722657

research-article2017
JOU0010.1177/1464884917722657JournalismBae

Article

Journalism

The social mediation of

2020, Vol. 21(10) 1522­–1538
© The Author(s) 2017
Article reuse guidelines:
political rumors: Examining sagepub.com/journals-permissions
DOI: 10.1177/1464884917722657
https://doi.org/10.1177/1464884917722657
the dynamics in social media journals.sagepub.com/home/jou

and belief in political rumors

Soo Young Bae

University of Massachusetts Amherst, USA

Abstract
Using survey data of social media users in South Korea, this study investigates the
dynamics of political rumors in online social networks. Findings of this study reveal the
significant connection between the users’ reliance on social media as a source for news
and their beliefs in political rumors. Taking a step further, this study underscores the
need to understand how users process misinformation they receive through online social
networks. Drawing attention to the role of network characteristics in the construction
of beliefs around political rumors, this study provides a more nuanced understanding
of the conditions under which rumors and misinformation can be regarded as more
believable.

Keywords
Information flow, network homogeneity, political rumors, quantitative, social media

Introduction
Fake news and unverified rumors are proliferating in online social networks. While the
circulation of gossip and rumor is not new, the recent explosion of online platforms that
facilitate social networking is considered to have opened up the pathway for rumors to
spread through these networks at an unprecedented speed. For instance, in April 2013,
the US stock market experienced an abrupt sell-off that caused the S&P500 to decline by
0.9%, which wiped out US$130 billion in stock value in a matter of seconds (Matthews,

Corresponding author:
Soo Young Bae, Department of Communication, University of Massachusetts Amherst, Integrative Learning
Center N366, 650 N. Pleasant St. Amherst, MA 01003, USA.
Email: sooyoungbae@umass.edu
Bae 1523

2013). This incident was caused by a false tweet from the Associated Press’s Twitter
account, which told its millions of followers that President Obama had been injured in an
explosion at the White House (Bradshaw et al., 2013). While the Associated Press
quickly addressed the situation by informing its readers that the Twitter account had been
hacked and that the information was false, the single tweet was enough to send a shock
wave through the stock market (Matthews, 2013). It was during the same month that
social networking sites played host to a series of rumors and speculations about the sus-
pects of the Boston marathon bombing (Bradshaw et al., 2013). The difference in the two
incidents is that the Associated Press case involved misuse of the ‘verified’ source that is
otherwise widely accepted as reporting confirmed facts, while the search for the bomb-
ing suspects involved the online crowd to take part in the ongoing speculations. Both
cases, however, clearly illustrate the risk and real-world consequences associated with
the dissemination of inaccurate information through social media.
While social media have grown as one of the most popular platforms where users can
freely interact with their friends connected through networks, the spread of falsehoods
and misinformation have also raised serious concerns. In an environment where the flow
of information is increasingly embedded in social relationships, it remains unclear how
individuals evaluate which piece of information is trustworthy or believable. Given this
context, this study focuses on the relationship between social media use and political
rumors, with the fundamental interest in clarifying the process through which informa-
tion circulated in online social networks is received and evaluated by individuals. Moving
this discussion forward, this study examines the conditions under which rumors can be
perceived as more believable, by taking into account the characteristics of the individu-
als’ online social networks.

Literature review
Media use and political rumors
Before we proceed with exploring the relationship between media use and political
rumors, it is useful to review the meaning of political rumors. Political rumors generally
refer to ‘unverified information – information that lacks a secure standard of evidence –
that spreads among a group of people because it promises to resolve uncertainty or pro-
vide new insight into important social or political phenomena’ (Garrett, 2011: 256).
While the study of rumors dates back to several decades, scholars agree that the field
needs more theorizing on political rumors (DiFonzo and Bordia, 2007; Donovan, 2007).
Given that a healthy participatory democracy depends on the active engagement of an
informed citizenry, the role of the media in providing accurate information becomes
particularly relevant. Research has shown that individuals’ use of different media outlets
can significantly impact their knowledge of public affairs (Moy et al., 2005), as well as
their perceptions around an issue (Pierce et al., 1990). With the rise of the new commu-
nication technologies, particularly the Internet, scholars have raised questions about their
potential impact on the dissemination of falsehoods and political rumors, drawing atten-
tion to the different mediation process that information undergoes prior to reaching an
audience.
1524 Journalism 21(10)

That is, while information in traditional mass media is selected, edited, and dissemi-
nated by professional gatekeepers and is identified with authoritative sources, such edi-
torial policies in many online information platforms are often nonexistent, unknown, or
ambiguous on various websites on the Internet (Flanagin and Metzger, 2000). The new
media environment has transformed the expressive potential of the average citizen, and
individuals are now in a position to post messages and images that can be viewed
instantly by global audiences at minimal costs (Lupia and Sin, 2003). The Internet offers
a variety of ways to disseminate news and information, and the process prior to publica-
tion differs significantly from that of the traditional media.
Such characteristics of the online environment led researchers to predict that the
Internet would serve not only as an information highway but also as a path for political
rumoring (Ayres, 1999). In a study of ‘legitimate media’ (daily newspapers and network
TV news) and ‘less legitimate media’ (blogs and grocery store tabloids), Stempel et al.
(2007) found that reading daily newspapers and their website versions were negatively
associated with beliefs about the conspiracies regarding the US government’s involve-
ment in the 9/11 attacks. In contrast, frequently consuming news from online blogs and
reading tabloids were positively associated with beliefs in the conspiracy, controlling for
uses of other media (Stempel et al., 2007). Moreover, investigations of federal officials
report that numerous websites and online newsgroups contain questionable information
about major medical conditions, and the Securities and Exchange Commission report
filing charges against stock promoters who created and circulated false rumors about
their own company’s stock prices on the Internet (Knight Foundation, 2001). These find-
ings suggest that while the Internet has certainly enhanced the ability of its users to
quickly access and disseminate information, it remains questionable whether it also
opened the door to misinformation and circulation of rumors (Garrett, 2011). The spread
and belief in unverified information in the online media landscape becomes particularly
more compounding as a growing number of Internet users embrace social media as alter-
native tools for acquiring news and information.

