Summarize the following article.

Media Psychology, 19:203–223, 2016

Copyright © Taylor & Francis

ISSN: 1521-3269 print/1532-785X online

DOI: 10.1080/15213269.2014.1002941

The Role of Partisan Sources and Audiences’ Involvement in Bias Perceptions of

Controversial News

MIHEE KIM Department of Journalism and Mass Communication, Sungkyunkwan University,

Seoul, South Korea

Based on the hostile media effect (HME), this 2 (audiences’ opin-

ion) � 2 (partisan source) � 2 (news valence) factorial experiment

(N D 229) investigated the effects of partisan sources and audi- ences’ different types of involvement on bias perceptions of slanted

news coverage regarding a controversial issue. The results show

that participants rated a news article, regardless of its valence

(congruent vs. incongruent), as less biased when it was attributed

to a partisan source that was consistent with their own position

than when it came from a partisan source that was inconsistent

with their position. Moderating effects of value-relevant involve-

ment on the source effects were found. The effects of partisan

sources on bias perceptions were only significant among those

with moderate or high levels of value-relevant involvement. The

implications of the source effects and the role of value-relevant

involvement as a moderator of such effects were discussed.

Audiences’ perceptions of source have been known to be an important factor in their judgments of a message (Hovland & Weiss, 1951). Audiences’ evaluations of news coverage also depend on which news outlet reports the coverage (Sundar & Nass, 2001). Due to the revolution of information technology, the number of available news outlets is dramatically increasing. The growth of cable television and Internet news outlets has provided a more fragmented media environment where audiences have more options for news consumption (Iyengar & Hahn, 2009). News outlets with certain partisan perspectives have emerged and competed for audiences’ attention in this new environment. These partisan news outlets ‘‘fit the news within a

Address correspondence to Mihee Kim, Department of Journalism and Mass Communi- cations, Sungkyunkwan University, 25-2, Sungkyunkwan-Ro, Jongno-Gu, Seoul, South Korea. E-mail: [email protected]

203

204 M. Kim

political narrative and create a coherent conservative or liberal interpretation of the day’s events’’ (Levendusky, 2013, p. 566). In this process, partisan sources, such as newspapers and television channels, which are known for being conservative or liberal, create certain political agendas by providing slanted news that is written in a biased way ( Jamieson, Hardy, & Romer, 2007). As a result, audiences of partisan sources often perceive the other side to be flawed or duplicitous, which decreases support for bipartisanship within the public (Levendusky, 2013).

With the rise of partisan media exposure, it is more important to under- stand the role of partisan sources in audiences’ perceptions of news content, especially in regard to their bias perceptions of news coverage of con- troversial issues. Audiences’ bias perceptions of news coverage have been examined within the framework of the hostile media effect (HME), which refers to the tendency for those who are highly involved in a controversial issue to perceive balanced news coverage of that issue as biased against their opinions (Gunther, Christen, Liebhart, & Chia, 2001). The concept of the HME has been extended to the concept of the relative HME, which focuses on bias perceptions of slanted news stories (e.g., Gunther & Chia, 2001; Gunther & Christen, 2002; Gunther et al., 2001). In a media landscape that contains partisan news outlets, people have more opportunities to be exposed to slanted news coverage of controversial issues. Therefore, examining how people perceive slanted news stories is more important than it has ever been before when exploring how controversial issues are communicated. The relative HME provides persuasive explanations for bias perceptions of slanted news stories. This suggests that the relative HME is increasingly relevant in contemporary media environments. However, very little research has been conducted regarding the relative HME. This study adds to the literature of the relative HME by exploring the role of partisan sources in bias perceptions of slanted news stories.

Moreover, this study extends past literature by examining certain mod- erators of source effects on the relative HME. Audiences’ involvement in a controversial issue is considered to be a required condition for the HME. Prior research has examined the role of two types of audiences’ involvement, value-relevant involvement and outcome-relevant involvement, in the HME by using balanced news stories that do not contain any source information. In order to extend the past literature, this study investigates the role of audiences’ involvement types as moderators of the effects of source on the relative HME. Most of the studies on the HME have utilized the concepts of both group membership and attitude extremity as indicators of audiences’ involvement (Gunther, Miller, & Liebhart, 2009); however, these concepts do not clearly differentiate the two different types of involvement (Choi, Yang, & Chang, 2009). None of the past studies have consistently conceptualized and operationalized audiences’ value-relevant and outcome-relevant involvement in the context of the effects of source on the relative HME. In order to

Partisan Source and Involvement 205

eliminate this dearth of information, this study explores the role of audiences’ value-relevant and outcome-relevant involvement in the effects of partisan sources (congruent vs. incongruent) on bias perceptions of slanted news (congruent vs. incongruent).

THE HOSTILE MEDIA EFFECT (HME)

Balanced messages are considered to be biased by certain audiences. This phenomenon has been explored within the framework of biased assimilation or the HME. Lord, Ross, and Lepper (1979) documented that biased assimi- lation provided evidence of those with opposing views perceiving research results to be in favor of their points of view. In their study, participants accepted congenial information but ignored uncongenial information that was provided by a research report. This biased assimilation of information resulted in the polarization, rather than the convergence, of opposing views. Regarding partisans’ polarized worldviews within media coverage, Vallone, Ross, and Lepper (1985) investigated how audiences with opposing positions perceive balanced news coverage of a controversial issue. Unlike the results of the biased assimilation studies, they found that partisan news consumers perceive balanced news coverage of a controversial issue to be biased against their own existing position, which is referred to as the HME. In the mass media context, many empirical studies have vividly demonstrated the HME across issues and populations (e.g., Christen, Kannaovakun, & Gunther, 2002; Giner-Sorolla & Chaiken, 1994; Gunther et al., 2001; Gunther et al., 2009; Gunther & Liebhart, 2006; Gunther & Schmitt, 2004; Kim, 2010).

