonalinteractionsappear quite bleak DISSIMILAR POLITICAL VIEWS indeed. The verdictwith respect to Americans'interpersonal information environments has become increasingly bleak in the last few decades.The kind of people with THE MEDIACONTRIBUTION TO whom any given individual discussespolitics is a funcEXPOSURE CROSS-CUTTING tion of two factors:the availability discussionpartof ners in one's immediateenvironment and the amount Few concepts have played as importanta role in the of selectivity exercisedin the choice of partners(Huck- history of research on mass communicationas the feldt and Sprague1995a).Researchsuggeststhat both notion that people selectively expose themselves to factors now operate primarily producegreater hoto like-minded media content (Katz 1981). Beginning in interpersonal interactions.With respect with the Erie Countyelection study (Lazarsfeld, Bermogeneity residentialpatternssuggestincreasingly elson, and Gaudet 1944),this assumption to availability, becamepart spatiallysegregatedliving even within the heteroge- of the conventionalview that the media have limited neous populationsof largecities, thusprompting many effects on political attitudes (Klapper 1960). That to argue that Americans are increasinglyseparated tradition is now considered pass6, but the issue of fromthose withpoliticalviewsdifferent fromtheirown whether and to what extent people may selectively does (e.g., Calhoun 1988).1Residentialbalkanization expose themselves to media content has never been not necessarilymean that hordes of Americans are fully resolved (e.g., Freedman and Sears 1965; Frey choosingto live amongpeople who sharetheirpolitical 1986;Katz 1968;Sweeneyand Gruber1984;Zillmann views because of those views. Indeed, few Americans and Bryant1985). Findingshave been so inconsistent assignpoliticssucha centralrole in theirlives.De facto as to discouragemuch research,although studies of selectivity is far more likely (Freedman and Sears selective attentionand bias in informationprocessing location have continued apace. Moreover, even if the many 1965),that is, people may choose a particular because it is convenient to local co-ops, or a golf studies of selective exposure had converged on a central finding, they might not apply today, due to course, or the schools they want their children to attend, and they find themselves among others who changesin the U.S. media environment. based selection on similar considerations.The initial relaEvidenceof selectiveexposurein interpersonal but goal may not have been politicallylike-mindedneigh- tions is incontrovertible, it is less clearwith respect bors, but that is achieved to the extent that lifestyle to the news media. Laboratory experimentsthat give correlatewith politicalperspectives. considerations people a choice of exposureto pro- or counterattitudinal media messageshave yielded mixed results (see views also is inhibitedby the Exposureto dissimilar such studiesare tendency for people to select politicallylike-minded Frey 1986for a review).Furthermore, discussion very limited in what they can reveal about life outside partners(HuckfeldtandSprague1995a).An the laboratory,where people do not always have a choice, are not alwaysforewarnedabout the political 1Because these trends have been documented strictly for residential content of a messagebefore exposure,and tend to use contexts, they do not necessarily point to greater homogeneity of a particular ratherthan on a storymediumhabitually views in people's larger social network. For example, political encounters in the workplace provide far more exposure to dissimilar in by-storybasis. Particularly light of residentialbalpolitical views than do contacts with neighbors (see, e.g., Mutz and this kanization, questionis worthreopening.Although Mondak 1998). Nonetheless, place of residence has become more of the media are often criticizedfor presenting a very a lifestyle choice (Katznelson and Weir 1986), and increasing resibiased(e.g., Schiller1986)or at leastverynarrow range dential segregation has been noted according to race (Harrison and of opinionsand arguments publicissues (e.g., Hallin on Bennett 1995), education (Frey 1995), age (Frey 1995), and income communication (Levy 1995). 1986), it is doubtfulthat interpersonal 98

AmericanPoliticalScience Review environments any less parochialor providegreater are diversity. The idea that the mediamayserveas an extensionof a geographically defined social context has been sugbut typicallythe emphasis is on how gested before, mass media exposurediffersfrom face-to-faceconversation (e.g., Sennett [1977] 1992). An obvious difference is the extent to which they allow interactivity. Nonetheless,if we allow that despite their noninteractive naturemedia mayhave the capacityto accomplish some of whatMill and othersconsiderbeneficialabout then exposureto dissimcommunication, cross-cutting ilar views via the media deservesattention.It is worth benefitsof exposure notingthat all three of the primary that Mill outlined-persuasion to a normatively better of view,deeperunderstanding one's ownviews,and the enhanced legitimacyof political decisions-are possible without face-to-faceinteraction.We do not mean to suggestthat interactivity of no value;rather,even is withoutit, the media may make a significant contribution. It is possible to separate the broader issue of whether interpersonalor mass communicationdoes more to advance the causes of democracyfrom the questionof whichkindsof channelsbest servethe need for exposureto cross-cutting We politicalperspectives. focus on this admittedly limitedbut nonethelessessential componentof politicalcommunication.

Vol. 95, No. 1 newspapershave readilyrecognizedpoliticalcomplexions that facilitate selective exposure to like-minded politicalviews. Likewise,the politicaltone of national television news is very similar across channels. Of course,the same cannotbe said aboutnewsmagazines, talk shows,and politicalweb sites. In more specialized media,people maywell be able to select a news source that sharestheir politicalbent. Aside fromthe abilityto choose, mediamayproduce less of a desire to exercise selective exposureas well. People often refrain from political discussionswith heterogeneousothers to avoid normativesocial pressure or the discomfortof public disagreement(Bennett, Fisher,and Resnick 1994;Ulbig and Funk 1999). These same people may be willing to expose themselves to media presentations,however,preciselybecause there is no personalinteraction. To summarize, hypothesizethat people are more we to expose themselves to dissonant opinions likely communithroughmediatedratherthan interpersonal cation,largelybecauseof the lack of selectiveexposure involved.After describing researchdesign,we will our evaluatecontributions the media and interpersonal of communication cross-cutting to exposureand present resultsfrom three tests of our general thesis.

RESEARCHDESIGN
We used datafroma representative nationaltelephone survey sponsored by the Spencer Foundation and executedby the University WisconsinSurveyCenter of in fall 1996, immediatelybefore the presidentialelection. This survey included a battery of items that tapped the frequencywith which respondentstalked about politics with up to three political discussants, plus five separateitems assessingthe extent to which they agreed with the views of each discussantnamed (see AppendixA). These five itemswere combinedinto an additive scale that measured the extent to which people'snetworksexposethemto politicalviewsunlike their own.2The sample of 780 respondentsprovided information more than 1,700discussants. obtain on To informationabout social contexts in which weak ties are especiallylikely, we asked respondentsthe same five questions with respect to people they know whichyielded associations, throughworkandvoluntary two similarindicesof exposureto politicaldifference in these contexts.Finally,the samebattery askedwith was reference to the views respondents encountered throughreadingnewspapers, watchingtelevisionnews, readingnews magazines,and watchingor listening to talk shows,afterinitialscreeningfor use of a particular medium.3
Eachof the five itemswas standardized then combinedinto an and additiveindex for each discussantand for each media source. To facilitate comparisonsacross informationsources, they were also with respect to the grand mean across all potential standardized sources of exposure to dissimilarviews. For the three primary Cronbach's thatthese fiveitems politicaldiscussants, alphaindicated scaledrelatively well:.78, .81, and .81 for the first,second,and third discussant, respectively. 3 Again,these itemsscaledacceptably: alphasof .73, .73, .69, and .81
2

IMPLICATIONS AVAILABILITY OF AND SELECTIVITY

We hypothesize that mainstream news media will in communication their capacity surpassinterpersonal to expose people to cross-cutting politicalperspectives for two reasons.First, there is a greateravailability of dissimilarviews in Americans' media environments thanin theirphysicalenvironments. Second,compared to personal interactions,people have less ability and desire to exercise selective exposure to news media content. With respect to the availability dissimilar of views, the media clearlyhave an advantageover faceto-face communication. Mainstream reportersare generally encouraged to illustrate stories with frequent referencesto people or groupswho expressconflicting views,in the typicalpoint-counterpoint format,and as an appeal to large audiencestend to cover a range of opinions (e.g., Zaller 1992). As businesses,American news media are certainlynot insulatedfrom pressures to reflectpublicopinion,but nationalpoliticalnews in local newspaperstends not to reflect local opinion (Dalton, Beck, and Huckfeldt 1998). In other words, the news media are not subject to the more narrow geographic constraints of face-to-face relationships, and they do not reflectthe structurally dictatedhomoor geneity of immediateneighborhoods communities. With respect to selective exposure, it is easier to avoid exposureto the views of personalacquaintances than to views expressedin national news. With daily newspapers and most national television news, the is abilityto exercisechoice on the basis of partisanship have more thanone severelylimited.Few communities daily newspaperfrom which to choose; moreover,few

99

acrossLines of PoliticalDifference Communication Facilitating Althoughthe sampleis relativelysmall,the Spencer survey provides a great depth of informationabout exposureto politicaldisagreementthroughboth mass and interpersonalchannels. Moreover,multiple indicators of the dependentvariablemake it possible to create indices that offermore reliablemeasuresof the extent to which a given source provides exposure to oppositionalviews, as well as the extent to which it provides dissonant contact independent of the frequencyof that contact. The survey results were supplementedwith data from the American and British components of the Cross-NationalElection Project (CNEP), obtained duringthe 1992 elections (for details on these studies, see Beck, Dalton, and Huckfeldt 1992; Heath et al. 1992). Although the CNEP data provide only one single-itemmeasureof the extentof expocomparable sure to disagreement throughmedia and interpersonal allow us to replicate and extend our channels, they initialfindings usingdatathatcapturebothperceptions of information sources' views (those of up to five and television)and indediscussants, plus newspapers of pendentassessments the extentof politicaldisagreement, which were made possible by a "snowball" sample of respondents' discussants,plus a content newspapers. analysisof respondents' Aside from the single-item (based on choice of presidentialcandidate)measure of political disagreement,the CNEPdiffersfromthe Spencersurveyin one other respect.The CNEP asked respondentsto volunteer the names of four people with whom they diswhereasthe Spencersurcussed "important matters," asked for people with whom they talked about vey "government,elections and politics." For the fifth discussant in the CNEP questionnaire,respondents were asked with whom they talked most "about the events of the recent presidentialelection campaign," which generateda more explicitlypolitical discussion partner.Research on name generatorssuggests that the explicitlypolitical frame will produce more nonwithwhomthere areweakties relativesand discussants and Sprague 1995b). Thus, the Spencer (Huckfeldt who will surveyis more likely to generate discussants to be politicallydissimilar the respondent.