Theoretical background of rumoring in online social networks

Originally created as a website for connecting social ties, social media has grown as a
central platform for sharing news. The boundaries between political and apolitical com-
municative spaces are blurring online (Brundidge, 2010), and news and political conver-
sations are flowing seamlessly into social networking sites (Bakshy et al., 2009). Whereas
the content shared in social networking sites is not necessarily expected to be about poli-
tics or social issues, users increasingly report that news and public affairs information
have become increasingly visible. In fact, recent statistics indicate that 62% of adults in
the United States get their news from social media (Pew Research Center, 2016). The
presence of news in social networking sites has grown to the extent that the window
through which the public views the world is increasingly the Facebook newsfeed or
Twitter timeline, rather than the front page of the New York Times, particularly for the 18
to 30 age group demographics (Messing and Westwood, 2012).
While the large-scale diffusion of news facilitated through social media can politi-
cally engage citizens to take part in collective action (Papacharissi and de Fatima
Bae 1525

Oliveira, 2012), the same network structure can also be leveraged to disseminate incor-
rect information or unverified rumors (Qazvinia et al., 2011). Even at the early develop-
ment stages of computer-mediated social networks, some researchers expressed fear that
these social networks would accelerate the flow of misinformation, including rumors,
complaints, jokes, and subversive communications (Finholt and Sproull, 1990). Despite
these general concerns about the potential outcomes of social media use, there is a need
for establishing a theoretical understanding about the relationship between online politi-
cal rumoring and social media platforms. In what follows, I will present a theoretical
framework that explains the process of information transmission and reception in online
social networks, and its implications on rumors and misinformation.
The creation and proliferation of social media represent a transformation of the online
information ecology, where information is increasingly merged with sociality. Since the
flow of information is engrained in existing social relations, the outcomes of information
exposure can vary depending on the social relationships between users in online net-
works. For instance, information can seem more credible when it comes from a trusted
source or an opinion-leader in the social network. This information can be perceived to
be even more credible when it resonates within the users’ network, as shown through
various indicators of endorsement. This prediction seems intuitive, as previous research
findings have established the effectiveness of source information and heuristic cues
(Chaiken, 1980; Kahneman and Tversky, 1979; Petty and Cacioppo, 1986).
Nevertheless, this mechanism provides an interesting theoretical framework to
explore the process through which misinformation and unverified rumors are propagated
in online social networks. While data scientists are detecting the presence of political
rumors and misinformation in networks through the use of machine learning, the field
lacks an understanding of the underlying mechanism that makes false information more
believable. Critics have raised caution that extant research has focused more on how
heuristic cues serve as rational strategies for effective decision making, overlooking the
potential problems associated with their inappropriate use (Kuklinski et al., 2000). The
enhanced visibility of social cues can lead individuals to automatically rely on others’
endorsement in their social networks, without worrying about the accuracy of informa-
tion, perhaps in the absence of any contextual knowledge.
Moreover, rumors do occur not in a vacuum but proliferate in the context of social
networks (DiFonzo and Bordia, 2007). Word-of-mouth referrals have been empirically
demonstrated to be extremely effective mechanisms of persuasion (Brown and Reingen,
1987), and the role of intermediary opinion leaders in shaping people’s understanding
of public affairs has been illustrated in previous studies (Lazarsfeld et al., 1944). In a
study of political rumors on the Internet, Garrett (2011) found that using emails with
friends and family members as a source for political information significantly increased
the likelihood of both exposure to and belief in political rumors. This finding was con-
trasted against the impact of using mainstream news media, which in fact increased
exposure to rumor rebuttal and showed no significant relationship with belief in rumors
(Garrett, 2011).
In light of these findings, this study predicts that users who rely more heavily on
social media as a source for news will experience political rumors differently from those
who also consume traditional news media. Although traditional media are faced with
1526 Journalism 21(10)

similar challenges of the changing news environment, the information published through
these outlets undergo a gatekeeping process, which differentiates it from the information
that is floating freely in online social networks. More specifically, this study predicts that
using social media as a news source will be positively associated with the users’ belief in
political rumors. When users who rely on social media for news also consume traditional
news media, it is expected to reduce the likelihood of believing in political rumors, and
weaken the relationship between social media use and rumor beliefs.

Hypothesis 1 (H1): Using social media as a news source more frequently will be
associated with belief in political rumors.
Hypothesis 2 (H2): The use of traditional news media will negatively moderate the
relationship between social media use as a news source and belief in political
rumors.