Additionally, the HME has been examined by using clearly slanted news content. In an experimental study, Gunther and associates (2001) found that when partisans are exposed to news that is unfavorable to their side, they perceive the news to be more biased than partisans on the other side of the issue do, which is referred to as the relative HME. Studies of the relative HME (Gunther & Chia, 2001; Gunther & Christen, 2002; Gunther et al., 2001) confirmed that ‘‘regardless of whether news content is objectively neutral, opposing parties would have divergent perceptions of news slant’’ (Choi et al., 2009, p. 56).

Previous studies of the relative HME (Gunther & Chia, 2001; Gunther & Christen, 2002; Gunther et al., 2001) have demonstrated that audiences’ bias perceptions of news content that contains a clear slant depend on whether the news valence is congruent or incongruent with their own opinions. This means that those who have opinions that are incongruent with the news valence perceive the news story to be more biased than those who have opinions that are congruent with the news valence. Therefore, the following hypothesis is proposed in order to test the relative HME of news content from partisan sources:

206 M. Kim

H1: Audiences who have opinions that are incongruent with the news va- lence perceive news stories from partisan sources to be more biased than audiences who have opinions that are congruent with the news valence.

SOURCE EFFECTS WITHIN THE HME

Gunther and Schmitt (2004) explained the influence of sources on the HME by using a reach concept. In their experiment, the HME was manifest when content was attributed to a newspaper article that had a high perceived reach, while biased assimilation was evident when the same content was attributed to a student essay that had a low perceived reach. Gunther and Schmitt’s study provided strong evidence that mass media source is a neces- sary condition for the HME; however, the source concept in their study was confounded with the reach concept.

In order to investigate this confound, Gunther and Liebhart (2006) at- tempted to separate source (journalist vs. college student) from reach (mass media vs. classroom composition) in their experimental design. They found that both reach and source independently produced the HME. However, the interaction between source and reach was found in manipulation checks, which suggests possible confounding effects of the two factors on actual audiences’ hostile media perceptions.

Unlike the above studies, which examined sources that had different perceived reach (mass media vs. student essay), certain scholars compared the effects of a mass media source with those of another mass media source, controlling for perceived reach. For example, Arpan and Raney (2003) ex- amined how different newspapers influence the HME in the context of sports news and fanship. College students were exposed to a balanced story about their hometown college football team in one of three newspapers: a hometown paper, a cross-state rival university town’s paper, and a neutral- town paper. The news from the rival town’s paper and the neutral paper was perceived to be more biased than the same news from the home town paper. In the same vein, Ariyanto, Hornsey, and Gallois (2007) exposed Muslims and Christians to an article about inter-religious conflict. People read the same article from three different newspapers: a Muslim newspaper, a Christian newspaper, and an unidentified newspaper. While participants perceived the article from the Muslim newspaper to be biased toward Mus- lims, they regarded the same article from the Christian newspaper to be biased toward Christians. These studies show that audiences’ prior beliefs of a source influence their bias perceptions of news coverage from that source.

A more recent study (Reid, 2012) tested the effects of source on the HME in the context of partisan authors. Participants who were either Democrats

Partisan Source and Involvement 207

or Republicans read a balanced CNN news report on their candidates that was written by either a member of a Democratic or Republican think tank. When participants saw the news story that was written by authors who had the same political position as their own, they believed that the news story contained more favorable information about their own candidates. In contrast, when they read the news story that was written by authors who had different political views than their own, they believed that the news story contained more favorable information about the other side’s candidates. These findings suggest that while partisan authors who have similar political positions to their audiences produce biased assimilation, partisan authors who have different positions than their audiences generate the HME. In other words, aspects of partisan authors, whose views are congruent or incongruent with the audiences’ political positions, induce different bias perceptions within the audiences regarding the same news coverage.

Further, Reid (2012) investigated how two partisan groups, Democrats and Republicans, perceive an attack against Democrats, based on the views of the authors of the attack. The attack was perceived to be less biased when the author of the attack was a Democrat than when the author was a Republican. By extending previous work that has shown the effects of partisan authors of news content on the HME (e.g., Reid, 2012), this study explores the effects of partisan sources, which are news outlets that have certain partisan perspectives (congruent vs. incongruent), on audiences’ bias perceptions of slanted news stories that are written in a biased way (con- gruent vs. incongruent). Anchored in the body of literature that reveals the effects of source on the HME, the following hypotheses are proposed:

H2a: Audiences perceive an incongruent news story as less biased when it comes from a source that is consistent with their own position than when it comes from a source that is inconsistent with their position.

H2b: Audiences perceive a congruent news story as less biased when it comes from a source that is consistent with their own position than when it comes from a source that is inconsistent with their position.

AUDIENCES’ INVOLVEMENT AND THE HME

A person’s involvement in an issue has been shown to be a predictor of the HME. Two types of involvement, value-relevant and outcome-relevant involvement, have been examined in HME studies. Value-relevant involve- ment, which is often termed ego involvement, refers to ‘‘the psychological state that is created by the activation of attitudes that are linked to important values’’ ( Johnson & Eagly, 1989, p. 290). Outcome-relevant involvement is conceptualized as ‘‘the relevance of an issue to one’s currently important goals or outcomes’’ ( Johnson & Eagly, 1989, p. 292). While people with

208 M. Kim

value-relevant involvement in an issue form their own opinions based on their social or personal values, people with outcome-relevant involvement form their attitudes based on the future consequences of the issue (Choi, Park, & Chang, 2011).

Social judgment theory provides a theoretical framework for under- standing the relationship between audiences’ involvement types and the HME. Social judgment theory posits that different levels of value-relevant involvement with a given issue are related to different levels of acceptance, rejection, or non-commitment regarding news coverage of an issue (Sherif, Sherif, & Nebergall, 1965). According to the theory, those who are more ego involved with an issue tend to have wider latitudes of rejection of statements about the issue, whereas less ego-involved people have relatively narrow latitudes of rejection but wider latitudes of acceptance or non-commitment. While messages that are located within a reader’s latitude of acceptance generate assimilation effects, messages that fall into the latitude of rejection produce the HME (Sherif et al., 1965).