March2001

SOURCES OF EXPOSURETO POLITICAL VIEWS CROSS-CUTTING

Figure 1, which is based on the combinedindex constructedfromthese five parallelmeasures,summarizes overall levels of exposureto dissimilarpoliticalviews by mass and interpersonal information sources. Whether the items are considered separately or in summary form, the findings are highly robust. As hypothesized, mainstream news media, especially newspapersand television,occupy the highest end of the disagreement continuum,followedcloselyby news magazines and more distantly by talk shows. This patternmakesa greatdeal of sense in light of people's
for newspapers, national television news, news magazines, and talk shows, respectively.

relativeabilityto exerciseselectiveexposureto each of that all the media these media. It is more surprising sources, including news magazines and talk shows, surpassinterpersonalsources in the extent to which respondentsperceive them to involve views substantiallydifferentfrom their own.4 The degree of intimacybetween main respondents and their discussantsfollows a highlypredictablepattern: Closeness is inversely related to exposure to conflictingpoliticalviews.Exposureto dissonantviews is most likely with casual acquaintances (mean exposure to dissimilarviews = .12), followed by friends (mean = -.24), close friends (mean = -1.11), and spouses or relatives(mean = -1.76). This significant linear trend indicatesthat cross-cuttingexposuredepends criticallyon contact with people who are not close friends or family (F = 18.35, p < .001). Moreover, as found in previous studies, the most frequent interactions tend to occur with the most politically homogeneous discussion partners (Huckfeldt and Sprague1995a). Even items that asked in very general terms about the people with whom respondentsdiscuss politicsthose known throughwork or a voluntaryassociation (rather than a named discussant)-did not generate levels of dissimilarity high as those for mediated as sources. These general items purposely directed reweakties thatwould attentionto potentially spondents' of have the greatestprobability puttingthem in contact with dissimilar views, but the responsesstill suggested that these sourceswere relativelyhomogeneouscomparedto the media.Likewise,an indexof disagreement drawnstrictlyfrom nonrelativediscussantsconfirmed that both newspapersand television expose respondents to significantlygreater political disagreement than do interpersonal discussants. The paired t-tests in the note to Figure 1 make it clear that these mean differencescannotbe explained by political or demographicdifferences among the users of variousinformation sources;in each case, the users of any two sources are comparedto themselves. In all possible comparisons of media sources and were less likely to expose discussants,the discussants politicalviews. respondentsto dissimilar Paired comparisonsmaximizethe sample size for for each of the tests,but they are inappropriate hypothesis testing because they do not adjust the observed significancelevels for the fact that so many comparisons are being made. Thus, to test our main hypothemedia providemore exposureto sis-that mainstream views than face-to-facecommunication-we dissimilar used a repeated measures analysis of variance;the
Given the progressive decline in concord with sequentially named discussants shown in Figure 1, it is possible that a fourth, fifth, or sixth discussant would have even more heterogeneous views, but it appears that few people can name that many political discussants. In our sample only 31% of respondents named a third person. In studies that ask for five political discussants, the proportion who do not name a fourth and fifth increases sharply (e.g., Huckfeldt and Sprague 1995b), thus making it unlikely that asking about additional discussants could overcome the large difference that we have observed.
4

100

AmericanPoliticalScience Review

Vol. 95, No. 1

1. FIGURE Perceived Exposureto DissimilarPoliticalViews by Source 2.00 --

>
O

1.50--

0
O

1.00

C') c)

0.55 -o o
a

0.00 Discussa
t

c)

Discussant

Discussant

olar Assocaon

Workplace

TalkShows

News

Magazines

Television News

Newspapers

-1.00-

-1.50
Source: Spencer survey, 1996. Note: Sample sizes are below in parentheses,withtwo-tailedsignificancelevels as noted. Statisticaltests are based on pairwisecomparisons.

Source Disc. 1 Disc. 2 Disc. 3 Vol. Assn. Workplace TalkShows News Mags. TV News Newspapers

(n) (715) (606) (446) (597) (502) (209) (151) (478) (640)

Disc. 1 .246 .069 .000 .000 .000 .000 .000 .000

Disc. 2 .453 .000 .000 .000 .000 .000 .000

Disc. 3

Vol. Assn.

Workplace

Talk Shows

News Mags.

TV News

.015 .003 .031 .000 .000 .000

.049 .028 .001 .000 .000

.829 .002 .000 .000

.246 .000 .001

.053 .061

.064

sourcewas designatedas a within-subjects information factor, and simple contrasts were done between a reference condition (interpersonalexposure to crosscuttingviews) and mediatedexposureto cross-cutting views.This procedureeliminatesthe problemof inflating the error rate with multiple t-tests, but its disadvantagewhen used to compareall meansin Figure1 is that the sample size for the analysisis constrainedto the smallestgroup of media users. In order to retain a more representative sample,we ran these comparisonsfor interpersonalcommunication versus newspaperreaders and television viewers more challenging only, andwe used threeprogressively standardsfor interpersonalexposure to cross-cutting

views:(1) the averagedissimilarity viewswithinthe of of network,(2) the dissimilarity respondent'spolitical the third(mostpolitically discussant named, dissimilar) and (3) the dissimilarity views among coworkers. of The omnibusF tests for all three analyseswere highly significant(F = 214.35, 54.77, and 40.39, respectively,p < .001 in all cases), as were the individual contrastsbetween interpersonal exposureversus television news (F = 263.30, 69.21, and 39.84, respectively,p < .001 in all cases) and interpersonal exposure versusnewspapernews (F = 301.65, 71.35, and 66.02, respectively, < .001 in all cases). Finally,we p also includedincome,education,politicalinterest,and partisanshipin those models to see whether the gap 101

acrossLines of PoliticalDifference FacilitatingCommunication

March2001

2. FIGURE Exposureto DissimilarPoliticalViews by Source and Partisanship

3.5-

> > 3.0 -O 3.0-2.5 -0

(D

tztsstttt

U Averagenameddiscussant U Voluntary association Workplace E] Television MNewspapers

2.0
Oo
... .. ..
?

w
0) a o
0)

........ ... .. ..

0.0-

0)

-0.5 t--~
-1.0 - -

x 0) -)

-10dte.5 -t

iI

Republicans

Independents

Democrats

Source: Spencer survey, 1996. sources (averagenamed or Note:Within each of the threesubgroups,allcomparisonsbetween mediasources (television newspapers)and interpersonal different and in the hypothesizeddirection(p < .05). association discussants) were significantly discussant, workplacediscussants, or voluntary

was especiallypronouncedin particular subsetsof the Unlike snowballsamples, "real"or largelyperceptual.6 The observed gap persisted in all cases, which ask named discussants about their political population. although partisanshipwas a particularlyinfluential views, or content analyses of media messages, our measures examined thus far do not tap the actual covariate. As Figure 2 illustrates,there are importantdiffer- similarityor dissimilarityof views but respondents' ences by party:Democrats tend to find mainstream perceptions.These perceptionsare subject to distornews sources more agreeable, whereas Republicans have significantlymore homogeneous interpersonal networks.These two tendencies combine to increase 6 Another possible challenge to our interpretation is that, for people are the mainstream, measures of disagreement the size of the gap between media and interpersonal whose views fromoutside or candidate constructed party preference questions may not but viewsamongRepublicans, it exposureto dissimilar adequately operationalize exposure to disagreement. Many critics of is both sizableand statistically the press highlight not so much an imbalance in presenting mainsignificant amongDemstream Republican or Democratic views as a dearth of more radical ocrats and independentsas well.5 In other words, the perspectives on either side of the spectrum. It should be noted, fundamental of mediain exposingpeople to advantage however, that three items in the index make no reference to parties dissonantinformation transcends partisanship. or candidates, and the pattern of findings is virtually identical using The differences betweenextentof exposureto polit- an index comprised of only these three items. Thus, we do not believe ical dissonancethroughmainstreamnews media and that the components of the index addressing only mainstream through interpersonalnetworksare clearly large, ro- political views account for the findings.the indices are comprised of It is also possible that, even though but bust, and statistically significant, one might ques- identical items for media and interpersonal communication, a word tion whether the pattern observed in these data is such as "often" means something very different when applied to
5 A repeated measures analysis of variance strictly among Democrats that simultaneously compared the interpersonal discussant average with exposure to difference through television and newspapers produced very strong findings (omnibus F = 51.68; contrast between personal network and TV: F = 62.39; contrast between personal network and newspapers: F = 77.06; all p < .001). media versus interpersonal political communication (Schaeffer 1991). Relative frequency is involved in only one of the five items in the index. In addition, this argument should work against our hypothesis, because phrases indicating relative frequency will tend to mean larger absolute frequencies when the activity is more frequent, and interpersonal exposure to political news is less frequent than mediated exposure.