The resonance of information in social networks

Research suggests that all interpersonal relations and constructions of a citizen’s social
network serve as a filter on the macro environmental flow of political information
(Huckfeldt and Sprague, 1995). The consequences of the larger information environment
often depend on the existence of microenvironments and social networks, which expose
citizens to surrounding opinion distributions. Since rumoring occurs in social contexts,
the characteristics of the network play a key role in determining the impact of the rumor
(Watts, 2003). Studies have shown that rumors circulate more within rather than across
homogeneous networks, and tend to produce disproportionate exposure to particular
rumors (DiFonzo and Bordia, 2007). Homogeneous networks also provide less informa-
tional diversity and fewer opportunities to detect inaccuracies (DiFonzo, 2010). In a
study of networks with different levels of homogeneity, the same rumor was more likely
to be dominant in a tightly knit and clustered network (Trpevski et al., 2010), showing
that individuals involved in homogenous social networks are more likely to have been
exposed to the same rumor (Festinger et al., 1948).
In addition, there is an increased risk of a reinforcing spiral in homogeneous networks
(Slater, 2007), and the positive feedback loop within closed groups is likely to invoke a
particular interpretation of events or issues to become more prevalent in the network.
Rumors tend to gradually acquire more credibility and appeal, as more network neigh-
bors acquire them (Ratkiewicz et al., 2011). Individuals’ perceptions around the credibil-
ity of information can be socially constructed in networks. Although traditional news
media can follow-up to verify and correct misinformation, these efforts can be overshad-
owed once the infiltrated content becomes widely circulated, and the users’ perceptions
are influenced by the apparent resonance of the information among their close friends.
As established in previous research, social consensus can significantly influence indi-
viduals’ perceptions to the extent that they sometimes lose confidence in their initially
accurate appraisals (Asch, 1956; Festinger, 1950). After rumors cross the threshold level
to become widespread in a network, they tend to be considered as ‘common knowledge’
among members of that network (Ratkiewicz et al., 2011). Thus, this study predicts that
Bae 1527

higher levels of homogeneity in online social networks will lead to a greater likelihood
of resonance of a particular rumor in the social network. More specifically, the present
study predicts that perceptions of political similarity of users’ online social networks will
strengthen the association between using social media as a news source and belief in
political rumors.

Hypothesis 3 (H3): Political similarity in online social networks will strengthen the
link between using social media as a news source and belief in political rumors.

The South Korean context

This study examined social media users in South Korea and their beliefs in two political
rumors. South Korea is known as one of the most wired countries in the world, ranking
among the top 10 with broadband connection worldwide, and a mobile phone penetration
rate that exceeds 100% (Organisation for Economic Co-operation and Development
[OECD], 2013). South Korean Internet users have actively adopted social media in their
daily lives and social media use expanded rapidly to engage not only the vast majority of
younger adults, but also a substantial proportion of older adults (Korea Internet &
Security Agency, 2016). With the widespread popularity of social media across different
age groups and a dynamic political landscape, the South Korean context offers an excel-
lent opportunity to explore the relationship between political rumors in social media.
The South Korean political landscape can be characterized by a continuous contesta-
tion between the left-wing progressive party and the right-wing conservative party. There
is a clear ideological divide between the two parties, and their close competition in elec-
tions at times leads to mudslinging. An important and unique factor that influences South
Korean politics is the tension with North Korea and its relationship with the United
States (Steinberg and Shin, 2006). While right-wing politicians support the military alli-
ance between South Korea and the United States, some skeptics raise doubts about the
role of the United States in resolving the tension between the two Koreas. The two
rumors in this study are about the sinking of a South Korean navy ship by North Korea
(Rumor 1), and about a candidate from the conservative party running for the Seoul
mayoral election, the capital of South Korea (Rumor 2; see the appendix for more details
on each rumor).

Method
Data
To evaluate the predictions about the dynamics of political rumors in the online informa-
tion environment, this study examined data collected through an online survey in South
Korea in March 2013. Participants were recruited through an international online survey
firm with a prerecruited national panel of approximately 960,000 members that mirror
the demographic characteristics of the 20- to 59-year-old population in South Korea.
From the panel, 4,203 respondents were randomly chosen and email solicitations were
sent out to participate in the survey. The sample (n = 205) was drawn via a quota sam-
pling method. Previous research has shown that older adults tend to be dramatically
1528 Journalism 21(10)

underrepresented in online surveys, which presents a methodological challenge to con-

struct a representative sample of the older population who are using social media. Thus,
this study defined the study population as those 18 to 59 years old, and used a quota-
stratified sampling method to equally represent four age groups, 20 to 29 (n = 47), 30 to
39 (n = 50), 40 to 49 (n = 50), and 50 to 59 (n = 58), and an equal ratio of gender, males
(n = 101) and females (n = 104). This stratified quota sampling method differs markedly
from more conventional probability sample procedures, yet produces highly comparable
data (Putnam, 2000; Putnam and Yonish, 1999).
The online survey was closed once each subgroup (age and gender) reached its quota,
and thus the response rate was not calculated. There were 137 additional respondents
who were originally included in the sample, but excluded later from it because they did
not meet the criteria for participation (i.e., not an active user of social networking sites).
Despite the limitations of a nonprobability sampling method, previous studies have uti-
lized the same quota sampling design across gender and age groups to enhance the rep-
resentativeness of the sample (see Kim, 2012).
Key demographic characteristics of the sample resemble the profiles of the population
figures reported by the South Korean government’s statistical agency, Statistics Korea,
with respect to age (mean age in both the sample and population is 39) and gender
(50.3% female in the sample and 49.0% in the population). The annual median house-
hold income of the population (KRW4,090,000) was within the same range as the sample
median (KRW4,000,000–KRW5,000,000). There was a greater proportion of respond-
ents with higher education in the sample (62.4% have a college degree) as compared
with the population (40.2% have a college degree).