Unlike value-relevant involvement, increasing outcome-relevant involve- ment does not raise the latitude of rejection ( Johnson, Lin, Symons, Camp- bell, & Ekstein, 1995), which suggests that outcome-relevant involvement activates different cognitive processes than value-relevant involvement does (Cho & Boster, 2005; Johnson & Eagly, 1989). People with high outcome- relevant involvement tend to adopt ‘‘a position that maximizes the immediate situational rewards’’ (Petty & Cacioppo, 1979, p. 1916). Thus, these people are known to be motivated to process messages systematically. Drawing on this logic, Petty and Cacioppo (1979) empirically examined how those with outcome-relevant involvement perceive disagreeable arguments. The researchers found that increasing outcome-relevant involvement enhanced the persuasion of counterattitudinal arguments when the arguments were strong and compelling. This suggests that counterattitudinal messages may not result in the HME, especially for individuals who have high outcome- relevant involvement.

Therefore, it was expected that those with value-relevant involvement, rather than outcome-relevant involvement, process messages in an opposi- tional way, which results in the HME (Choi et al., 2011). Choi et al. (2009) pro- vided empirical evidence for this prediction. In their study, value-relevant in- volvement was a critical predictor of the HME, rather than outcome-relevant involvement. The current study extends prior research by testing the effects of these two involvement types in the context of the relative HME. Based on social judgment theory and previous studies (e.g., Choi et al., 2009; Choi et al., 2011), the following hypothesis is proposed:

H3: Value-relevant involvement has a greater effect on the perceived bias of a slanted news story than outcome-relevant involvement.

Partisan Source and Involvement 209

AUDIENCES’ INVOLVEMENT TYPES AND THE

EFFECTS OF SOURCE ON THE HME

Few studies have examined the different influences of the two types of involvement on the effects of source on the HME. Given the disparate nature of value-relevant and outcome-relevant involvement, these two types of involvement may have different effects on the role of partisan sources in the HME. Thus, this study explores which type of involvement is more sensitive to the effects of variations in source factors on bias perceptions of slanted news.

Value-Relevant Involvement

Sherif and Sherif (1967) theorized the relationship between the source of content and readers’ value-relevant involvement with a topic of the content. The researchers argued that because people who are not highly ego involved with an issue lack the internal standards to judge content, they use other cues, such as the identity of the source. In other words, for those who are less ego involved in an issue, the source of content may become a major anchor for the evaluation of the content (Sherif & Sherif, 1967). Consistent with this view, social judgment theory predicts that source effects will be manifest among those with low levels of value-relevant involvement, rather than those with high levels of value-relevant involvement (Sherif et al., 1965).

Previous attitude change studies have provided evidence that supports social judgment theory’s prediction. Sereno (1968) investigated the relation- ship among value-relevant involvement, the high credibility of a source, and attitudinal responses to a belief-discrepant message. When the message was attributed to a highly credible source, participants with high value-relevant involvement changed their attitude on the topic in the direction that was advocated in the message less than participants with low value-relevant involvement did. Along similar lines, Johnson and Scileppi (1969) found a significant interaction between the source and value-relevant involvement. There was greater attitude change in the low value-relevant involvement and high source credibility condition than in the other three combinations of source credibility and value-relevant involvement. The researchers ar- gued that because participants with high value-relevant involvement evaluate content more critically, source differences tend to disappear. Rhine and Severance (1970) also found that more attitude change occurred for those with low, rather than those with high, value-relevant involvement. For those with high value-relevant involvement, source credibility had little effect on attitude change.

Certain HME studies examined the relationship between audiences’ value- relevant involvement and source effects. Gunther et al. (2009) investigated

210 M. Kim

the effects of source on the HME in the context of genetically modified wild rice. Two groups (non-Native Americans vs. Native Americans) were exposed to news from different sources (friendly vs. non-friendly newspapers). A main effect of source on the perceived bias of the news was not found; however, an interaction effect between source and group was significant. Source effects were only manifest in the non-Native American group. Only non-Native Americans judged the news to be more favorable to their posi- tion when it came from a friendly source than when it came from a non- friendly source. Native Americans perceived the same news to be neutral, regardless of the source. The researchers argued that group membership is conceptually related to audiences’ value-relevant involvement. If this is true, the results suggest that those with high value-relevant involvement may not be influenced by sources regarding their hostile media perception. The effects of source on the HME may only be manifest among those with low- levels of value-relevant involvement. This is consistent with the idea of social judgment theory.

A more recent study by Reid (2012) explored how partisanship (Demo- crats vs. Republicans) influences the effects of source on the HME. The effects of source on the HME were amplified by partisanship, which, according to the author, reflects audiences’ value-relevant involvement. This suggests that when people have higher value-relevant involvement, the effects of source on the HME will be greater. This is contrary to social judgment theory’s prediction.

As shown above, previous studies provide conflicting evidence for the relationship between value-relevant involvement and the effects of source on the HME. On the one hand, when value-relevant involvement was op- erationalized as group membership, the effects of source on the HME were manifest among those with low levels of value-relevant involvement (Gun- ther et al., 2009). On the other hand, when value-relevant involvement was measured by partisanship, high levels of value-relevant involvement increased the effects of source on the HME (Reid, 2012).

Partisanship and group membership do not explain what type of in- volvement underlies audiences’ attitudes. A person may decide to belong to one group because the group is related to one’s personal values or because the group is currently related to one’s important goals or outcomes (Choi et al., 2009). Partisanship can also be developed based on either value- relevant or outcome-relevant involvement (Cho & Boster, 2005). Therefore, it is unclear whether either partisanship or group membership distinctly in- dicates audiences’ ego involvement with an issue. Accordingly, the influence of different levels of ego-involvement (high vs. low) on the effects of source on the HME needs to be investigated. This study attempts to measure distinct constructs of audiences’ value-relevant involvement and then tests how this involvement moderates the effects of source on bias perceptions of clearly slanted news by examining the following research question:

Partisan Source and Involvement 211

RQ1: How does an audience’s value-relevant involvement in a news topic moderate the effects of source on the relative HME?