102

AmericanPoliticalScience Review

Vol. 95, No. 1

TABLE Accuracyof Respondent Perceptions 1.

Respondents Perceive Discussants to IndependentAssessment by Favor Actual Discussant or Newspaper No One Bush Clinton Perot Coders A. Perceived versus IndependentlyAssessed Agreementand Disagreement 7% 12% 43% 83% Bush 7 35 19 81 Clinton 14 9 9 63 Perot 2 4 6 8 No one 100% 100% 100% 100% Percentage Numberof cases (65) (186) (400) (458) B. Accuracy and Inaccuracyby Respondents' Views 8% Inaccurate/strongprojection 4% Inaccurate/weakprojection 78% Accurate perceptions 4% Inaccurate/weakcontrast 5% contrast Inaccurate/strong 99% Percentage Numberof cases (1,036) Respondents Perceive Newspaper to Favor Bush Clinton Perot No One

0% 71 4 25 100% (112)

0% 72 1 27 100% (364)

0% 67 0 33 100% (3)

0% 61 2 37 100% (367)

4% 17% 48% 27% 4% 100% (788)

ElectionProject,American Source: Cross-National 1992). component (Beck, Dalton,and Huckfeldt, was fromboth mainrespondentand discussant Note: ForPartA, the unitof analysison the left is the total pool of dyads forwhichinformation available on interviews; the rightthe pool is all mainrespondentswho read newspapersfor whichcontent analyses were available(see AppendixB). ForPartB, the unitof analysis is the same but sample sizes are slightlylower because the perceptionsof respondentswith inaccurateperceptionswho did not In when a respondentmisperceives express a preferencecould not be categorizedas contrastor projection. PartB, we code a perceptionas projection this the newspaper/discussant agree withhis/herown views but independentassessments suggest thatthey do not agree. We characterize as strong to when the respondent prefers one candidate but the newspaper/discussantactually prefers a completely differentcandidate, and weak when the to candidate.We code a perceptionas contrastwhen a respondent newspaper/discussant favorhis/herown preferred respondentmisperceivesa neutral this to misperceivesa newspaper/discussant disagree when independentassessments suggest that they agree. We characterize as strong when they favor the same candidate but the respondent perceives the newspaper/discussantto favor a differentcandidate, and weak when the respondent misperceives a neutralnewspaper/discussantto favor a differentcandidate from his/her own preference.Percentages may not add to 100 due to rounding.

tion and inaccuracy;for example, a great deal of interpersonaldisagreement goes unrecognized, and errors sometimes occur in recognizingagreement as well (Huckfeldtet al. 1995). For our purposes,we considerperceptualmeasures of disagreementmore appropriatethan actual meato sures. In order for disagreement stimulatethe kind citedby Milland of thoughtprocessesandreevaluation others, it must be perceived as such. If people so misperceiveothers'views as to considerthem to be in agreement,then they will not be promptedto reconand sidertheiropinions,broadentheirperspectives, so forth. Likewise, if people perceive that others disagree-even if this is not the case-then the same benefits may derive despite the lack of objectivedisagreement.Thus, in one importantsense, it may be meaningless to ask whether people are "really"exposed to views differentfrom their own. If discussants are hesitantto maketheir differenceof opinionknown to friends,or if people persistentlyignore differences to that are communicated them, then such interactions contributelittle to the kinds of benefitspoliticaltheoristshaveproposed.Moreover,Huckfeldtand Sprague were influencedexclu(1995a) found that respondents sively by the views they perceivedtheir discussantsto have, not by their actualviews. Nevertheless,have we observedanythingmore than of a conglomeration systematic perceptualdistortions?

The question is importantbecause research in two areashighlights systematic the distortionin the perceptions of others' views that can result from strong partisanleanings.False consensusstudiessuggestthat the people will overestimate extent of their agreement with others, and studiesof the hostile media phenomenon indicatethat people sometimesoverestimatethe with the mass media extent of their true disagreement and Chaiken1994;Vallone, Ross, and (Giner-Sorolla Lepper 1985). Separately or in combination, these perceptualdistortionsmay account for our findingof greater exposure to perceived political disagreement through the mass media than through interpersonal communication. Drawingon the CNEPdata,Table 1 showsthat most perceptionsof interpersonalagreementare accurate; overall, for all respondent-discussant dyads, respondents correctlynamed the politicalviews of over 70% of their discussants. sorted based on whetherthe We was distortion in the directionof the mainrespondent's view (projection)or awayfrom it (contrast).We then classifiedthese cases as strong, if the misperception was the oppositeof the discussant's preference,or true weak, for example,when a true independentor undecided discussantwas perceived as a Democrat by a Democraticrespondent. As the bottomleft panel of Table 1B shows,roughly 12%of respondentserredin the directionof their own 103

acrossLines of PoliticalDifference FacilitatingCommunication

March2001

3. FIGURE Perceived and Independently Measured Exposure to Dissimilar Political Views by Source 2.00 -1.80

MPerceiveddisagreement
1 1.60-

(.
iu

Independentassessment

1.40

CO

Q 0.800 0

0.20

0C x 0.40--

0.20-0.00 Discussant Discussant Discussant Discussant Political Television Newspapers 1 2 4 3 Discussant News
Source: Cross-National ElectionProject,American component (Beck, Dalton,and Huckfeldt1992). Note: No independentassessment of televisionnews content was available.

views, and 9% in the direction opposite their own views. The results suggest significantprojection(t = 2.72, p < .01), consistentwith much of the psychological literature on this topic (e.g., Fabrigar and Krosnick1995;Kruegerand Clement 1994).7In other more errorsoccurredin the diwords, systematically rection of perceivinggreaterinterpersonal agreement (projection)thanwas actuallythe case,just as the false consensushypothesispredicts.The bottom line is that people are exposed to more interpersonaldisagreement than they recognize, but because they fail to recognizeit, it probablyhas little capacityto produce the beneficial effects of cross-cuttingexposure.Conthere is versely,as shownin Table 1A for newspapers, a tendencyto perceivegreaterdisagreement (contrast) withthe mediathancontentanalysessuggeststo be the case.8 For newspapers,21% of respondentsshowed signs of projection, and a significantlygreater 31% showedsigns of contrast(t = 3.56, p < .001).
7 To test whetherthe errorswere randomor systematically the in direction respondents' views,we assigned of own valuesof -2 to +2 to the contrast projection and scores,withaccurate perceptions equal to 0. We then tested the hypothesis that the meanwas equalto 0. for thiscomparison becausecontentanalysis is 8 We use newspapers available.

In Figure3, the CNEPU.S. dataseem to confirmthe patternobservedin the Spencersurvey,in that people perceive media to expose them to more oppositional As viewsthando theirpersonalnetworks. shownby the solid bars, this is true across the four "important as matters"discussants well as for the explicitly"political"discussant. The stripedbarsin Figure3 represent independentassessmentsprovidedby discussantsand newsdrawnfrom independentcodingof respondents' papers.9 Figure 3 suggests that the true extent of views maybe slightlyoveresexposureto oppositional timated in the case of newspapers and somewhat in underestimated the case of interpersonal relations. But most important,even if one uses the "real"measuresbased on discussant reportsand contentanalyses, still providesignificantly greaterexposure newspapers to viewsdifferent fromthe respondents' own (t = 5.74, p < .001). In other words, the advantageof news media over interpersonal channelsin relayingpolitical
favoredone candidateor none was estabpublication lished by runninga repeated-measures analysisof varianceon the articles coded for each newspaperto see whether there was a and, if so, in what statistically significantdifferencein favorability direction.If no statistically difference significant emerged,the newsno paperwas coded as favoring one.
9 Whethera

104

AmericanPoliticalScience Review is disagreement real, not simplya matterof perceptual distortion. Althoughthe Spencerdata do not includeindependent measuresof exposureto disagreement source, by the results are consistentwith the CNEP findingthat is the relativemediaadvantage not simplya functionof distortion.The perceptualdistortioninherperceptual ent in both false consensusand hostile media effectsis driven by partisanship;in the former case, it leads people to perceivegreatersupportfor their own views and than trulyexistsin their environment, in the latter case, it leads them to perceive greater opposition. Anotherway to graspthe extent of partisandistortion in the perceptualdata is to examinewhetherthe same relative advantageof news media over interpersonal sources exists among political moderatesor nonpartisans.'0Figure2 illustratedthat even independents(or those withoutideologicalleanings),that is, people with little or no partisan predisposition to distort their of impressions conflictand consensus,believe the news media expose them to dissimilarpoliticalviews more t than interpersonal sources(newspapers: = 6.97, p < television:t = 6.61, p < .001). .001; Figure 4 makes an additionalpoint concerningthe relationshipbetween partisanextremityand exposure to cross-cuttingpolitical perspectives.If we were to plot the hostile media hypothesis on one of these a graphsit would be U-shaped;studies demonstrating hostile mediaeffectshow that partisansat eitherend of will the spectrum reportmore exposureto oppositional views, even when the exact same media content is involved (Vallone, Rose, and Lepper 1985). If a medium were instead perceivedto be more congenialto one side of the political spectrumthan the other, we wouldexpecta linearpattern:People at one end of the spectrumwould find it least dissimilarto their own views, people at the other end would consider it dissimilar,and most independents and nonpartisans would fall somewherebetween. Figure4 plots all the availableindicesof exposureto disagreement,for both mediated and interpersonal In channels,across levels of partisanship. additionto showing the means for each level of partisanship,it revealseither linear or curvilinear patternsdepending on whether trend tests across the means indicated a significantlinear or curvilinearcomponent,with the highest order significantpatterndisplayed.In the top panel of Figure 4, for all media sources there is a significant linear pattern; compared to Democrats, newsas Republicans perceivetelevisionandnewspaper exposingthem to a greatmanymoreviewsunliketheir own. For talk radio, the pattern is reversed but still to linear:Democratsare more likelythan Republicans with find that the views expressedare in disagreement fall theirown, and independents in the middle.Despite the relative consonance of views that Republicans perceivein talk shows,they believe this mediumoffers
10 Independentsand those who refuse to identify themselves in partisanterms may be quite different,but examinationof these groups separatelyand collectivelyproducedthe same pattern of results.

