Numerous studies connect media use and interpersonal discussion networks to political participation (e.g., Gil de Zúñiga et al., 2014; H. Lee et al., 2015; S. Lee et al., 2019; Valenzuela et al., 2018), but few analyze communicative paths to engaged citizenship (exceptions exist; e.g., Eveland et al., 2015; Hopmann, 2012) distinctively focusing on political network diversity and its interaction with social media news access. In particular, cross-national research performed outside the boundary of Western democracies is still lacking despite the meaningful differences in frequency of political discussion and discussion network dis/agreement (Eveland et al., 2015; Nir, 2012).

The current study examines relationships among personal network composition, social media news access, political discussion, and political participation in the United States and South Korea. We specifically theorize that the understudied interplay between political network characteristics (i.e., diversity) and social media usage, and also between political network characteristics (i.e., strong ties political homogeneity) and cross-cutting discussion, will varyingly influence political participation in the two nations. This theorization is based on the following interrelated assumptions: Culture shapes communication and relationship formation (Emirbayer, 1997; Hofstede, 2001; Mische, 2011), political systems shape public discourses (Nir, 2012), and media environments shape information access (Prior, 2007).

Personal Network Composition, Communication, and Political Behaviors

The composition of personal networks influences the type of information access, the frequency of political discussion, and even voting propensities in elections. Larger, more diverse networks tend to give access to a wider range of information (Granovetter, 1983). Greater access to diverse information and frequent political discussion facilitates participatory outcomes such as voting or protesting (Huckfeldt & Sprague, 1987). Politically homogenous close-tie networks provide a psychological buffer to ease discomfort associated with political disagreement (H. Lee et al., 2015; S. Lee et al., 2019). Thus, political networks’ composition substantially shapes political participation.

Personal networks can be understood as cultural processes manifested through communication (Mische, 2011). People from collectivistic cultures value group identity and build lifelong relationships, with their personal networks consisting mostly of strong ties such as ties with family, close friends, and members of their community (Jackson & Wang, 2013). In contrast, people from individualistic cultures develop more fragmented and short-term relationships (i.e., weak ties), prioritizing personal independence (Hall, 1981; Y. Kim et al., 2011). Americans tend to be more individualistic and engage in low-context communication (Jackson & Wang, 2013) compared to Koreans who are more collectivistic and rely more on high-context communication (Y. Kim et al., 2011). As a result of distinct social networking patterns shaped by culture, Americans may have larger, more diverse personal networks (Y. Kim et al., 2011), and Koreans may have smaller, more homogenous strong-tie networks. Larger personal networks tend to be more diverse (Granovetter, 1983), and so do weak ties, such as those with acquaintances, who can have varying social orientations, including political orientations (Huckfeldt & Sprague, 1987; S. Lee et al., 2019).

The composition of personal networks influences uses of social media where users communicate with members of their personal networks (Ji et al., 2010). Previous research found Korean social network site (SNS) users had fewer friends overall but more close friends, kept their public profiles anonymous, engaged in self-disclosure less frequently, and communicated more nonverbally through graphics and icons compared to Americans (Ji et al., 2010). A comparison between Chinese and American SNS showed that Chinese SNS (i.e., Renren) was shown to have a more collectivistic culture due to the members on the platform sharing more frequently within ingroups and the American SNS (i.e., Facebook) was shown to reflect a more individualistic culture because members engaged in more self-expression and self-interested behaviors (Qiu et al., 2012). Tan (2008) found Eastern-based SNSs (e.g., Renren, Weibo) demonstrated tighter relationships, with more high-context communication and less self-disclosure while Western-based SNSs (e.g., Facebook, Twitter) had wider social networks, with more low-context communication and frequent self-disclosure. Based on the findings of cross-national variation in personal networks, Americans may have larger SNS networks and mobile phone contacts than Koreans. Following Haythornthwaite (2002), we termed those contacts mediated “latent-tie” networks. These latent ties are not considered strong or weak but are connected through personal media and can be activated and mobilized when necessary (S. Lee et al., 2019).

