Bursting the filter bubble: the
mediating effect of discussion
Smith School of Enterprise and the Environment, University of Oxford, Oxford, UK
Department of Communication Sciences, Universitat Jaume I, Castell
o, Spain, and
Homero Gil de Z
Department of Political Science, University of Salamanca, Salamanca, Spain;
Department of Film Production and Media Studies,
Pennsylvania State University, State College, Pennsylvania, USA and
Facultad de Comunicaci
on y Letras Universidad Diego Portales, Santiago, Chile
Purpose –The purpose of this study is to identify the structural processes that lead citizens to escape their
common social circles when talking about politics and public affairs (e.g. “filter bubbles”). To do so, this study
tests to what extent political attitudes, political behavior, news media consumption and discussion frequency
affect discussion network heterogeneity among US citizens.
Design/methodology/approach –Supported by the polling group Nielsen, this study uses a two-wave panel
online survey to study the antecedents and mechanisms of discussion network heterogeneity among US
citizens. To test the hypotheses and answer the research questions, ordinary least squares (OLS) regressions
(cross-sectional, lagged and autoregressive) and mediation analyses were conducted.
Findings –The findings imply that political discussion frequency functions as the key element in explaining
the mechanism that leads politically interested and participatory citizens (online) as well as news consumers of
traditional and online media to seek a more heterogeneous discussion network, disrupting the so-called “filter
bubbles.”However, mediation analyses also showed that discussion frequency can lead to more homogenous
discussion networks if people score high on political knowledge, possibly reflecting the formation of a close
network of political-savvy individuals.
Originality/value –The survey data give important insights into the 2016 pre-election situation, trying to
explain why US citizens were more likely to remain in homogenous discussion networks when talking about
politics and public affairs. By using two-wave panel data, the analyses allow to draw tentative conclusions
about the influential and inhibiting factors and mechanisms that lead individuals to seek/avoid a more
heterogeneous discussion network.
Keywords Network heterogeneity, Discussion frequency, Political interest, News use, Filter bubble
Paper type Research paper
The US Presidential Election 2016 and the pervasive partisan flow of information about the
presidential candidates online have revived the notion of “filter bubbles”(Pariser, 2011)and
“information cocoons”(Sunstein, 2006). Journalists, critics and opinion makers were quick in
identifying the scapegoats of the surprising turnout of the election, blaming social media for
spreading false information (Baer, 2016;El-Bermawy, 2016) and creating an information
environment that has reinforced existing beliefs and strengthened political polarization (cf.,
Stroud, 2010;Knobloch-Westerwick and Meng, 2011). What is more, European politicians and
advisors have expressed their concerns about increased personalized media content and its
danger for a healthy democracy (V
ık¸e-Freiberga et al.,2013). It is in this vein that social media
and personalized algorithms (e.g. on Twitter and Facebook) have been suspected of enforcing
The current issue and full text archive of this journal is available on Emerald Insight at:
Received 6 November 2019
Revised 11 May 2020
Accepted 21 June 2020
Online Information Review
© Emerald Publishing Limited
people’s tendency to solely reside in information environments and social circles that affirm
their personal opinions, attitudes and points of views (cf. Pariser, 2011).
However, not only has recent research casted doubt on the so-called “filter bubble”(Nelson
and Webster, 2017;Zuiderveen Borgesius et al., 2016), part of the theoretical rational is also
imperfect: citizens and voters are not active on the Internet or social media 24/7. Surely, recent
figures show that 77% of Americans go online every day and 26% of Americans are online
almost constantly (Perrin and Jiang, 2018). Yet these findings do not rule out that American
citizens also get exposed to views and opinions about politics and current affairs when
interacting with their peers in real-life settings. After all, the majority of people go online for
leisure and entertaining purposes rather than for political reasons (Park et al., 2009;Quan-
Haase and Young, 2010). And political beliefs, attitudes and voting choices are still vastly
influenced by face-to-face conversations with family members, friends and neighbors (Berger
et al., 2008;Huckfeldt and Sprague, 1995).
Thus, the discussion about the prevalence of like-minded information cocoons (Sunstein,
2006) needs to better incorporate the offline scenery theoretically and empirically. Furthermore,
research needs to better explain why and under what circumstances people are more likely to
encounter opinions and viewpoints that differ from their own—both online and offline (cf. Choi
and Lee, 2015;Lee et al., 2014;Sunstein, 2018). What we know from past research is that social
media and the internet, just as the offline scenery, offers information and exchange
opportunities for both: people who seek information that reaffirms their existing beliefs; and
people who seek a more diverse and heterogeneous information environment (Festinger, 1957;
Knobloch-Westerwick and Meng, 2009,2011;Sears and Freedman, 1967).
Facing an age with information ubiquity both offline and online, the question therefore
remains: What are the structural processes that lead citizens to escape their common social
circles when talking about politics and public affairs? Using a two-wave panel survey among
US citizens in 2013/2014, we find that political discussion frequency functions as the key
mediator in explaining the mechanism that leads politically interested and participatory
citizens (online) as well as news consumers of traditional and online media to seek a more
heterogeneous discussion network. However, discussion frequency can also lead to more
homogenous discussion networks if people score high on political knowledge, possibly
reflecting the formation of political-savvy elite.
1. Literature review
1.1 Network heterogeneity and democracy
We are currently living in a political environment that is characterized by political
polarizations, extremism and violent rioting (e.g. Charlottesville 2017; Gunter and Hughes,
2018). All around the world, but particularly in the US, we have witnessed clashes among
society and politics with regards to controversial topics (e.g. climate change, refugee policy
and vaccines). Although political debates and disagreement are necessary and useful aspects
of a well-functioning democracy (Sorensen, 2018;Chadwick et al., 2017), it has been reported
that not only political debates have become more hostile, disrespectful and polemic over time
(Sood and Iyengar, 2016), the strong divide between ideological political stances has also
made reconciliations between parties more difficult (Iyengar et al., 2019;Davis and Dunaway,
2016). Given these recent developments, it becomes of paramount interest to study the
underlying mechanisms that foster a citizenship that is open-minded and willing to listen to
opinions and voices that differ from their own, opening up possibilities for fruitful
compromises that create solutions for a harmonious life together. In fact, it has long been
argued by sociologists, political scientists and communication scholars that being exposed to
a diverse set of viewpoints is the nucleus for deliberative democracies (Habermas, 1989;Price
et al., 2002) and democratic citizenship (e.g. Gil de Z
niga and Valenzuela, 2011;Mutz, 2006;
However, recent detrimental societal movements, such as the rise of the political right
around the globe (e.g. Berlet and Lyons, 2018) as well as hate speech and discrimination on the
net (e.g. Gerstenfeld et al., 2003), show that our democratic societies are far away from
enacting deliberative discourses. Moreover, the increasing fragmentation and polarization
within societies pose a real threat for the functioning of democracies. Individuals have
become less likely to identify common ground and understanding with the “other group”
(Sunstein, 2001). It is in this vein that human interactions and discussions between
individuals who differ in their ideas, viewpoints as well as ethnical and social backgrounds
gain increasing relevance. Scholars in communication science have identified the discussion
among people with diverse viewpoints and political attitudes as discussion network
heterogeneity (Kwak et al., 2005;Mutz, 2002;Scheufele et al., 2004,2006). These unfamiliar
encounters are believed to increase the identification of shared experiences, and thereby the
likeliness of acceptance and tolerance which work as “social glue”(Sunstein, 2002,p.9;Mutz
and Mondak, 2006;Price et al., 2002) for diverse democratic societies. As Sunstein (2018)
contends, “in a democracy...lives—including digital ones—should be structured so that
people frequently come across views and topics that they have not specifically selected”(p. 7).
1.2 Network heterogeneity and endogeneity problems
However, the vast majority of research on discussion network heterogeneity has
concentrated on the concept as a direct or indirect antecedent for political behavior such
as political participation, political knowledge or civic engagement (e.g. Choi et al., 2017;Kim
and Chen, 2015;McLeod et al., 1999b;Mutz, 2002;Scheufele et al., 2004,2006). Thus, rather
than providing another study that aims at testing the positive outcomes of political
discussions among a diverse network of people for democracy, we strive for empirical
evidence that informs us about the underlying mechanisms that explain why individuals are
more likely to expose themselves to a heterogeneous discussion network. Reviewing previous
research on network heterogeneity, we have identified various behavioral and attitudinal
variables that can be hypothesized to contribute to individuals’likeliness to reside in non-like-
minded discussion networks. These include political participation, political efficacy, political
interest, political knowledge, ideology and news use (Brundidge et al., 2014). What is more, we
position discussion frequency about politics and public affairs—which has so far been mostly
considered as a control variable—at the center of our analysis, arguing that it works as a
conduit between political attitudes, political behavior, news use and discussion network
heterogeneity. The conceptual model is shown in Figure 1.