Measures
Belief in political rumors. The survey asked respondents about their belief in two different
political rumors that were circulated online. First a screening question was asked to
ensure that only participants familiar with the rumor proceeded to the next question, and
then they were asked whether they believe the rumor to be true. If the participants were
not familiar with the rumor, they were not further asked to indicate their belief. Each
rumor contained controversies that were officially refuted by the involved stakeholders
(Rumor 1: the South Korean government, Rumor 2: the attacked candidate and the
Supreme Court) as blatant falsehoods. While the two rumors included in this study are
not representative of all unverified political information, the high-profile nature of these
political rumors circulated on the web allow for an initial exploration of their relation-
ship with media use. Respondents were asked to indicate their belief in each political
rumor on a 3-point scale, ranging from 1 (not true) to 3 (true; see Appendix 1 for ques-
tion wording and background descriptions).

Traditional news media use. Studies report that traditional news sources and their online
counterparts are subject to professional and social standards and constraints (Johnson
and Kaye, 1998). Thus, respondents were asked to report their weekly exposure to news
from three traditional news media outlets: watch national TV news (M = 3.77, SD =
1.35), read daily major newspapers in print (M = 2.19, SD = 1.40), and read the daily
major newspapers online (M = 2.08, SD = 1.15). Responses were coded on a 5-point
Bae 1529

scale, ranging from 1 (never) to 5 (everyday). A composite index was created by adding
up the responses (Cronbach α = .50).

Social media use for news. The use of social networking sites was measured with three
items asking respondents about their use of social media, such as Facebook, Twitter,
Cyworld, and others, as a source for news. Two items tapped into their motivation of
social media use on a 5-point scale: ‘I use social media to learn about politics and current
affairs issues’ (M = 3.47; SD = 0.88), and ‘I use social media to receive news from my
friends’ (M = 3.31; SD = 0.86). An additional item measured the frequency of getting
their news from online social networking sites. The same 5-point scale was used for a
weekly basis, ranging from 1 (never) to 5 (everyday; M = 2.90; SD = 1.41). A composite
index was created by adding up the responses to the three items (Cronbach α = .75).

Perceived political similarity. The respondents’ perception of political similarity in net-

works was measured with two items that asked respondents how similar they felt to their
friends in online social networks with regard to two items. First, respondents were asked
to indicate how much they agree with the statement, ‘my friends in online social net-
works and I have similar political viewpoints’ on a 5-point scale, ranging from 1 (strongly
disagree) to 5 (strongly agree). Second, respondents were asked to rate on a 5-point
scale, ranging from 1 (completely different) to 5 (almost the same), how similar they felt
to their friends about the candidate favored in the 2012 presidential election. The
responses were summed to create an additive index (M = 6.0; SD = 1.55, Spearman
Brown coefficient = .55).

Control variables. A set of demographic variables served as controls in the analysis: gen-
der (49.3% males), age (mean = 40, SD = 10.97), education (median = college degree),
and household income (median = KRW4,000,000–KRW5,000,000). In addition, this
study controlled for the respondents’ political predisposition with two items: interest in
politics and political ideology and support for candidate in the 2012 South Korean presi-
dential election. Respondents were asked to indicate their level of political interest on an
8-point scale, ranging from 1 (definitely not interested) to 8 (definitely interested; M =
5.0, SD = 1.47).
Given the nature of political rumors, it is likely that the respondents’ belief in them
might be significantly influenced by their political leanings. To focus on the variance in
rumor belief that relate to news media use and social networking, the respondents’ politi-
cal ideology and their support for the presidential candidate in the 2012 election were
controlled in the analyses. This sample was composed of liberals (24.9%), moderates
(45.4%), and conservatives (29.8%), and 36.6% of respondents reported having voted
for Park (conservative), 38.5% for Moon (liberal), and 17.1% for Ahn (withdrawn).

Analytic procedure
To examine the hypothesized relationships between the use of traditional news media
and social media as news sources and the users’ perceptions around political rumors,
multivariate regression analyses were run with belief in each political rumor as the
1530 Journalism 21(10)

outcome variable. The survey did not explicitly ask respondents to indicate whether each
news source directly contributed to rumor exposure or belief, as this level of detail would
be difficult for respondents to self-report with accuracy (Garrett, 2011; Schwarz, 1999).
Instead, the influence of news media use on rumors was analyzed by accounting for the
variance explained by control variables and testing the statistical significance of each
relationship. In addition to testing the direct relationship between social media use and
rumor beliefs, this study also explored how the similarity of users’ online social networks
moderates the relationship. To carry out this analysis, interaction terms were created
between social media use for news and perceived political similarity. In order to avoid
potential multicollinearity problems between the interaction term and its components, all
of the component variables were standardized prior to the formation of the interaction
terms (Cronbach, 1987).

