Latest Issue

Rumors refer to fact claims about individuals, groups, events and institutions that have not yet been validated (Sunstein, 2014) but have earned credibility, not because direct evidence is accumulated to endorse them, but because they are shared among a wide range of individuals (Fine & Ellis, 2013). Rumors are characterized by two features (Berinsky, 2017). First, rumors are statements that lack specific standards of evidence (Fine & Ellis, 2013; Jerit & Zhao, 2020). Political rumors, therefore, are distinguished from “warranted beliefs” (Keeley, 1999). Second, rumors nonetheless acquire notable influence through widespread social transmission (Berinsky, 2017). The development of communication technologies has made the expansion and reproduction of political rumors easier than ever. As political rumors are replicated and spread through an extensive range of social networks, political rumors can have the noteworthy potential to sway and create public opinion.

Recently, SNS has become a powerful vehicle to accelerate the spread of political rumors. Numerous studies have theorized the role of technological affordances and audiences in how information on SNS is transmitted and accepted (Sundar, 2008; Walther, 2017). More germane to the current research, Oh et al. (2013) noted the remarkable influence of source ambiguity, content ambiguity, and anxiety on rumor mongering on media with a surprisingly little comparable effect of social ties on the spreading of rumors. As such, it is conceivable that the ambiguity of the source as well as sheer volume of content on SNS may have fueled the dissemination of malicious rumors.

According to Entman (1993), “[t]o frame is to select some aspects of a perceived reality and make them more salient in a communicating text in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation” (p. 52). Framing pertains to how an issue or an event is presented by the media, thereby engendering a certain interpretation (Scheufele, 2000). Thus, news framing can influence the way audience members appraise an issue or an event by highlighting certain aspects (Scheufele, 2000). In this vein, how a political candidate is framed does not only paint citizens’ assessment (Cho, 2005; Shen, 2004) but also affect their tendency to share news about that candidate (Bøggild et al., 2021).

Recent advances in framing theory underscore its critical role in political communication, particularly in shaping not only what audiences think about, but also how they think about it (Chong & Druckman, 2007). In addition, Scheufele and Iyengar (2017) stress that framing is not static but adapts to evolving media landscapes and audience perceptions, underscoring the importance of considering contemporary dynamics in the analysis of media effects. Meanwhile, news media can cover stories about a political candidate in light of two distinct aspects, namely, image or issue (Roddy & Garramone, 1988; Schenck-Hamlin et al., 2000). The image coverage focuses on character-related qualities, whereas the issue coverage emphasizes candidates’ positions on policy, governmental action, and civic affairs (Kaid & Johnston, 1991). Similarly, traits of a political candidate can be divided into two dimensions: personality-based and performance-based (Kinder et al., 1980). Johnston (2002) also identified two essential personality considerations of political actors: “Character, the political actor’s integrity and empathy; and competence, the actor’s intellectual and inspirational attributes” (p. 166). Bittner (2011) aptly summarized trait qualifications of political leaders: character-related traits such as empathy, warmth, and compassion and competence-related traits such as intelligence and strength of leadership.

These previous studies lend plausibility to the idea that news coverage of political rumors can be classified according to its emphasis either on a politician’s character or competence; further, these different foci of news reports can embody distinctive frames deeply entrenched in media coverage of political rumors. In line with this notion, we categorize news media’s frames of rumor coverage either into character-emphasized political rumors or competence-emphasized political rumors. Drawing on this typology, we strive to uncover differential effects of characteror competence-framed rumors on their diffusion. On the other hand, theories of framing explain the role of selection and salience in promoting a specific interpretation or a moral evaluation of the item described (Entman, 1993). Taken together, it is conceivable that a distinctive frame of political rumor can lead to a distinguishable interpretation, as it can make certain aspects of a candidate more salient and nonequivalent.