1.3 Political behavior, attitudes and network heterogeneity
1.3 1. Political participation. Although the conceptualization of network heterogeneity differs
quite substantially among scholars (Eveland and Hively, 2009), the main focus of previous
research has lied on the role of network heterogeneity for political participation (e.g. Choi et al.,
2017;Lu et al., 2018). Two camps have emerged in this field of research. Representative of the
first camp posit that individuals embedded in a heterogeneous social network are more likely
to retreat from political activity due to so-called “cross-cutting pressures”and the need for
harmony with their social relationships (Lazarsfeld et al., 1944;Mutz, 2002;Nir, 2005). It has
been argued that individuals belonging to social groups, characterized by conflicting
interests and opinions, have difficulties in making up their political mind, resulting in delayed
participation or no participation in elections at all. However, research that has tried to
replicate the findings could not find evidence for the cross-pressure hypothesis, pointing to
methodological errors and theoretical misconceptions (Eveland and Hively, 2009;Horan,
In fact, more recent work (second camp) suggests that heterogeneous discussion networks
lead to positive effects in terms of political engagement, civic citizenship and deliberative
democracy overall (Cappella et al., 2002;Choi et al., 2017;Kim and Chen, 2015;Lu et al., 2018;
McLeod et al., 1999b;Scheufele et al., 2004,2006). For example, individuals who are embedded
in a heterogeneous network are not only more knowledgeable in politics (e.g. Scheufele et al.,
2004), they are also reported to be more likely to increase political tolerance (Mutz and
Mondak, 2006;Price et al., 2002) and decrease polarization (e.g. Lee and Choi, 2020).
Furthermore, previous research that has shown that network heterogeneity is positively
related to political participation (McLeod et al., 1999b;Scheufele et al., 2004,2006), has been
confirmed by a current stream of research on online social networks that has found positive
effects of network heterogeneity on political participation (Choi et al., 2017;Kim and Chen,
2015;Lu et al., 2018), particularly when news use is involved (e.g. Kim, 2018). Despite the
plethora of research on the relation between network heterogeneity and political
participation, none of the research cited above has focused on the reversed direction.
Hence, to get more insights into what determines the constitution of heterogeneous networks,
we want to inquire: (RQ1): How does political participation (offline, online and social media)
relate to discussion network heterogeneity?
1.3 2. Political efficacy. Given that political efficacy is an important predecessor for political
participation (Hayes et al., 2006) and voting (Verba and Nie, 1972), it might also play a crucial
role for discussion network heterogeneity. However, as Eveland and Hively (2009) have
pointed out, there is a lack of research that has investigated the potential influence of political
efficacy on diverse networks. Even up to now, research on discussion network heterogeneity
has forgone to include political efficacy in conceptual and analytical models, or only used it as
a control variable (Kim et al., 2013). However, reversely, research has shown that political
efficacy is positively influenced by the frequency of political talk and the orientation toward
common understandings within political conversations (Rojas, 2008).
Hence, citizens who encounter more political talk are more likely to believe that they have
a good understanding of politics and feel qualified to participate in politics (cf. internal
efficacy: Niemi et al., 1991). Likewise, citizens might also perceive themselves to be more
politically efficacious if they get regularly exposed to a broad range of political ideas,
arguments and issues that can be considered representative of the electorate. Reversely, it is
also likely that individuals who are exposed to diverse political attitudes become disillusioned
about their own ability to participate in politics and more skeptical with regards to their
model of network
perception about their understanding of the political sphere (cf. Eveland and Hively, 2009).
The sheer diversity of viewpoints, argumentations and positions that are difficult to
understand—and maybe even to tolerate for some—might discourage individuals to feel
politically efficacious. Given the scarce body of research on this matter, we pose the following
research question: (RQ2): How is political efficacy related to individuals’discussion network
1.3 3. Political interest. The fact that motivation and interest in politics play crucial roles in
explaining political behavior has long been accepted in the academic community, but
particularly since the publication of prior work on media choice and inequality in political
involvement and elections (2007). Focusing on network heterogeneity, Kwak et al. (2005) have
shown that attention paid to discussions is a crucial factor in explaining the positive
relationship between network heterogeneity and political participation. In other words,
individuals who are entrenched in a heterogeneous discussion network, and who are attentive
to what their discussion partners have to contribute to the conversation, are more likely to
become politically engaged. Thus, the extent to which somebody is interested in politics might
also be strongly linked to the extent to which individuals seek diverse discussion networks.
However, political interest has mostly been ignored (e.g. McLeod et al., 1999b;Scheufele
et al., 2004,2006) or simply been treated as a control variable in past research about
discussion network heterogeneity (e.g. Kwak et al., 2005). Investigating disagreement within
communication networks, Huckfeldt and Morehouse Mendez (2008) concluded that political
interest, knowledge and the education of individuals are strong predictors for discussion
frequency. A more recent paper by Choi and Lee (2015) has studied the reverse relationships
and provided evidence that political interest moderates the relationship between news
sharing online and network heterogeneity. Thus, we hypothesize (H1): Individuals who have a
strong interest in politics are more likely to be exposed to a heterogeneous discussion network.
1.3 4. Political knowledge. Long ago, political interest has already been found to be strongly
linked to political knowledge (Seeman, 1966). Since then, a wide range of research has
highlighted that the more people know about politics, the more likely they are to actively
participate in politics (e.g. Bennett, 1986;McLeod et al., 1999a;Neuman, 1986). At the same
time, it has been argued that this relationship might also be triggered by other intervening
factors, such as the frequency of political discussions or with whom people actually discuss
politics. Scheufele et al. (2006), for example, found evidence that network heterogeneity is
positively related to political participation, but mediated through political knowledge. Hence,
how political knowledge contributes to political participation (in various forms) might also
depend on the diversity of political opinions and viewpoints individuals encounter.
In fact, previous research has shown that network heterogeneity enhances political
knowledge (Huckfeldt et al., 2004;McLeod et al., 1999b;Scheufele et al., 2004). More
specifically, it has been argued that individuals exposed to disagreement and opposing views
have to use more cognitive activity to reflect upon new arguments they encounter, and in
order to find (new) ways to defend their position—possibly also revising their own opinion
(Levine and Russo, 1995). More recent studies support this mechanism. Eveland and Hively
(2009), distinguishing between issue stance knowledge and knowledge structure density,
found that both dangerous and diverse discussion networks are positively related with
knowledge structure density.
Hence, when considering the reverse direction, it might also be possible that individuals
with strong political knowledge are more likely to seek various political viewpoints to
maintain their knowledge and get to know to various perspectives toward an issue. Given the
absence of evidence for the reversed relationship between political knowledge and discussion
network heterogeneity, we assume: (H2): The more individuals know about politics, the more
they are exposed to a heterogeneous discussion network.
1.3 5. Political ideology. A related stream of research investigating information selection in
a diverse information environment has focused on the phenomenon of selective exposure and
how it reinforces or alters political ideologies and attitudes. Grounded in cognitive dissonance
theory (Festinger, 1957) and motivated reasoning (Kraft et al., 2015), scholars argue that
people are likely to avoid media content that is not in line with their own beliefs and tend to
consume information that confirms their already existing opinions and ideologies (Taber and
Lodge, 2006), albeit not necessarily avoiding counter-attitudinal information (Garrett et al.,
2013). Scholars in political science have found evidence that exposure to diverse political
views and ideologies indeed brings about detrimental effects, such as the reinforcement of
pre-existing political attitudes and opinions, thus leading to even more extreme positions (e.g.
on gun control: Taber and Lodge, 2006; same-sex marriage and sexual minority rights:
Wojcieszak and Price, 2010; or attitude toward political candidates: Meffert et al., 2006).