Outcome-Relevant Involvement

Audiences’ outcome-relevant involvement was investigated in relation to persuasive messages. Petty, Cacioppo, and Goldman (1981) found that for those with high outcome-relevant involvement, the quality of arguments in a message significantly influenced their attitudes. Attitudes of those with low outcome-relevant involvement were primarily influenced by the expertise of the source. This difference suggests that the effects of source on persua- sive messages are greater among those with low outcome-relevant involve- ment. In the same vein, Chaiken (1980) examined the relationship between outcome-relevant involvement (high vs. low) and the information process (heuristic vs. systematic). The results show that those with high outcome- relevant involvement employed a systematic information processing strategy in which message-based cognitions mediated persuasion. In contrast, partic- ipants with low outcome-relevant involvement used a heuristic processing strategy in which non-content cues, such as the source’s identity (likable vs. unlikable), mediated persuasion. This difference in involvement suggests that for those who employ the heuristic strategy, source characteristics may exert a greater impact on the perceptions of messages than message character- istics (Chaiken, 1980). These persuasion studies demonstrate that outcome- relevant involvement moderates the effects of source on attitude changes. This means that source effects may be maximized when the message is on a topic that is not related to message receivers’ self-interest or important goals (Petty & Cacioppo, 1979).

In regard to HME research, only Gunther et al.’s (2009) study examined the role of outcome-relevant involvement in source effects, investigating the relationship between attitude involvement and the effects of source on the HME. In their study, as noted earlier, involvement that was based on group membership interacted with source, but involvement that was based on attitude measures did not interact with source. The researchers argued that the attitude measures of their study shared elements with outcome- relevant involvement. If this is true, outcome-relevant involvement may not moderate the effects of source on the HME. This finding is not consistent with the findings of persuasion studies. These conflicting results demonstrate the need for further research that examines whether outcome-relevant involve- ment moderates the effects of source on the HME. Therefore, the following research question is proposed:

RQ2: Does an audience’s outcome-relevant involvement in a news topic moderate the effects of source on the relative HME?

212 M. Kim

METHOD

This study’s focal issue was the dissolution of the Unified Progressive Party in South Korea. The South Korean government accused the Unified Progres- sive Party of participating in pro-North Korean activities and petitioned the Constitutional Court to dissolve this group early in November 2013. Since this occurrence, controversy (pro or con) over this issue has been accelerated in South Korea. Supporters of the dissolution of the party welcomed the gov- ernment’s move as a way to enhance national security, whereas opponents claimed that the compulsory dissolution of the political party may take away people’s political freedom.

This study employed an online experiment that had a 2 (audiences’ opinion: pro or con) � 2 (news valence: congruent or incongruent) � 2 (partisan source: congruent or incongruent) factorial design. In this experi- ment, while audiences’ opinion variable was measured, both news valence and partisan source variables were manipulated. This web-based experiment was conducted during the last week of November 2013.

Participants

An online survey company in South Korea recruited participants. The survey company sent email invitations to its 980,000 panel members. Initially, a total of 483 participants signed up for this study. Because audiences who have opinions that support or oppose the focal issue have been regarded to be a necessary condition for the HME (Gunther et al., 2009), those who reported holding a neutral position (N D 226) were excluded. In addition, data from 28 participants who did not complete this experiment were eliminated. This selection produced a total sample of 229 participants (150 men, 79 women; age: M D 42.17, SD D 10.63).

Procedure

Those who voluntarily wished to participate in this experiment were pro- vided with a link to a website. They completed a pretest survey that asked about their opinions, value-relevant involvement, and outcome-relevant in- volvement in the focal issue. After that survey, they were randomly assigned to one of four conditions, which included being given a news article (con- gruent or incongruent) from a certain source (congruent or incongruent). Random assignment to the conditions was automatically conducted by the online survey tool. After reading the news article, they completed a post survey that measured their perceived bias regarding the news article that they had just read.

Partisan Source and Involvement 213

Stimuli

The valence (pro-dissolution of the Unified Progressive Party or anti-dissolu- tion of the Unified Progressive Party) of the news articles was manipulated based on the supporters’ or opponents’ arguments that were represented in previously published news stories. The valence of the news articles was previously tested by 18 college students. They all indicated that the news articles were explicitly slanted as intended.

Two sources, Chosun Daily and Hankyoreh Newspaper, were selected for this experiment. These two newspapers are considered to be partisan news outlets that have opposite political ideologies in South Korea. While Chosun Daily has a distinctive reputation for providing conservative or pro- government perspectives, Hankyoreh Newspaper is widely known to deliver more liberal or anti-government points of view. The news article page that was attributed to Chosun Daily or Hankyoreh Newspaper was meant to resemble an actual online news page from each newspaper.

Pretest Measures

Participants were asked to indicate their preexisting opinions on the issue (i.e., ‘‘How strongly do you support or oppose the dissolution of the Unified Progressive Party?’’) by using a scale from 1 (strongly oppose) to 7 (strongly support). Two groups were finally identified: a supporter group (N D 117, M D 6.13, SD D .87) and an opponent group (N D 112, M D 2.10, SD D .83). There was a significant difference in opinions between the two groups, t (227) D 36.66, p < .001.

Two types of involvement were measured by the pre-survey. Value- relevant and outcome-relevant involvement are distinct attitude constructs (Choi et al., 2009; Choi et al., 2011). This study conducted a confirmatory factor analysis (CFA) in order to confirm the validity of the two types of involvement constructs. As expected, the two-factor model was supported (�2(4) D 8.29, p D .08, CFI D .994, RMSEA D .06). The one-factor model did not fit the data (�2(5) D 257.26, p < .001, CFI D .656, RMSEA D .47).

In order to measure value-relevant involvement, participants were asked to indicate how much they agreed with the following statements: ‘‘My posi- tion on the dissolution of the Unified Progressive Party has a lot to do with my beliefs about how life should be lived,’’ ‘‘My position on the dissolution of the Unified Progressive Party reflects who I am,’’ ‘‘The values that are most important to me are what determine my stand on the dissolution of the Unified Progressive Party issue’’ (1 D strongly disagree, 7 D strongly agree). The reliability was ˛ D .89 (M D 4.74, SD D 1.40).