The third-person effect (TPE) posits that individuals tend to believe others are more influenced by media messages than themselves, particularly for socially undesirable content (Davison, 1983; Tsfati & Cohen, 2005). Recent studies have explored TPE in various contexts. In the realm of social media, Lee and Yeon (2022) found that individuals with low political trust and high political self-efficacy are more likely to exhibit stronger third-person perception (TTP) of fake news effects and are less likely to share fake news than others. Understanding privacy risks, Chen and Atkin (2020) revealed users perceive greater privacy risks for others than for themselves, with social distance, perceived privacy knowledge, and online privacy experiences serving as significant predictors of these perceptions. In health communication, Siegenthaler and Fahr, (2023) explored the effects of narrative voice in explainer videos about work-related stress, finding that narrative voice impacts identification with characters and attitudes towards health recommendations.

As described above, framing plays a crucial role in shaping public perception (Entman, 1993). Political rumors can be framed in terms of a candidate’s character or competence (Johnston, 2002; Kinder et al., 1980). Characterframed rumors focus on personal traits, while competence-framed rumors emphasize leadership and policy-related skills. Research has shown that framing can influence the magnitude of TPE. For example, Lancendorfer (2005) explored how political ads framed around character (personal) vs. competence attitudes elicit different behavioral responses among voters. This study’s findings suggest that character-based attributions, which elicit emotional reactions, lead to stronger actions, aligning with TPE where individuals act based on their perception that others are more influenced by emotionally charged rumors. In the similar context, Tutmann (2016) demonstrated how emotional framing (character-based) tends to invoke stronger behavioral responses compared to competence-based framing, which may relate to the stronger TPT seen in emotionally charged character-based rumors. Similarly, Wei et al. (2011) demonstrated that the framing of political attack ads influenced the extent of TPE. Building on these findings about framing effects on TPE, it is important to consider how different types of political information, particularly those related to character versus competence, may differentially impact third-person perceptions.

The psychological mechanisms underlying TPE, such as ego enhancement (Perloff, 1989) and biased optimism (Gunther & Mundy, 1993), may be more pronounced when dealing with character-related information. This aligns with research by Hoorens and Ruiter (1996), who found that people tend to believe they are less susceptible to persuasion attempts targeting personality traits compared to those targeting competence-related attributes. Moreover, Funk (1999) demonstrated that character assessments often play a larger role in voters’ overall evaluations of political candidates than issue-based assessments. However, voters typically claim to prioritize competence-related factors in their decision-making (Fiorina, 1988). This discrepancy suggests that individuals may underestimate the influence of character-related information on themselves while overestimating its effect on others. Jang and Kim (2018) also found that the perceived influence of rumors on others was greater for character-attacking rumors compared to policy-related rumors. This finding directly supports the notion that character-framed political information may elicit stronger TPE.

We argue that when political rumors are framed negatively about a candidate’s trait qualification, exposure to those rumors can trigger TPE, perhaps due to ego enhancement or biased optimism. This notion is reinforced by the findings of Tal-Or and Tsfati (2016), who showed that TPE can be intensified when individuals perceive the content to be particularly biased. Character framing can prompt a higher level of TPE than the competence counterpart, given that individuals who think themselves to be more rational than others may believe that the former is less relevant for the evaluation of a candidate as compared with the latter and thus conclude that the former is more likely to affect others than themselves.

While character and competence are both critical considerations in selecting candidates, voters may be inclined to believe that character affects others more, whereas competence influences themselves more (Cappella & Li, 2023; Green, 2024). This perception may stem from a desire to present their decisions as rational and based on logical evaluation, rather than being driven by emotional considerations. This bias is also consistent with the ego-enhancement strategy, in which individuals strive to maintain a positive self-image by emphasizing their rationality (Gunther & Mundy, 1993; Perloff, 1989). Supporting this line of reasoning, Lau and Redlawsk (2001) found that voters often employ cognitive heuristics to simplify political decisionmaking, frequently focusing on perceived competence to reinforce an image of rational decision-making. Similarly, Fiorina (1988) observed that voters who consider themselves rational tend to prioritize candidate qualities such as competence and leadership over personality. Moreover, Sun et al. (2008) suggests that TPE is often heightened when content is perceived as undermining social norms, a phenomenon particularly pertinent to character-based political rumors. As such, we hypothesize that:

• H1. The character-framed political rumor will prompt a higher level of TPE than the competence-framed counterpart.