At the same time, empirical findings have also casted doubt on the black/white news and
information consumption behavior (e.g. Nelson and Webster, 2017), pointing to various
underlying factors that influence the extent to which people expose themselves to opposing
views and ideologies. Lee et al. (2014) have shown, for example, that individuals with a more
heterogeneous network on social network services are more polarized in ideology, but only if
they discuss politics with others more often. Similarly, a more recent study suggests that the
effect of heterogeneous discussion networks on opinion polarization might be moderated by
political orientation such that more liberal individuals (i.e. South Koreans) show higher levels
of polarization when being exposed to opposing opinions than individuals with lower liberal
attitudes (Lee and Choi, 2020). To the best of our knowledge, Lee et al. (2014) have provided
the only study that has investigated the reversed direction, and they found that political
ideology is positively related to network heterogeneity. Given that previous research about
network heterogeneity has only controlled for political ideology (Brundidge, 2010) or strength
of partisanship respectively (e.g. Kim et al., 2013), we hypothesize based on Leeet al.’s (2014)
findings: (H3): The stronger individuals’political ideology, the more they are exposed to a
heterogeneous discussion network.
1.4 News use and network heterogeneity
Scholars disagree whether being embedded in a heterogeneous discussion network leads
individuals to consume more (hard) news media in order to learn about diverse viewpoints,
political arguments and topics (McLeod et al., 1998;Nisbet et al., 2003), or whether the
consumption of news and the encountering of various positions and issues in the news lead
people to seek a more diverse social network (Brundidge, 2010). It has been argued based on
the uses and gratifications theory that individuals will pay more attention to diverse topics in
the news if they also anticipate to interacting with a heterogeneous pool of individuals with a
variety of interests, backgrounds and viewpoints (Eveland, 2004;Scheufele et al., 2004).
McLeod et al. (1999b), for example, concluded that discussions about politics in heterogeneous
networks may enhance reflective thinking about issues. And Scheufele et al. (2004), indeed,
found that individuals who have reported to have a more diverse discussion network were
more likely to use hard news content in newspapers and television—news formats that
demand more cognitive effort when processing information compared to soft news.
Brundidge (2010) has confirmed the reversed hypothesis, showing that online news use is
positively related with a heterogeneous political discussion network, albeit the relationship
was rather small. Looking at news sharing activities but not consumption directly, Choi and
Lee (2015) have shown that news sharing mediates the relationship between social
networking services and network heterogeneity. Similarly, Choi et al. (2017) provided
evidence that news sharing positively moderates the relationship between network
heterogeneity and political participation. However, given that previous research has
neither differentiated between traditional, online or social media news use nor did not
measure active news consumption per se (cf. Choi and Lee, 2015), the relationship between
news use and network heterogeneity is still unclear. We thus pose the third research question:
(RQ3): How is news use (RQ3a: traditional, RQ3b: online, RQ3c: social media) related to
discussion network heterogeneity?
1.5 The mediating role of discussion frequency
The more citizens talk about politics and public affairs, the more they become politically
informed and the more likely they take responsibility as participatory democratic citizens
(e.g. Scheufele et al., 2004;Wyatt et al., 2000). A wide range of research has shown that political
discussion is a determining factor for political participation in various forms (e.g. Gastil et al.,
2002;McLeod et al., 1999a;Scheufele et al., 2004;Wyatt et al., 2000). Furthermore, more
political discussion has been found to enhance understanding for political topics (e.g.
Scheufele et al., 2002) and the integration into a community (Stamm et al., 1997). In addition,
scholars have identified political discussion as a crucial moderator in strengthening political
efficacy when being exposed to political campaigns, either in the newspapers or on the
internet (Nisbet and Scheufele, 2004).
However, the findings reported above solely focus on the uni-directional effect of
discussion frequency on political behavior, but not on mediation processes that are indicative
for a deliberative democracy, such as the extent to which people discuss politics with a
diverse network of opinions and attitudes. What is more, the seminal studies (Choi et al., 2017;
Kim and Chen, 2015;Kwak et al., 2005;Mutz, 2002;Scheufele et al., 2004) in the field of
discussion network heterogeneity have not consistently considered discussion frequency as a
key variable that affects network heterogeneity. While Choi et al. (2017),Mutz (2002) and
Scheufele et al. (2004;2006) simply controlled for the variable when predicting political
participation, Kim and Chen (2015) and Kwak et al. (2005) did not even include discussion
frequency in their analyses. Thus, there is a conceptual and empirical lack of research that
investigates the pivotal role that discussion frequency might play with regard to network
heterogeneity. As follows, this study seeks to investigate the mediating effect of discussion
frequency by posing the final research question: (RQ4): To what extent is the relationship
between political behavior, political attitudes, news use and network heterogeneity mediated by
2.1 Sample and data
The data for this study are based on a two-wave panel online survey which was administered
in the United States. The distribution of the survey was supported by the media polling group
Nielsen. Nielsen uses a stratified quota-sampling method to recruit respondents from a pool of
over 200,000 pre-registered US citizens to reach a sample that is most likely to resemble the
demographic distribution as reported by the US Census. Wave 1 of the survey was carried out
in December 2013. From an initial sample of 5,000 participants, 2,060 individuals responded,
resulting in a relatively high response rate of 34.6%. Data for Wave 2 were collected in March
2014. 1,024 respondents filled out the survey, which indicates an acceptable retention rate of
57% (Watson and Wooden, 2006).
The resulting sample of the panel survey is diverse and comparable with the US national
population as well as similar surveys that use random sampling strategies (e.g. Pew Research
Center for the People and the Press, 2018). Respondents varied demographically regarding
age (mean[M]552.71, standard deviation [SD]514.77), education (range of scale: 1–8,
M53.61, SD 51.44, median [Mdn]5some college), income (ranger of scale 1–8, M54.46,
SD 51.44, Mdn 5US$50,000 –US$59,000), sex (49.9% female) and race (78,1% white).
However, slight differences compared to the US Census were prevalent. The sample of this
study is slightly younger, more educated and includes fewer Hispanics. See the Appendix for
the formulation of the following measures.
2.2.1 Discussion network heterogeneity. Discussion network heterogeneity has been
conceptualized and measured in various ways in past research (Eveland and Hively, 2009).
We followed Moyand Gastil’s (2006) approach in considering network heterogeneity as a
multifaceted concept (i.e. ethnic, social and political), using four distinct items for the
measurement (see also Diehl et al., 2016). Respondents were asked how often (1 5never,
10 5all the time) they talk about politics and public affairs online and offline with people (1)
who disagree with [them], (2) whose political views are different from [theirs], (3) from a
different race or ethnicity and (4) from a different social class (Wave 1: Cronbach’s
alpha 50.94, M53.58, SD 52.45; Wave 2: Cronbach’s alpha 50.93, M53.41, SD 52.36).
2.2.2 Discussion frequency. Discussion frequency was measured by using nine items that
asked respondents how often (1 5never, 10 5all the time) they talk about politics or public
affairs online and offline with their spouse or partner, family relatives, friends, acquaintances,
strangers, neighbors they know well and don’t know well and co-workers they know well and
don’t know well (Cronbach’s alpha 50.87, M53.27, SD 51.74).
2.2.3 Political participation. Political participation was differentiated between offline,
online and social media. For offline political participation, respondents were asked to indicate
on a 10-point Likert scale (1 5never,105all the time), how often [they] have (1) contacted an
elected public official, (2) attended a political rally, (3) participated in any demonstrations,
protests or marches, (4) donated money to a campaign or political cause, (5) participated in
groups that took any location action for social or political reform and f) been involved in
public interest groups, political action groups, political clubs, political campaigns or political
party committees (Cronbach’s alpha 50.89, M52.15, SD 51.78). Regarding online political
participation, respondents were questioned how often [they] (1) signed or shared an online
petition, (2) participated in online polls, (3) participated in an online question and answer
session with a politician or public official, (4) created an online petition and (5) signed up
online to volunteer to help with a political cause (Cronbach’s alpha 50.78, M52.29,
SD 51.70). Finally, social media participation was gauged by inquiring how often [they] (1)
joined a political or cause-related group on a social media site and (2) started a political or
cause-related group on a social media site (Pearson 50.66, M51.56, SD 51.46).