Vol. 95, No. 1 dissimilar views to a greater extent than do their interpersonal networks (repeated-measuresanalysis for televisionversuspersonalnetworkaverageamong Republicans,t = 2.44, p < .05). Althoughthe earlierfindingof greatercontrastthan of contentseemed projectionin perceptions newspaper consistentwith the hostile media hypothesis,Figure4 makes the furtherpoint that it would be erroneousto describethis patternas one in whichpartisanson both ends of the political spectrumsee the news media as more hostile to their views than do independents(cf. DemocratsperDalton, Beck, and Huckfeldt1998).11 ceive newspapersand television news to be relatively agreeablebut, as shown in Figure 2, Democrats still considerthese sourcesmore dissonantthantheirinterpersonalnetworks. As shownin the bottompanelof Figure4, in contrast assessto the linear patternsfor media, interpersonal ments of exposureto dissimilar viewsconsistently yield an invertedU-shape.This indicatesthat, as compared to independents,strong partisansare less exposed to views unliketheir own. The patternholds for both the index for average discussants and contacts in the workplaceor throughvoluntaryassociations. If people successfullyexercise selective exposure, then one wouldexpectto see a patternin whichstrong partisansare most likely to be exposed to consonant information,and weaker or nonpartisansless so. Indeed, the fact that this patternis true for interpersonal medialends supportto sourcesbut not for mainstream our argumentthat selectiveexposureis less influential in shaping people's exposure to mediated than to information. There are, of course, other interpersonal betweenstrongandweakpartisans, it is but differences worth noting that the generallynarrowerlatitudesof partisanshouldworkagainst acceptanceof the strongly this finding;strong partisansshould find more views if expressedby mediaandby other people disagreeable only because not much of a differenceof opinion is to requiredfor strongpartisans perceivea viewpointas unlike their own (e.g., Sherif and Hovland 1961). Findings from the CNEP replicate these patterns. For television news and newspapers,perceived disagreement is a linear function of partisanship.For on communication, strongpartisans both interpersonal sides of the political spectrum are exposed to less disagreementthan independents.Moreover,this pattern remains curvilinearwhether based on respondents' perceptionsor the independentreports of discussants. In sum, our evidence suggests that most people in politicalharmony theirinterperperceivesubstantial sonal associationsand tend to misperceivedisagree11We use the same data as these authorsfor muchof our analyses, but we arriveat differentconclusionsabout a hostile media effect because the single variablefor partisanship (rangingfrom strong on Democratto strongRepublican) whichthey base their interprebetween a hostile media patternand the tation cannot distinguish we mediaperception observe,i.e., onlyRepublicans not (but partisan Democrats)perceivemediacoverageas more hostile to theirviews and Chaiken1994;Vallone, Ross, and Lepper (see Giner-Sorolla 1985).

105

acrossLines of PoliticalDifference FacilitatingCommunication FIGURE Perceived Dissimilarity of Views by Source and Strength of Partisanship 4.

March2001

353.0

MediaSources
1 %%6Talk

2.5
2.0 -oi1.5 o

(R2= 0.73) S.
+ +
*go **. A4-

Shows

Newspapers
(R2= 0.82)

W 0.5
-o

S+
?-?

TelevisionNews

0.0--0.5-1.0 -1.5 -2.0 +

-,,(R2

=0.95)

Strong Republicans

Independents

Strong Democrats
+ TalkShows + Workplace

X Newspapers *Television News Association Average Discussant 0 Voluntary 3.5 3.0

2.5 _Interpersonal

Sources

_g S2.0
0.0

1.5

so5
0.0 1

Workplace

-1. - (R2=0.99)
-1.5
-2.0

Discussant.
(R2 = 0.94)

Average

::/
I

:'-.
Independents Strong Democrats

Strong Republicans

106

AmericanPoliticalScience Review ment that may exist. Politicaldiversityis clearlynot a for people goal in social relationships most Americans; tend to cultivatehomogeneousinterpersonal networks, and those with strongpartisanattitudesare particularly by likely to be surrounded similarothers. In comparison, the mainstreammedia expose people to more political disagreement,regardlessof partisanshipor the exremityof their views.

Vol. 95, No. 1

TABLE The Effectof Availability 2. of Local Newspapers on Exposureto Multiple Dissimilar Views in the UnitedStates
Coefficient (s.e.) Morethan one newspaper availablein area RepublicanParty identification DemocraticParty identification
Education

t-value 2.20 .30 3.25

.26

-.43* (.20) -.07

-.01

(.22)
(.04)

THE MEDIAADVANTAGE EXPLAINING

Whatis it about media that enablesthese communication channels to more successfullyexpose people to cross-cuttingpolitical viewpoints? We suggest that people often seek political reinforcementfrom their informationsources, whether mass or interpersonal, but that the extent to which people are exposed to dissimilarviews througha given source reflectsvariation in their motivationsto exerciseselectiveexposure and the ease with which it is possible to do so. This hypothesiscan be tested in three differentways with the Spencer data. First, choice should facilitate less exposure to dissonant views. Therefore, one would to expectpeople with a choice of local dailynewspaper more with the views in the one selected than do agree those with no choice. Second, our theoryalso impliesthat findingsshould time and dependupon the mediacultureof a particular One wouldexpect,for example,that our findings place. would be time-boundeven in the United States;in the era of the partisan press it is unlikely that people received much except partisan reinforcement from of theirdailypapers(Schudson1981).Newspapers that period would not have fared any better than face-toface communicationin exposingpeople to cross-cutting views.Whilewe have no suitablehistoricaldata to such a comtest this hypothesis,we can approximate the contemporary United States parisonby comparing to a countrywhere the press is more closely aligned with partiesand politicalviews.Drawingon the British component of the CNEP, we compare exposure to cross-cuttingviews in American and British newspaWe networks. expectthat selecpers and interpersonal tive exposure is much easier when the media are should play a overtlypartisan,and Britishnewspapers less importantrole than the Americanpress in exposing people to dissimilar politicalviews. Third,we examine individualdifferencesin the exwith face-to-face tent to whichpeople are comfortable motivatedto exercise conflictand thusare differentially selective exposure.Althoughpeople in general dislike variationsin how strongly conflict,there are individual people are motivated to avoid it (Ulbig and Funk 1999). We hypothesizethat the media will be responsible for a particularly large proportionof a person's total exposure to dissonant views if he or she is with face-to-facedisagreement. uncomfortable

-.72** (.22)

Age Sex Race Income Constant

R2

(n)

.00 (.01) -.25 (.19) .37 (.28) .01 (.01) 1.56* (.71) .06 (460)

.49 1.34 1.34 1.80 2.19

Source: Spencer survey, 1996. is Note:Availability coded 1 forzip code areas withonlyone newspaper, 2 for areas with more than one. Entriesare ordinaryleast-squares regressioncoefficients,withstandarderrorsin parentheses.*p < .05,
**p < .01.

Local News Options

If the motivationto selectivelyexpose one's self influences political informationconsumption(despite the difficultyin fully realizingthis goal), then those who live in areas with a choice of local daily newspapers shouldbe exposedto fewer dissonantviews in the one they choose. Of course,the differencein politicalslant between local newspapers can be very subtle, but nonetheless the population in multiple newspaper towns has some degree of choice. To test our hypothesis, respondentsin the Spencersurveywere matched with informationabout the circulationof daily newspapers in their zip code area, as obtained from the Audit Bureauof Circulation. code areaswith only Zip one dailywere coded 1; those with more thanone were coded 2. Becausecorrelatessuch as income,education, or partisanship might produce a spuriousrelationship or maska significant between exposureto relationship dissimilarviews and access to multiple newspapers, controlvariableswere included.For example,markets with multiplenewspapersare likely to be better educated and urban,and these demographic characteristics may be associatedwith the likelihoodof exposure to diverseviewpoints. in As shownin Table 2, respondents areaswith more than one newspaperreported significantly less exposure to dissimilar views throughtheir dailynewspaper, and even after controllingfor partisanship other variables, althoughthe model as a whole accountsfor very little variancein exposureto disagreement.Given the often subtle differencesin the politicalcomplexionof newspaperswithin a locale, the modest size of the effect is not surprising.

107

acrossLines of PoliticalDifference FacilitatingCommunication

March2001

Views by Source and Country TABLE Influenceof Partisanshipon Exposureto Dissimilar 3.