Outcome-relevant involvement was measured by asking the extent to which they agreed with the following statements: ‘‘Whether the Unified

214 M. Kim

Progressive Party is dissolved or not has little impact on my life,’’ ‘‘All in all, the effect of the dissolution of the Unified Progressive Party on my life is little,’’ (1 D strongly disagree, 7 D strongly agree). These two items provided high reliability (˛ D .91, M D 4.10, SD D 1.57).

For source manipulation checks, participants were asked to evaluate two newspapers that were employed as partisan sources in the subsequent exper- iment, Chosun Daily and Hankyoreh Newspaper: ‘‘How would you describe Chosun Daily’s (Hankyoreh Newspaper’s) political ideology?’’ (1 D strongly conservative, 7 D strongly liberal), ‘‘How would you describe Chosun Daily’s (Hankyoreh Newspaper’s) political position?’’ (1 D strongly against the gov- ernment, 7 D strongly in favor of the government). In order to reduce any priming effects, the items regarding this study’s target sources were included with questions for other newspapers, such as Joongang Daily, Donga Daily, and Kyunghyang Newspaper, which are leading newspapers in South Korea.

Posttest Measures

In order to measure participants’ perceived bias, three items were adapted from previous studies (e.g., Arpan & Raney, 2003). Participants were prompted to indicate how much they agreed with the following statements using a 7- point scale (1 D strongly disagree, 7 D strongly agree): ‘‘The portrayal of the dissolution of the Unified Progressive Party in the news article is biased,’’ ‘‘Chosun Daily (Hankyoreh Newspaper), which reported the news article, is a biased news outlet,’’ ‘‘The reporter of the news article is biased.’’ All three items, which produced high reliability (˛ D .91), were combined in order to create a bias index.

As a manipulation check for content, participants were asked to report their thoughts about the valence of the news article (i.e., ‘‘How strongly does the news article support or oppose the dissolution of the Unified Progressive Party?’’) by using a scale from 1 (strongly oppose) to 7 (strongly support).

RESULTS

Manipulation Checks

An independent t-test showed that the news article manipulated in favor of the dissolution of the Unified Progressive Party was considered to support the dissolution (M D 5.57, SD D 1.60). In contrast, the article that was designed to be against the dissolution was perceived to oppose it (M D 3.46, SD D 2.07). The difference between the perceived valences of the news articles was significant (t (227) D 8.62, p < .001). This check confirmed that participants perceived the valence of the news articles as intended.

Partisan Source and Involvement 215

A repeated-measures analysis of variance (ANOVA) was employed in order to test the source manipulations. Chosun Daily was perceived to be a conservative outlet (M D 2.05, SD D .08), while Hankyoreh Newspaper was judged to be a liberal channel (M D 5.32, SD D .09). This difference was significant (F (1,228) D 449.09, p < .001, �2p D .66). In addition, participants considered Chosun Daily to be in favor of the government (M D 5.73, SD D .10), while they regarded Hankyoreh Newspaper to be against the govern- ment (M D 2.95, SD D .09). This difference was also significant (F (1,228) D 352.15, p < .001, �2p D .61). As a whole, participants evaluated Chosun Daily to be a conservative and pro-government news outlet, whereas they considered Hankyoreh Newspaper to be a liberal and anti-government news source. This suggests that in the context of this study’s focal issue, Chosun Daily would be perceived to be in favor of the dissolution of the Unified Progressive Party, whereas Hankyoreh Newspaper would be considered to be against the dissolution.

Hypothesis Tests

Hypothesis 1 proposed that audiences with opinions that were incongruent with the news valence would perceive news stories from the partisan sources to be more biased than audiences with opinions that were congruent with the news valence. The results of a one-way ANOVA demonstrated a significant main effect of whether a participant’s opinion is consistent with the news valence (F (1,227) D 29.39, p < .001, �2p D .12). As predicted in Hypothesis 1, those with an opinion that was incongruent with the news valence (M D 4.81, SD D .15) perceived the news to be more biased than those with a stance that was congruent with the news valence did (M D 3.69, SD D .14). Thus, Hypothesis 1 was supported.

Hypotheses 2a and 2b focused on the source effects on audiences’ bias perceptions of clearly slanted news. Hypothesis 2a predicted that audiences would perceive an incongruent news story as less biased when it came from a source that was consistent with their own position than when it came from a source that was inconsistent with their position. In support of Hypothesis 2a, the one-way ANOVA confirmed a statistically significant difference (F (1, 109) D 11.06, p < .01, �2p D .10), which suggests that participants perceived a news story that was inconsistent with their opinions to be less biased when it was attributed to a congruent source (M D 4.33, SD D .17) than when it was attributed to an incongruent source (M D 5.27, SD D .19). Therefore, Hypothesis 2a was supported.

Hypothesis 2b proposed that audiences would perceive a congruent news story as less biased when it came from a source that was consistent with their own position than when it came from a source that was inconsistent with their position. The results of a one-way ANOVA confirmed a significant main effect of whether a participant’s opinion is consistent with the source’s

216 M. Kim

position (F (1,116) D 5.35, p < .05, �2p D .04). Participants considered a congruent news story to be less biased when it was delivered by a congruent source (M D 3.36, SD D .21) than when it came from an incongruent source (M D 4.03, SD D .21). Thus, Hypothesis 2b was supported.

Hypothesis 3 proposed that value-relevant involvement would have a greater effect on the perceived bias of clearly slanted news than outcome- relevant involvement. A hierarchical regression was conducted in order to test Hypothesis 3. Demographics, content (congruent vs. incongruent), and source (congruent vs. incongruent) were entered in the first block, and value-relevant involvement and outcome-relevant involvement were entered in the second block (see Table 1). In this regression model, content and source (Block 1) were the manipulated factors, and the two types of in- volvement (Block 2) were the measured factors. Collinearity statistics (value- relevant involvement: tolerance D .92, Variance Inflation Factor (VIF) D 1.09; outcome-relevant involvement: tolerance D .97, VIF D 1.03) indicated that the two involvement variables could be entered into the regression without any collinearity concerns. The results of the regression showed that neither value-relevant involvement (ˇ D �.03, p D .71) nor outcome- relevant involvement (ˇ D �.02, p D .81) was related to the participants’ bias perceptions. Therefore, Hypothesis 3 was not supported.