Results
The central motivation for this study was to explore how individuals, situated in an envi-
ronment where information reaches them increasingly through online social networks,
understand and process political rumors. Toward this end, this study began by examining
the direct relationships between citizens’ use of social media as a source for news and
their belief in political rumors (H1). Findings reported in Table 1 show the relationship
between the block of control variables, media use, and the users’ beliefs in the two politi-
cal rumors, respectively. Notably, the significant result of presidential candidate support
(Rumor 1: β = .31, p < .01; Rumor 2: β = .25, p < .01) for both rumors demonstrates that
individuals’ belief in political rumors are most significantly shaped by their preexisting
political leanings. Yet even after controlling for the respondents’ political predisposition
as well as a set of demographic variables, findings highlight that relying on social media
use as a news source is a positive predictor of rumor beliefs (Rumor 1: β = .14, p < .05;
Rumor 2: β = .17, p < .01). Given that this positive relationship was significant after
accounting for the variance explained by the respondents’ presidential candidate support,
it can be inferred that their news media consumption pattern is related to their under-
standing of political rumors.
The next hypothesis predicted that using traditional news media would negatively
moderate the relationship between social media use as a source for news and belief in
political rumors (H2). Table 2 shows findings related to H2, and illustrates the significant
interaction between the two variables in predicting beliefs in political rumors (Rumor 1:
β = –.12, p < .05; Rumor 2: β = –.11, p < .10). This finding suggests that when individuals
who use social media as a source for news also frequently consume traditional news
media, the relationship between social media use and rumor beliefs was significantly
weaker.
Finally, interested in the conditions under which rumors mediated through social net-
works can be regarded as more believable, the next analysis focused on the role of homo-
geneity in online social networks (H3). As shown in Table 3, a positive interaction term
was found between perceived political similarity in online social networks and using
social media for news to predict belief in political rumors (Rumor 1: β = .11, p < .10;
Rumor 2: β = .13, p < .05). This result indicates that the positive relationship between
Bae 1531

Table 1. Predictors of belief in online rumors – Traditional news media versus social media
use for news.

Control variables Belief in Rumor 1 Belief in Rumor 2

β t value β t value
Age .09 1.2 –.05 –0.67
Gender (high: female) .02 0.34 .16** 2.34
Education .47 0.47 .07 1.07
Household income –.06 –0.91 .08 1.14
Political interest .07 0.96 .02 0.32
Political ideology .02 0.21 .08 1.12
Presidential candidate support .31** 3.9 .25** 3.21
R2 (%) 10.4% 15.2%
News source
Traditional news media –.02 –0.28 –.12 –1.58
Social media .14* 1.99 .17** 2.43
Total R2 (%) 12.4% 18.2%

Entries are standardized final regression coefficients.

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

Table 2. Two-way interaction – Social media use for news × Traditional news media use.

Belief in Rumor 1 Belief in Rumor 2

β t value β t value
Prior blocks (R2, %) 12.4% 18.2%
Traditional news media use × –.12* 1.73 –.11# 1.64
Social media Use
Inc. R2 (%) 1.3%* 1.1%#
Total R2 (%) 13.7% 19.3%

Prior blocks include age, gender, education, household income, political interest, political ideology, presiden-
tial candidate support, social media use for news, and traditional news media use. Entries are standardized
regression coefficients after controlling for the prior blocks.
#p < .10; *p < .05; **p < .01.

using social media as a news source and believing in the two political rumors becomes
stronger for those who find themselves to be in a politically homogeneous social network
that aligns with their own viewpoints. For a better illustration of the interactive relation-
ship, the predicted values for Rumor 2 are plotted in Figure 1. The two different slopes
for high and low levels of social network similarity demonstrate that the association
between social media use for news and belief in political rumors is stronger when users
feel that their friends hold similar political views as themselves. In fact, the slope for
those with low levels of social network similarity is almost flat, suggesting that the posi-
tive relationship between social media for news and rumor beliefs may be of greater
1532 Journalism 21(10)

Table 3. Two-way interaction – Social media use for news × Perceived political similarity.

Belief in Rumor 1 Belief in Rumor 2

β t value β t value
Prior blocks (R2,
%) 12.4% 18.2%
Perceived political similarity .14* 2.00 .16* 2.28
Inc. R2 (%) 2.5% 3.4%
Social media use for news × .11# 1.65 .13* 1.9
Perceived political similarity
Inc. R2 (%) 1.2%# 1.5%*
Total R2 (%) 16.1% 23.1%

Prior blocks include age, gender, education, household income, political interest, political ideology, presi-
dential candidate support, social media use for news, traditional news media use. Entries are standardized
regression coefficients after controlling for the prior blocks.
#p < .10. *p < .05. **p < .01.

Figure 1. Predictors of political rumor beliefs – Social media use for news × Perceived political
similarity (Rumor 2).

concern for individuals in more politically similar or homogeneous networks. In addi-

tion, the perceived political similarity of the respondents’ online political networks was
a positive predictor of believing in the two political rumors (Rumor 1: β = .14, p < .05;
Rumor 2: β = .16, p < .05), suggesting that network characteristics are a significant factor
in understanding the way information is transmitted and received by social media users.

