Computers in Human Behavior 27 (2011) 971–977

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier .com/locate /comphumbeh

The contribution of social network sites to exposure to political difference:The relationships among SNSs, online political messaging, and exposure tocross-cutting perspectives

Yonghwan Kim ⇑School of Journalism, College of Communication, The University of Texas at Austin, 1 University Station A1000, Austin, TX 78712, USA

a r t i c l e i n f o a b s t r a c t

Article history:Available online 8 January 2011

Keywords:Social network sitesInternetPolitical differenceExposure to cross-cutting viewpointsOnline political messaging

0747-5632/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.chb.2010.12.001

⇑ Tel.: +1 512 825 9772.E-mail address: y.kim@mail.utexas.edu

The increasing popularity of social network sites (SNSs) has raised questions about the role of social net-work media in the democratic process. This study explores how use of SNSs influences individuals’ expo-sure to political difference. The findings show a positive and significant relationship between SNSs andexposure to challenging viewpoints, supporting the idea that SNSs contribute to individuals’ exposureto cross-cutting political points of view. Partisanship was not found to interact with SNS use, suggestingthat SNSs contribute to expanding exposure to dissimilar political views across individuals’ partisanship.Online political messaging also has a direct effect on exposure to dissimilar viewpoints, and it mediatesthe association between SNSs and exposure to cross-cutting political views.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction for a summary of research on SNSs). While researchers have begun

Recent years have witnessed the coming to prominence of so-cial media such as social network sites within the political realmas well as everyday life (Lenhart, 2009; Smith, 2009b). Social net-work sites (SNSs, e.g., Facebook and MySpace) have been used bycampaign managers and ordinary citizens both as sources of infor-mation and as tools to broaden their campaign and viewpoints(e.g., Gueorguieva, 2008; Nielsen Reports, 2009).

The role of SNSs in providing information about politics or pub-lic affairs and providing an online space for citizens to expressopinions and participate in various activities has been increasing.In the 2008 US presidential elections, many people got informationabout candidates and the campaign from social network sites (ThePew Research Center for the People, 2008); nearly 1 in 10 of peopleunder age 30 signed up with or ‘‘friended’’ one of the candidates onsuch a site. Users not only got news and campaign informationfrom these SNSs during the election, they were able to post theirthoughts and comments. Thus citizens displayed a more active rolein the political process (Smith, 2009a).

The increasing popularity of the SNSs as information sourcesand space for political activities and social interactions necessitatesfurther research on the influence of such media on democratic pro-cesses. Studies have examined diverse influences of SNS useincluding impression management, network structure, privacy is-sues, and online/offline connections (see boyd and Ellison (2007)

ll rights reserved.

to explore intersections of SNSs and politics, extant scholarship hasfocused primarily on citizens’ participatory activities such as polit-ical and civic engagement and social capital (e.g., Ellison, Steinfield,& Lampe, 2007; Park, Kee, & Valenzuela, 2009).

One of the major academic debates over the role of the Internetis whether it contributes to or harms pluralistic democracy. Schol-ars have debated whether the Internet contributes to people’sexposure to diverse viewpoints, and whether it facilitates chancesof exposure to varied, ‘‘cross-cutting’’ political views (Mutz, 2002a,2002b; Wojcieszak & Mutz, 2009) or heterogeneous contacts (Bru-ndidge, 2010; Scheufele, Nisbet, Brossard, & Nisbet, 2004).

Exploring whether and how communication technologies affectindividuals’ exposure to political difference is important because ofits democratic implications. While exposure to similar points ofview may have important political consequences such as enhanc-ing attitude strength and mobilizing citizens in political activities(e.g., Schudson, 1995; Stroud, 2006), avoiding political differenceand ignoring opinion-challenging information may hurt democ-racy. If individuals expose themselves only to similar points ofview and avoid contrasting information and perspectives, theyare less likely to be tolerant of challenging viewpoints (Mutz,2002b), and society will be fragmented and polarized (Sunstein,2007).

This debate can be extended to emerging social network media.As use of the SNSs grows and integrates with citizens’ politicalactivities, the question becomes, what do people do on SNSs forpolitics, and what would be the consequences of those activities?Whether and how SNS use influences people’s exposure to politicalviewpoints is one of those considerations. While SNSs may make it

972 Y. Kim / Computers in Human Behavior 27 (2011) 971–977

easier for citizens to either broaden their exposure to a variety ofpolitical viewpoints or seek information that supports their pointsof view, little is known about whether SNS use facilitates individ-uals’ exposure to political difference. Thus, the present researchaims to examine the relationship between SNS use and individuals’exposure to political difference.