Research Question 1 explored whether an audience’s value-relevant involvement in a news topic would moderate the source effects on the bias perceptions of the slanted news. A moderation test was conducted by using the Hayes PROCESS macro (Hayes, 2013), treating sources (congruent vs. incongruent) as an independent variable and value-relevant involvement as a moderator. The interaction between source (incongruent was coded as high) and value-relevant involvement was significant (b D .43, SE D .16, t D 2.67, p < .01). In order to probe this interaction, values of the moderator were sorted into three groups using a sample mean, as well as plus and minus

TABLE 1 Hierarchical Regression with Bias Perception as a Dependent Variable

Predictor Model 1 Model 2

Gender �.067 �.071 Age �.055 �.048 Source .239* .238* Content .328* .326* Value-relevant involvement �.026 Outcome-relevant involvement �.016

R2 .179* .179* Adjusted R2 .164 .157

Note. Standardized coefficients are reported.

*p < .001.

Partisan Source and Involvement 217

one standard deviation from the mean. The three groups were represented as low, moderate, and high on the moderator. Figure 1 shows that the interaction between source and value-relevant involvement approaches a transverse interaction because the regression lines for the congruent and incongruent sources move in opposite directions (Eveland, 1997). An effect of source on participants’ perceived bias was not significant at low levels of value-relevant involvement (�1 SD from the mean) (b D .23, SE D .27, t D .84, p D .40), but the effect was significant at moderate (mean) (b D .81, SE D .21, t D 3.84, p < .01) and high levels of value-relevant involvement (C1 SD from the mean; b D 1.39, SE D .33, t D 4.24, p < .001). The results showed that an effect of source (congruent vs. incongruent) on perceived bias of a news article was only manifest among those with moderate or high levels of value-relevant involvement in the focal issue of the article.

Research Question 2 explored whether an audience’s outcome-relevant involvement in a news topic would moderate the source effects on the bias perceptions of the slanted news. A second moderation test was conducted by treating source as an independent variable and outcome-relevant involve- ment as a moderator. The results indicated that there was no significant interaction between source (incongruent was coded high) and outcome- relevant involvement (b D �.27, SE D .15, t D �1.77, p D .08). The results showed that outcome-relevant involvement did not moderate the source effects on participants’ bias perceptions of the slanted news.

FIGURE 1 Conditional effects of congruent versus incongruent sources on bias perception among those who have relatively low, moderate, or high levels of value-relevant involvement.

218 M. Kim

DISCUSSION

The present study explored the role of partisan sources (congruent vs. in- congruent) in audiences’ bias perceptions of slanted news (congruent vs. incongruent). Particularly, source effects were examined in the context of the audiences’ two types of involvement, value-relevant and outcome-relevant involvement.

Consistent with findings of prior research on the relative HME, the interaction between news valence and the audiences’ opinions had great effects on the audiences’ bias perception of slanted news stories. Audiences with opinions that were incongruent with the news valence perceived news stories from partisan sources as more biased than those with opinions that were congruent with the news valence did. Source effects on perceived bias of slanted news stories were also found. Audiences rated the news article, regardless of its valence (congruent vs. incongruent), as less biased when it was attributed to a source that was consistent with their own position than when it came from a source that was inconsistent with their position. Contrary to the expectation, all types of audiences’ involvement in the focal issue were not related to the audiences’ bias perceptions of the slanted news stories. While value-relevant involvement moderated the effects of partisan sources on the bias perceptions of the slanted news stories, outcome-relevant involvement did not. Source effects were only significant among those with moderate or high levels of value-relevant involvement in the issue.

The results of this study have several practical and theoretical impli- cations. First, this study suggests that partisan sources might amplify or attenuate the relative HME. According to the relative HME, audiences with prior opinions on an issue consider news stories that are incongruent with their own opinions to be biased (Gunther & Chia, 2001; Gunther & Christen, 2002; Gunther et al., 2001). This bias perception could inhibit audiences from ‘‘giving reasonable consideration’’ (Gunther et al., 2001, p. 316) to news coverage that is inconsistent with their points of view, which enhances opinion polarization regarding controversial issues. The results of this study indicate that audiences perceive opinion-incongruent content that comes from a source that is consistent with their own position to be relatively unbiased. This suggests that if partisan sources continue to provide one-sided content that only reflects their political position (conservative or liberal), this may amplify the relative HME. In contrast, if partisan sources deliver content whose valence is inconsistent with the sources’ existing political or ideological disposition (e.g., when a conservative news outlet reports a news story that contains a few liberal perspectives), this may mitigate the relative HME.

The role of partisan sources in reporting controversial issues is particu- larly important in South Korea. South Korea has a different journalism history than that of Western societies (Oh & Park, 2005). For example, the institution

Partisan Source and Involvement 219

of journalism in the United States adopted a norm of objective reporting in the turn of the twentieth century (Schudson, 2001). Although partisan media have emerged particularly on cable news and the Internet, most mainstream news outlets in the United States are considered to be objective and balanced (Levendusky, 2013). In contrast, mainstream news outlets in South Korea are developed based on their political or ideological positions, rather than on professional ethics such as objectivity (Oh & Park, 2005). As a result, South Korea’s mainstream news outlets are sharply polarized and only focus on their political positions (conservative vs. liberal).

Ideologically polarized mainstream news organizations have been criti- cized for aiding opinion polarizations in South Korea. In this situation, it is necessary for news outlets to address this criticism by sharing a professional norm of objectivity and fairness. This study suggests that news sources that provide multiple perspectives, rather than only providing their parti- san views, offer audiences opportunities to give consideration to alternate perspectives of a controversial issue, which may contribute to the attenuation of opinion polarization within that issue.

Second, this study extends the discussion of the effects of source on the HME by examining the moderating functions of audiences’ different types of involvement. Two types of involvement, value-relevant and outcome- relevant involvement, were often intertwined in prior studies, leading to theoretical ambiguity in understanding the relationship between audiences’ involvement and the source effects. This study attempts to reduce this theo- retical ambiguity by exploring the distinct roles of different types of involve- ment in the effects of source on the relative HME.