Discussion
The exponential growth of social networking sites in recent years has brought them to the
center of scholarly research, as illustrated in the sharp increase in the number of
Bae 1533

published journal articles. This study focuses on the changes brought to the information
environment by social media and draws attention to the importance of understanding the
process through which users receive and evaluate information in online social networks.
Although scholars have begun to investigate the diffusion of information online, how
social media use relates to the individuals’ perceptions and beliefs around unverified
information and rumors remains largely unexplained. This study adds valuable insight
about the conditions under which political rumors can be regarded as more believable, by
drawing attention to the users’ news media consumption habits as well as the political
homogeneity of their online social networks.
As one might expect, individuals’ belief in political rumors tend to be closely aligned
with their political ideology. The data of this study confirm that the respondents’ political
ideology was a strong determinant of their belief in the two political rumors. Yet the
significant positive relationship between using social media as a source for news and
beliefs in political rumors even after accounting for the variance explained by as political
ideology, suggests the centrality of individuals’ information environment.
One possible explanation is that the social mediation of information through collabo-
rative filtering can provide a less accurate picture about politics and current affairs, com-
pared with the traditional news media. Studies of the blogosphere and citizen journalism
have shown that many who participate lack the basic traits associated with journalists,
such as reporting with objectivity and double-checking information (Gil de Zúñiga et al.,
2011; Tremayne, 2007). Instead, they have been known to take on different roles as
observers, analyzers, and commentators who provide an interpretation of current affairs
events (Andrews, 2003). The news frames constructed by citizens contributing to the
newsfeed in online social network networks may thus reflect enduring ideologies of
individuals or particular social groups. As a result, news can become part of a storytelling
process in social media (Papacharissi and De Fatima Oliveira, 2012), which could con-
sequently serve to perpetuate existing stereotypes and biases embedded in political
rumors. Any intermediary in social media can blend emotions with opinions, drama with
fact, and reflect deeply subjective values and interpretations.
Recently, some online media and social networking sites have begun to realize their
responsibility in combating the spread of rumors, in the face of increasing criticism for
circulating falsehoods and misinformation. For instance, Facebook announced that it
would introduce new policies to enhance the fact-checking process in its networks, by
partnering with third-party fact-checkers, as well as increasing human resources in
checking the information (Issac, 2016; Pierson and Etehad, 2016). Despite these efforts,
the filtering and detecting mechanism is still largely dependent on human capacity,
which is proportionately much smaller than the vast amount of information that is shared
per minute in any given social media platform. Combating fake news in social media will
become the biggest challenge that service providers must confront, as their repercussions
have been shown to be detrimental to our democracy as evidenced by the most recent
2016 US presidential election.
The fundamental motivation of this research is to understand the way misinformation
and rumors transmitted through online social networks are received by the users, and
evaluated as believable. Previous research on audience reception studies informs us that
the reactions of information consumers is necessarily subjective (Graber, 1988), and that
1534 Journalism 21(10)

the reception of a message is affected by varying degrees of noise or interference (Bird,

2003). As a beginning step toward explicating how users react to misinformation circu-
lated in social media and form opinions around rumors, this study first examined the
resonance of political rumors in homogeneous online social networks. Findings confirm
that higher levels of political similarity in social networking sites predict a stronger rela-
tionship with relying on social media for news and believing in political rumors. The
results of the two-way interaction provide suggestive evidence of how relying on news
and information mediated through an online social network of like-minded individuals
can intensify individuals’ beliefs about controversial issues or unfounded rumors. Thus,
while the stream of information in social networking sites can facilitate encounters with
news and politics, the contribution to an informed democratic citizenry remains depend-
ent not only on the individual users’ use of diverse news sources but also the character-
istics of the network.
What further complicates this process is that the information curated in social media
newsfeeds is not only a product of the users’ collaborative filtering, but also the result of
an algorithmic function unique to each social networking service provider. Although the
backbone technology of social media appears to be relatively consistent, each social net-
working site operates on its own platform with a unique algorithm and provides a unique
user interface. While these algorithms generally remain obscure, it is known that they can
determine the visibility of information in social media newsfeeds (Gillespie, 2014). It may
be that the inclusion or promotion of information is a reflection of the users’ record of
previous interactions in social networks in which case the newsfeed can contribute to a
reinforcing spiral (Slater, 2007). Moreover, some social networking sites require users to
build reciprocated friendships, while others allow users to maintain unreciprocated fol-
lowerships. Accordingly, particular cultures and practices can be developed in different
social networking sites boyd and Ellison (2007), competing for and attracting different
social media users. Thus, in order to precisely understand the impact of the social element
engrained in the information as it is mediated through interpersonal networks, it is neces-
sary to take into account features that are specific to the platforms.
Finally, there are a few limitations in this study to be discussed. First, the findings in
this study are based on data from a cross-sectional online survey, which allows for
exploring and clarifying structural relationships, yet provides weak support for causal
inferences. Given that rumors constantly evolve, longitudinal research that observes
rumor beliefs over time could provide more robust evidence to understand the role of
different news sources during the process. A longitudinal design could also provide clear
evidence to understand the over time impact of relying on social media as a source for
news, particularly in comparison with users with a more diverse palette of news sources.
The second limitation regards the measurement of some concepts in the study. This
study measured the respondents’ belief in two political rumors in Korea. Including more
items can better tap into the respondents’ belief in political rumors. In addition, an in-
depth content analysis of the nature of news and information that gain currency in social
media to large-scale audiences will enrich the understanding of the dynamics of social
networks.
Despite its limitations, the present study illustrates a meaningful step toward under-
standing the potential consequences of information diffusion in online social networks.
Bae 1535

What can start out as a mistake or a prank can be compounded within the network and
overshadow the truth of the information. Considering that the long-term viability of par-
ticipatory democracy depends on an informed citizenry with accurate information (Delli
Carpini and Keeter, 1996), future research examining the growing complexities of shar-
ing, commenting, and linking news and information in social platforms will contribute to
further clarifying their role as disseminators of news.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article.