Ego enhancement, which motivates individuals to evaluate themselves positively, is one of the main facilitators of TPE (Perloff, 1999). This selfoverestimation bias is also likely to reinforce TPE of political rumors. Given that others are deemed less reasonable than themselves, people may be driven to believe that political rumors on social media—i.e., those highlighting presumably less relevant information in determining a candidate’s capacity—have a greater impact on others than on themselves. Recent research suggests that TPE is particularly potent when content is perceived as threatening or socially undesirable, thereby further amplifying the belief that others are more susceptible to its influence (Jang & Kim, 2018).

Meanwhile, the elaboration likelihood model (ELM) is a dual-process theory of persuasion that explains how attitudes are formed and changed (Petty & Cacioppo, 1986a). The model posits two routes to persuasion: the central route, involving high elaboration and critical thinking, and the peripheral route, relying on simple cues or heuristics. In applying this model to research, it’s crucial to understand that certain commonly measured variables do not directly indicate the processing route. Perceived argument strength, which measures how persuasive participants find a message, does not necessarily indicate central route processing. Similarly, emotional responses to a message can occur in both central and peripheral processing and don’t inherently indicate engagement in either route. Therefore, researchers must be cautious not to infer processing routes solely from these measures, as neither perceived argument strength nor emotional responses directly indicate which route a participant is using.

The ELM framework provides a valuable lens through which to examine the spread and impact of misinformation, particularly in the context of fake news and political rumors, as demonstrated by several recent studies. For example, Osatuyi and Hughes (2018) analyzed that fake news often uses peripheral cues and negative affective content, potentially influencing readers through the peripheral route of persuasion. Weeks and Garrett (2014) found that political rumors significantly influenced the 2008 U.S. presidential election by reinforcing biases through motivated reasoning. Seo et al. (2012) developed an approach to identify rumors and their sources in social networks, emphasizing the need for interventions to counteract misinformation spread. Kwak et al. (2020) also examined instant messaging platforms increased political participation but not political knowledge, potentially facilitating rumor spread through the peripheral route. Hosni and Li (2020) proposed a dynamic blocking period approach to minimize the influence of malicious rumors during breaking news events, aiming to disrupt rapid dissemination of unverified information.

The synthesis of the literature on TPE and ELM suggests that the perception about rumors’ effects on others can influence which information processing route is taken. Given that characterframed political rumors tend to exaggerate relatively less integral aspects of candidates such as trustworthiness or likability (Homer & Batra, 1994), those cues may fall short of forming solid reasonable arguments (Bittner, 2011). In line with TPE’s role in ego enhancement (Perloff, 1989), individuals may believe that emotionally charged, less rational political rumors will have a greater impact on others than on themselves, as such rumors are perceived to be less relevant to their own voting considerations. TPE can then significantly shape how these cues are processed.¹ In particular, when individuals view themselves as less impacted by a certain message, their motivation to engage deeply with the content diminishes, leading them to favor the peripheral route of processing. This relationship between perceived relevance and processing route is critical because it underscores how TPE may predispose individuals to engage superficially with political rumors, thereby reinforcing reliance on peripheral cues over central processing (Tormala & Petty, 2004). For example, younger generations might immediately experience TPE in response to a message about healthcare policy, deeming it less relevant to them. This perception may then prompt them to process the message heuristically via the peripheral route. Consequently, TPE can play a critical role in shaping how individuals assess a message or rumor. Specifically, when people perceive a message as likely to have a greater impact on others than on themselves, they may process the information less critically, aligning more with peripheral route processing as they allocate fewer cognitive resources to analyze content deemed less personally relevant.

However, the relationship between TPE and central route processing in ELM is more nuanced and warrants further exploration. Notably, the literature suggests that central and peripheral routes are not mutually exclusive. Petty and Cacioppo (1986a) recognized that while heuristic cues often drive initial message processing, deeper, more effortful cognitive engagement can occur when the message aligns with personally significant attitudes. Slater and Rouner’s (2002) extension of the ELM further emphasizes that engagement with a narrative itself can facilitate more thorough processing, even via the central route, when the content resonates meaningfully with the audience. Therefore, while we hypothesize a positive association between TPE and peripheral processing, this does not imply a straightforward inverse or absence of association with central route processing. Rather, the link between TPE and central route processing may vary based on additional factors, such as personal relevance or motivation, which extend beyond the scope of TPE alone. Thus, we propose the following hypothesis and research question:

• H2. TPE of the character-framed political rumor compared to competence-framed political rumor will be more associated with a response indicative of more superficial processing linked to the peripheral route.