Unlike previous studies, which reveal that value-relevant involvement ‘‘leads people to process media coverage as more biased’’ (Choi et al., 2009, p. 70), this study did not find any direct impacts of value-relevant involve- ment on the perceived bias of slanted news. Value-relevant involvement only influenced the bias perceptions by moderating the source effects. Those with moderate or high levels of value-relevant involvement were significantly influenced by partisan sources (congruent vs. incongruent), whereas those with low levels of value-relevant involvement were not vulnerable to such source effects. This is inconsistent with what social judgment theory predicts. Social judgment theory suggests that because people with high levels of value-relevant involvement have a wider latitude of rejection, they perceive information to be more biased or hostile. Moreover, the hostile perceptions of those with high levels of value-relevant involvement are not expected to be easily influenced by source characteristics (Sherif et al., 1965; Sherif & Sherif, 1967). Several attitude change studies (e.g., Johnson & Scileppi, 1969; Rhine & Severance, 1970; Sereno, 1968) also suggest that for those with high value-relevant involvement, the sources of messages may have little effect on attitude changes. In contrast, this study found that those with high levels of value-relevant involvement evaluated news coverage to be

220 M. Kim

less biased or hostile when the coverage was attributed to a source that was congruent with their own opinion. These findings provide confirmatory evidence for the self-categorization explanation of the HME. A recent study (Reid, 2012) suggests that self-categorization theory, rather than social judg- ment theory, can explain the effects of source on the HME. According to self-categorization theory, audiences’ internalized group membership (e.g., Democrats vs. Republicans) is a basis for their bias perceptions of the news (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Therefore, a news story from an ingroup source is perceived to be less biased than the same news from an outgroup source. When the position of individuals within a group is more extreme, it is more likely that they will use group membership as a critical element of their bias perceptions of the news (Reid, 2012). The current study extends prior research on the self-categorization explanation of the HME by focusing on the involvement types that underlie audiences’ group membership. As noted earlier, group membership can represent either value-relevant or outcome-relevant involvement. The results of this study show that only value-relevant involvement that underlies group member- ship moderates the effects of source on the HME. Moreover, this study demonstrates that the self-categorization explanation of the HME makes sense under certain conditions of audiences’ value-relevant involvement. Only those with moderate or high levels of value-relevant involvement seem to categorize congruent and incongruent sources as ingroup and outgroup sources, respectively. This means that self-categorization theory may only explain the source effects (ingroup vs. outgroup) for people with high levels of value-relevant involvement in an issue.

Third, because those who show greater value-relevant involvement tend to hold a more fixed and polarized position on an issue, they are less likely to be persuaded than those who show less involvement ( Johnson & Eagly, 1989). Contrary to this information, the current study suggests that if an outgroup message is delivered by an ingroup source, it might be highly persuasive for those with high-levels of value-relevant involvement. For example, if ingroup news sources (conservative or liberal) present messages from all of the sides fairly, including counterattitudinal messages, when reporting on controversial subjects, this would provide their audiences with a chance to rethink and understand the other party. An outgroup message from an ingroup news source may have an especially striking effect on changing attitudes of people with high ego involvement, who play a leading role in opinion polarization within an issue by joining advocacy groups and collective actions (Grunig, 1989).

The present study is subject to several limitations. This study did not recruit participants from existing group memberships, instead grouping par- ticipants based on their prior opinions (pro vs. con) on an issue. If future studies examine source effects in the context of actual group memberships, as many HME studies have done, this would be more helpful in under-

Partisan Source and Involvement 221

standing the role of participants’ social identity in their bias perceptions. The issue that was used in this study, the dissolution of the Unified Progressive Party, is more likely to be related to participants’ values rather than their everyday lives. Therefore, it is required to verify the findings of this study by using highly outcome-relevant issues. This study excluded participants with a neutral position on the issue: In this study, 46.79% of participants who were initially recruited had indicated that they held a neutral position on the issue. Given the large portion of the ambivalent public, it is necessary to examine how those who have a neutral position perceive slanted news and how partisan sources influence their bias perceptions. In this study, participants’ bias perceptions were measured as either neutral or hostile. The direction of the perceived bias was not precisely measured. If future studies explore audiences who have a neutral position on an issue, a directional bias measure would be more beneficial.

In conclusion, this study found different effects of source (congruent vs. incongruent) on the HME and identified audiences’ value-relevant in- volvement as a moderator of such effects. This study expands self-categori- zation explanations for the HME, providing strong evidence that the self- categorization of sources (ingroup vs. outgroup) plays a role in the bias perceptions of news stories about a controversial issue only for audiences with moderate or high levels of value-relevant involvement in that issue. This study also provides empirical evidence that outgroup message from ingroup source may attenuate the HME. These findings highlight the potential role of partisan news outlets in a divided society where partisan audiences are kept apart. Partisan news outlets may contribute to mitigate partisanship or opinion polarization in the society by providing multiple perspectives beyond their own ideological position.

REFERENCES

Ariyanto, A., Hornsey, M. J., & Gallois, C. (2007). Group allegiances and perceptions of media bias: Taking into account both the perceiver and the source. Group Processes & Intergroup Relations, 10, 266–279. doi:10.1177/1368430207074733

Arpan, L. M., & Raney, A. A. (2003). An experimental investigation of news source and the hostile media effect. Journalism & Mass Communication Quarterly, 80, 265–281. doi:10.1177/107769900308000203

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, 752–766. doi:10.1037/0022-3514.39.5.752

Cho, H., & Boster, F. J. (2005). Development and validation of value-, outcome-, and impression-relevant involvement scales. Communication Research, 32, 235– 264. doi:10.1177/0093650204273764

Choi, J., Park, H. S., & Chang, J. J. C. (2011). Hostile media perception, involvement types, and advocacy behaviors. Journalism & Mass Communication Quarterly, 88, 23–39. doi:10.1177/107769901108800102

222 M. Kim

Choi, J., Yang, M., & Chang, J. (2009). Elaboration of the hostile media phenomenon: The roles of involvement, media skepticism, congruency of perceived media influence, and perceived opinion climate. Communication Research, 31, 54– 75. doi:10.1177/0093650208326462