PoliticalViews Exposureto Dissimilar UnitedStates Great Britain Discussions ThroughInterpersonal
Republican/Conservative Democrat/Labour -.34*** (.04) -.43*** (.04) -.27*** (.05) -.25*** (.05)

Frequencyof discussion Politicalinterest Education

Age Gender

.01 .04 -.01

-.00 -.08

(.06) (.04) (.01)

(.00) (.07)

-.08* (.03) .00 (.03) .04* (.02)

.00 -.09 (.00) (.07)

Initial log-likelihood Finallog-likelihood (n) ThroughNewspapers Republican/Conservative

Democrat/Labour Political interest Education

-1556.77 -1477.30

158.95***
(998) .14*** (.04) (.02)
(.05) (.01) (.07)

-1436.15 -1408.56

9*** 55.1
(1,074) -.28*** (.04)
-.10** (.04)

-.27*** (.04)

Frequencyof reading

.04

-.02

(.05) (.00)
(.06)

-.04 -.01 .03

Age
Gender

-.01** (.00) -1048.27 -991.84 112.86*** (1,033)

-.07** (.03) -.03* (.02)

-.00

-.01

Initial log-likelihood Finallog-likelihood X2 (n)

-1586.51 -1542.03 88.94*** (1,687)

and American ElectionProject,British Source: Cross-National components, 1992. views throughall the respondent'sdiscussants; for Note: For interpersonal discussion, the dependent variableis the average exposure to dissimilar are views throughthe respondent'smost often read newspaperon the same scale. Entries political newspapers,it is the extent of exposureto dissimilar < orderedprobitcoefficients,with standarderrorsin parentheses.Cutpointsare not shown. *p < .05, **p< .01, ***p .001.

betweenthe amount difference panel,the hypothesized of selective exposureexercisedin the two news enviwith more sharplydefinedpolitA media environment ronmentsis confirmed.As anticipated,British partiical differencesbetween newspapersprovidesa better sans on both sides are significantlyless likely than test of whether selective exposure is exercisedwhen Americansto be to exposedin their newspapers politconditionsmake it easier to do so. Britishvoters are ical views with which they disagree. In other words, well aware of the partisanslant in their newspapers, with their and there are many competingproductsfrom which they can and do find news sourcesthat mesh Justas with face-to-faceexpothey can choose (Curtice, Schmitt-Beck,and Schrott politicalpredispositions. in Britain promotes greater 1998, 12). If we are correctthat the largelynonpartisan sure, strong partisanship U.S. press makes it difficultfor people to selectively selective exposure. In contrast, the American case reveals a simple expose themselves to media content, then we would linear Democrats pattern:comparedto nonpartisans, a markedly different pattern among British expect are less likely to encounter dissimilarviews in their respondents.Because British partisansshould find it are but thanare nonpartisans, Republicans relativelyeasy to exerciseselective exposure,the rela- newspapers to more likelythan are nonpartisans do so. To confirm and tionshipbetweenpartisanship exposureto dissimilarviewsshouldtake the same curvilinear shape as for a general pattern of selective exposure, both party and need to be significant negative,but the that is, strongpartisanson both coefficients sources, interpersonal coefficientfor Republicansis significantand positive. sides should be less exposed to dissimilarpolitical In other words, given the opportunityto exercise views. between strength selective exposurethroughtheir choice of newspaper, Table 3 examinesthe relationship of partisanship and exposure to dissimilar views respondentscould do so in the highlypartisanBritish but throughthe media and interpersonalcommunication news environment, fewercould do so in the American case. The differingresults can be seen in the In for BritishandAmericanrespondents. the top panel we find in both samples the same pattern shown in marginals for these dependent variables as well. 4 for interpersonaldiscussantsin the Spencer Among newspaper readers, 51% in Britain read a Figure paper that shares their political leanings, but in the negativecoefficientsin the survey.As the significantly 35% in both United States the figureis only 16%.Conversely, top two rowsof Table 3 indicate,partisanship in the United States read a paper that explicitlydiscountriesreducesthe amountof cross-cutting interpersonal exposurethat people experience.In the bottom agrees with their political leanings,whereas the com-

A Partisan Media Environment

108

AmericanPoliticalScience Review parablefigurein Britainis only 20%.In both countries there is more cross-cutting exposurethroughnewspapers than through interpersonalnetworks, but the and differencebetweenthe extentto whichnewspapers communication provideexposureto disinterpersonal similarviews is significantly largerfor Americans(t = < .001). Selective exposure clearly occurs 4.58, p under the right real world conditions;when people have a choice, they tend to use it to reduce their politicalviews. exposureto cross-cutting

Vol. 95, No. 1

4. TABLE RelativeImportanceof the Media in Exposureto Dissimilar Views by Comfort with Controversy
Coefficient (s.e.) Comfortwith face-to-face controversy Total exposure to dissimilarviews (media average + average) interpersonal Politicalinterest (high) RepublicanParty identification DemocraticParty identification
Education

t-value 2.35

-.37*

(.16)

Comfort with Face-to-Face Controversy

In additionto limited opportunitiesto exercise selective exposure,the media advantageover interpersonal sources may be due to its impersonalnature, which of the circumvents discomfort face-to-facecontroversy. to Beyondfewer opportunities exerciseselectiveexpoto sure,Americansmayhaveless motivation do so with mass media. We hypothesizethat people will obtain a larger relative proportionof their total exposure to if dissimilarviews from interpersonalcommunication they are comfortablewith face-to-face conflict. As a test we used two surveyitems designed to assess the communication patternsencouragedduringearly political socializationin the family (see AppendixA for questionwordingand McLeodand Chaffee1972 for a reviewof the literature).For our purposes,it matters little whetherthese items tap accuraterecollectionsof parental behavioror reflect current attitudes toward In controversy. either case, we anticipatethat people with face-to-faceconflict are who are uncomfortable to get more of their dissonantinformationfrom likely with the media relativeto people who are comfortable conflict. interpersonal We compared the average level of exposure to disagreementin respondents'interpersonalnetworks with the averagelevel for television and newspapers, the subtracting formerfromthe latter.A highscore on this measureindicatesgreatermedia exposureto dissonance, and a low score indicates greater interpersonal exposure.Because these differencemeasuresare not independentof overallexposurelevels,we included a controlfor the total amount,thatis, the sumof media and interpersonalexposure. This allowed us to estimate the relativeimportanceof the media,controlling for the effect of overall exposure levels. Due to the obviousrole of generalpoliticalinterestin motivating exposure,it was also included. As shownin Table4, the total amountof exposureto dissimilarviews positivelypredicts the importanceof media relative to interpersonalsources. The relative importanceof media is especiallygreat for Republimore homogeneousintercans, who have significantly personalnetworksthan Democrats.In contrast,a high level of education makes the media less important sources.Most importantfor our vis-a-visinterpersonal hypothesisabout why people turn to the media is the controversyscale, which is a significantnegative preof dictorof the relativeimportance the media.In other words, as expected,people who are comfortablewith face-to-faceconflict get a relativelylarger amount of

.47*** (.05) .35 (.34) 3.74*** (.70) .58

-.26* - .65

9.21 1.02 5.33 .89

2.06

(.66)
(.13)

Age Income
Race

.01 (.02) .01* (.00)

(.82)

.40 2.30
.79

Gender Constant R2 (n)

-.67 (.56) 5.59* (2.36) .38 (305)

1.19 2.37

Source: Spencer survey, 1996. Note: The dependent variableis the relativeimportanceof the media network(averaged level). (averaged level) compared to interpersonal Entries ordinary are least-squaresregressioncoefficients,withstandard < < .05, ***p .001. errorsin parentheses.*p

ratherthan theirdissonantinformation interpersonally from the media. The opposite is true for people who find controversy unpleasant; they are more likelyto be exposed to dissimilarviews through the media. This finding supports our proposition that impersonality in Variability selechelps explainthe mediaadvantage. of tive exposurereflectsnot only the availability partisan information sources,as shownin tables2 and3, but also individual differencesin the desire to avoid interpersonalconflict.

DISCUSSION
across lines of political differenceis Communication basis of a pluralist essentialto the social psychological Pluralismrequires that a society be able to society. endureongoingpoliticaland moraldisagreement (e.g., Berlin 1969). Furthermore, accordingto most democratic theorists, the expressionof opposing political views is integral to the democraticprocess. Yet, the desire for harmony in interpersonal relationships makespoliticaldisagreement both relatively infrequent and particularlydifficult to convey successfully. In in addition,residentialbalkanization the United States is decreasingthe likelihoodof heterogeneousface-toface interaction.These factors make it all the more importantto understandthe avenues throughwhich people obtain exposure to dissimilarpolitical views. Ourfindingssuggestthatmediaare far moreimportant than interpersonalnetworks in exposing people to viewsunliketheirown. As a result,the mediahave the contribution potentialto make an extremelyimportant