1.1. New media technology, political fragmentation, and socialnetwork sites

The fragmentation thesis, which presents ‘‘the challenge tocoherent societies and effective governments presented by thebreakdown of broadly shared social and political experience’’ (Ben-nett, 1998, p. 741), has been prominent among debates over the roleof the Internet in the contemporary democratic process. Its propo-nents are suspicious of users’ increased control over communicationonline. They see such power as possibly heightening selective expo-sure, in which people tend to expose themselves to like-minded per-spectives, leading to the fragmentation of public opinion and thepolarization of politics, to the detriment of democracy (Stroud,2008; Sunstein, 2007). Others have argued that the Internet is a placewhere people can more freely interact with diverse others and par-ticipate in the political process. The former scenario results in expo-sure limited to similar viewpoints, while the latter results in morediversified exposure, which leads to hearing and talking to others.

Sunstein (2007) proposes mechanisms by which the Internetcontributes to fragmentation of communities in the public sphere.He expresses concern that unrestricted, free media and channelchoice undermines the preconditions of deliberative democracy,which he believes to be that citizens should be exposed to politicaldiversity and should have a range of common experiences (Sun-stein, 2007). These democratic foundations may be eroded by thechoices offered by the Internet, which greatly increases people’sability to ‘‘filter’’ what they want to read, and whom they wantto see, hear, and even meet. This allows for the bypassing of con-trasting viewpoints, thus hindering the ability to develop a broadercommon basis on which to relate, resultant in the fragmentationand polarization of society.

Psychologically, people tend to seek out perspectives like theirown and information that supports their points of view for cogni-tive consistency (Festinger, 1957) and efficient information pro-cessing (Smith, Fabrigar, & Norris, 2008; Stroud, 2007). However,decreased exposure to political difference that may be driven byselective exposure – what Sunstein (2007) refers to as enclave com-munication – may widen political polarization to extremes. Thisruns directly counter to the norms of deliberative democracy: plu-rality and diversity of public opinion. Studies provide evidence ofselective exposure on the Internet. Iyengar and Hahn (2009) foundevidence of ideological selectivity in news media use. Republicanspreferred to view Fox News and to avoid news from CNN and NPR,while Democrats exhibited the opposite pattern (2009). A studyfound that blog users tended to avoid information that challengedtheir points of view and to seek out information that supportedtheir political beliefs; and individuals’ political website use wasamong the most important predictors of selective exposure topolitical blogs (Johnson, Bichard, & Zhang, 2009).

However, there are arguments against selective exposure on theInternet; and studies have found divergent results. Bimber (2004)is less pessimistic than Sunstein (2007) with his suggestion thatthe Internet holds the potential to create opportunities againstselective exposure or fragmentation. According to Bimber (2004),the most relevant dimension for our understanding of politicalfragmentation is the reach of the political communication system– that is, the extent to which messages span political cleavages.While he warns that failing to span the most important politicalcleavages will foster fragmentation, he also presents possibilities

for the Internet to create a broader reach for political communica-tion systems. He argues that ‘‘the Internet does not represent a sin-gular mode of communication, but a flexible and adaptable set ofopportunities for communication that can be exploited by individ-uals and groups in many ways’’ (Bimber, 2004, p. 31). This suggeststhat the Internet may contribute to integration of the public sphereas well as political fragmentation.

A recent study provides further demonstration. According toBrundidge (2010), the psychological and structural mechanisms in-volved with online selectivity may be somewhat weak; onlinenews use and online political discussion contribute to expandingthe heterogeneity of political discussion networks. It is importantto note that inadvertency of media may facilitate exposure to di-verse discussion networks. Brundidge argues that ‘‘inadvertencyis facilitated online through (1) less than perfect online selectiveexposure strategies (2) non-avoidance of encounters with politicaldifference, (3) weakened social boundaries between far flung geo-graphic locations, between one discursive space and the next(blurred and porous boundaries creating increased interspatiality),between political and apolitical spaces of communication, and be-tween the private and the public spheres’’ (2010, p. 687). This inad-vertency thesis indicates that, even with less purposive seeking outof political difference, people are still likely to be exposed to polit-ical difference, at least to some extent, especially through inadver-tency (Brundidge, 2010). In sum, a potential mechanism that mayfacilitate exposure to political difference is that people can inad-vertently be exposed to cross-cutting perspectives and challenginginformation on the Internet because of the blurring of boundariesthat cyberspace provides.