Christen, C. T., Kannaovakun, P., & Gunther, A. C. (2002). Hostile media perceptions: Partisan assessments of press and public during the 1997 United Parcel Service strike. Political Communication, 19, 423–436. doi:10.1080/10584600290109988

Eveland, W. P. (1997). Interactions and nonlinearity in mass communication: Con- necting theory and methodology. Journalism & Mass Communication Quar- terly, 74, 400–416. doi:10.1177/107769909707400211

Giner-Sorolla, R., & Chaiken, S. (1994). The causes of hostile media judgments. Jour- nal of Experimental Social Psychology, 30, 165–180. doi:10.1006/jesp.1994.1008

Grunig, J. E. (1989). Sierra Club study shows who become activists. Public Relations Review, 15, 3–24. doi:10.1016/S0363-8111(89)80001-3

Gunther, A. C., & Chia, S. C. (2001). Predicting pluralistic ignorance: The hostile media perception and its consequences. Journalism & Mass Communication Quarterly, 78, 688–701. doi:10.1177/107769900107800405

Gunther, A. C., & Christen, C. T. (2002). Projection or persuasive press? Contrary effects of personal opinion and perceived news coverage on estimates of pub- lic opinion. Journal of Communication, 52, 177–195. doi:10.1111/j.1460-2466. 2002.tb02538.x

Gunther, A. C., Christen, C. T., Liebhart, J. L., & Chia, S. C. (2001). Congenial public, contrary press, and biased estimates of the climate of opinion. Public Opinion Quarterly, 65, 295–320. doi:10.1086/322846

Gunther, A. C., & Liebhart, J. L. (2006). Broad reach or biased source? Decomposing the hostile media effect. Journal of Communication, 56, 449–466. doi:10.1111/ j.1460-2466.2006.00295.x

Gunther, A. C., Miller, N., & Liebhart, J. L. (2009). Assimilation and contrast in a test of the hostile media effect. Communication Research, 36, 747–764. doi:10.1177/ 0093650209346804

Gunther, A. C., & Schmitt, K. (2004). Mapping boundaries of the hostile media effect. Journal of Communication, 54, 55–70. doi:10.1111/j.1460-2466.2004.tb02613.x

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford.

Hovland, C. I., & Weiss, W. (1951). The influence of source credibility on communi- cation effectiveness. Public Opinion Quarterly, 15, 635–650. doi:10.1086/266350

Iyengar, S., & Hahn, K. S. (2009). Red media, blue media: Evidence of ideological se- lectivity in media use. Journal of Communication, 59, 19–39. doi:10.1111/j.1460- 2466.2008.01402.x

Jamieson, K. H., Hardy, B., & Romer, D. (2007). The effectiveness of the press in serving the needs of American democracy. In K. H. Jamieson (Ed.), A republic divided: The Annenberg Democracy Project (Pp. 21–51). New York, NY: Oxford University Press.

Johnson, B. T., & Eagly, A. H. (1989). Effects of involvement on persuasion: A meta- analysis. Psychological Bulletin, 106, 290–314. doi:10.1037/0033-2909.106.2.290

Johnson, B. T., Lin, H. Y., Symons, C. S., Campbell, L. A., & Ekstein, G. (1995). Initial beliefs and attitudinal latitudes as factors in persuasion. Personality and Social Psychology Bulletin, 21, 502–511. doi:10.1177/0146167295215008

Partisan Source and Involvement 223

Johnson, H. H., & Scileppi, J. A. (1969). Effects of ego-involvement conditions on at- titude change to high and low credibility communicators. Journal of Personality and Social Psychology, 13, 31–36. doi:10.1037/h0027992

Kim, K. S. (2010). Public understanding of the politics of global warming in the news media: The hostile media approach. Public Understanding of Science, 20, 1– 17. doi:10.1177/0963662510372313

Levendusky, M. (2013). Partisan media exposure and attitudes toward the opposition. Political Communication, 30, 565–581. doi:10.1080/10584609.2012.737435

Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098–2109. doi:10.1037/0022- 3514.37.11.2098

Oh, T. S., & Park, S. H. (2005). Hostile media perception: Message or messenger? Korean Journal of Journalism and Communication Studies, 49, 135–290.

Reid, S. A. (2012). A self-categorization explanation for the hostile media effect. Journal of Communication, 62, 381–399. doi:10.1111/j.1460-2466.2012.01647.x

Petty, R. E., & Cacioppo, J. T. (1979). Issue involvement can increase or decrease per- suasion by enhancing message-relevant cognitive responses. Journal of Person- ality and Social Psychology, 37, 1915–1926. doi:10.1037/0022-3514.37.10.1915

Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of argument based persuasion. Journal of Personality and Social Psychology, 41, 847–855. doi:10.1037/0022-3514.41.5.847

Rhine, R. J., & Severance, L. J. (1970). Ego-involvement, discrepancy, source cred- ibility, and attitude change. Journal of Personality and Social Psychology, 16, 175–190. doi:10.1037/h0029832

Sereno, K. K. (1968). Ego-involvement, high source credibility, and response to a belief-discrepant communication. Speech Monographs, 35, 476–481.

Sherif, C. W., & Sherif, M. (1967). Attitude, ego-involvement, and change. New York, NY: Wiley.

Sherif, C. W., Sherif, M., & Nebergall, R. E. (1965). Attitude and attitude change: The social judgment-involvement approach. Philadelphia, PA: Saunders.

Schudson, M. (2001). The objectivity norm in American journalism. Journalism, 2, 149–170.

Sundar, S. S., & Nass, C. (2001). Conceptualizing sources in online news. Journal of Communication, 51, 52–72. doi:10.1111/j.1460-2466.2001.tb02872.x

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford, UK: Black- well.

Vallone, R. P., Ross, L., & Lepper, M. R. (1985). The hostile media phenomenon: Biased perception and perceptions of media bias in coverage of the Beirut mas- sacre. Journal of Personality & Social Psychology, 49, 577–587. doi:10.1037/0022- 3514.49.3.577