109

Communication acrossLines of PoliticalDifference Facilitating to awarenessof diversepoliticalperspectivesand thus to nationalpoliticalintegration.12 Ourfindingsalso haveimplications manydebates for the media and publicopinion in the United regarding States. On the one hand, the stereotypesthat Lippmann(1922) fearedcapableof eliminating exposureto dissonantviewpointsaltogether appear to be potent but limitedforces.Selectiveexposureis exercisedwith contactsin the contemporary respect to interpersonal United States, to be sure, but at least as of the 1990s, of the structure mainstream mediamakesit difficult for to screen out large amounts of oppositional people information. the otherhand,cross-cutting On exposure of and intrapersonal awareness oppositional viewsdoes not necessarilytranslateinto the kind of face-to-face deliberation that theoristssuch as Dewey (1927) had in mind. Nonetheless, traditionalnews media such as and televisionnews hold the potentialfor newspapers views. creatinggreaterawarenessof cross-cutting like Dewey,have suggestedthat only Manytheorists, face-to-facedeliberationcan bring about the kind of democratic ends implied by the notion of a public sphere (e.g., Barber 1984; Fishkin 1991). Others emphasizethat, "in modernsocieties,publicdeliberation is (and probablymust be) largely mediated" (Page 1996).The usual rationaleis that the mass media may but be fine for relayinginformation, they are not very useful for fostering public discourse and collective action,becausethey are essentiallyone-waycommunication to people in spatiallydispersedand privatized settings (Carey1987).13 This argumentmakes considis erablesense if collectiveactionor consensus-building the outcome of interest.But if one focuses insteadon in the importance exposureto politicalperspectives of order to promote awareness of diverse viewpoints, politicaltolerance,and benefits of this kind, then the are inherent limitationson the media's contributions less obvious. The sheer idea that mass media might serve to the benefit of the public sphere strikesmost as heretical. there is a tenBut, in assessingmedia'scontribution, to rely on an ideal standarddrawnfrom interdency personal communication,typicallyakin to the ideal
12 Our findings are consistent across two data sets and two presidential elections, but this could be a function of similarities between the 1992 and 1996 elections. For example, the perception among Republicans and Democrats of media favoritism toward Clinton may be a function of his front-runner status in both cases. Nonetheless, our findings are consistent with many previous studies in suggesting that Americans perceive media content as to the left of their positions and consider journalists more liberal than the general population (Weaver and Wilhoit 1996). This does not necessarily mean that the content journalists produce favors one side, but the perception is that the mainstream media lean left if they lean at all (Watts et al.). 13Much has been made of the importance of direct interpersonal relationships in nurturing a sphere of politically oriented discourse, but a growing number of perspectives challenge that idea. Sanders (1997), for example, points out that the formal standards for the deliberative ideal are seldom, if ever, met. As usually practiced, face-to-face deliberative decision making may essentially perpetuate and exacerbate social inequalities. Schudson (1997) disputes the conversational ideal that pervades current writing on the quality of public life, arguing that democracy has little to do with the kind of intimacy and community that characterize this romanticized view of face-to-face interaction.

March2001

speech situation (Habermas1989) in which everyone has an equal opportunity participate,everyonelisto tens to and carefullyconsidersothers' views, and so forth (e.g., Fishkin 1991; Knightand Johnson 1994). This "conversational ideal" is not drawn from realworld experience,but it is often used as the standard against which media are judged and found wanting: "We are not really interested in what face-to-face communicationis like: rather,we have developed a notion that all communication shouldbe like a certain modelof conversation, whetherthatmodelreallyexists or not" (Schudson1982,43, emphasisin original). Instead, if we compare the reality of face-to-face with the realityof media popoliticalcommunication liticalcommunication, have a moresolidfoundation we on which to base conclusionsabout the contributions that both formsmaymake to a publicsphere.There is a disappointing tendencytowardhomogeneityin faceto-face political communication, whereas the media can transcend interpersonalgeography and expose people to views unlike their own. Regardlessof one's views on whether the media present a broad enough range of perspectives,it is clear that for manyAmericans they are the main source of exposureto crosscutting political views. Comparedto most interpersonal networks, they are hotbeds of diversity, not because the media are doing such an exemplary job pursuingdiversity,but because individualsare doing such a poor one, in partbecauseof the desireto avoid conflict in interpersonalsituations. The media send multiple conflictingmessages, and in so doing they advance an importantaspect of the democraticprocess. Of course, the patterns we have observed could simplybe interpretedas evidence of the media'slimited influenceon political attitudes.That is, if homogeneous interpersonalchannels persuade people to hold certain views, and the media are not equally persuasive,then the media will, as a result, appearto views.Even if this providemore exposureto dissimilar were the case, it still could be arguedthat media play an importantrole in furthering exposure cross-cutting in a way that interpersonalcommunication does not and perhapscannot.Previousstudies have found limited supportfor the socialcohesionviewof the effectof politicaldiscussants Huckfeldtand Sprague1995a; (cf. Kenny 1994), that is, that the cohesivenessof social But relationshipsdeterminestheir persuasiveness. reto gardlessof the relativepowerone attributes massor interpersonalcommunicationto persuade people to viewsthat changetheirmind,anyexposureto differing does not produceinstantaneous complianceis a valuable componentof publicopinionformationand of the viewswithina pluralist processof legitimizing differing system. In the contemporary United States,the majorproblem with looking to media to fulfill this role is the current trend toward highly specialized rather than mass channels.The breakupof the networkbroadcast monopoly,the proliferationof Internetnews sources, and the potentialthey offerfor tailoringnews to one's own interestsandprejudices only a few signsof this are

110

AmericanPoliticalScience Review
trend. As the number of potential news sources multiplies, consumers must choose among them, and that exercise of choice may lead to less diversity of political exposure. Although media observers have long lamented the lack of choice in news sources, the proliferation of choice creates new problems. Advertisers and media firms are now working together to create "electronic equivalents of gated communities" (Turow 1997, 2). The market segments they identify often are not based explicitly on political views, but "lifestyle" categories are hardly independent of political leanings. Turow (1997, 3) argues that "segment-making media," those which "encourage small slices of society to talk to themselves," are on the rise, whereas "societymaking media," which "have the potential to get all those segments to talk to each other," are on the decline. This leaves the future of the mass audience in question (cf. Neuman 1991). These developments obviously do not bode well for the future of communication across lines of political difference. In addition to reducing the amount of direct exposure to dissonant views, specialized media and fragmented audiences may have secondary negative effects on interpersonal communication, since shared viewing or reading may encourage conversations across partisan lines (Katz 1996). Ironically, the mass media's greatest strength may not be facilitation of the "Great Community" of solidarity envisioned in the early twentieth century by media philosophers such as Cooley and Dewey (Simonson 1996). Instead, media's greatest potential lies in its impersonal exposure of audiences to cross-cutting views, an essential form of communication in a highly pluralistic society. In order to sustain this benefit, however, news media must be structured so as to limit the public's capacity for selective exposure.

Vol. 95, No. 1

else. Who is the belongto, or theymightbe fromsomewhere person you've talked with most about politics? [Discussant #1] Aside from this person,who is the personyou'vetalked with most aboutpolitics?[Discussant Aside fromanyone #2] you'vealreadymentioned,is there anyoneelse you'vetalked with about politics? [Discussant#3]" If at any point the respondentcould not give a name:"Wellthen, can you give the firstname of the personwithwhomyou were most likely to have informalconversations duringthe courseof the past few months?"

MediaSource Generators
the Newspapers: "During pastweek,didyou readone or more Whichnewspaper was that?" newspapers? Television News:"During pastweek, did you watchany the nationalnews programson television?What nationalnews show was it that you watched?Whichnews programdo you watchmost often [if respondentnamesmore than one]?" NewsMagazines: "Duringthe past week, did you read any news or currentevents magazines,such as U.S.News, Time, or Newsweek, some other news magazine?Whichmagazine was that?" Political Talk Shows: "During the past week, did you happento see a talkshowon televisionor heara talkshowon the radiothat includedsome discussion politicalor social of issues?What talk showwas that?"

Associations and Workplace Voluntary Generators

and Respondentswere firstaskedaboutmembership particsimilarto associations, ipationin a list of types of voluntary those used in the GeneralSocialSurvey. Peoplewho claimed membershipin more than one such group were randomly assignedto one of the groupsfor a follow-upsequence,which asked identicallyworded questionsparallel to those asked about the politicaldiscussants. These same five items were also askedwith referenceto coworkers, clients,or customers that employed respondentsencounteredin the workplace. The five items askingaboutvoluntary associations produced an alphaof .74, andthe sameitemsreferring the workplace to scaled at .72.

APPENDIX THESPENCER A. SURVEY Design

This nationaltelephonesurvey conductedby the Univerwas sity of Wisconsin Survey Center from September 1996 throughelection eve using randomdigit dialing.Each number was screened to verify that it was associated with a household. The person selected for the interviewwas randomly chosen from among householdmembersat least 18 allowed.The response rate years old, with no substitutions was 47%, calculatedas the proportionof completedinterviews divided by total sample (which includes those who never answered and all other nonresponsesand refusals) minus the nonsamplenumbers.This is virtuallyidenticalto the rateobtainedin the CNEPsurvey. Interviews 25 averaged of minutes.A maximum 30 calls were made to each nonansweringor otherwiseunresolvedtelephonenumber.

Five-ItemIndexof Exposureto Dissimilar Views

1. "Compared with [nameddiscussant, media outlet, volunwould you say that your tary association,or workplace], politicalviewsare muchthe same [low],somewhatdifferent, or very different[high]?" or favorsRepublicans or association, workplace] normally same as responDemocrats,or both,or neither?" Scoring: dent'spartisanship fromrespondent's (low), different partisanship(high), or neither. 3. "Whichpresidentialcandidate,if any, does [named disor cussant,mediaoutlet,voluntary association, workplace] favor? Clinton,Dole, Perot, or some other candidate?" Scoring:same as respondent's preference(low), different from respondent's preference(high), or neither. 4. "Overall,do you feel [named discussant,media outlet, voluntaryassociation,or workplace]sharesmost of your views on political issues [low], opposes them [high], or doesn't [it/person's name] do either one?" 5. Mediaversion:"Howoften do you disagreewith political views you [read/hear]about in/on the [media outlet]?
2. "Do you think [named discussant, media outlet, voluntary

Discussant Generator
"Fromtime to time, people discussgovernment,elections, and politicswith other people. We'd like to know the first namesor just the initialsof people you talkwith aboutthese matters.These people mightbe fromyourfamily,fromwork, from the neighborhood, from some other organization you

111

Communication acrossLines of PoliticalDifference Facilitating

Often [high],sometimes,rarely,or never [low]?"Discussant version: "When you discuss politics with [named discussant,voluntaryassociation,or workplace],do you disagreeoften [high],sometimes,rarelyor never [low]?"

March2001

Closeness of Relationship
"Is [nameddiscussant]a close friend,just a friend, or just someonethatyou regularly comeinto contactwith?"Spouses and relativeswere coded into a fourthcategory.