• RQ1. Will TPE of the character-framed political rumor compared to competence-framed political rumor be more associated with a response indicative of more effortful processing linked to the central route?

The extant literature on ELM suggests that individuals tend to treat political rumors as peripheral cues, given that those are usually challenging and ambiguous to authenticate in fuzzy situations (Petty & Cacioppo, 1986a). As individuals engage in less scrutiny of message content, they tend to focus on peripheral cues such as source credibility and attractiveness when processing that information (Pornpitakpan, 2004). In particular, voters typically put limited cognitive effort toward political rumors focusing on the personality or character aspect of a candidate. This is consistent with the findings of Albarracín and Vargas (2010), who noted that individuals are less likely to engage in in-depth processing when the content lacks direct relevance to their immediate concerns or interests. As such, it can be inferred from ELM that political rumors emphasizing distinct dimension of a candidate may be handled through a different message processing route. Particularly germane to the present study, Homer and Batra (1994) noted that individuals process character-based political messages via the peripheral route, given that those tend to necessitate an individual’s vigorous curiosity or complex cognitive abilities to a lesser degree.

Meanwhile, numerous studies have highlighted the significance of source credibility and attractiveness, key catalysts of the peripheral route, in individual’s intention to disseminate (Shin et al., 2018) and forward message on SNS (Yan & Huang, 2014). ELM has identified source credibility (Pornpitakpan, 2004) and attractiveness (Shavitt et al., 1994) as important cues of the peripheral route. Individuals who process information peripherally are more susceptible to social validation cues and emotional responses (Petty & Cacioppo, 1986a), which have been identified as crucial determinants of SNS sharing behavior (Heimbach & Hinz, 2018). This creates a self-reinforcing cycle, as described by Vosoughi et al. (2018). The peripheral processing of political rumors facilitates their rapid dissemination through SNS, while social proof generated by this sharing further encourages peripheral processing and subsequent dissemination. The immediacy and emotional nature of peripheral processing thus serves as facilitating factor for the dissemination of rumors in SNS environments where rapid, affect-driven decisions are the norm (Qiu et al., 2017).

In accordance with this line of reason, political rumors that prioritize peripheral cues, such as a candidate’s credibility and attractiveness, are included to be processed through peripheral route and subsequently shared on SNS. This is due to the fact that these elements directly engage with the heuristic processing mechanisms that drive SNS sharing behavior (Pennycook & Rand, 2019). In brief, the peripheral route may play a pivotal role in the dissemination of political rumors on SNS. Yet, the dissemination of rumors via the central route remains uncertain. Thus, we put forth the following hypothesis and the research question:

• H3. The peripheral route is significantly more linked to the intention to disseminate the character-framed political rumor compared to competence-framed political rumor on SNS.

• RQ2. Will the central route be more related to the intention to disseminate the characterframed political rumor compared to competence-framed political rumor on SNS?

Collectively, the synthesis of framing, TPE, and ELM allows us to postulate the following hypothesis and the research question concerning the indirect effects of rumor frame emphasizing a candidate’s character via TPE and dual routes of message processing:

Figure 1.  Conceptual Model

• H4. The character-framed political rumor compared to competence-framed political rumor will be more linked to the intention to disseminate it on SNS indirectly via TPE and a response indicative of more superficial processing linked to the peripheral route.

• RQ3. Will character-framed political rumor compared to competence-framed political rumor be more linked to the intention to disseminate it on SNS indirectly via TPE and a response indicative of more effortful processing linked to the central route?