2.2.4 Political efficacy (internal). Internal political efficacy was measured by using three
items derived from previous research (Niemi et al., 1991). Ranging on a 10-point Likert scale
(1 5strongly disagree, 10 5strongly agree), respondents had to indicate to what extent they
think (1) “People like me can influence government”, (2) “I consider myself well qualified to
participate in politics”and (3) “I have a good understanding of the important political issues
facing our country”(Cronbach’s alpha 50.78, M55.17, SD 52.24).
2.2.5 Political interest. Political interest was measured by asking respondents two
questions, following prior research (Niemi et al., 1991). Respondents had to indicate on a 10-
point scale (1 5not at all,105a great deal) (1) how interested [they] are in information about
what’s going on in politics and public affairs and (2) how closely [they] pay attention to
information about what’s going on in politics and public affairs (Pearson 50.93,
M56.67, SD 52.70).
2.2.6 Political knowledge. To measure political knowledge, the respondents were asked eight
questions regarding politics in the US After recoding right and wrong answers (1 5right,
05wrong), the eight items were averaged (Cronbach’s alpha: 0.75, M50.57, SD 50.27).
2.2.7 Political ideology. To measure political ideology, respondents were asked to place
themselves on 11-point scales (0 5strong conservative,115strong liberal) for a) social issues
and b) economic issues. The two items were recoded to form a four-point scale (1 5neutral,
45strong ideology) (Pearson 50.75; M52.36, SD 51.00).
2.2.8 News use. News use was distinguished between traditional news media, online news
media and social media news. Traditional news use (e.g. Gil de Z
niga et al., 2010) was
captured by asking respondents nine items that asked how often they get news (1 5never,
10 5all the time) from network TV news (e.g. ABC, CBS and NBC), local television news (local
affiliate stations), national newspapers (e.g. New York Times, Washington Post and US
Today), local newspapers (e.g. Oregonian, Houston Chronicle and Miami Herald), cable news
(e.g. CNN, Fox News and MSNBC), radio news (e.g. NPR and talk shows), print media,
television media and radio (Cronbach’s alpha 50.77, M55.26, SD 51.86).
Online news use was measured using eight items, asking respondents how often they get
news (1 5never, 10 5all the time) from online news sites (e.g. Gawker, Politico and
BuzzFeed), citizen journalism sites (e.g. CNN’s iReport, Examiner.com), hyperlocal news sites
(e.g. Patch.com or other sites dedicated to news in [their] local community), computer web
browser (laptop or desktop), tablet app or browser (iPad, 7 inches or larger), smartphone app
or browser (handheld mobile device smaller than 7 inches), news aggregators (e.g. Google
News, etc.) and sites and apps that collect news (e.g. Flipboard or Pulse) (Cronbach’s
alpha 50.78, M52.97, SD 51.55).
The variable for social media news use was constructed based on four items (e.g. Gil de
niga et al., 2010), asking respondents how often they use Facebook for getting news,
Twitter for getting news, use social media to stay informed about current events and public
affairs and use social media to get news from mainstream media. Answer categories ranged
on a 10-point scale (1 5never, 10 5all the time; Cronbach’s alpha 50.82,
M52.67, SD 52.06).
2.2.9 Controls. Following previous research on network heterogeneity (e.g. Lee et al., 2014),
we also controlled for daily social media use by asking respondents how often (1 5never,
10 5all the time) they use social media on a typical day (M54.13, SD 52.99). Furthermore,
following Mutz (2002) and others (e.g. Eveland and Hively, 2009), we measured network size to
control for the amount of people with whom respondents usually have political discussions.
We asked respondents about how many people they would say they have talked to a) face-to-
face or over the phone about politics or public affairs and b) via the internet, including e-mail,
chat rooms, social network sites and micro-blogging sites. Given the highly skewed variable
when averaged, a constant (1) was added before applying a log transformation to the variable
(M50.89, SD 50.99, min 50.00, max 55.62). As for demographics, we included age, gender
(1 5male), race (1 5white), annual household income and education in our analyses.
To answer the research questions and test the hypotheses, including the mediating effect of
discussion frequency, we conducted a series of ordinary least squares regression analyses
and causal mediation analyses (Baron and Kenny, 1986) with the computing software R.To
control for non-normal data, the mediation analyses were conducted with 1,000 bootstrap
samples. What is more, the two-wave survey design allowed us to test for tentative causal
directions of the assumed relationships. To do so, we estimated three OLS regression models
(cross-sectional, lagged and autoregressive). It is particularly the autoregressive model that
allows for more confident causal assumptions as it regresses the independent variable (IV)
from wave one on the dependent variable (DV) from wave two while, at the same time,
controlling for the DV from wave one (Finkel, 1995).
The first research question dealt with the relationship between political participation and
network heterogeneity. The OLS results (see Table 1) indicate that none of the forms of
political participation (offline, online and social media) is significantly related to network
heterogeneity. In the second research question we inquired how political efficacy is associated
with network heterogeneity. We equally find no significant relationship in the OLS models.
The first hypothesis posed in this study assumed that individuals who have a strong interest
in politics are more likely to be exposed to a heterogeneous discussion network. The results
show that there is a direct significant positive relationships between political interest and
discussion network heterogeneity in the cross-sectional model (OLS: β50.082, p< 0.05).
Hence, the more people are interested in politics and public affairs, the more likely they seek a
Block 1: Autoregressive term
Constant 0.000 (0.372) 0.000 (0.489) 0.000 (0.570) 0.000 (0.533)
Block 2: Demographics and controls
Age 0.059 (0.004) 0.015 (0.005) 0.024 (0.006) 0.030 (0.005)
Gender (1 5male) 0.011 (0.097) 0.029 (0.123) 0.007 (0.149) 0.020 (0.139)
Race (1 5white) 0.014 (0.122) 0.024 (0.160) 0.022 (0.187) 0.011 (0.175)
Income 0.080** (0.034) 0.057* (0.046) 0.063 (0.053) 0.038 (0.050)
Education 0.030 (0.034) 0.026 (0.044) 0.067 (0.052) 0.056 (0.048)
Social media use 0.019 (0.021) 0.053 (0.027) 0.089* (0.032) 0.066 (0.030)
Network size (log) 0.385*** (0.056) 0.126*** (0.081) 0.213*** (0.094) 0.158*** (0.089)
Block 3: Political antecedents
0.045 (0.049) 0.039 (0.064) 0.093 (0.075) 0.076 (0.070)
0.144** (0.051) 0.062 (0.068) 0.107 (0.079) 0.080 (0.074)
0.058 (0.049) 0.033 (0.064) 0.074 (0.075) 0.060 (0.070)
0.050 (0.029) 0.015 (0.038) 0.074 (0.045) 0.068 (0.042)
Political interest 0.182*** (0.028) 0.082* (0..037) 0.084 (0.043) 0.049 (0.040)
Political knowledge 0.098** (0.232) 0.067 (0.306) 0.019 (0.357) 0.048 (0.334)
Political ideology 0.030 (0.050) 0.012 (0.065) 0.019 (0.076) 0.014 (0.071)
Block 4: News media use
Traditional news 0.135*** (0.030) 0.029 (0.040) 0.077* (0.047) 0.089* (0.044)
Online news 0.095** (0.039) 0.004 (0.052) 0.014 (0.060) 0.016 (0.056)
Social media news 0.037 (0.034) 0.053 (0.045) 0.020 (0.052) 0.043 (0.049)
Block 5: Mediator
–0.596*** (0.053) 0.286*** (0.061) 0.028 (0.069)
0.549 0.629 0.457 0.526
Observations 641 641 641 641
Note(s): OLS regression models predicting discussion frequency and network heterogeneity (cross-sectional,
Cell entries are final-entry ordinary least squares (OLS), standardized Beta (β) coefficients; standard errors in
parentheses; *p< 0.05, **p < 0.01, ***p < 0.001
OLS regression models
frequency and network
heterogeneous discussion network. In the second hypotheses we asserted that individuals
who know more about politics, are more likely to be exposed to a heterogeneous discussion
network. However, we did not find support for this assumption in the OLS models.
The third hypothesis dealt with the positive relationship between political ideology and
network heterogeneity. Here, no significant correlations weredetected in the OLS models either.
In addition, weinquired how news media consumption relates to network heterogeneity (RQ3).
