Selective Exposure and Political Polarization of Public ...dcontent.dkyobobook.co.kr/genomad_gift/001/article/1/03/81/10381508.pdfSelective Exposure and Political Polarization of Public

Selective Exposure and Political Polarization of Public Opinion on the Presidential Impeachment in South Korea 137

* This study was supported by the National Research Foundation of Korea Grant funded by the Korean Government

(NRF-2016-S1A3A2925063)

** Kyung Hee University; E-mail: [email protected]

*** Corresponding author. Kyung Hee University; E-mail: [email protected]

KOREA OBSERVER, Vol. 49, No. 1, Spring 2018, pp.137-159

© 2018 by INSTITUTE OF KOREAN STUDIES.

https://doi.org/10.29152/KOIKS.2018.49.1.137

Selective Exposure and Political Polarization of Public

Opinion on the Presidential Impeachment in South Korea:

Facebook vs. KakaoTalk*

Hee Min**, Seongyi Yun***

Many previous studies have proven that social media is more likely to promote selective

exposure in the acquisition of political information and to escalate political polarization than

is older media. However, this study argues that the impact of social media on selective

exposure and political polarization occurs differently according to the characteristics of their

platforms. To prove this argument, we firstly compared old and social media in terms of

their impact on selective exposure. Secondly, we compared the two different types of social

media, Facebook and KakaoTalk, to examine their influences on selective exposure and the

political polarization of public opinion during the presidential impeachment in South Korea.

As a result, we observed that people using social media were more selective in choosing

news sources than those using television. We also found that users of KakaoTalk were more

selective in acquiring political information on the presidential impeachment than were Facebook

users, and the same consequences appeared in political polarization. Therefore, we determined

that the political and social influence of social media varies according to its platform

characteristics.

KeyWords: Old media, Social media, Facebook, KakaoTalk, Selective exposure, Political polarization

I. Introduction

This study compares the impact of different types' social media use on political

engagement. Specifically, we focus on the two different social media of Facebook and

138 Hee Min, Seongyi Yun

KakaoTalk in terms of their influences on selective exposure and the political polarization

of public opinion in the process of presidential impeachment in South Korea. On March

10, 2017, the South Korean Constitutional Court, in a unanimous ruling, upheld the

National Assembly's decision to impeach the president. This marked the first impeachment

of a president of South Korea. Starting on October 29, 2016, candlelight rallies were

held every weekend, demanding the president's resignation. They continued for 20 weeks,

until March 11, 2017, the day after the Constitutional Court's decision to uphold the

impeachment. In total, more citizens participated in the demonstrations than in the

democratization uprising of June 1987. Every weekend, hundreds of thousands of furious

citizens poured onto the streets holding candles.

Meanwhile, the protests against the impeachment of the president also continued every

week. They gathered in the plaza shaking the South Korean flag and called their gatherings

the "Taegeukgi (National Flag) Rally." According to polls, the overwhelming majority

favored the impeachment of the president, while those in opposition to the impeachment

only accounted for about 15 percent. Nevertheless, the protests against impeachment lasted

for 19 weeks, with hundreds of thousands of people participating each week. Although

those against impeachment were in the minority, the public opinion seemed to be much

more polarized than it actually was.

In this process, the social media, KakaoTalk, played a large role in forming and

maintaining the minority opinion of objection to the impeachment as a part of the public

opinion. KakaoTalk, which is a leading Korean mobile-based social media, was launched

in March 2010. People not only communicate privately but also consume various contents

through KakaoTalk. This service forms an online network centered on existing offline

friends, and it is more closed than Facebook. These two services show a clear difference

in their platforms. Facebook is an example of open social media. KakaoTalk, on the

other hand, is a typical example of closed social media. In other words, one links friends,

friends of friends, or people who do not know the user well, while the other generally

connects people the user already knows.

Opponents of the impeachment could mobilize supporters by convincing them and

spreading their views through a KakaoTalk network. Participants in the protest against

impeachment learned how to use smartphones to share information through KakaoTalk

(Kim & Lee 2017). Opponents of the impeachment propagated and shared information

stating that former President Park Geun-hye (hereafter "Park") had made no errors and

was the victim of political intrigue. Protests against the impeachment could be organized,


spread, and continued using KakaoTalk. For example, the pastor who participated in the

rallies against impeachment suggested that they drive impeachment support groups by

mobilizing them through a KakaoTalk group chatroom in each region of the country

(Yonhapnews 2017). We believe that people acquiring information about Park's

impeachment via KakaoTalk are more selective and polarized in their political attitudes

than those using Facebook during the presidential impeachment.

Media platforms have an effect on the way in which people obtain political information

and form opinions in certain situations. In light of this perspective, prior studies focused

on the media's effect on users of older media platforms (Neuman et al. 1992). Scholars

have mostly compared old media and social media. They have found that those who

use social media are more selective when acquiring information than those who use old

media. The social media platforms enhance the ability to accept information selectively

and efficiently (Hwang 2013). Moreover, scholars believe that selective exposure to

political information is a problem that ends up polarizing communities (An et al. 2013).

Ample evidence suggests that partisan selectivity and ideological segregation is present

in online communication (Baum and Groeling 2008; Hahn et al. 2013; Hwang 2013; Roh

and Min 2012). Notably, political expressions on social media reinforce the expressers'

pre-existing political preferences (Cho et al. 2018).

Meanwhile, the literature on selective exposure and political polarization when using

new media is confined mostly to research on internet use as a whole, or it treats different

platforms as one category of social media (Nam and Stromer-Galley 2012; Swigger 2012;

Yoo and De Zúñiga 2014). In addition, all analyses have been conducted by focusing

on only a single type of social media (Buccafurri et al. 2015). Despite these research

trends, this study does not think it is appropriate to treat social media use as one type.

We are aware of the differences among types of social media and treat them separately.

During the presidential impeachment, Facebook and KakaoTalk exhibited different ways

in which to circulate information and mobilize people. We consider these social media

outlets to be distinctive in their impacts on political engagement.