Wojcieszak and Mutz (2009) support this notion that cross-cut-ting political exchange is likely to take place where politics comesup incidentally and is not the main purpose of online discussionspaces. They found that apolitical space was more likely to contrib-ute to cross-cutting political discussion, while political chat roomsand message boards made only limited contributions to cross-cut-ting discussion (Wojcieszak & Mutz, 2009). These studies suggestthat exposure to diverse and heterogeneous networks may happensomewhat accidentally in places where politics and non-politics ex-ist together.

In a similar fashion, SNS users may be more likely to be exposedto cross-cutting perspectives because of characteristics of SNSsthat contribute to inadvertency, such as heterogeneity of SNS pop-ulations, hyperlinks, and interactive communication applications.In the context of SNSs, users have ample opportunity to becomeexposed to a variety of beliefs and perspectives, including contrast-ing viewpoints. Because electronic media have increased the diver-sity of available ideas (Page, 1996), and difference is not that hardto find in cyberspace (Dahlberg, 2001), especially where character-istics of politics and non-politics exist together (Wojcieszak &Mutz, 2009), SNSs may make it more likely that people will be ex-posed to diversity across political difference that otherwise mightnot be available. Therefore, it is reasonable to expect that individ-uals’ SNS use may positively contribute to increasing their expo-sure to cross-cutting political views.

Based on the previous literature review, the current study pro-poses the following hypothesis:

Hypothesis 1. Use of social network sites will be positively relatedto exposure to cross-cutting political viewpoints.

1.2. Online political messaging

In addition to exploring the association between SNS use andexposure to political difference, this study investigates the role ofonline political messaging. Literature suggests that another possi-ble means of inadvertent online exposure to political difference

Y. Kim / Computers in Human Behavior 27 (2011) 971–977 973

is online political discussion. Brundidge (2010) found convincingevidence for positive effects of online discussion on exposure topolitical difference. She found that online political discussion is di-rectly and positively associated with discussion network heteroge-neity and, furthermore, that it mediates the influence of onlinenews use on individuals’ exposure to political difference. Similarreasoning can be applied to online political messaging – postingor commenting one’s thoughts and opinions through interactivemessaging technologies such as on electronic bulletin boards andonline chats (Price & Cappella, 2002; Shah, Cho, Eveland, & Kwak,2005) – because online political discussion and political messagingshare some of the characteristics of inadvertency or weakened so-cial boundaries. Thus, this study proposes:

Hypothesis 2. Online political messaging will be positively asso-ciated with exposure to cross-cutting political viewpoints.

Hypothesis 3. Online political messaging will mediate the influ-ence of SNS use on exposure to cross-cutting political viewpoints.

1.3. Moderating effects of partisanship

It has been demonstrated that individual differences, includingpreexisting psychological difference, may interact with effects ofcommunicatory behaviors (i.e., media use and interpersonal com-munication) on individuals’ behavioral consequences. This studyfocuses on the role of partisanship, one of the most significant con-cepts that may affect exposure to political difference. Partisanshipmitigates media effects (Iyengar & Kinder, 1987; Young, 2004). Par-tisans are more likely to expose themselves to information thatsupports their points of views because they are psychologicallymotivated to process information that is similar to their personalbeliefs (Donsbach, 1991; Johnson et al., 2009; Stroud, 2008). Thus,it is reasonable to expect that partisanship might moderate the ef-fects of online political messaging and SNS use on exposure topolitical difference. Thus, this study proposes:

Hypothesis 4. The influence of online political messaging onexposure to political difference will be moderated by partisanship.

Hypothesis 5. The influence of SNS use on exposure to politicaldifference will be moderated by partisanship.

2. Method

2.1. Data

Data for this study were collected by the Pew Internet & Amer-ican Life Project, which was conducted by Princeton Survey Re-search Associates between November 20 and December 4, 2008.Using a random-digit sample of telephone numbers representingall continental Unites States telephone households, the surveywas conducted through interviews of a sample of 2254 respon-dents ages 18 and older. The response rate was 23%.

2.2. Measurements

2.2.1. SNS useThe use of social network sites is the independent variables of

this study. SNS use was measured by an additive index compiledfrom five items gathered by asking the respondents about theiruse of network sites such as Facebook and MySpace. They wereasked whether they (1) had gotten any campaign or candidateinformation from these sites; (2) started or joined a political groupor a group supporting a cause on a social networking site; (3) re-

vealed on a social networking site which presidential candidatethey voted for this year; (4) discovered from the sites which pres-idential candidate their friends voted for this year; and (5) signedup as a ‘‘friend’’ of any candidates on a social networking site.The items were each dummy-coded (yes = 1, no = 0) and addedto create an index (Cronbach’s a = .80).