The results reveal that traditional news use is positively related with network heterogeneity,
both in the lagged and autoregressive OLS model (lagged: β50.077, p< 0.05; autoregressive:
β50.089, p< 0.05).Thus, the more people consume traditional news (e.g. newspapers, TV and
radio), the more likely they reside in heterogeneous networks. However, no other form of news
media use has been found to be related with discussion network heterogeneity.
The final research question dealt with the mediating effect of discussion frequency for the
relationship between political attitudes, behavior (see Table 2), news use and discussion
network heterogeneity. The causal mediation analyses have revealed that the positive effect
of political participation online on discussion network heterogeneity is largely dependent on
discussion frequency. Accordingly, when frequently discussing politics, the relationship
between online political participation and network heterogeneity becomes apparent for
individuals. Second, political interest can also lead to a more diverse discussion network over
time, but only if politics is discussed more frequently. Third, while there was no direct effect
of political knowledge on network heterogeneity in the OLS models, discussion frequency
works as a negative mediator; hence decreasing discussion network heterogeneity.
Eventually, the positive effects of traditional news consumption on network heterogeneity
were found to be largely contingent on whether individuals frequently talk to other people
about politics and public affairs. Positive mediation effects were also detected for online news
consumption. For social media news use, the mediation effect was absent, however.
Having recently faced widely debated issues around “information cocoons’,“echo chambers”
and “filter bubbles”, this study aimed at shedding light on what political attitudes, political
behavior as well as news consumption behaviors are indicative for US citizens to talk to
people from diverse backgrounds (e.g. social, ethnical and political). What scholars in political
science and communication science have called “discussion network heterogeneity”
(Mutz, 2002;Scheufele et al., 2004,2006), which has been ascribed a pivotal role for
Discussion frequency →Network heterogeneity
Political participation offline 0.039 0.018 0.002
Political participation online 0.128** 0.059** 0.006
Political participation social media 0.064 0.030 0.003
Political efficacy internal 0.033 0.015 0.001
Political interest 0.096*** 0.046*** 0.04
Political knowledge 0.522* 0.241* 0.023
Political ideology 0.044 0.020 0.002
News media use
Traditional news 0.104*** 0.048*** 0.005
Online news 0.093* 0.043* 0.004
Social media news 0.026 0.012 0.001
Note(s): Cell entries are unstandardized coefficients of the average causal mediation effect (ACME, indirect
effect); nonparametric bootstrap confidence intervals with 1,000 iterations; *p< 0.05, **p < 0.01, ***p < 0.001
deliberative democracies (Habermas, 1989), has so far been mainly considered as an
exogenous variable in theoretical and empirical models (except: Brundidge, 2010;Lee et al.,
2014;Choi and Lee, 2015). However, in this study we sought out to identify the major
antecedents for network heterogeneity, while at the same time, also accounting for the
mediating role of discussion frequency.
Findings revealed that the most important factors that lead people to seek a heterogeneous
discussion network are political interest, political participation online as well as news use
(traditional, online). Political knowledge, on the other hand, has been found to rather attenuate
network heterogeneity. However, while the direct relationship between these factors were
found to be limited (e.g. only present for political interest and traditional news use), it became
clear that citizens need to foster frequent talks about politics and public affairs in order to get
exposed to counter-attitudinal political opinions. Hence, the more politically interested
individuals talk to others about politics and public affairs, the more likely they are to
encounter alternative viewpoints and explanations. Research on selective exposure has
indeed shown that politically interested individuals are not dismissive about opinions and
attitudes that differ from theirs, but that they seek opposing arguments in order to construct
counter-arguments and become prepared for political discussions with the opposition (e.g.
Garrett, 2009;Valentino et al., 2009).
While this mechanism seems to be at work for politically interested citizens, political
intelligence combined with discussion frequency might function as a damper. The results
have shown that politically savvy people are less likely to encounter heterogeneous
discussion networks if they talk about politics and public affairs frequently. Rather, it seems
that politically knowledgeable people—especially if they talk with others about politics and
public affairs more often—might be less inclined to listening to counter-attitudinal
viewpoints. These findings point to a dangerous formation of a politically sophisticated
elite that becomes increasingly delineated from the broader, average political forum
(cf. Putnam, 1976).
What is more, having researched news media effects for a variety of news platforms
(traditional, online and social media), it became apparent that traditional news use can be
considered the strongest predictor for discussion network heterogeneity—and even over
time. Furthermore, the mediation analyses showed that frequent discussions with others
about politics and public affairs do not only mediate the positive effect of traditional news use
on network heterogeneity, it also enhances citizens who consume online news to talk more
often with people from diverse backgrounds. In other words, the more individuals consume
news via traditional and online media outlets and discuss politics with others, the more likely
they are to seek out conversation partners that differ from themselves socially, politically
These findings are not only in line with Brundidge (2010), who has shown that online news
consumption leads to network heterogeneity through discussion frequency; the results also
provide evidence that the previously identified uni-directional relationship between
traditional news use and network heterogeneity (Scheufele et al., 2004,2006) also holds in
the reverse order. Eventually, the reason that we do not find a mediation effect of discussion
frequency for the relationship between social media news use and network heterogeneity can
be well-reasoned. Given that consuming news via social media already implies that people
have a large network and frequent discussions in the online sphere (in most cases), it is
plausible that the mediating effect of discussion frequency seems to become obsolete in our
Lastly, we found that discussion frequency works as a positive mediator for political
participation online affecting network heterogeneity positively. Hence, while political
participation on social media and offline might not necessarily increase the likeliness to be
exposed to different viewpoints, online political participation (e.g. signing online petitions or
participating in a Q&A with politicians online) might represent actions that could help to
burst the so-called “filter bubble.”Explanations for this finding can be found when
considering the nature of the different forms of political participation. Offline political
participation (e.g. participation in rallies, demonstrations or political interest groups)
oftentimes takes place within a common social community (e.g. political party, friends and
social circles who stand up for a shared cause). The same can be said about political
participation on social media. Here, the activity remains in the social media sphere, such as
starting or joining a cause-related group but does not extend beyond the familiar social online
circles. Yet, online political participation, which includes creating, sharing, signing a petition
or signing up to volunteer for a political cause, implies that one does not necessarily need to be
exposed to the same group of people. Petitions can take various forms and can focus on a
diversity of topics. What is more, online tools nowadays enable people to become exposed to a
variety of political opinions and backgrounds that one would not have necessarily
encountered offline or on social media (e.g. emails, newsletters, forums, WhatsApp groups).
Although this study has given new insights into the antecedents and mechanisms of
discussion network heterogeneity, there are certain limitations that need to be taken into
account. First, the data could be considered outdated as the surveys were conducted back in
2013 and 2014, respectively. However, as we are still facing many questions regarding the
outcome of the US elections in 2016, the findings of this study can be deemed indicative for
why US citizens have turned away from seeking alternative political views and rather resided
in social circles that (re-)confirmed their existing political beliefs. Second, we had to rely on
survey data which are widely known in the field to be prone to social desirability,
untrustworthy answers and missing values (cf. decreases the sample for final analyses).
However, rather than relying on a small dataset that might be less generalizable, we opted for
a large-scale, representative survey in order to make more compelling conclusions about the
US population. Future research should, however, focus on tracking-data and survey-
techniques that use smartphone and other intrusive online techniques.
Third, and lastly, we were only able to identify a direct relationship between traditional
news use, political interest and discussion network heterogeneity respectively. All other
variables that measured political attitude and behavior as well as news use did not yield
positive results, albeit the mediation effects revealed more interesting insights. It is in this
vein that we reason that network heterogeneity is part of a recursive process in which
political attitudes, behavior and news consumption mutually influence and reinforce each
other. As a result, this study has given useful insights into the antecedents and mechanisms
of network heterogeneity when considered as an endogenous construct.
What is more, follow-up studies should investigate discussion network heterogeneity as a
social constructive process. More specifically, future research could take a look at the
individual, cognitive processes that take place when citizens encounter counter-attitudinal
political viewpoints, and how these conversations affect their political attitudes and
behaviors (over time). Another possibility could be an intervention study, focusing on the
micro-level processes of promoting network heterogeneity. A prime example in this regard is
the initiative by the Germany newspaper ZEIT “Deutschland spricht” since 2017. A similar
project could be conducted in the US, engaging citizens to meet and talk about politics with
fellow citizens who differ in their political ideas. The results of the personal exchange, the
regular meetings and conversations could be scientifically accompanied by means of follow-
up interviews and surveys, providing insights into the effects of heterogeneous discussion
networks on beliefs and attitudes toward democratic principles.