Based on these perspectives, we assume that each media type has its own communication

technology. In addition, the characteristics of this technology can lead to differentiated

effects on selective exposure and political polarization. To test our assumption, we

compared the two most popular social media platforms in Korea: Facebook and KakaoTalk.

As mentioned above, Facebook makes it easy to connect with many people, regardless

of whether the user knows them well. On the contrary, KakaoTalk is a communication


service based on a limited network formed by offline social and geographical relationships

(Ha 2014). Therefore, it is easier to gather groups that have a similar background in the

KakaoTalk network than in the Facebook network. The group polarization phenomenon,

which refers to members' having one-sided opinions, is also likely to appear more clearly

in the KakaoTalk.

In pursuing these objectives, the rest of this paper is structured as follows: Section

2 gives an overview of existing arguments regarding selective exposure and political

polarization and makes hypotheses. In addition, we compare the communication

technology used on Facebook and KakaoTalk. Section 3 explains the methods used to

support our arguments. Section 4 is devoted to an empirical analysis of different media

platforms' effect on selective exposure and political polarization. In Section 5, this study

discusses the results and conclusions are drawn in section 6.

II. Literature review and hypothesis

A. Selective exposure: old media vs. social media

As the Internet age comes, there is an expectation that people will acquire much

political information. The cost of information consumption has drastically decreased. In

particular, the Internet contributed significantly to the diversity in media environment.

While the Internet-based media channels appeared, people came to access news anytime,

anywhere through new media platforms such as online newspapers, blogs, portals, and

social media so on. New media environment helps to resolve disparities of information

access. Meanwhile, this media environment allows people to become more selective in

their choice of political information. The social media environment makes it easier for

people to acquire political information similar to their predisposition (Mutz 2006). When

people access public issues via social media, they are rarely exposed to different views

(Baumgartner and Morris 2010). In other words, either people could choose to read any

news that is accessible to them, or they can selectively read the news that is preferable

to them in the new media environment (Choi 2016, 818).

Political scientists, however, are concerned that social media platforms are a space

where selective exposure to political information is more prominent than old media

platforms. They focus their interests on the question of whether old media and social


media platforms show a distinctive effect on political information access or political

participation. This study suggests two factors of reciprocity and content control, explaining

the difference between the two media platforms. Reciprocity is about whether or not

we can differentiate between the sender and recipient of a message (Rafaeli 1988; Yoo

and De Zúñiga 2014, 35). TV as a representative of old media has the distinction between

an information provider and a recipient is evident. On the other hand, communications

on social media platforms are more interactive and reciprocal among participants than

old media platforms. Reciprocal and interactive message exchanges on social media

platforms may bring the users closer to the experience of a higher quality of

communication. However, the reciprocity of social media is likely to strengthen the impact

of partisan disposition in a politically similar context (Cho 2005).

The second factor, content control, is related to the amount and complexity of

information that a medium can convey as well as its ease of adding information (Heeter

1989; Yoo and De Zúñiga 2014, 35). Additionally, this study regards content control

as being whether media users can easily choose their preferred content. Old media

environments have a limited number of channels, so that people are more likely to be

exposed to opinions inconsistent with their views, or unfavorable content, regardless of

their intent. People are likely to chance upon counter-attitudinal content in the old media

platforms (Mutz and Martin 2001). Meanwhile, people who use a social media platform

can more easily select content based on their preferences than those who use an old

media platform (Prior 2005; 2007).

In the social media space, people can select specific media channels, and access

preferred news in real time. It is possible that people will use a social media platform

to fragment into more specific like-minded groups. Twitter users tend to construct virtual

networks based on their partisan positions and interact with like-minded others (Colleoni

et al. 2014; Hahn et al. 2013; Himelboim et al. 2013; Roh and Min, 2012; Yun 2013).

Conservative Twitter users form a network with the Twitter of the conservative press

while liberal Twitter users form a network with the Twitter of the liberal media (Hwang

2013). According to the study by Yun (2013), 60 percent of Twitter users read more

tweets congenial to their political disposition. As expected, Facebook users show patterns

divided by political ideologies in posts on candidates' Facebook "walls" during election

campaigns. Robertson, Vatrapu, and Medina found that Facebook users who posted on

a candidate's wall mostly only referred other users to the same candidate's website

(Robertson et al. 2010). Based on survey data about people's political information use,


this study analyzes the following hypothesis:

H1. People exposed to social media platforms are more likely to be selective in

acquiring political information than those who are exposed to old media platforms

B. Selective exposure and political polarization: Facebook vs. KakaoTalk

As a representative of the new media platform, social media has reached tremendous

popularity around the world. Currently, the two most popular social media in South Korea

are Facebook and KakaoTalk. Of the two, KakaoTalk is more popular than Facebook.

According to one survey (KISA 2017), KakaoTalk ranked first as the most popular social

media, followed by Facebook and KakaoStory. KakaoStory is an additional service of

KakaoTalk. People share their daily lives through this service, mainly by photo sharing

via KakaoStory. Moreover, most instant messenger users use KakaoTalk (99.2%),

followed by Facebook Messenger (29.7%), Line (13%), and Nate on (6.9%) according

to a 2016 survey (KISA 2017). Social media also takes a role as a news channel. People

read news most frequently via KakaoTalk (50.6%), followed by Facebook (27%) and

KakaoStory (19.6%) according to a 2016 survey (KPF 2017).

Facebook and KakaoTalk have different characteristics in terms of communication

technology. To be more specific, there are two ways of establishing and recommending

connections in social media (Lee 2013). One method is that when people sign up, they

use their mobile phones' contact lists to connect with people they already know. Another

way is to connect based on the social media usage pattern of users. It recommends friends

who are familiar with each other or who are likely to have similar interests. KakaoTalk

networks are composed by the way of the former while Facebook networks are made

by the latter way. Moreover, these features are even more pronounced with the two

distinctive services of each medium. Facebook uses the Thumbs up symbol as a

recommendation sign, and KakaoTalk uses the Dantalkbang as a group chatroom.