Frequency of Political Discussion

Discussant Version: "Whenyou talkwith [nameddiscussant], do you discusspoliticsa lot, some, a little, or very rarely?" Media Version:"When you read/watch[named media storiesaboutpoliticsa lot, some, outlet], do you read/watch a little, or not at all?"

of 27 days were coded for each paper, depending on its circulation pattern.On each codingday all storiesthat dealt withthe presidential and and/orcandidates directly campaign that appearedin the firsthalf of the first news sectionwere coded, as well as any special election pages and campaignrelatedstorieson the editorialpages.Coderswere instructed as follows: "Code the overall content of the article that involves Bush, Quayle, or the Bush/Quaylecampaign in terms of its favorability unfavorability the Bush camor to of paign.Evaluatethe articlefrom the perspective the Bush campaignand assess the content of the article from this perspective.In other words,would the Bush campaignlike seeing this article in print?"(Dalton, Beck, and Huckfeldt 1992). The same instructions appliedfor Clinton/Goreand for Perot.We aggregated acrossall articlesin a newspaper to obtain an averagefavorability/unfavorability for each score candidate. were coded as favoring candidateif a Newspapers his scorewashigherthanthatof the othertwocandidates and the differenceswere statistically significantat the p < .05 level.

Partisanship
"Generally speaking,do you consideryourselfa Republican, or Democrat,independent, what?"

Discussant Generator
"Nowlet's shift our attentionto anotherarea. Fromtime to time, most people discuss importantmatters with other people. Looking back over the last six months, I'd like to know the people you talked with about matters that are importantto you. Can you think of anyone?What is this person'sfirstname?Is there anyoneelse you talkwith about mattersthat are importantto you?"Up to four names are accepted,then: "Asidefrom anyoneyou have alreadymentioned, who is the person you talked with most about the events of the recentpresidential election campaign?"

Political Interest
"Somepeople seem to followwhat'sgoingon in government and public affairs most of the time, whether there's an electiongoingon or not. Othersaren'tthatinterested. Would and you sayyou followwhat'sgoingon in government public affairs mostof the time, some of the time,only now andthen, or hardlyat all?"

Comfort with Controversy

"When you were growingup, about how often did your take the position that certaintopics [parentsor guardians] are better left undiscussed?Often, sometimes, rarely, or
4 = never.

Perceived Disagreement with the Media

were asked:"Inthe presidential election,which Respondents one of the candidates you prefer?" did Theyalsowere asked: "Which candidatedo you think[the newspaper] presidential favored duringthe campaign,or didn't it favor any candidate?"The two responses were compared, scoring as and was follows:0 = absoluteagreement(respondent newspaper and concur), 1 = mixed (either respondent or newspaperis and 2 = disagreement(respondent independent/neutral), and newspaper disagree).Televisionnewsmagazineand talk show scoringwas the same. For the British sample, similar items were used with or respectto the Conservative LabourParty,and scoringwas the same.

"Howoften did they have spiriteddiscussionson controversial matterslike politics or religion?Often, sometimes, rarely,or never?"Scoring:1 = never... 4 = often. The scalebasedon these two measuresrangedfrom2 to 8.

never?" Scoring: 1 = often...

APPENDIXB. CROSS-NATIONAL ELECTION AND PROJECTSURVEY: AMERICAN BRITISH COMPONENTS Design

The Americancomponentof the CNEPcombineda national named by main responsurveywith a surveyof discussants dents and a content analysisof 40 newspapersconducted election(for detailsand reliabilduringthe 1992presidential ity of content measures,see Dalton, Beck, and Huckfeldt 1998; Dalton et al. 1998). The national surveywas drawn from40 countyclustersaroundthe United States(48 states), and the major newspaperread by residentsin each of the countieswas selectedfor contentanalysis. The Britishsurvey componentwas includedas a modulein the BritishGeneral Election Cross-Sectionsurveyof 1992. Full details of the are surveydesign and administration given in Beck, Dalton, and Huckfeldt(1992)for the Americansurvey,and Heath et al. (1992) for the Britishsurvey. The newspapersamplingwas done every third day and every Sundayfrom LaborDay to Election Day; a maximum

Perceived Disagreement with Discussant

A comparisonwas made between self-reportedcandidate supportand the perceivedsupportof Clinton,Bush,or Perot "Whichcandidatedo you think by the named discussants: in electionthisyear?" [discussant] supported the presidential 0 and Scoring: = absoluteagreement(respondent discussant concur), 1 = mixed (either respondent or discussant is and 2 = disagreement(respondent independent/neutral), and discussant disagree). A similar scale with respect to the Conservativeand Labourpartieswas used for the Britishsample,and scoring was the same.To comparea singleindicator interpersonal of was exposureto disagreement, perceiveddisagreement avera aged acrosshowevermanydiscussants respondentnamed.

112

AmericanPoliticalScience Review

Vol. 95, No. 1

the Benhabib,Seyla.1992.Situating Self.New York:Routledge. Bennett,Stephen,BonnieFisher,and DavidResnick.1994."Speaking of Politicsin the UnitedStates:WhoTalksto Whom,Why,and WhyNot."Paperpresentedat the annualmeetingof the Midwest PoliticalScienceAssociation,Chicago. Berlin, Isaiah. 1969. Four Essays on Liberty.New York: Oxford UniversityPress. Mediaand Calhoun,Craig.1988."Populist Politics,Communication 6 LargeScale SocietalIntegration." Theory (3): 219Sociological 41. Carey,JamesW. 1987."The Press and the PublicDiscourse."The Center 20 Magazine (2): 4-32. and Curtice,John,RiidigerSchmitt-Beck, Peter Schrott.1998."Do the MediaMatter?" Paperpresentedat the annualmeetingof the MidwestPoliticalScienceAssociation,Chicago. Dalton, Russell J., Paul Allen Beck, and Robert Huckfeldt.1992. CNEP-USA CommonCoreMediaCodebook. Typescript. Dalton, Russell J., Paul Allen Beck, and Robert Huckfeldt.1998. "Partisan Cues and the Media: InformationFlows in the 1992 Presidential Election." American Political Science Review 92 (March):111-26. Dalton,RussellJ., PaulAllen Beck, RobertHuckfeldt,and William Koetzle.1998."A Test of Media-Centered NewsAgendaSetting: paper Content and Public Interest in a PresidentialElection." PoliticalCommunication (November): 15 463-81. New York:Henry Dewey, John. 1927. ThePublicand Its Problems. Holt. Eliasoph, Nina. 1998. AvoidingPolitics:How AmericansProduce Press. Apathyin Everyday Life. New York:Cambridge University LeandreR., and Jon A. Krosnick. 1995."AttitudeImporFabrigar, tance and the False Consensus Effect." Personality Social & Bulletin21 (5): 468-79. Psychology as In Fearon,James. 1998. "Deliberation Discussion." Deliberative Democracy,ed. Jon Elster. New York: CambridgeUniversity Press.Pp. 44-68. and New Fishkin,JamesS. 1991.Democracy Deliberation: Directions New Haven,CT:Yale University Press. for Democratic Reform. Freedman, Jonathan L., and David O. Sears. 1965. "Selective In in SocialPsychology, 2, vol. Exposure." Advances Experimental ed. LeonardBerkowitz. New York:AcademicPress.Pp. 58-98. Frey, Dieter. 1986. "Recent Research on Selective Exposure to Information." Advancesin Experimental In SocialPsychology, vol. Berkowitz. New York:AcademicPress.Pp.41-80. 19, ed. Leonard of Frey, WilliamH. 1995."TheNew Geography PopulationShifts: TrendsTowardBalkanization." Stateof the Union: In Americain the1990s,vol. 2, ed. ReynoldsFarley.New York:RussellSage.Pp. 271-336. Giner-Sorolla, Roger, and Shelly Chaiken.1994. "The Causes of HostileMediaJudgments." Journal Experimental SocialPsycholof ogy 30 (2): 165-80. Mark.1973. "The Strengthof Weak Ties."American Granovetter, 78 Journal Sociology (May):1360-80. of 1989.TheStructural Habermas, Jiurgen. Transformation thePublic of MA: MIT Press. Sphere. Cambridge. War":The Media and Hallin, Daniel C. 1986. The "Uncensored New York:OxfordUniversity Vietnam. Press. RoderickJ., and ClaudetteE. Bennett.1995."Racialand Harrison, EthnicDiversity." Stateof the Union: In Americain the 1990s,vol. 2, ed. ReynoldsFarley.New York:RussellSage. Pp. 141-210. Heath, Anthony,RodgerJowell,John Curtice,JackA. Brand,and James C. Mitchell. 1992. BritishGeneralElection Cross-Section Survey,1992 (computer file) (Study #6453). ICPSR version. London, England: Social and CommunityPlanning Research ESRCDataArchive/Ann Colchester, [producer]. Arbor, England: MI:Inter-university for Consortium Politicaland SocialResearch 1996. [distributors], Huckfeldt,Robert,Paul Allen Beck, RussellJ. Dalton, and Jeffrey Levine. 1995. "PoliticalEnvironments, Cohesive Social Groups, and the Communication PublicOpinion." of American Journal of PoliticalScience39 (4): 1025-54. Huckfeldt,Robert,and John Sprague.1995a.Citizens, Politics,and Social Communication: and Information Influencein an Election New York:Cambridge Press. Campaign. University Information Huckfeldt, Robert,andJohn Sprague.1995b."Political and Communication amongCitizens:HumanCapitalin an Elec-

Assessed Independently Agreement/Disagreement

For newspapers,assessmentis based on the candidatethat the respondentsupportedand the favorability score for the relevantnewspaperas determinedby content analysis.Scoring: 0 = absolute agreement (respondent and newspaper concur), 1 = mixed (either respondent or newspaper is and 2 = disagreement(respondent independent/neutral), and newspaper disagree). For discussants,assessment is

based on the respondent's self-report and the named discussants' self-reports, scored on the same three-point scale.