All in all, the findings of this study lend support that political discussion can be considered
the “social glue”(Sunstein, 2002, p. 9) that keeps democratic societies together and that might
lead to the “bursting”of the alleged “filter bubbles”and “information cocoons”these days.
Thus, there is a need to create platforms in societies to talk about politics and public affairs
more frequently—either privately, at home, in the public sector, in schools, at universities, or
at work. Furthermore, educational institutions and lecturers are in demand to arouse interest
among pupils and students to become engaged with politics and public affairs. Reading the
news regularly (e.g. in the newspapers or online) is part of the process to stay informed and
equip oneself with the necessary tools to participate in political discussions. In short, the
results of this study have shown that strengthening political interest, political participation
online and news media consumption, coupled with more political talks, is the key to ensure a
thriving democracy that is based on citizens who seek to encounter diverse opinions, show
interest in various standpoints and are open to listen to alternative viewpoints.
1. More information on the project (in German): https://www.zeit.de/serie/deutschland-spricht.
Baer, D. (2016), “The ‘filter bubble’explains why trump won and you didn’t see it coming”,The Cut,
available at: https://www.thecut.com/2016/11/how-facebook-and-the-filter-bubble-pushed-
trump-to-victory.html (accessed 4 November 2019).
Baron, R.M. and Kenny, D.A. (1986), “The moderator-mediator variable distinction in social
psychological research: conceptual, strategic, and statistical considerations”,Journal of
Personality and Social Psychology, Vol. 51 No. 6, pp. 1173-1182.
Bennett, S.E. (1986), Apathy in America 1960-1984: Causes and Consequences of Citizen Political
Indifference, Transnational Publishers, Dobbs Ferry, NY.
Berger, J., Meredith, M. and Wheeler, C. (2008), “Contextual priming: where people vote affects how
they vote”,Proceedings of the National Academy of Sciences, Vol. 105 No. 26, pp. 8846-8849.
Berlet, C. and Lyons, M.N. (2018), Right-Wing Populism in America: Too Close for Comfort, The
Guilford Press, New York, NY.
Brundidge, J. (2010), “Political discussion and news use in the contemporary public sphere: the
‘accessibility’and ‘traversability’of the Internet”,Javnost–The Public, Vol. 17 No. 2, pp. 63-82.
Brundidge, J., Garrett, R.K., Rojas, H. and Gil de Z
niga, H. (2014), “Political participation and
ideological news online: “Differential gains”and “differential losses”in a presidential election
cycle”,Mass Communication and Society, Vol. 17 No. 4, pp. 464-486.
Cappella, J.N., Price, V. and Nir, L. (2002), “Argument repertoire as a reliable and valid measure of
opinion quality: electronic dialogue during campaign 2000”,Political Communication, Vol. 19
No. 1, pp. 73-93.
Chadwick, A., O’Loughlin, B. and Vaccari, C. (2017), “Why people dual screen political debates and
why it matters for democratic engagement”,Journal of Broadcasting and Electronic Media,
Vol. 61 No. 2, pp. 220-239.
Choi, J. and Lee, J.K. (2015), “Investigating the effects of news sharing and political interest on social
media network heterogeneity”,Computers in Human Behavior, Vol. 44, pp. 258-266.
Choi, J., Lee, J.K. and Metzger, E. (2017), “Investigating effects of social media newssharing on the
relationship between network heterogeneity and political participation”,Computers in Human
Behavior, No. 75, pp. 25-31.
Davis, N.T. and Dunaway, J.L. (2016), “Party polarization, media choice, and mass partisan-ideological
sorting”,Public Opinion Quarterly, Vol. 80 No. 1, pp. 272-297.
Diehl, T., Weeks, B.E. and Gil de Z
niga, H. (2016), “Political persuasion on social media: Tracing
direct and indirect effects of news use and social interaction”,New Media and Society, Vol. 18
No. 9, pp. 1875-1895.
El-Bermawy, M.M. (2016), “Your filter bubble is destroying democracy”,Wired, available at: https://
www.wired.com/2016/11/filter-bubble-destroying-democracy/ (accessed 4 November 2019).
Eveland, W.P., Jr and Hively, M.H. (2009), “Political discussion frequency, network size, and
‘heterogeneity’of discussion as predictors of political knowledge and participation”,Journal of
Communication, Vol. 59 No. 2, pp. 205-224.
Eveland, W.P., Jr (2004), “The effect of political discussion in producing informed citizens: the roles of
information, motivation and elaboration”,Political Communication, Vol. 21 No. 2, pp. 177-193.
Festinger, L. (1957), A Theory of Cognitive Dissonance, Stanford University Press, Stanford, CA.
Finkel, S.E. (1995), Causal Analysis with Panel Data, Sage, Thousand Oaks, CA.
Garrett, R.K., Carnahan, D. and Lynch, E.K. (2013), “A turn toward avoidance? Selective exposure to
online political information, 2004-2008”,Political Behavior, Vol. 35 No. 1, pp. 113-134.
Garrett, K. (2009), “Politically motivated reinforcement seeking: reframing the selective exposure
debate”,Journal of Communication, Vol. 59 No. 4, pp. 676-699.
Gastil, J., Deess, E.P. and Weiser, P. (2002), “Civic awakening in the jury room: a test of the connection
between jury deliberation and political participation”, The Journal of Politics, Vol. 64 No. 2,
Gerstenfeld, P.B., Grant, D. and Chiang, C.-P. (2003), “Hate online: a content analysis of extremist
internet sites”,Analyses of Social Issues and Public Policy, Vol. 3 No. 1, pp. 29-44.
Gil de Z
niga, H., Veenstra, A., Vraga, E. and Shah, D. (2010), “Digital democracy: reimagining
pathways to political participation”,Journal of Information Technology & Politics, Vol. 7 No. 1,
Gil de Z
niga, H. and Valenzuela, S. (2011), “The mediating path to a stronger citizenship: online and offline
networks, weak ties, and civic engagement”,Communication Research, Vol. 38 No. 3, pp. 397-421.
Gunter, J. and Hughes, R. (2018), “Charlottesville remembered: ‘A battle for the soul of America”,BBC,
available at: https://www.bbc.com/news/world-us-canada-44619374 (accessed 4 November 2019).
Habermas, J. (1989), The Structural Transformation of the Public Sphere: An Inquiry into a Category of
Bourgeois Society, MIT Press, Cambridge, MA.
Hayes, A.F., Scheufele, D.A. and Huge, M.E. (2006), “Nonparticipation as self-censorship: publicly
observable political activity in a polarized opinion climate”,Political Behavior, Vol. 28 No. 3,
Horan, P.M. (1971), “Social positions and political cross-pressures: a re-examination”,American
Sociological Review, Vol. 36 No. 4, pp. 650-660.
Huckfeldt, R. and Morehouse Mendez, J. (2008), “Moths, flames, and political engagement: managing
disagreement within communication networks”,The Journal of Politics, Vol. 70 No. 1, pp. 83-96.
Huckfeldt, R. and Sprague, J.D. (1995), Citizens, Politics, and Social Communication: Information and
Influence in an Election Campaign, Cambridge University, Cambridge, UK.
Huckfeldt, R., Morehouse Mendez, J. and Osborn, T. (2004), “Disagreement, ambivalence, and
engagement: the political consequences of heterogeneous networks”,Political Psychology, Vol. 25
No. 1, pp. 65-95.
Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N. and Westwood, S.J. (2019), “The origins and
consequences of affective polarization in the United States”,Annual Review of Political Science,
Vol. 22, pp. 129-146.
Kim, Y. and Chen, H.T. (2015), “Discussion network heterogeneity matters: examining a moderated
mediation model of social media use and civic engagement”,International Journal of
Communication, Vol. 9, pp. 2344-2365.
Kim, Y., Hsu, S.H. and Gil de Z
niga, H. (2013), “Influence of social media use on discussion network
heterogeneity and civic engagement: the moderating role of personality traits”,Journal of
Communication, Vol. 63 No. 3, pp. 498-516.