(1) The recommendation sign of Facebook: Thumbs up

Facebook, an open social media network, has the advantage of its users being able

to connect with many people. Not only friends or friends of friends, but also people

who have no relationship to one another can easily become friends. Information circulated

on Facebook is likely to be exposed to the unspecified majority. Facebook users can

cultivate weak ties in an informal manner (Wilson et al. 2012).


Social media users no longer consider news sources like media brands. Instead, they

select stories themselves that are deemed by friends or friends' friends to be important.

They are more likely to follow the majority of their friends through passive means such

as viewing their newsfeeds, clicking "like" as a recommendation sign, or browsing (Burke

et al. 2010). In particular, Facebook's distinctive feature, the recommendation sign of

Thumbs up, reduces the probability that people will select contents based on partisan

biases (Messing and Westwood 2012). A recent study on Facebook has shown that

although people are most likely to click on links shared by those with whom they have

the closest relationships, weaker relationships generate newer recommendations (Bakshy

et al. 2012; Garrett 2009, 252). Messing and Westwood proved that within the confines

of a web interface similar to Facebook and Twitter, social endorsement that ranges from

0 to 100 recommendations reduced partisan selectivity to levels indistinguishable from

change (Messing and Westwood 2012, 15). In other words, the Facebook platform helps

users crystallize ephemeral relationships and broaden their worldviews (Ellison et al. 2009;

West et al. 2009; Wilson et al. 2012).

(2) The group chatroom of KakaoTalk: Dantalkbang

As a kind of closed social media, KakaoTalk is a service for communicating primarily

with people who are on one's mobile phone' contact list. It also has the function of

implementing from many to many group chats. KakaoTalk is not a system that is open

to everyone from the beginning but one of sharing daily life by grouping friends,

colleagues, and people from specific groups. People can only participate if someone

invites them to KakaoTalk. It looks like a little mobile-based internet cafe. In addition,

information circulated in KakaoTalk cannot be retrieved from outside the network.

Among services provided by KakaoTalk, the Dantalkbang (group chatroom) is its

distinctive feature. An abbreviation for Danche KakaoTalk Bang in Korea, Dantalkbang

is widely used as a means of information circulation. Once a Dantalkbang is created,

the space continues unless someone intentionally leaves it or quits KakaoTalk. KakaoTalk

is also online 24 hours. It is quite common for most Koreans to make a Dantalkbang

to exchange information in private and public relations. Moreover, KakaoTalk, unlike

other social media, has the function of showing the number of people who do not read

a posting uploaded in the Dantalkbang. For example, assuming that there are three

members in the Dantalkbang, if the number two is marked on a post, this means that

two members have not yet read it. By using this system, Dantalkbang members can


form common beliefs. The priority condition for building these beliefs among members

is confirming whether they read specific information in the online space (Kim 2016).

Likewise, participants of the candlelight rallies shared information about candlelight

rallies or the presidential impeachment via Dantalkbang during the presidential

impeachment process. Dantalkbang is based on offline-centered intimacy, so this service

promotes to strengthen the cohesion of inner groups (Ko 2017). For this reason,

Dantalkbang worked as a catalyst for mobilizing people who supported or opposed the

impeachment of the president.

Especially, Dantalkbang was considered the highlight of mobilizing protesters against

Park's impeachment (Bae 2017). Participants opposing the presidential impeachment were

primarily over 60-years of age and conservative. They hardly ever take part in street

rallies. Moreover, they are not accustomed to using the Internet and communicating with

unfamiliar people in online spaces. However, the automatic registration of people on their

mobile phones' contact lists created KakaoTalk networks. Most members of Dantalkbang

already knew one another. Additionally, the Internet usage of older people is sharply

increasing. The Internet usage rate of people over 65-years of age was only 13.4% in

2011, but their Internet usage rate was 38.4% in 2016 (KISA 2017). In particular, 61.4%

of elderly people use social media services like KakaoTalk (KISA 2017). Only 17.3%

of them use services such as KakaoStory, Facebook, Instagram, and so on (KISA 2017).

In the presidential impeachment, conservative politicians who opposed Park's

impeachment encouraged supporters to mobilize more people via the social media activity,

KaokaoTalk. The online community, the opposition group of Park's impeachment, announced

the following: "Please do your best to encourage a lot of patriotic people to join the

rallies by posting on the patriotic group's KakaoTalk" (Song 2017). One protester against

Park's impeachment expressed his opinion that the proactive Dantalkbang activities affect

an increase in the number of participants (Song 2017). KakaoTalk takes a critical role

in enabling seniors to exchange political views as well as deepen their friendships.

(3) Hypotheses 2 and 3

Within this context, we examined the two most popular social media, Facebook and

KakaoTalk, and developed hypotheses regarding how each platform might affect selective

exposure and political polarization. We assumed that those who obtain political

information mainly via closed social media networks would be more selective in acquiring

political information than those using open social networks. As mentioned above, closed


social media like KakaoTalk are services that allow users to communicate with each

other by creating groups of family, friends, or colleagues. In the KakaoTalk space, people

tend to communicate with people they already intimately know.

A study on selective exposure states that people tend to actively seek information

that corresponds to their opinions and avoid information that challenges their opinions

(Festinger 1957). From this perspective, members of a Dantalkbang on KakaoTalk,

generally composed of a relatively homogeneous group, are more likely to share political

information that is favorable or congenial to their political predispositions than those

who use Facebook. In short, the KakaoTalk platform is more suitable for maintaining

strong ties, whereas the Facebook platform is more suitable for cultivating weak ties.

Finally, this study focuses on the political polarization of KakaoTalk and Facebook

users. According to Stroud (2010, 557), polarization is "the strengthening of one's original

position or attitude by consuming only media outlets featuring a congruent political

ideology." The structure of social media is deemed important in explaining polarization

because it tends to function as individuals' information environment, where various

political information is diffused (Huckfeldt and Sprague 1987, 1995; Huckfeldt et al.