References
Andrews P (2003) Is blogging journalism? Nieman Reports 57(3): 63–64.
Asch SE (1956) Studies of Independence and conformity: I. A minority of one against a unani-
mous majority. Psychological Monographs-General and Applied 70(9): 1–70.
Ayres JM (1999) From the streets to the Internet: The cyber-diffusion of contention. The Annals of
the American Academy of Political and Social Science 566(1): 132–143.
Bakshy E, Karrer B and Adamic, L A (2009) Social influence and the diffusion of user-created con-
tent. In: EC’09 Proceedings of the 10th ACM conference on Electronic commerce, Stanford,
California, USA, 6–10 July 2009, pp. 325–334. New York, NY, USA: ACM.
Bird SE (2003) Audience in Everyday Life: Living in a Media World. New York: Routledge.
boyd d and Ellison NB (2007) Social network sites: Definition, history, and scholarship. Journal
of Computer-Mediated Communication 13: 210–230.
Bradshaw T, Massoudi A and Scannell K (2013) Bogus terror tweet sparks shares blip. Financial
Times. Available at: http://www.ft.com/cms/s/0/33685e56-ac3d-11e2-a063-00144feabdc0.
html#axzz2wMi1Hqsh
Brown JJ and Reingen PH (1987) Social ties and word-of-mouth referral behavior. Journal of
Consumer Research 14: 350–362.
Brundidge J (2010) Encountering “difference’’ in the contemporary public sphere: The con-
tribution of the internet to the heterogeneity of political discussion networks. Journal of
Communication 60(4): 680–700.
Chaiken S (1980) Heuristic versus systematic information processing and the use of source
versus message cues in persuasion. Journal of Personality and Social Psychology 39(5):
752–766.
Cronbach LJ (1987) Statistical tests for moderator variables: Flaws in analyses recently proposed.
Psychological Bulletin 102: 414–417.
Delli Carpini MX and Keeter S (1996) What Americans Know about Politics and Why It Matters.
New Haven, CT: Yale University Press.
Difonzo N (2010) Ferreting facts or fashioning fallacies? Factors in rumor accuracy. Social and
Personality Psychology Compass 4: 1124–1137.
Difonzo N and Bordia P (2007) Rumors influence: Toward a dynamic social impact theory of
rumor. In: Pratkanis AR (ed.) The Science of Social Influence: Advances and Future Progress.
Philadelphia, PA: Psychology Press, pp. 271–296.
Donovan P (2007) How idle is idle talk? One hundred years of rumor research. Diogenes 54(1):
59–82.
Festinger L (1950) Informal social communication. Psychological Review 57(5): 271–282.
1536 Journalism 21(10)

Festinger L, Cartwright D, Barber K, et al. (1948) A study of a rumor: Its origin and spread.
Human Relations 1(4): 464–486.
Finholt T and Sproull LS (1990) Electronic groups at work. Organization Science 1(1): 41–64.
Flanagin AJ and Metzger MJ (2000) Perceptions of Internet information credibility. Journalism &
Mass Communication Quarterly 77(3): 515–540.
Garrett RK (2011) Troubling consequences of online political rumoring. Human Communication
Research 37(2): 255–274.
Gil de Zúñiga H, Lewis SC, Willard A, et al. (2011) Blogging as a journalistic practice: A model
linking perception, motivation, and behavior. Journalism 12(5): 586–606.
Gillespie T (2014) The relevance of algorithms. In: Gillespie T, Boczkowski PJ and Foot KA (eds)
Media Technologies – Essays on Communication, Materiality, and Society. Cambridge, MA:
The MIT Press, pp. 167–192.
Graber DA (1988) Processing the News: How People Tame the Information Tide. New York:
Longman.
Huckfeldt RR and Sprague JD (1995) Citizens, Politics and Social Communication: Information
and Influence in an Election Campaign. Cambridge; New York: Cambridge University Press.
Issac M (2016) Facebook mounts effort to limit tide of fake news. The New York Times. Available
at: https://www.nytimes.com/2016/12/15/technology/facebook-fake-news.html?_r=0
Johnson TJ and Kaye BK (1998) Cruising is believing? Comparing Internet and traditional sources
on media credibility measures. Journalism & Mass Communication Quarterly 75(2): 325–
340.
Kahneman D and Tversky A (1979) Prospect theory: An analysis of decision under risk.
Econometrica 47(2): 263–291.
Kim SH (2012) Testing fear of isolation as a causal mechanism: Spiral of silence and genetically
modified (GM) foods in South Korea. International Journal of Public Opinion Research 24:
306–324.
Knight Foundation (2001) Promoting credible Internet journalism. John S and James L Knight
Foundation Informed & Engaged Communities. Available at: http://www.knightfoundation.
org
Korea Internet and Security Agency (2016) Survey on the Internet usage. Available at: http://isis.
kisa.or.kr/
Kuklinski JH, Quirk PJ, Jerit J, et al. (2000) Misinformation and the currency of democratic citi-
zenship. The Journal of Politics 62: 790–816.
Lazarsfeld PF, Berelson B and Gaudet H (1944) The People’s Choice. New York: Columbia
University Press.
Lupia A and Sin G (2003) Which public goods are endangered? How evolving communication
technologies affect the logic of collective action. Public Choice 117: 315–331.
Matthews C (2013) How does one fake tweet cause a stock market crash? Time, 24 April.
Available at: http://business.time.com/2013/04/24/how-does-one-fake-tweet-cause-a-stock-
market-crash/
Messing S and Westwood SJ (2012) Selective exposure in the age of social media: Endorsements
trump partisan source affiliation when selecting news online. Communication Research 41:
1042–1063.
Moy P, Torres M, Tanaka K, et al. (2005) Knowledge or trust? Investigating linkages between
media reliance and participation. Communication Research 32(1): 59–86.
Organisation for Economic Co-operation and Development (OECD) (2013) Broadband and tele-
com statistics. Available at: http://www.oecd.org/sti/broadband/broadband-statistics-update.
htm
Bae 1537