2.2.2. Online political messagingOnline political messaging was measured with two items asking

respondents whether they have posted comments, queries or infor-mation about the campaign or the elections (1) in an online discus-sion, a listserv, or other online group forum and (2) on a blog, theirown or someone else’s. The items were each dummy-coded(yes = 1, no = 0) and converted into an additive scale (Cronbach’sa = .68).

2.2.3. Exposure to cross-cutting point of viewThe dependent variable of this study is exposure to cross-cut-

ting political points of view. In the survey, respondents were askedto indicate whether most of the sites they visit to get political orcampaign information online challenge their own point of view,share their point of view, or do not have a particular point of view.The dependent variable was dummy-coded ‘‘1’’ if most sites chal-lenged their point of view.

2.2.4. Control variablesDemographic variables were included as control variables in the

analyses: age, sex, education, ethnicity, and income. Respondents’party identity – whether they identified themselves as partisan orindependent – was also introduced as a control variable. Respon-dents’ use of traditional news and online news was measured.Respondents were asked whether they read a printed copy of thenewspaper yesterday and watched television news yesterday. Thesetwo items were added to create an index of traditional news use. On-line news use was measured by asking how often respondents wentonline to get news about the election, with 1 indicating never and 6indicating more than once a day. Another control variable includedin the analysis was partisan media use. The survey included itemsasking whether respondents had visited websites related to Obamaand the Democratic Party or the Republican Party and GOP candi-dates. This partisan Internet use variable was dummy-coded ‘‘0’’ ifthe respondent reported that they had never visited either kind ofwebsite, and ‘‘1’’ if they had.

2.3. Data analysis

In order to test the Hypotheses 1 and 2, logistic regression analy-ses were employed on the dichotomous dependent variable. In addi-tion, ordinary least squares (OLS) regression analysis with acontinuous dependent variable – online political messaging – wasemployed to test a mediating effect of online political messaging be-tween SNS use and exposure to cross-cutting viewpoints (Hypothe-sis 3). Two interaction terms (online messaging � partisanship andSNS use � partisanship) were created and used in the analysis to testthe moderating effects of partisanship (Hypotheses 4 and 5). Therespondents’ demographics, partisanship, news media use, and par-tisan Internet use were entered as control variables in the all regres-sion models to assess the influence of each variable on the dependentvariable and to examine the effects of the independent variables con-trolling for a set of confounding variables.

3. Results

Fifty-three percent of the sample was female and 79.5% wereWhite; the median age group was 35–44. Their median education

Table 2Logistic regression analysis investigating whether SNS use influences exposure tocross-cutting viewpoints.

B SE Wald Exp (B)

Age .041 .069 .354 1.042Gender (male = 1) .156 .180 .754 1.169Education �.304*** .110 7.589 .738Ethnicity (White = 1) �.127 .228 .313 .881Income .061 .048 1.601 1.063Partisanship �.082 .192 .182 .921Traditional news .069 .124 .307 1.071Online news �.008 .059 .021 .992Partisan Internet use �.385* .228 2.867 .680SNS use .200** .083 5.822 1.221N 760Chi square 16.87*

* p < .10.** p < .05.

*** p < .01.

Table 3Logistic regression analysis investigating whether online political messaging influ-ences exposure to cross-cutting viewpoints.

B SE Wald Exp (B)

Age .000 .064 .000 .000Gender (male = 1) .142 .180 .620 .620Education �.303*** .111 7.486 7.486Ethnicity (White = 1) �.109 .229 .225 .225Income .057 .049 1.361 1.361Partisanship �.043 .192 .051 .051Traditional news .056 .124 .201 .201Online news �.020 .060 .110 .110Partisan Internet use �.427* .230 3.451 3.451Online political messaging .440*** .159 7.674 7.674N 758Chi-square 19.05**

* p < .10.** p < .05.

*** p < .01.

974 Y. Kim / Computers in Human Behavior 27 (2011) 971–977

level was some college including vocational school and 2- or 4-yearcollege degree; median income level was $40,000 to under$50,000. Sixty-four percent of the respondents identified them-selves as partisan and 11% of the sample reported that they hadvisited party-oriented websites. More than a quarter (25.3%) ofthe respondents reported that most of the sites they visit to getpolitical or campaign information challenge their point of view.