2004). Prior studies emphasized the structure of social media as a factor that has

significant effects on the level of polarization (Mutz 2002; Sunstein 2000).

Simultaneously, prior studies showed that selective exposure according to individuals'

political predispositions, such as partisan affiliation or their media use patterns, resulted

in political polarization (DiMaggio et al. 1996 Stroud 2007, 2008, 2010). People acquiring

political information selectively tended to perceive the distance of the opinion difference

between themselves and others as closer than it actually was (Roh and Min 2012). As

a result, a positive perception of public opinion in a group with similar views is

strengthened and their political attitude is polarized. For example, according to the results

of the 2004 US presidential election panel survey, selective exposure based on political

predispositions was consistently found to promote polarization of public opinion (Roh

and Min 2012). Therefore, we assume that political polarization will happen more on

the KakaoTalk platform than on the Facebook platform. To analyze the selective exposure

(H2) and political polarization (H3) differences between two social media platforms, we

applied the specific political issue of the last presidential impeachment. Additionally,

when people interact with like-minded people on social media, their pre-existing political

preference is reinforced. Given this consideration, we regard interaction behaviors on

Facebook and KakaoTalk to be the sharing of posts as well as writing on those platforms.


This study analyzes the following hypotheses:

H2. People exposed to the KakaoTalk platform are more likely to be selective in

acquiring political information than those who are exposed to the Facebook

platform.

H3. People involving in online activities on the KakaoTalk platform are more polarized

in their political attitudes than those participating in online activities on the

Facebook platform.

III. Method

A. Sample

The data was collected from March 23 to March 31, 2017, using an online survey

administered by Macromill Embrain, an online research and panel company. Among

respondents, men (50%) and women (50%) were equally distributed. With regard to age

distribution, respondents in their 40s accounted for the highest proportion (23.5%),

followed by those in their 50s (23.3%), 30s (20.1%) 60s (15%) and 20s (18%),

demonstrating a balanced age distribution without much difference between younger and

older generations. Of the respondents, 73% had pursued higher education, a statistic

broken down into college students or graduates (61.3%) and graduate school students

or higher (11.8%).

B. Measures

(1) Selective exposure

The most important criterion for politically motivated selective exposure is personal

political ideology (Roh and Min 2012). We focused on the consumption of news from

partisan sources. To gain this perspective, this study employed self-reports of political

ideology (i.e., conservative, moderate, or liberal). These reports were measured on a

11-point scale with values ranging from 1 (strong conservative) to 11 (strong liberal),

with moderate views at the center point. We then considered the issue of politically

motivated selective exposure in two dimensions, one between old media and social media

platforms, the other between Facebook and KakaoTalk platforms.


a. Selective exposure across old media and social media use. In order to operationalize

politically motivated selective exposure, two steps were taken. The first step was

to operationalize the selection of politically congenial media outlets. The second

step was to measure the difference in selective exposure between the liberal and

the conservative group. Both of these steps are discussed in more detail below.

We first distinguished the conservative and the liberal group. We then got the scores

of selective exposure for each group, which range from 0 to 4. Selective exposure scores

for the liberal/conservative group were set by whether or not liberals/conservatives used

the media outlets congenial with their political predispositions more often. For this

purpose, we categorized media outlets as liberal and conservative daily newspapers,

general cable television networks, political podcasts, and internet newspapers. Liberals

are more likely to use liberal newspapers, general cable television networks, podcasts

and internet newspapers, and conservatives vice versa.

We classified media outlets as liberal and conservative according to their political

leanings. Among daily newspapers, The Kyunghyang Shinmun and The Hankyoreh are

considered to be consistent exposure for liberals, while Joong Ang Daily, The Dong-A

IIbo and The Chosun Ilbo are considered to be consistent exposure for conservatives.

JTBC, a general cable television network and broadcasting company is classified as liberal

media. JTBC coverage, as a result, tended to be more supportive of liberals and political

views in favor of candlelight rallies or presidential impeachment in comparison to other

channels. Among political podcasts, Kim Eo-joon's Papa is, Chung Bong-ju's National,

and Kim Yongmin's Briefing are treated as consistent exposure for liberals, while Patriotic

Channel SNS TV, Choice of God, Sesame Broadcasting, and Chung Kyujae TV are

consistently conservative. Among internet newspapers, Ohmynews, Pressian, and DDanzi

Ilbo are consistently liberal, while Dailylian, New Daily, Media Watch, and Cho

Kabjae.com are consistently conservative.

Respondents were separately asked, "Which of the following media outlets did they

use most often to get information about presidential impeachment during the last rallies?"

Consequentially, the scores of selective exposure strength were measured by subtracting

the liberal from the conservative group scores. Then absolute values were calculated.

(M=1.33, SD=1.08)

b. Selective exposure across Facebook and KakaoTalk use. In these instances, selective

exposure was measured by how often they read news articles congenial with their

political views about presidential impeachment on portal sites (search engine sites


like Naver or Daum). Liberal people tend to be more supportive of news articles

in favor of candlelight rallies and presidential impeachment. On the other hand,

people with more conservative beliefs are more likely to read news articles that

are about Taegeukgi rallies and opposed to presidential impeachment. Respondents

were separately asked, using a 11-point scale, about how often in the past five

months they read news articles in favor of/against presidential impeachment or

in favor of candlelight/Taegeukgi rallies on portal sites (in favor of presidential

impeachment or candlelight rallies: M=8.05, SD=2.44, against presidential

impeachment or in favor of Taegeukgi rallies: M=6.11, SD=2.79).

(2) Political polarization

Political polarization was measured by the difference between the feeling thermometer

indexes of Moon Jae-in and Hong Joon-pyo, which range from 1 (not at all) to 11 (very

favorable). Moon Jae-in and Hong Joon-pyo were presidential candidates for the

Democratic Party and the Liberty Korea Party in the 19th presidential election. The

Democratic Party and the Liberty Korea Party represent liberal and conservative party.