Accuracy, Projection,and Contrast

Accuracy,projection,and contrastwere measured by the to the abilityof mainrespondents identifycorrectly candidate supportedby either their named discussants revealedin (as the discussantsurvey) or by their local newspaper(as revealed in content analysis).If the perceptionsof the main respondentmatchedthe independentassessmentby coders
or discussants, then they were scored as accurate. If the respondent judged the discussant or newspaper to be in more agreement than was the case, the respondent was scored as projecting. If the respondent judged the discussant or newspaper to be in less agreement than was the case, the respondent was scored as contrasting.

AmericanPartisanship
(1) "Many people lean towarda particular politicalpartyfor a long time, although they may occasionallyvote for a differentparty. Do you generallylean toward a particular
party?" (2) "If you had to choose, do you think of yourself as closer to the Republican or the Democratic party?" (3) "Taken altogether, how strongly or weakly do you lean toward the [Republican, Democratic] party? Would you say

very strongly,fairly strongly,moderately,fairly weakly, or

very weakly?"

BritishPartisanship
Form A: "Generally speaking, what do you think of yourself as [names of political parties]? Do you think of yourself as closer to one party? Which? How strong [name party] are you?" Form B: "Generally speaking, what do you think of yourself as?" "Which of the following parties do you feel a little closer to?" "How strong [name party] are you?"

REFERENCES
MA: Altemeyer,Bob. 1996. TheAuthoritarian Specter. Cambridge, Harvard Press. University Past and Future: in Arendt, Hannah.1968.Between EightExercises PoliticalThought. New York:VikingPress. 1984.Strong Politics a Barber, Benjamin. Democracy: Participatory for NewAge. Berkeley:University California of Press. Beck, Paul Allen, Russell J. Dalton, and Robert Huckfeldt.1992. Cross-National Election Studies: United StatesStudy, 1992(computer file) (Study#6541). 2d ICPSRversion.Columbus, OH: PaulA. CA: Beck, Ohio State University/Irvine, Russell J. Dalton, UniIN: versityof California/Bloomington, RobertHuckfeldt,Indiana University [producers],1993. Ann Arbor, MI: Inter-university for Consortium Politicaland SocialResearch[distributor], 1995. Bellah, Robert N., Richard Madsen, William M. Sullivan,Ann IndividSwidler,and StevenM. Tipton.1985.Habitsof theHeart: ualismand Commitment American in of Life. Berkeley: University California Press.

113

acrossLines of PoliticalDifference FacilitatingCommunication

tion Campaign."Proposal submittedto the National Science Foundation. Katz, Elihu. 1968. "On Reopening the Question of Selectivityin In Exposure to Mass Communications." Theoriesof Cognitive WilliamMcGuire,TheodoreNewcomb,Milton Rosenberg,and Chicago: Rand-McNally. 788-96. PercyTannenbaum. Pp. and Notes on the Katz,Elihu.1981."Publicity Pluralistic Ignorance: Weisbaden: Westdeutscher plinger,andKurtReuman. Verlag.Pp. 28-38. Annalsof Katz,Elihu. 1996."AndDeliverUs from Segmentation." Weir. 1986."Working-Class FormaKatznelson, and Margaret Ira, tion: Constructing Cases and Comparisons." Working In Class ed. Formation, Ira Katznelsonand Aristide Zolberg.Princeton, NJ: PrincetonUniversity. 3-41. Pp. B. of Kenny, Christopher 1994. "The Microenvironment Attitude Glencoe, IL: Free Press. and Knight,Jack,and JamesJohnson.1994."Aggregation Deliberation: On the Possibilityof Democratic Legitimacy." Political Joachim,and RussellW. Clement.1994."TheTrulyFalse Krueger, and ConsensusEffect:An Ineradicable EgocentricBias in Social 596-610. PaulF., Bernard Lazarsfeld, Berelson,and Hazel Gaudet.1944.The People'sChoice.New York:Duell, Sloanand Pearce. and In Levy,Frank.1995."Incomes IncomeInequality." Stateof the Farley.New York:RussellSage. Pp. 1-58. New York:Free Press. Walter.1922.PublicOpinion. Lippmann, and Marcus,George E., John L. Sullivan,ElisabethTheiss-Morse,
Sandra L. Wood. 1995. With Malice Toward Some: How People Make Civil LibertiesJudgments. New York: Cambridge University Union:America in the 1990s, vol. 1: Economic Trends,ed. Reynolds Perception." Journal of Personality and Social Psychology 67 (4): Theory 22 (May): 277-96. Change." Journal of Politics 56 (August): 715-28. Klapper, Joseph T. 1960. The Effects of Mass Communications. the American Academy of Political and Social Science 546: 22-33. Spiral of Silence." In Public Opinion and Social Change: For Elisabeth Noelle-Neumann, ed. Horst Baler, Hans Mathias KepA Consistency: Sourcebook, ed. Robert P. Abelson, Elliot Aronson,

March2001

Page, Benjamin I. 1996. Who Deliberates? Mass Media in Modem

of Democracy. Chicago: University ChicagoPress. Deliberation." PoliticalTheory 25 Sanders,LynnM. 1977."Against (3): 347-76. Schaeffer,Nora Cate. 1991. "HardlyEver or Constantly?Group PublicOpinionQuarterly Comparisons Using VagueQuantifiers." 55 (Fall):395-423. York:OxfordUniversity Press. the Schudson,Michael. 1981. Discovering News. New York: Basic Books. Michael.1982."TheIdealof Conversation the Studyof in Schudson, MediaWorld, GaryGumpert RobertCathcart. ed. and New York: Oxford.Pp. 41-8. and Schudson,Michael. 1984. "Embarrassment ErvingGoffman's Ideaof HumanNature." and 633-48. Theory Society13 (October): Schudson,Michael. 1995. The Power of News. Cambridge,MA: Harvard Press. University Is Schudson,Michael. 1997. "WhyConversation Not the Soul of 297-309. Sennett,Richard.[1977] 1992. TheFall of PublicMan. New York: AlfredA. Knopf. and AssimSherif,Muzafer, CarlI. Hovland.1961.SocialJudgment: New Haven,CT:Yale University Press. Peter.1996."Dreams Democratic of ComSimonson, Togetherness: Studiesin Mass munication Hope from Cooley to Katz."Critical New York:Doubleday. Sullivan,John L., James Piereson, and George E. Marcus.1982. of ChicagoPress. Sweeney,Paul D., and KathyL. Gruber.1984."Selective Exposure: VoterInformation Affair." and Journal Preferences the Watergate of Turow,Joseph. 1997.Breaking America.Chicago:University Up ChicagoPress. Avoidanceand Ulbig, StacyG., and CaryL. Funk. 1999."Conflict Vallone, Robert P., Lee Ross, and Mark R. Lepper. 1985. "The HostileMediaPhenomenon: and BiasedPerception Perception of Media Bias in Coverage of the Beirut Massacre." Journalof Weaver,David H., and G. ClevelandWilhoit.1996. TheAmerican NJ: Erlbaum. Mahwah, Lawrence Watts,MarkD., DavidDomke,DhavanV. Shah,andDavidP. Fan. 1999. "Elite Cues and Media Bias in PresidentialCampaigns: of Press." Communication PublicPerceptions a Liberal Explaining
Research 26 (2): 144-75. Zaller, John. 1992. The Nature and Origins of Mass Opinion. New Journalist in the 1990s: U.S. News People at the End of an Era. Personalityand Social Psychology 49 (3): 577-85. Political Participation." Political Behavior 21 (3): 265-82. of Personalityand Social Psychology 46 (6): 1208-21. Political Tolerance and American Democracy. Chicago: University Communication 13 (4): 324-42. Stouffer, Samuel. 1955. Communism, Conformity,and Civil Liberties. ilation and ContrastEffects in Communication and Attitude Change. Democracy." Critical Studies in Mass Communication 14 (4): Mass Media." In Inter/Media:Interpersonal Communication in a Schiller, Herbert I. 1986. Information and the Crisis Economy. New

Press.

BasicBooks. McLeod,JackM., and StevenH. Chaffee.1972."TheConstruction Aldine-Atherton. 50-99. Tedeschi.Chicago: Pp. ed. Mill, John Stuart.[1859] 1956.On Liberty, C.V. Shields.Indianapolis:Bobbs-Merril. Social Networks:Testing Mutz, Diana C. 1999. "Cross-Cutting DemocraticTheoryin Practice." Paper presentedat the annual meetingof the MidwestPoliticalScienceAssociation,Chicago. J. at Mutz,Diana C., andJeffery Mondak.1998."Democracy Work: of Towarda PublicSphere."Paper Contributions the Workplace presentedat the annualmeetingof the MidwestPoliticalScience Association,Chicago. York:Cambridge. Nunn, Clyde Z., HarryJ. Crockett,and J. Allen Williams.1978.
Tolerancefor Nonconformity. San Francisco: Jossey-Bass. Neuman, W. Russell. 1991. The Future of the Mass Audience. New of Social Reality." In The Social Influence Process, ed. James

Mansbridge, Jane. 1980. Beyond Adversary Democracy. New York:

York:Cambridge Press. University Zillmann,Dolf, and JenningsBryant.1985. Selective Exposureto Communication. Erlbaum. Hillsdale,NJ: Lawrence

114