News Access via Social Media

People rely on news media to access political information. Increased digitization, struggling legacy media, and declining trust in news characterize the news media environments in both South Korea and the US (Newman et al., 2021), but with important nuances. Increasing demand for digital content threatens print newspaper circulation and advertising revenue. A handful of U.S. legacy newspapers have successfully monetized their digital content (Newman et al., 2021), but Korean legacy news organizations struggle to achieve stable income from digital content (J. Kim, 2020).

The proliferation of information and communication technologies has shifted news consumption patterns for both countries, but Koreans access more digital news than Americans do. In 2021, more than 80% of Koreans reported going online for news, compared to 66% of Americans (Newman et al., 2021). Interestingly, both countries had the same rate of social media usage for news access (42%). Koreans rely heavily on the online news aggregating and search engine website Naver, which offers digital news content for free. Two-thirds of Koreans report using Naver for news on a weekly basis. However, less than 20% of Americans reported going online for news on a weekly basis (Newman et al., 2021).

Media trust also shapes news consumption patterns (Tsfati & Cappella, 2003). U.S. and South Korean citizens share a deep distrust for news media (Newman et al., 2021). Low trust in legacy news media may increase reliance on trusted personal networks for news and political information. Social media provide access to political information shared among strong ties, facilitate exposure to a wide range of viewpoints via weak ties and latent mediated ties, and provide a platform for political discussion among network members (Chan et al., 2017; Gil de Zúñiga et al., 2014; Yamamoto & Morey, 2019). Koreans tend to prefer Korean social media, such as Kakao Talk or Line, over American social media (Newman et al., 2021) due to distinct cultural values and preferences (Choi et al., 2013; Park et al., 2018). Due to collectivism and indirect ways of communicating and forming relationships, the display interface of Facebook’s friend list may be too direct and blunt for Koreans (Choi et al., 2013).

Political Discussion and Participation

Interpersonal political discussion is a key political process as it increases exposure to diverse viewpoints (Moy & Gastil, 2006; Price et al., 2002) and facilitates cognitive elaboration, which can enhance political learning and participation (Eveland et al., 2003; H. Lee et al., 2015). This study focuses on a particular type of political discussion, namely cross-cutting discussion (CCD) that involves disagreement so as to clarify its relationship with political participation. Many scholars argue CCD will reduce participation because of its causing ambivalence of attitudes and negative emotions (Mutz, 2002; Torcal & Maldonado, 2014). However, other scholars present counter evidence that shows CCD does not deter participation (Huckfeldt et al., 2004; Matthes et al., 2019), and sometimes, with a particular network composition, CCD can even facilitate political participation (H. Lee et al., 2015; Nir, 2011).

Large online networks facilitate exposure to diverse political opinions only when individuals are embedded in politically diverse networks. Network scholars examine the processes of a society’s cultural and political division (e.g., Keijzer et al., 2018; Yarchi et al., 2021). Political communication online illustrates such a polarized structure based on homophily (networking with similar others; McPherson et al., 2001), and social media’s and search engines’ embedded algorithms exacerbate the polarization by exposing people to like-minded others and agreeable content, creating echo chambers (Pariser, 2011). Keijzer et al. (2018) demonstrated how the one-to-many communication form of social media fosters cultural isolation over time because people become more similar with ingroup members and strengthen ties with them, rather than reaching out to strangers with completely different cultural traits. Thus, it is necessary to consider the composition of personal networks when examining communicative paths toward political participation.

Social media enable political communication with personal network members, but some platforms facilitate interactions more effectively with either strong ties or weak ties. Valenzuela et al. (2018) found both Twitter and Facebook facilitated political participation, but interactions with strong ties on Facebook and interactions with weak ties on Twitter were highlighted. Yarchi et al. (2021) also found notable cross-platform differences in political discourse and their structures (e.g., polarization) from a computational analysis of data collected over 16 months. While Twitter showed homophilic interaction structures with aggravating polarization, WhatsApp showed de-polarization over time; Facebook showed the least amount of homophilic interactions and polarized structures.