This election was held on 9 May 2017, after the impeachment of Park Geun-hye. Voters

preferring Moon Jae-in received the formal presidential impeachment while voters

preferring Hong Joon-pyo denied Park's impeachment during election campaign period.

Hong Joon-pyo scores were subtracted from Moon Jae-in scores, and absolute values

were calculated (M=4.34, SD=3.30). The greater the difference between Moon Jae-in

and Hong Joon-pyo scores, the greater the political polarization.

(3) Media use

a. Television use and social media use. For the old media use variable, respondents

were asked about how often in a day they received political information from

terrestrial television over the past week. For the social media use variable,

respondents were asked about how often in a day they got political information

on social media via PCs or smartphones. The usage time question was an

open-ended question.

b. Facebook use and KakaoTalk use. For Facebook and KakaoTalk use variables,

respondents were separately asked, using a 11-point scale, about how often they

viewed postings on presidential impeachment in the past five months. (Facebook

use: M=5.49, SD=3.45, KakaoTalk use: M=5.35, SD=3.29)


c. Online activities on Facebook and KakaoTalk. For the variables of online activities

on Facebook and KakaoTalk, respondents were separately asked how often they

shared (wrote) news (opinions) about the presidential impeachment on Facebook/

KakaoTalk in the past five months, using a 11-point scale (Facebook sharing:

M=3.46, SD=2.91; Facebook writing: M=2.92, SD=2.81; KakaoTalk sharing:

M=3.63, SD=3.16; KakaoTalk writing: M=3.09, SD=2.96).

(4) Control variables

As control variables, this study included age, gender, education, income, political

interest, and political ideology strength. To measure political interest, respondents were

asked to state how much they agreed with the statement, "I am very interested in politics,"

on a 5-point scale ranging from definitely disagree to definitely agree (M=3.36, SD=0.84).

In addition, we controlled for political ideology strength. According to prior studies,

people tend to filter political information based on perceived ideological congruence

(Bennett and Iyengar 2008; Iyengar and Hahn 2009). Political ideology was first measured

with respondents' self-reports. One question was asked: "In politics today, do they consider

yourself liberal, conservative, or moderate?" with answers ranging from 1 (definitely

conservative) to 11 (definitely liberal) (M=6.12, SD=2.04). Then, the answers to this

question were recoded to give a measure of ideological strength ranging from 0

(moderate), 1 (weak liberal and conservative) to 5 (strong liberal and conservative)

(M=2.0, SD=1.79).

IV. Results

The first step we employed was to analyze whether media platforms influence the

increase of politically motivated selective exposure. Table 1 reports the results from a

regression analysis predicting selective exposure from television and social media

platforms, while accounting for control variables. The results showed that social media

use predicted a significant increase in politically motivated selective exposure, whereas

television did not significantly increase it. The more frequently people used social media

to acquire political information, the more likely they were to become selective (β = .102,

p < .01). Age, education, political interest were also positively related to selective exposure.

Highly educated people are more likely to become selective about political information


Unstandardized coefficient B(SE) Standardized coefficient β

(Constant)-.135

(.229)

Gender

(Male = 1)

-.041

(.065)-.019

Age.062

(.025).075*

Education.094

(.035).082**

Household income-.009

(.016)-.017

Political interest.282

(.039).221***

Political ideology strength.017

(.018).028

Television use.000063

(.000).005

social media use .001

(.000).102**

Adjusted R-Squared .075

(d, F)(8, 12.076)

***

Note: *** p < 0.001, ** p < 0.01, * p <0.05

Table 1. Media Platforms Predicting Selective Exposure

(β = .082, p < .01). Older people are more likely to become selective (β = .075, p <

.05). Additionally, the more interested people were in politics, the more likely they were

to become selective in the acquisition of political information (β = .221, p < .001). This

study found that social media use has a stronger association with politically motivated

selective exposure than television use. Thus, there is support for Hypothesis 1.

These results are a concern for the development of Korean democracy in two respects.

The first is that the acquisition of political information through social media is becoming

increasingly common, and the selective exposure of social media is likely to lead to

more political polarization. Second, ideology and generation conflicts in Korean society

are very serious, and these two conflicts overlap each other. While younger generations

are generally progressive, older generations support conservative forces. It is highly likely

that selective exposure through social media will deepen social conflicts by interacting

with the overlap of ideology and generation conflicts.

Table 2 outlines the results of the regression analysis regarding factors affecting

selective exposure across social media platforms. If politically motivated selective


Conservative group

(N=356)

Liberal group

(N=422)

Unstandardized

coefficient

B(SE)

Standardized

coefficient

β

Unstandardized

coefficient

B(SE)

Standardized

coefficient

β

(Constant)5.917

(.1.071)

4.934

(.706)

Gender

(Male = 1)

.216

(.264).042

.322

(.188).079†

Age-.042

(.099)-.022

-.131

(.075)-.079†

Education-.342

(.138)-.130**

.018

(.106).008

Household income -.106

(.063)-.087†

.015

(.049).014

Political interest.594

(.175).183**

.567

(.119).232***

Political ideology

strength

-.304

(.111)-.139**

.145

(.084).079†

Facebook use.090

(.045).115*

.065

(.028).111*

KakaoTalk use .166

(.046).202***

.159

(.039).261***

Adjusted R-Squared .154 .204

(d, F)(8, 9.091)

***

(8, 14.527)

***

Note: *** p < 0.001, ** p < 0.01, * p <0.05, † p <0.1

Table 2. Social Media Platforms Predicting Selective Exposure

exposure is occurring, the political beliefs of people should significantly predict news

consumption about presidential impeachment. The first column in Table 2 reports the

conservative group. A closer look reveals that social media use exerted a statistically

significant influence on selective exposure. Table 2 shows that the more conservatives

read news articles about presidential impeachment through Facebook or KakaoTalk, the

more frequently they read news articles approving of Taegeukgi rallies or against

presidential impeachment on portal sites (Facebook use: β = .115, p < .05, KakaoTalk

use: β = .202, p < .001). The second column in Table 2 reports the liberal group. Likewise,

the more liberals who read news articles about the presidential impeachment through

Facebook or KakaoTalk, the more frequently they read news articles in favor of

candlelight rallies or presidential impeachment on portal sites (Facebook use: β = .111,

p < .05, KakaoTalk use: β = .261, p < .001). Compared to Facebook use, KakaoTalk

has a greater impact on selective exposure in both groups. This is the evidence that


shows that the KakaoTalk platform is more closely related to higher levels of selective

exposure than Facebook. People consuming more KakaoTalk platforms display more

selective exposure behavior than Facebook users.