We conducted an online experiment² in South Korea. The total sample size was 800, collected by an established research firm, Macromill Embrain. The demographic characteristics of 800 respondents were as follows. 50% (n = 400) were males. Their ages were stratified to match 20% for the respective age group: 19–29 20% (n=160), 30–39 20% (n = 160), 40–49 20% (n = 160), 50–59 20% (n = 160), more than 60 (n = 160). More than half of the respondents had a bachelor’s degree (56.8%; n = 454), followed by some college 17.4% (n = 139), high school 16.1% (n = 129), beyond bachelor’s degree 8.1% (n = 65), and less than high school 1.6% (n = 13).

This study’s online experiment was based on a 2 × 2 (rumor frame type x ideological stance of the rumor target) factorial design experiment, allowing participants to be randomly assigned to one of the four experimental conditions (see Figure 2 for the stimuli used): the competenceframed rumor about a conservative candidate (n = 200); the competence-framed rumor about a liberal candidate (n = 200); the character-framed rumor about a conservative candidate (n = 200); and the character-framed rumor framed negatively about a liberal candidate (n = 200).

Figure 2.  Stimuli

Two distinct frames of rumors were manipulated, i.e., competence and character framing, with a format of a newspaper article. Each article featured a fictitious candidate’s rumor about a housing redevelopment project in a fictional city, Annam, South Korea, but details varied depending on the type of frame. The article with competence framing stated that “Candidate Park misled finances of the Annam city with a wrong redevelopment policy at the time he served as the mayor of the city,” using the tone of “failure of redevelopment project,” “pushing the project inappropriately,” and “he left trillions of dollars in debt due to this policy failure.” On the other hand, character framing incorporates an article stating that candidate Park falsely claimed his family’s residence for real estate speculation during the reconstruction plan when he served as the city’s mayor, using the sentence such as “candidate Park attempted to fraudulently register the address of his spouse and daughter in the redevelopment area for property speculation.”³

Each of these manipulated articles were produced in two versions to take into account the target candidate’s political ideological position, thereby constituting a total of four articles. To manipulate the target candidate’s political ideological position, the name of the candidate in the article’s title was prefixed with the words “Conservative Icon” or “Liberal Icon.” Also, the first sentence of each article included phrases such as “the conservative party’s next presidential candidate Park” or “the liberal party’s next presidential candidate Park.” Finally, all articles added photo of a fictitious apartment building related to the redevelopment project within the range of 630 to 635 characters (excluding space) to control for extraneous variables, except rumor framing and candidate’s ideology. Participants read one of the four articles that matched the experimental condition to which they were allotted to read for about 30 seconds and answered the corresponding questions. It took about 15 to 20 minutes until the final response.

After participants read the assigned article, they answered the questions asking about weaknesses of the candidate covered in the article. A total of six items were measured on a seven-point scale, ranging from strongly disagree to strongly agree. Among those, three items were related to the lack of the candidate’s competence such as “poor job performance” (M = 5.22; SD = 1.24; α = .83). The remaining three items considered the lack of a candidate’s character traits such as “the lack of morality” (M = 5.53; SD = 1.19; α = .86). The result of an independent T-test indicates that the mean of perceived candidate’s weaknesses concerning his competency is higher in the competence framing group (M = 5.69; SD = .94) than in the character framing counterpart (M = 4.74; SD = 1.32); t(798) = -11.73; p = .001. Those who are assigned to the character framing group (M = 5.80; SD = 1.08) showed a higher level of perception that the article they read addressed the lack of a candidate’s character as compared to those who read the competence frame article (M = 5.26; SD = 1.24); t(798) = 6.64; p = .001. These results confirm that each article was perceived to deal with the weaknesses of the candidate according to the assigned type of rumor framing, thus confirming that the stimuli were successfully manipulated.

To probe the effect of rumor frame on its dissemination on SNS through TPE and ELM, this study conducted a path analysis utilizing lavaan package in the R software while using boot package to conduct bootstrapping with 5,000 samples in the same software. In a path model, the dummy variable for character framing (competence framing as a reference group) is our independent variable while TPE, argument quality and emotional responses are mediators. Our outcome variable is rumor dissemination. Two dummy variables for conservative and liberal voters were entered as covariates in this model.