Before addressing the primary aims of the study with multivar-iate analysis, a partial correlation analysis was conducted. A matrixof partial correlations between exposure to cross-cutting views,SNS use, online political messaging, traditional news use, onlinenews use, and partisanship is provided in Table 1. SNS use and on-line political messaging were significantly correlated with individ-uals’ exposure to cross-cutting political views. None of the newsmedia use, however, was correlated with exposure to political dif-ference. Online news use was significantly associated with individ-uals’ SNS use and online political messaging. There were positiveand significant associations between respondents’ partisanshipand their use of both SNSs and traditional news. To test hypothe-ses, more stringent analyses are in order since exposure to politicaldifference may be correlated with news media use and partisan-ship as well as some demographic variables.

3.1. Tests of hypotheses

Hypothesis 1 predicted that SNSs use will be positively relatedto exposure to cross-cutting points of view. Table 2 presents the re-sults of the logistic regression analysis investigating whether SNSuse influences respondents’ exposure to cross-cutting viewpoints.Two control variables – education and partisan Internet use – werefound to predict exposure to political difference (see Table 2). Peo-ple having higher levels of education and reporting more partisanInternet use tended to have less exposure to cross-cutting points ofview. Other demographic variables, partisanship, and news mediause did not exert a significant effect on exposure to cross-cuttingviewpoints. Logistic regression analysis revealed that the use ofSNSs was significant in predicting respondents’ exposure tocross-cutting political viewpoints even after controlling for demo-graphic variables (age, gender, education, ethnicity, and income),partisanship, and media use variables (B = .200, p < .05). Thus,Hypothesis 1 was supported.

Hypothesis 2, online political messaging will be positively asso-ciated with exposure to cross-cutting political viewpoints, also wassupported. As Table 3 shows, online political messaging was signif-icant after controlling for demographic variables, partisanship, andmedia use variables (B = .440, p < .01). People exhibiting higher de-grees of online political messaging had significantly more exposureto political difference.

In order to test a mediation hypothesis, which states that onlinepolitical messaging will mediate the influence of SNS use on expo-sure to cross-cutting political viewpoints, additional regressionanalyses were conducted (see Table 4). The following conditions

Table 1Partial correlations between exposure to cross-cutting views, SNS use, online messaging,

Variables M SD 1

1. Cross-cutting exposure .25 .43 –2. SNS use .47 1.10 .075*

3. Online messaging .21 .53 .086*

4. Traditional news 1.12 .75 .0165. Online news 3.37 1.60 �.0106. Partisanship .64 .47 �.014

Note: Cell entries are partial correlation coefficients, controlling for age, gender, educati* p < .05.�� p < .01.

*** p < .001.

have to be met for a variable to function as a mediator (Baron &Kenny, 1986): (a) a significant effect of the independent variable(SNSs) on the presumed mediator (online political messaging)(B = .166, p < .001); (b) a significant effect of the mediator on thedependent variable (exposure to cross-cutting point of view,B = .440, p < .01); (c) adding the mediators reduces a previous sig-nificant relation between the independent and dependent vari-ables (from B = .200, p = .016 to B = .144, p = .103). Thus,Hypothesis 3 was supported (see Fig. 1).

Turning to the moderating role of partisanship, two interactionterms were entered in the logistic regression. The results are pre-sented in Table 4 (see the right column). Hypothesis 4 predictedthat the influence of online political messaging on exposure to

traditional news use, online news use, and partisanship.

2 3 4 5 6

–.371*** –.020 .029 –.162*** .216*** .041 –085* �.012 .079* .057 –

on, race, and income (N = 758); two-tailed.

Table 4OLS regression predicting online political messaging and logistic regression predictingexposure to cross-cutting perspectives.

Online politicalmessaging

Exposure to cross-cuttingperspectives

Age �.006 (.012) .039 (.069)Gender (male = 1) .012 (.030) .160 (.181)Education �.006 (.019) �.305 (.111)**

Ethnicity (White = 1) �.033 (.040) �.115 (.229)Income �.004 (.008) .062 (.049)Partisanship �.066 (.032)* �.071 (.193)Traditional news .023 (.021) .057 (.125)Online news .042 (.010)*** �.025 (.060)Partisan Internet use .184 (.039)*** �.475 (.233)*

Online political messaging – .352 (.168)*

SNS use .166 (.016)*** .144 (.088)

Interaction termsMessaging � Partisanship – �.180 (.082)*

SNS use � Partisanship – .008 (.067)N 967 758R2/Chi-square .211*** 21.65*

Note: Cell entries for online political messaging are unstandardized OLS regressioncoefficients. Cell entries for exposure to cross-cutting perspectives are logisticregression coefficients. All entries in parentheses are standard errors.