Papacharissi Z and De Fatima Oliveira M (2012) Affective news and Networked publics: The
rhythms of news storytelling on #Egypt. Journal of Communication 62(2): 266–282.
Petty RE and Cacioppo JT (1986) The Elaboration Likelihood Model of Persuasion (In
Communication and Persuasion). New York: Springer, pp. 1–24.
Pew Research Center (2016) News use across social media platforms 2016. Available at: http://www.
journalism.org/2016/05/26/news-use-across-social-media-platforms-2016/#fn-55250-1
Pierce JC, Lee-Sammons L, Steger MAE, et al. (1990) Media reliance and public images of envi-
ronmental politics in Ontario and Michigan. Journalism & Mass Communication Quarterly
67: 838–842.
Pierson D and Etehad M (2016) Facebook is working to stop fake news by tapping human fact-
checkers. Los Angeles Times. Available at: http://www.latimes.com/business/la-fi-tn-face-
book-fake-news-20161215-story.html
Putnam RD (2000) Bowling Alone: The Collapse and Revival of American Community. New York:
Simon & Schuster.
Putnam RD and Yonish S (1999) How important are random samples? Some surprising new evi-
dence. Paper presented to the annual meeting of the American association of public opinion
research, St. Petersburg, FL, 13–16 May 1999.
Ratkiewicz J, Conover M, Meiss M, et al. (2011) Detecting and tracking political abuse in social
media. In: Proceedings from ICWSM’11: The association for the advancement of artificial
intelligence, Barcelona, 17–21 July 2011. Menlo Park, CA: AAAI Press.
Schwarz N (1999) Self-reports: How the questions shape the answers. American Psychologist
54(2): 93–105.
Slater MD (2007) Reinforcing spirals: The mutual influence of media selectivity and media effects
and their impact on individual behavior and social identity. Communication Theory 17(3):
281–303.
Steinberg DI and Shin M (2006) Tensions in South Korean political parties in transition: From
entourage to ideology? Asian Survey 46: 517–537.
Stempel C, Hargrove T and Stempel GH (2007) Media use, social structure, and belief in 9/11
conspiracy theories. Journalism & Mass Communication Quarterly 84(2): 353–372.
Tremayne M (2007) Harnessing the active audience: Synthesizing blog research and lessons for
the future of media. In: Tremayne M (ed.) Blogging, Citizenship, and the Future of Media.
New York: Routledge, pp. 262–272.
Trpevski D, Tang WKS and Kocarev L (2010) Model for rumor spreading over networks. Physical
Review E 81: 056102.
Watts D (2003) Six Degrees: The Science of a Connected Age. New York: W. W. Norton &
Company.

Author biography
Soo Young Bae is an assistant professor of Communication at the University of Massachusetts,
Amherst. She is interested in the social and political impact of new communication technologies,
with a particular focus on the dynamics of user interaction and information flow in social media
platforms.
1538 Journalism 21(10)

Appendix 1
Rumor question wording
Please tell us about your judgment on the truths about the following statements.

1. North Korea is not responsible for the sinking of the South Korean Navy ship
Cheonan.

*Author’s Note: The South Korean navy ship carrying 104 personnel, Cheonan, sank on
the country’s west coast on 26 March 2010, killing 46 sailors. South Korea’s military
intelligence gathered with the United States, Britain, Australia, and Sweden revealed that
a torpedo fired from a North Korean submarine sank this navy ship. This attack was one
of the deadliest military provocations since the Korean War, which ended in a truce in
1953. North Korean officials responded denying responsibility. Rumors emerged in
South Korea that evidence was manipulated to attribute the cause of the sinking to North
Korea. Various rumors about the cause of the ship’s sinking appeared. Among others, a
theory developed that Cheonan crashed into an American warship, despite clear evi-
dence that the damage done to the ship would have been considerably different. South
Korean officials took steps to crack down the rumors that were proliferating on the web,
and accused an Internet political magazine for spreading inaccurate information about
the sinking and arousing public mistrust in the investigation. The South Korean Defense
Minister officially debunked these rumors as being unfounded and confirmed that the
traces of torpedo suggest that the attack was clearly carried out by North Korea.

2. Accusations against politicians brought to attention by the political talk show

podcast Naneun Ggomsuda are mostly true.

*Author’s Note: In 2011, ‘Naneun Ggomsuda,’ (translated as ‘I’m a Petty-Minded

Creep’) was one of the most popular podcast shows in South Korea. The hosts of this
podcast reported that Na Kyung-won, the governing party’s candidate in the Seoul may-
oral race, frequented a luxury skin care clinic with an annual membership fee of 100
million won, or US$90,000. Ms. Na admitted using the clinic but denied membership.
This mayoral election was characterized as an election in which the gap between rich and
poor played a critical role. Ms. Na has sued the program for libel, and the host of the talk
show has been ruled by the Supreme Court to violate the country’s election and defama-
tion laws when he spread unconfirmed rumors.