Meanwhile, political interest predicted a significant increase in the politically

motivated selective exposure of both groups. The more interested conservatives/liberals

were in politics, the more likely they were to become selective about political information

(conservative group: β = .183, p < .01, liberal group: β = .232, p < .001). The fact

that high level of political interest lead to more selective exposure is a very worrisome

aspect of Korean democracy development. This is because the people with high political

interest are more active in political participation and consequently the increase in political

participation can lead to deepening political polarization. Political ideological strength,

however, predicted a significant increase in the politically motivated selective exposure

of only liberals. The stronger liberals were among liberals, the more frequently they

read news articles supporting the candlelight rallies or presidential impeachment on portal

sites (β = .079, p < .1). Contrarily, the stronger conservatives were among conservatives,

the less frequently they read news articles supporting Taegeukgi rallies or against

presidential impeachment on portal sites (β = -.139, p < .01). That is why candlelight

rallies were of interest to the entire country, regardless of political ideology, as the

conservative also might have read news about the candlelight rallies or the presidential

impeachment. In addition, conservatives with a low level of education are more likely

to become selective about political information (β = -.130, p < .01). The younger people

are more likely to become selective in liberal groups (β = -.079, p < .1). These results

reflect that conservatives are composed mainly of older people while liberals are

composed of younger people. Thus, there is support for Hypothesis 2.

Table 3 outlines the results of the regression analysis of factors regarding political

polarization across social media activities. It shows that politically motivated selective

exposure predicted a significant increase in political polarization. The more people were

selective about acquiring political information, the more likely they were to become

polarized (β=.249, p < .001). In addition, education, political interest, and political

ideology strength were also positively related to political polarization. Highly educated

people are more likely to become polarized in the formation of their political attitudes

(β = .061, p < .05). The more interested people were in politics, the more likely they

were to become polarized (β = .311, p < .001). In addition, the stronger people are

ideologically orientated, the more they become polarized (β = .073, p < .01).


Notably, we found that the online activity on KakaoTalk has more of an impact on

the increase of political polarization than online activity on Facebook. The more people

write on Facebook, the less that people show a preferential difference between Moon

Jae-in and Hong Joon-pyo (β = -.121, p < .05). In contrast, the more people share

presidential impeachment news on KakaoTalk, the more pronounced the difference in

their preferences become (β = .081, p < .1). In support of Hypothesis 3, there is evidence

that KakaoTalk users are more likely to be polarized politically than those who use

Facebook.

Unstandardized coefficient

B(SE)

Standardized coefficient

β

(Constant)-1.353

(.641)

Gender

(Male = 1)

-.142

(.185)-.021

Age-.072

(.069)-.029

Education .213

(.098).061*

Household income.048

(.046).030

Political interest1.210

(.117).311***

Political ideology strength.136

(.051).073**

Selective exposure.757

(.086).249***

Facebook sharing-.008

(.059)-.007

Facebook writing-.143

(.069)-.121*

KakaoTalk sharing.084

(.049).081†

KakaoTalk writing-.026

(.058)-.023

Adjusted R-Squared .214

(d, F)(11, 27.885)

***

Note: *** p < 0.001, ** p < 0.01, * p <0.05, † p <0.1

Dependent variable=absolute score of Moon-Hong preference

Table 3. Social Media Activities Predicting Political Polarization


V. Discussion

This paper aimed to analyze media platforms' effect on selective exposure to political

information and on political polarization of public opinion. The first step we took was

to analyze whether media platforms have an effect on the aggravation of politically

motivated selective exposure. To accomplish this, we compared old media and social

media platforms. Then we observed that people who use social media become more

selective than those who engage in television use. This results from the differences in

interactivity or reciprocity and content control between the two media platforms.

Communications via social media platforms are more interactive than are those of old

media. Social media users are also likely to select preferential contents. Therefore, despite

the fact that the social media environment provides people with various political

information, this tends to generate fragmented users divided by political orientations or

interests. As shown in recent studies, people are likely to accept like-minded people

as their friends and to form online networks on Facebook or Twitter spaces (Hahn et

al. 2013; Hwang 2013; Yun 2013).

However, we directed our attention toward the fact that various social media platforms

can be classified using different aspects of communication technology, and we focused

on the Facebook and KakaoTalk platforms, which are the two most popular social media

networks in Korea. It is good that social media as an individuals' communication

environment shows their information consume behaviors (Huckfeldt et al. 2004).

Therefore, we argue that the impact of social media on selective exposure and political

polarization occurs differently according to the characteristics of their platforms. In the

latter investigation, we found that more people who use KakaoTalk become selective

in acquiring political information than those who use Facebook. The last analysis is

regarding the effect of KakaoTalk and Facebook activities on political polarization. It

shows that online activity on KakaoTalk has more of an impact on the increase of political

polarization than online activity on Facebook.