* p < .05.** p < .01.

*** p < .001.

Fig. 2. Interaction effects of online messaging and partisanship on exposure tocross-cutting viewpoints.

Y. Kim / Computers in Human Behavior 27 (2011) 971–977 975

political difference will be moderated by partisanship. Partisanshipturned out to significantly moderate the relationship between on-line political messaging and exposure to political difference(B = �.180, p < .05); thus Hypothesis 4 was supported. The interac-tion between online political messaging and partisanship on expo-sure to cross-cutting points of view is plotted in Fig. 2. The effectsof online messaging on exposure to political difference were stron-ger for nonpartisans than for partisans.

Hypothesis 5 predicted that the influence of SNS use on expo-sure to political difference would be moderated by partisanship.Hypothesis 5 was not supported. These results suggest that the ef-fects of SNS use on exposure to political difference were stableacross partisan status.

4. Discussion

This paper aims to understand how social network sites influ-ence individuals’ exposure to political difference. First, this studyexamined the association between SNS use and exposure tocross-cutting political viewpoints. It found a positive and signifi-cant relationship between SNS use and exposure to cross-cuttingpoint of view. In addition, SNS use indirectly influences exposureto cross-cutting points of view through online political messaging.

Note: The number in the parentheses is the logthe mediating variable (online messaging) as a*p < .05 **p < .01 ***p < .001

SNS

Online Messag

.144 (.200*

.166***

Fig. 1. Path diagram of mediating effe

Another aim of this study is to explore the role of online polit-ical messaging in facilitating cross-cutting exposure. Findingsdemonstrate that citizens’ online political messaging not only di-rectly affects exposure to political difference, but it mediates therelationship between SNS use and exposure to cross-cuttingperspectives.

Demographic variables (i.e., age, gender, ethnicity, and income)were not predictors of exposure to cross-cutting perspectiveswhile education turned out to be a significant variable affectingexposure to political difference. More educated people are less ex-posed to cross-cutting points of view than less educated people,which is consistent with previous research showing that exposureto disagreement in political discussion was lowest among thosewho have attended graduate school (e.g., Mutz, 2006). Interest-ingly, partisanship did not significantly influence individuals’exposure to dissenting political views while it moderated the asso-ciation between online political messaging and exposure to cross-cutting perspectives. Regression analyses consistently showed thenegative effects of partisan Internet use on individuals’ exposure topolitical difference. Those who visited partisan websites were lesslikely to encounter cross-cutting views than those who did not usepartisan-based Internet. Previous studies demonstrate that indi-viduals’ partisan media use (e.g., reading newspapers endorsing acandidate and visiting a candidate website) leads to polarizationof political attitudes (Stroud, 2008, 2010). Theoretically, therefore,partisan media use should be related to exposure to cross-cuttingperspectives. Those who use partisan media more frequently maybe more likely to be exposed to like-minded perspectives becauseit is much easier to find similar views through partisan media.

Overall, findings imply that social network sites contribute toexpanding individuals’ exposure to political difference in general,which is consistent with optimistic views on the role of the Inter-net in enhancing democracy by, for instance, increasing the heter-ogeneity of political discussion network (Brundidge, 2010),

istic regression coefficient without adding control variable.

ing

Exposure to cross-cutting perspectives )

.440**

cts of online political messaging.

976 Y. Kim / Computers in Human Behavior 27 (2011) 971–977

informing citizens, and facilitating political engagement (e.g., Ken-ski & Stroud, 2006). The Internet has been regarded as an impor-tant source of political knowledge and engagement in this line ofresearch because the Internet has easily and cheaply availableinformation (Bimber, 2001). Indeed, social network services suchas Facebook also increase the opportunities for individuals toencounter news and information (Nielsen Reports, 2009). Thisstudy provides evidence that SNSs may also expand individuals’exposure to political difference.

More importantly, the contributing role of SNSs to expandingexposure to dissimilar political views works across individuals’ par-tisanship. While this study hypothesized that partisanship wouldmoderate the effect of SNSs on exposure to political difference, itfound instead that partisanship did not moderate the associationsbetween SNSs and exposure to cross-cutting perspectives. This indi-cates that inadvertent exposure to political difference facilitated bySNSs may happen regardless of individuals’ political orientations.That is, growing social network media can be helpful both for parti-sans and nonpartisans to expand their social boundaries by provid-ing opportunities for information exchange, social interactions, andinadvertent exposure to dissimilar political views.