Kim, M. (2018), “How does Facebook news use lead to actions in South Korea? The role of Facebook
discussion network heterogeneity, political interest, and conflict avoidance in predicting
political participation”,Telematics and Informatics, Vol. 35 No. 5, pp. 1373-1381.
Knobloch-Westerwick, S. and Meng, J. (2009), “Looking the other way: selective exposure to attitude-
consistent and counterattitudinal political information”,Communication Research, Vol. 36 No. 3,
Knobloch-Westerwick, S. and Meng, J. (2011), “Reinforcement of the political self through selective
exposure to political messages”,Journal of Communication, Vol. 61 No. 2, pp. 349-368.
Knoke, D. (1990), Political Networks: The Structuralist Perspective, Cambridge University Press, New
Kraft, P.W., Lodge, M. and Taber, C.S. (2015), “Why people ‘don’t trust the evidence’: motivated
reasoning and scientific beliefs”,The Annals of the American Academy of Political and Social
Science, Vol. 658, pp. 121-133.
Kwak, N., Williams, A.E., Wang, X. and Lee, H. (2005), “Talking politics and engaging politics: an
examination of the interactive relationships between structural features of political talk and
discussion engagement”,Communication Research, Vol. 32 No. 1, pp. 87-111.
Lazarsfeld, P.F., Berelson, B. and Gaudet, H. (1944), The People’s Choice: How the Voter Makes up His
Mind in a Presidential Campaign, Columbia University Press, New York, NY.
Lee, J. and Choi, Y. (2020), “Effects of network heterogeneity on social media on opinion polarization
among South Koreans: focusing on fear and political orientation”,International Communication
Gazette, Vol. 82 No. 2, pp. 119-139.
Lee, J.K., Choi, J., Kim, C. and Kim, Y. (2014), “Social media, network heterogeneity, and opinion
polarization”,Journal of Communication, Vol. 64 No. 4, pp. 702-722.
Levine, J.M. and Russo, E. (1995), “Impact of anticipated interaction on information acquisition”,Social
Cognition, Vol. 13 No. 3, pp. 293-317.
Lu, Y., Lee, J. and Kim, E. (2018), “Network characteristics matter in politics on Facebook: evidence
from a US national survey”,Online Information Review, Vol. 42 No. 3, pp. 372-386.
McLeod, J.M., Sotirovic, M. and Holbert, R.L. (1998), “Values as sociotropic judgments influencing
communication patterns”,Communication Research, Vol. 25 No. 5, pp. 254-480.
McLeod, J.M., Scheufele, D.A. and Moy, P. (1999a), “Community, communication, and participation: the
role of mass media and interpersonal discussion in local political participation”,Political
Communication, Vol. 16 No. 3, pp. 315-336.
McLeod, J.M., Scheufele, D.A., Moy, P., Horowitz, E.M., Holbert, R.L., Zhang, W., Zubric, S. and Zubric,
J. (1999b), “Understanding deliberation. The effects of discussion networks on participation in a
public forum”,Communication Research, Vol. 26 No. 6, pp. 743-774.
Meffert, M.F., Chung, S., Joiner, A.J., Waks, L. and Garst, J. (2006), “The effects of negativity and
motivated information processing during a political campaign”,Journal of Communication,
Vol. 56 No. 1, pp. 27-51.
Moy, P. and Gastil, J. (2006), “Predicting deliberative conversation: the impact of discussion networks,
media use, and political cognitions”,Political Communication, Vol. 23 No. 4, pp. 443-460.
Mutz, D.C. and Mondak, J.J. (2006), “The workplace as a context for cross-cutting political discourse”,
The Journal of Politics, Vol. 68 No. 1, pp. 140-155.
Mutz, D.C. (2002), “The consequences of cross-cutting networks for political participation”,American
Journal of Political Science, Vol. 46 No. 4, pp. 848-855.
Mutz, D.C. (2006), Hearing the Other Side: Deliberative versus Participatory Democracy, Cambridge
University Press, New York, NY.
Nelson, J.L. and Webster, J.G. (2017), “The myth of Partisan selective exposure: a portrait of the online
political news audience”,Social Media þSociety, Vol. 3 No. 3, pp. 1-13.
Neuman, W.R. (1986), The Paradox of Mass Politics: Knowledge and Opinion in the American
Electorate, Harvard University Press, Cambridge, MA.
Niemi, R.G., Craig, S.C. and Mattei, F. (1991), “Measuring internal political efficacy in the 1988 national
election study”,American Political Science Review, Vol. 85 No. 4, pp. 1407-1413.
Nir, L. (2005), “Ambivalent social networks and their consequences for participation”,International
Journal of Public Opinion Research, Vol. 17 No. 4, pp. 422-442.
Nisbet, M.C. and Scheufele, D.A. (2004), “Political talk as a catalyst for online citizenship”,Journalism
and Mass Communication Quarterly, Vol. 81 No. 4, pp. 877-896.
Nisbet, M.C., Moy, P. and Scheufele, D.A. (2003), “Religion, communication, and social capital”, Paper
Presented to the Political Communication Division of the International Communication
Association (ICA), May, San Diego.
Pariser, E. (2011), The Filter Bubble: What the Internet Is Hiding from You, Penguin, New York, NY.
Park, N., Kee, K.F. and Valenzuela, S. (2009), “Being immersed in social networking environment”,
CyberPsychology and Behavior, Vol. 12 No. 6, pp. 729-733.
Perrin, A. and Jiang, J. (2018), “Almost a quarter of US adults say they are ‘almost Constantly”, Online,
World Economic Forum, available at: https://www.weforum.org/agenda/2018/03/about-a-
quarter-of-u-s-adults-say-they-are-almost-constantly-online/ (accessed 4 November 2019).
Pew Research Center for the People and the Press (2018), “Methodology”,The American Trends Panel
Survey Methodology, available at: http://www.people-press.org/2018/10/15/methodology-207/
(accessed 4 November 2019).
Price, V., Cappella, J.N. and Nir, L. (2002), “Does more disagreement contribute to more deliberative
opinion?”,Political Communication, Vol. 19 No. 1, pp. 95-112.
Prior, M. (2007), Post-broadcast Democracy: How Media Choice Increases Inequality in Political
Involvement and Polarizes Elections, Cambridge University Press, New York, NY.
Putnam, R.D. (1976), The Comparative Study of Political Elites, Prentice Hall, New Jersey, NJ.
Quan-Haase, A. and Young, A.L. (2010), “Uses and Gratifications of Social Media”,Bulletin of Science,
Technology and Society, Vol. 30 No. 5, pp. 350-361.
Rojas, H. (2008), “Strategy versus understanding: how orientations toward political conversation
influence political engagement”,Communication Research, Vol. 35 No. 4, pp. 452-480.
Scheufele, D.A., Shanahan, J. and Kim, S. (2002), “Who cares about local politics? Media influences on
local political involvement, issue awareness, and attitude strength”,Journalism and Mass
Communication Quarterly, Vol. 79 No. 2, pp. 427-444.
Scheufele, D.A., Nisbet, M.C., Brossard, D. and Nisbet, E.C. (2004), “Social structure and citizenship:
examining the impact of social setting, network heterogeneity, and informational variables on
political participation”,Political Communication, Vol. 21 No. 3, pp. 315-338.
Scheufele, D.A., Hardy, B.W., Brossard, D., Waismel-Manor, I.S. and Nisbet, E.C. (2006), “Democracy
based on difference: examining the links between structural heterogeneity, heterogeneity of
discussion networks, and democratic citizenship”,Journal of Communication, Vol. 56 No. 4,
Sears, D.O. and Freedman, J.L. (1967), “Selective exposure to information: a critical review”,Public
Opinion Quarterly, Vol. 31 No. 2, pp. 194-213.
Seeman, M. (1966), “Alienation, membership, and political knowledge: a comparative study”,Public
Opinion Quarterly, Vol. 30 No. 3, pp. 353-367.
Sood, G. and Iyengar, S. (2016), “Coming to dislike your opponents: the polarizing impact of political
campaigns”,Stanford University, SSRN, Rochester, NY, available at: https://ssrn.com/
abstract52840225 (accessed 10 May 2020).
Sorensen, G. (2018), Democracy and Democratization: Processes and Prospects in a Changing World,
Routledge, New York, NY.
Stamm, K.R., Emig, A.G. and Hesse, M.B. (1997), “The contribution of local media to community
involvement”,Journalism and Mass Communication Quarterly, Vol. 74 No. 1, pp. 97-107.