The reason for these is their distinct communication structures: Facebook is an open

social media network, while KakaoTalk is a closed social media network. In addition,

there are two unique communication tools: the Thumbs up icon that serves as a

recommendation sign on Facebook and the group chatroom of KakaoTalk known as

Dantalkbang. Facebook users can access political information that is not only similar

to their political predispositions purposefully but also unintentionally. In particular,


clicking "Like" on Facebook connects the majority of one's friends and increases the

possibility of obtaining varied information.

Meanwhile, KakaoTalk users chiefly share and disseminate political information with

specific members via Dantalkbang. The Dantalkbang members already know each other

and have a social relationship that is based offline. For this reason, older conservatives

who are pro-Park protesters used Dantalkbang on KakaoTalk as a mobilization tool for

the last Taegeukgi rallies. They distrusted the mainstream media that criticized the formal

president. Rather, they relied on non-journalistic channels like KakaoTalk for news on

presidential impeachment issues, and then their voices grew rapidly. As a result, Korean

society showed the deep political divide between the anti-Park, candlelight rally group

and the pro-Park, Taegeukgi rally group during the presidential impeachment period.

VI. Conclusion

As has been shown in many previous studies, social media create a more selective

exposure to information acquisition and consequently enhance stronger political

polarization than do old media. However, not all social media encourage the same

level of selective exposure and political polarization. In Korea's presidential

impeachment phase, KakaoTalk, which has a relatively closed platform, promoted

selective exposure and political polarization, but Facebook did not encourage as

selective an exposure polarization.

The results of this study have little positive implications for the maturity of Korean

democracy. Though the spread of social media gives a positive effect on increasing

political participation, it also creates a problem of intensifying political polarization while

enhancing selective exposure. It is a serious problem that the higher the level of political

interest, the higher the level of education, and the more obvious the ideological tendency

lead to the more selective exposure and political polarization. What makes it worse is

that these groups are not only active in political participation but also play a leading

role in shaping public opinion. Those political circumstances contribute to grow the

distortion of democratic representation. Our democracy might be representing only citizens

who are interested in politics.

Nevertheless, we can expand our social networks through many different kinds of

social media, which, in turn, contributes to the improvement of democratic quality. Each


types of social media may play its own role in developing democracy. Facebook is more

useful for bridging heterogeneous social groups due to its open platform, while closed

platform of KakaoTalk is helpful for bonding homogeneous peoples in a group. Therefore,

we need to understand different characteristics of each social media in studying its political

and social influence.

References

An, Jisun, Daniele Quercia, and Jon Crowcroft. 2013. "Fragmented social media: a look into

selective exposure to political news." Proceedings of the 22nd International

Conference on World Wide Web. 51-52.

Bae, Young. 2017. "Why do I see the same side on SNS." Hankookkyungjae, April 27,

2017. <http://news.hankyung.com/article/201704272677g>

Bakshy, Eytan, Itamar Rosenn, Cameron Marlow, and Lada Adamic. 2012. "The role of social

networks in information diffusion." Proceedings of the 21st international conference

on World Wide Web (April): 519-528.

Baum, Matthew A., and Tim Groeling. 2008. "New media and the polarization of American

political discourse." Political Communication 25(4): 345-365.

Baumgartner, Jody C., and Jonathan S. Morris. 2010. "My Face Tube politics: Social

networking web sites and political engagement of young adults." Social Science

Computer Review 28(1): 319-338.

Bennett, W. Lance, and Shanto Iyengar. 2008. "A new era of minimal effects? The changing

foundations of political communication." Journal of communication 58(4): 707-731.

Buccafurri, Francesco, Gianluca Lax, Serena Nicolazzo, and Antonino Nocera. 2015.

"Comparing Twitter and Facebook user behavior: Privacy and other aspects."

Computers in Human Behavior 52: 87-95.

Burke, Moira, Cameron Marlow, and Thomas Lento. 2010. "Social network activity and social

well-being." Proceedings of the SIGCHI conference on human factors in computing

systems.

Cho, Jaeho. 2005. "Media, interpersonal discussion, and electoral choice." Communication

Research 32(3): 295-322.

Cho, Jaeho, Saifuddin Ahmed, Heejo Keum, and Yun Jung Choi, and Jong Hyuk Lee. 2018.

"Influencing myself: Self-reinforcement through online political expression."


Communication Research 45(1): 83-111.

Choi, Jihyang. 2016. "News internalizing and externalizing: the dimensions of news sharing

on online social networking sites." Journalism & Mass Communication Quarterly

93(4): 816-835.

Colleoni, Elanor, Alessandro Rozza, and Adam Arvidsson. 2014. "Echo chamber or public

sphere? predicting political orientation and measuring political homophily in Twitter

using big data." Journal of Communication 64(2): 317-332.

DiMaggio, Paul, John Evans, and Bethany Bryson. 1996. "Have American's social attitudes

become more polarized?" American journal of Sociology 102(3): 690-755.

Ellison, Nicole B., Cliff Lampe, and Charles Steinfield. 2009. "Social network sites and society:

Current trends and future possibilities." Interactions 16(1): 6-9.

Festinger, Leon. 2016. A theory of cognitive dissonance. USA: Stanford university press.

Garrett, R. Kelly. 2009. "Politically motivated reinforcement seeking: Reframing the selective

exposure debate." Journal of Communication 59(4): 676-699.

Ha, Jung Bin. 2014. "The rise of closed SNS and prospects." Trend and Prospect: Broadcast,

Telecommunication, and Transmission 71. <http://www.kca.kr/open_content/bbs.do?a

ct=file&bcd=research&msg_no=118&file_no=1>

Hahn, Kyu S., Ju-Yong Park, Deok-Jae Lee, and Lee, Hye-Lim Lee. 2013. "A test of

representativeness and polarization in Twitter followership: A cross-national assessment

of legislators' Twitter followers in the U.S. and South Korea." Journal of

Cybercommunication Academic Studies 30(1): 295-336.

Heeter, Carrie. 1989. "Implications of new interactive technologies for conceptualizing

communication." In media use in the information age: Emerging patterns of

adoption and consumer use, edited by J.L. Salvaggio & Bryant, Hillsdale, NJ:

Lawrence Erlbaum Associates, Inc.