However, partisanship was found to significantly moderate theeffect of online political messaging on exposure to political differ-ence. Partisans tend to expose themselves to dissimilar politicalviews online neither more nor less even if they post political mes-sages, comments, and opinions, while nonpartisans are more likelyto expose themselves to political difference if they post such com-ments more often. These mixed results of the moderating role ofpartisanship suggest that different kinds of media environmentsmay have different consequences for exposure to politicaldifference.

4.1. Limitations and future studies

There were several limitations to this study. Because of thecross-sectional data analyzed, the study could not be fully confi-dent in causal relationships among variables. The study also reliedon respondents’ self-reports to measure exposure to cross-cuttingpolitical views online, not on actual measurements of respondents’activities. An experimental setting would allow researchers tomeasure participants’ actual site-hits, which could be operational-ized as exposure to political views. To overcome these limitations,longitudinal studies and experimental designs should be consid-ered. In addition, this study focused on social network sites suchas Facebook and MySpace as a social medium that may facilitateinadvertent exposure to political difference. Future research couldalso investigate the role of other social media such as Twitter, oneof the rapidly growing applications.

Furthermore, although this study took several variables (i.e.,demographic variables, news use, partisanship, and partisan Inter-net use) into account, future studies could expand the currentstudy by examining whether and how other variables relate toindividuals’ exposure to political difference. Existing studies hasdemonstrated that political knowledge and political efficacy aresignificantly associated with individuals’ heterogeneity of discus-sion networks and exposure to disagreement in political discussion(Brundidge, 2010; Mutz, 2006; Wojcieszak & Mutz, 2009). In par-ticular, activities on the Internet would be an important consider-ation to understand individuals’ exposure to cross-cutting views asshown by Wojcieszak and Mutz (2009) that online group activities– especially, the leisure chat rooms or message board – may con-tribute to increasing exposure to dissimilar political perspectives.Thus individuals’ various online activities that may be associatedwith exposure to political difference as well as political-orientationvariables (e.g., political efficacy and political knowledge) should betaken into consideration for future studies.

Acknowledgements

The author would like to thank the Pew Internet and AmericanLife Project (Pew) for making their data available, and acknowl-edges that Pew bears no responsibility for the results and conclu-sions of this study based on the analyses of the data.

References

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction insocial psychological research: Conceptual, strategic, and statisticalconsiderations. Journal of Personality and Social Psychology, 51, 1173–1182.

Bennett, W. L. (1998). The uncivic culture: Communication, identity, and the rise oflifestyle politics. PS: Political Science and Politics, 31(4), 740–761.

Bimber, B. (2001). Information and political engagement in America: The search foreffects of information technology at the individual level. Political ResearchQuarterly, 54(1), 53–67.

Bimber, B. (2004). The internet and political fragmentation. Paper presented at thedemocracy in the 21st century conference, Urbana-Champagne, IL.

boyd, d.m., Ellison, N.B. (2007). Social network sites: Definition, history, andscholarship. Journal of Computer-Mediated Communication, 13(1), article 11.Available from: <http://jcmc.indiana.edu/vol13/issue1/boyd.ellison.html>.

Brundidge, J. (2010). Encountering ‘‘difference’’ in the contemporary public sphere:The contribution of the Internet to the heterogeneity of political discussionnetworks. Journal of Communication, 60, 680–700.

Dahlberg, L. (2001). Computer-mediated communication and the public sphere: Acritical analysis. Journal of Computer-Mediated Communication, 7(1), 8.

Donsbach, W. (1991). Exposure to political content in newspapers: The impact ofcognitive dissonance on readers’ selectivity. European Journal of Communication,6, 155–186.

Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The benefits of facebook ‘‘Friends’’:Social capital and college students’ use of online social network sites. Journal ofComputer-Mediated Communication, 12(4), 5–55. article 1. <http://jcmc.indiana.edu/vol12/issue14/ellison.html>.

Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: StanfordUniversity Press.

Gueorguieva, V. (2008). Voters, MySpace, and YouTube: The impact of alternativecommunication channels on the 2006 election cycle and beyond. Social ScienceComputer Review, 26(3), 288–300.

Iyengar, S., & Hahn, K. S. (2009). Red media, blue media: Evidence of ideologicalselectivity in media use. Journal of Communication, 59(1), 19–39.