Stroud, N.J. (2010), “Polarization and partisan selective exposure”,Journal of Communication, Vol. 60
No. 3, pp. 556-576.
Sunstein, C.R. (2001), Republic.com, Princeton University, Princeton, NJ.
Sunstein, C.R. (2002), “The law of group polarization”,The Journal of Political Philosophy, Vol. 10 No. 2,
Sunstein, C.R. (2006), Infotopia: How Many Minds Produce Knowledge, Oxford University Press, New
Sunstein, C.R. (2018), #Republic: Divided Democracy in the Age of Social Media, Princeton University
Press, Princeton, NJ.
Taber, C.S. and Lodge, M. (2006), “Motivated skepticism in the evaluation of political beliefs”,
American Journal of Political Science, Vol. 50 No. 3, pp. 755-769.
Valentino, N.A., Hutchings, V.L., Banks, A.J. and Davis, A.K. (2009), “Selective exposure in the Internet
age: the interaction between anxiety and information utility”,Political Psychology, Vol. 30 No. 4,
Verba, S. and Nie, N.H. (1972), Participation in America: Political Democracy and Social Equality,
University of Chicago Press, Chicago, IL.
ık¸ e-Freiberga, V., D€
aubler-Gmelin, H., Hammersley, B. and Pessoa Maduro, L.M.P. (2013), “A free and
pluralistic media to sustain European democracy”,Report, High Level Group on Media Freedom
and Media Pluralism, available at: http://static.eu2013.lt/uploads/documents/Programos/
20European%20Democracy.pdf (accessed 4 November 2019).
Watson, N. and Wooden, M. (2006), “Modelling longitudinal survey response: the experience of the
HILDA survey”,inACSPRI Social Science Methodology Conference, pp. 10-13.
Wojcieszak, M. and Price, V. (2010), “Bridging the divide or intensifying the conflict? How
disagreement affects strong predilections about sexual minorities”,Political Psychology, Vol. 31
No. 3, pp. 315-339.
Wyatt, R.O., Katz, E. and Kim, J. (2000), “Bridging the spheres: political and personal conversation in
public and private spaces”,Journal of Communication, Vol. 50 No. 1, pp. 71-92.
Zuiderveen Borgesius, F., Trilling, D., M€
oller, J., Bod
o, B., de Vreese, C. and Helberger, N. (2016),
“Should we worry about filter bubbles?”,Internet Policy Review, Vol. 5 No. 1.
1. Discussion network heterogeneity
How often do you talk about politics and public affairs online and offline with...
(1 5never, 10 5all the time)
(1) Who disagree with you
(2) Whose political views are different from your
(3) From a different race or ethnicity
(4) From a different social class
2. Discussion frequency
How often do you talk about politics and public affairs online and offline with...
(1 5never, 10 5all the time)
(1) Spouse or partner
(2) Family relatives
(6) Neighbors you know well
(7) Neighbors you don’t know well
(8) Co-workers you know well
(9) Co-workers you don’t know well
3. Political participation offline
How often have you...
(1 5never, 10 5all the time)
(1) Contacted an elected public official
(2) Attended a political rally
(3) Participated in any demonstrations, protests, or marches
(4) Donated money to a campaign or political cause
(5) Participated in groups that took any location action for social or political reform
(6) Been involved in public interest groups, political action groups, political clubs, political
campaigns, or political party committees
4. Political participation online
How often have you...
(1 5never, 10 5all the time)
(1) Signed or shared an online petition
(2) Participated in online polls
(3) Participated in an online question and answer session with a politician or public official
(4) Created an online petition
(5) Signed up online to volunteer to help with a political cause
5. Political participation social media
How often have you...
(1 5never, 10 5all the time)
(1) Joined a political or cause-related group on a social media site
(2) Started a political or cause-related group on a social media site
6. Political efficacy (internal)
To what extent do you agree with the following statements...
(1 5strongly disagree, 10 5strongly agree)
(1) People like me can influence government
(2) I consider myself well qualified to participate in politics
(3) I have a good understanding of the important political issues facing our country
7. Political interest
(1) How interested are you in information about what’s going on in politics and public affairs?
(1 5not at all, 10 5a great deal)
(2) How closely are you paying attention to information about what’s going on in politics and public
affairs? (1 5not at all, 10 5a great deal)
8. Political knowledge
(answers recoded as: 1 5right, 0 5wrong)
(1) What job or political office does Joe Biden currently hold?
(2) For how many years is a United States Senator elected –that is, how many years are there in one
full term of office for a US Senator?
(3) What job or political office does John Roberts currently hold?
(4) On which of the following does the US federal government currently spend the least (a: Foreign
Aid; b: Medicare; c: National defense; d: Social Security; e: Don’t know)
(5) Do you happen to know whether the immigration bill before Congress was introduced by:
(6) Do you happen to know whether the ruling of the Supreme Court about Obamacare was?
(7) Which organization’s documents were released by Edward Snowden?
(8) Recently, the UN and US were in negotiations with the Syrian government over the removal of:
9. Political ideology
(1) On social issues, where would you place yourself on a scale of 0–10, where 0 5strong
conservative and 10 5strong liberal?
(2) On economic issues, where would you place yourself on a scale of 0–10, where 0 5strong
conservative and 10 5strong liberal?
10. Traditional news use
How often do you get news...
(1 5never, 10 5all the time)
(1) From network TV news (e.g. ABC, CBS, NBC)
(2) Local television news (local affiliate stations)
(3) National newspapers (e.g. New York Times, Washington Post, US Today)
(4) Local newspapers (e.g. Oregonian, Houston Chronicle, Miami Herald)
(5) Cable news (e.g. CNN, Fox News, MSNBC)
(6) Radio news (e.g. NPR, talk shows)
(7) Print media
(8) Television media
11. Online news use
How often do you get news...
(1 5never, 10 5all the time)
(1) From online news sites (e.g. Gawker, Politico, BuzzFeed)
(2) Citizen journalism sites (e.g. CNN’s iReport, Examiner.com)
(3) Hyperlocal news sites (e.g. Patch.com or other sites dedicated to news in your local community)
(4) Computer web browser (laptop or desktop)
(5) Tablet app or browser (iPad, 7 inches or larger)
(6) Smartphone app or browser (handheld mobile device smaller than 7 inches)
(7) News aggregators (e.g. Google News etc.)
(8) Sites and apps that collect news, such as Flipboard, or Pulse
12. Social media news use
(1 5never, 10 5all the time)
(1) How often do you use Facebook for getting news?
(2) How often do you use Twitter for getting news?
(3) How often do you use social media to stay informed about current events and public affairs?
(4) How often do you use social media to get news from mainstream media?
About the authors
Nadine Straußis currently a Marie Sklodowska-Curie Research Fellow at the Smith School of Enterprise
and the Environment, University of Oxford. Previously she has worked as a postdoc at the Media
Innovation Lab of the Department of Communication at the University of Vienna. She obtained her Ph.D.
from the Amsterdam School of Communication Research (ASCoR), University of Amsterdam. Her
research interests include political communication, journalism studies, financial communication,
sustainability and news use. Nadine Straußis the corresponding author and can be contacted at: nadine.
noz is a lecturer in Journalism at the Department of Communication Sciences at the
Universitat Jaume I of Castell
o (UJI) in Spain. She has a degree in Journalism and a Master’s Degree in
New Trends and Innovation Processes in Communication and holds a PhD from UJI. She also graduated
in Political and Administration Sciences from the Universitat Pompeu Fabra of Barcelona. Her research
focuses on the impact of the Internet on democracy and political activism, on the use of social media by
political actors, citizens and the media and on the study of the populist phenomena in the digital
Homero Gil de Z
niga, Ph.D. in Politics at Universidad Europea de Madrid and Ph.D. in Mass
Communication at University of Wisconsin - Madison, serves as Distinguished Research Professor at
University of Salamanca where he directs the Democracy Research Unit (DRU), as Professor at
Pennsylvania State University, and as Senior Research Fellow at Universidad Diego Portales, Chile. His
research addresses the influence of new technologies and digital media over people's daily lives, as well
as the effect of such use on the overall democratic process.
For instructions on how to order reprints of this article, please visit our website:
Or contact us for further details: firstname.lastname@example.org