Himelboim, Itai, Stephen McCreery, and Marc Smith. 2013. "Birds of a feather tweet together:

integrating network and content analyses to examine cross-ideology exposure on

Twitter." Journal of Computer-mediated Communication 18(2): 154-174.

Huckfeldt, Robert, Jeanette Morehouse Mendez, and Tracy Osborn. 2004. "Disagreement,

ambivalence, and engagement: The political consequences of heterogeneous networks."

Political Psychology 25(1): 65-95.

Huckfeldt, Robert, and John Sprague. 1987. "Networks in context: the social flow of political

information." American Political Science Review 81(4): 1197-1216.

Hwang, Yoosun. 2013. "Ideological polarization of Twitter space through selective exposure


behavior." The Korean Society for journalism & communication Studies 57(2): 58-79.

Iyengar, Shanto, and Kyu S. Hahn. 2009. "Red media, blue media: Evidence of ideological

selectivity in media use." Journal of Communication 59(1): 19-39.

Kim, Boyoung and Lee, Seulki. 2017. "Taegeukgi wind blowing on SNS." Edaily, March

20, 2017.

Kim, Mihee. 2016. "The effects of news recommendation from Facebook friends on news

credibility perceptions and behaviors for news engagement: Focusing on cognitive

elaboration and opinion leadership." Korean Journal of Journalism & Communication

Studies 60(5): 176-202.

KISA. 2017. 2016 Survey on the Internet usage. Seoul: Korean Internet & Security Agency.

Ko, Tae-Kyung. 2017. "Flag Flapped in Two Names in the Square." Literature 3 (February).

KPF. 2017. 2016 Press audience awareness survey. Seoul: Korea Press Foundation.

Lee, Seok-Pil. 2013. "Personalized contents recommendation system based on social network."

Journal of Broadcast Engineering 18(1): 98-105.

Messing, Solomon, and Sean J. Westwood. 2012. "Selective exposure in the age of social

media: Endorsements Trump partisan source affiliation when selecting news online."

Communication Research 41(8): 1042-1063.

Mutz, Diana C. and Paul S. Martin. 2001. "Facilitating communication across lines of political

difference: The role of mass media." American Political Science Review 95(1):

97-114.

Mutz, Diana C. 2002. "The consequences of cross-cutting networks for political participation."

American Journal of Political Science 46(4): 838-855.

Mutz, Diana C. 2006. Hearing the other side: Deliberative versus participatory democracy.

New York: Cambridge University Press.

Nam, Taewoo, and Jennifer Stromer-Galley. 2012. "The democratic divide in the 2008 U.S.

presidential election." Journal of Information Technology & Politics 9(2): 133-149.

Neuman, W. Russell, Marion R. Just and Ann N. Crigler. 1992. Common knowledge. Chicago:

University of Chicago Press.

Prior, Markus. 2005. "News vs. entertainment: how increasing media choice widens gaps in

political knowledge and turnout." American Journal of Political Science 49(3):

577-592.

Prior, Markus. 2007. Post-broadcast democracy: How media choice increases inequality in

political involvement and polarizes elections. New York: Cambridge University Press.

Rafaeli, Sheizf. 1988. "From new media to communication." Sage annual review of


communication research: Advancing communication science 16: 110-134.

Robertson, Scott P., Ravi K. Vatrapu, and Richard Medina. 2010. "Online video "Friends"

social networking: Overlapping online public spheres in the 2008 U.S. presidential

election." Journal of Information Technology & Politics 7(2): 182-201.

Roh, Jeong-Kyu and Young Min. 2012. "Effects of politically motivated selective exposure

on attitude polarization: A study of non-political online community users." Korean

Journal of Journalism & Communication Studies 56(2): 226-248.

Song, Wha Sun. 2017. "Public sphere or their own fort: 2017 KakaoTalk sphere in South

Korea." Weekly dong-a, March 8, 2017. <http://weekly.donga.com/Main/3/all/11/862528/1>

Stroud, N. Jomini. 2007. "Media effects, selective exposure, and Fahrenheit 9/11." Political

Communication 24(4): 415-432.

Stroud, N. Jomini. 2008. "Media use and political predispositions: Revisiting the concept of

selective exposure." Political Behavior 30(3): 341-366.

Stroud, N. Jomini. 2010. "Polarization and partisan selective exposure." Journal of

Communication 60(3): 556-576.

Sunstein, Cass R. 2000. "Deliberative trouble? why groups go to extremes." The Yale Law

Journal 110(1): 71-119.

Swigger, Nathaniel. 2012. "What you see is what you get: drawing inferences from campaign

imagery." Political Communication 29(4): 367-386.

West, Anne, Jane Lewis, and Peter Currie. 2009. "Students' Facebook 'friends': Public and

private spheres." Journal of youth studies 12(6): 615-627.

Wilson, Robert E., Samuel D. Gosling, and Lindsay T. Graham. 2012. "A Review of Facebook

research in the social sciences." Perspectives on psychological science 7(3): 203-220.

Yonhapnews. 2017. "Rally against impeachment in Seoul plaza." Hankook Ilbo, February

11, 2017. http://www.hankookilbo.com/v/83cad4310b834043a01b0832d03eeb39

Yoo, Sung Woo, and Homero Gil De Zúñiga. 2014. "Connecting blog, Twitter and Facebook

use with gaps in knowledge and participation." Comunicación y Sociedad 27(4): 33-48.

Yun, Seongyi. 2013. "Political use of tweet and its impact on voting behavior: The 19th national

assembly election." 21st Century Political Science Review 23(3): 225-247.

Received 11 December 2017

Received in revised form 25 February 2018

Accepted 28 February 2018

Selective Exposure and Political Polarization of Public ...dcontent.dkyobobook.co.kr/genomad_gift/001/article/1/03/81/10381508.pdfSelective Exposure and Political Polarization of Public

Documents