Iyengar, S., & Kinder, D. R. (1987). News that matters: Television and Americanopinion. Chicago: The University of Chicago Press.

Johnson, T. J., Bichard, S. L., & Zhang, W. (2009). Communication communities or‘‘CyberGhettos?’’: A path analysis model examining factors that explainselective exposure to blogs. Journal of Computer-Mediated Communication, 15,60–82.

Kenski, K., & Stroud, N. J. (2006). Connections between Internet use and politicalefficacy, knowledge, and participation. Journal of Broadcasting & ElectronicMedia, 50, 173–192.

Lenhart, A. (2009). Adults and social network Web sites. Pew Internet and Americanlife project report. <http://www.pewinternet.org/Reports/2009/Adults-and-Social-Network-Websites.aspx> (retrieved 20.10.09).

Mutz, D. C. (2002a). The consequences of cross-cutting networks for politicalparticipation. American Journal of Political Science, 46(4), 838–855.

Mutz, D. C. (2002b). Cross-cutting social networks: Testing democratic theory inpractice. American Political Science Review, 96(1), 111–126.

Mutz, D. C. (2006). Hearing the other side: Deliberative versus participatory democracy.New York: Cambridge University Press.

Nielsen Reports. (2009). Social networks & blogs now 4th Most popular online activity,ahead personal email. <http://en-us.nielsen.com/content/nielsen/en_us/news/news_releases/2009/march/social_networks__.html> (retrieved 20.03.10).

Page, B. I. (1996). Who deliberates?: Mass media in modern democracy. Chicago:University of Chicago Press.

Park, N., Kee, K. F., & Valenzuela, S. (2009). Being immersed in social networkingenvironment: Facebook groups, uses and gratifications, and social outcomes.CyberPsychology & Behavior, 12, 729–733.

Price, V., & Cappella, J. N. (2002). Online deliberation and its influence: Theelectronic dialogue project in campaign 2000. IT & Society, 1(1), 303–329.

Scheufele, D. A., Nisbet, M. C., Brossard, D., & Nisbet, E. C. (2004). Social structureand citizenship: Examining the impacts of social setting, networkheterogeneity, and informational variables on political participation. PoliticalCommunication, 21(3), 315–338.

Schudson, M. (1995). The power of news. Cambridge, MA: Harvard University Press.Shah, D. V., Cho, J., Eveland, W. P., & Kwak, N. (2005). Information and expression in

a digital age: Modeling Internet effects on civic participation. CommunicationResearch, 32(5), 531–565.

Smith, A. (2009a). The Internet’s role in campaign 2008. <http://www.pewinternet.org/’/media//Files/Reports/2009/The_Internets_Role_in_Campaign_2008.pdf(retrieved 15.04.09).

Smith, A. (2009b). Online participation in the social media era. Pew internet &American life project. <http://www.pewinternet.org/Presentations/2009/RTIP-Social-Media.aspx>.

Y. Kim / Computers in Human Behavior 27 (2011) 971–977 977

Smith, S. M., Fabrigar, L. R., & Norris, M. E. (2008). Reflecting on six decades ofselective exposure research: Progress, challenges and opportunities. Social andPersonality Psychology Compass, 2, 464–493.

Stroud, N. J. (2006). Selective exposure to partisan information. Unpublisheddissertation, University of Pennsylvania.

Stroud, N. J. (2007). Media effects, selective exposure, and Fahrenheit 9/11. PoliticalCommunication, 24, 415–432.

Stroud, N. J. (2008). Media use and political predispositions: Revisiting the conceptof selective exposure. Political Behavior, 30, 341–366.

Stroud, N. J. (2010). Polarization and partisan selective exposure. Journal ofCommunication, 60, 556–576.

Sunstein, C. R. (2007). Republic.com 2.0. Princeton, NJ: Princeton University.The Pew Research Center for the People & the Press (2008). Internet’s broader role

in campaign 2008. <http://people-press.org/report/2384/internets-broader-role-in-campaign-2008> (retrieved 11.01.08).

Wojcieszak, M. E., & Mutz, D. C. (2009). Online groups and political discourse: Doonline discussion spaces facilitate exposure to political disagreement? Journal ofCommunication, 59(1), 40–56.

Young, D. G. (2004). Late-night comedy in Election 2000: Its influence on candidatetrait ratings and the moderating effects of political knowledge and partisanship.Journal of Broadcasting & Electronic Media, 48(1), 1–22.