Therefore, examining the interaction between different social media news access and personal network composition may reveal distinct cross-national mechanisms that facilitate political participation. While an echo chamber of political discussion on social media may work as a facilitator of political participation, such as mobilizing people of the same orientation, those whose political networks consist of people with diverse orientations will have more opportunities to be exposed to news that may contradict their existing views and engage in CCD, leading to more political knowledge and participation (S. Lee et al., 2019; McLeod et al., 1999; Price et al., 2002). Thus, we ask:

RQ1: How does political network diversity interact with social media news access to increase political participation of (a) U.S. and (b) South Korean citizens?

While strong and weak ties have differential characteristics and most people have a mix of both relationships and mediated latent ties, strong ties’ political homogeneity works as a psychological buffer for engaging in CCD (H. Lee et al., 2015; S. Lee et al., 2019). Being surrounded by strong ties sharing similar political views feels safe and confirms one’s existing view, thus making one less afraid of engaging in CCD. Whether this buffering effect holds cross-nationally will also be examined in this study:

RQ2: How does CCD interact with strong ties’ political homogeneity to increase political participation of (a) the US and (b) South Korean citizens?

While U.S. and Korean citizens may both engage in political discussion or use social media to facilitate political participation, we expect communicative paths to political participation, especially the extent and strength of relationships between relevant components, to vary cross-nationally (Eveland et al., 2015). Importantly, this research acknowledges the established theoretical model, O-S-R-O-R of campaign communication mediation (Cho et al., 2009), highlighting the mediating role of communication (e.g., interpersonal discussion) between campaign exposure and political knowledge or participation. It also follows the model’s logic in considering various preexisting individual orientations (the first “O”; e.g., demographics, social networks) and news access (“S,” i.e., stimulus) that may spur political discussions (the first “R,” i.e., reasoning) leading to participation (the second “R,” i.e., response). While we do not aim to test the mediation model, this study examines whether personal network composition (O), social media news access (S), and political discussions (R) influences political participation (R) differently for U.S. and Korean citizens. The study complements the model by testing two interaction effects: one between political network diversity and social media news access and another between CCD and close tie networks’ political homogeneity.

Method

Professional survey panel services (i.e., Qualtrics for the U.S. and Macromill Embrain for Korean samples) were utilized to collect data; and relatively large convenient samples regarding gender, age, and residential distributions were gathered from South Korea (N = 1166) and the US (N = 1001) during August 2015. Quota sampling was used based on gender, age, and residential area. Participants who completed the survey were paid at a rate determined by the survey companies and we paid $5 per response.

Sample

The U.S. sample was evenly split between male (49.8%) and female (50.2%) participants, whose ages ranged between 18 and 90 (M = 43.3, SD = 16.6) years. One-fifth of the U.S. participants reported their highest education level as high school graduate or below, 29.5% had some college education or were currently in college, 33.5% were college graduates, and 16.7% had some graduate education or a professional degree. About one-third of the U.S. participants reported earning $2000 or less per month, 33.1% reported $2000 to $4000, 18.7% reported $4000 to $6000, and the rest reported over $6000. All U.S. states and territories except for Alaska, South Dakota, Vermont, and Puerto Rico were represented as the current state of residence in the sample. While 18.3% of participants lived in the Northeast region, 22.5%, 36.4%, and 22.7% in the Midwest, South, and West, respectively.

The Korean sample was also evenly split between male (50.7%) and female (49.3%) participants, with their ages ranging from 20 to 64 (M = 42.2, SD = 12.1) years. Many Korean participants had a college degree (62.3%), 27.3% were high school graduates or currently in college, and 10% had completed some graduate school or graduated with degrees. Over 24% of the Korean sample reported earning $2000 or less per month, and approximately 40% reported between $2000 and $4000. A quarter of the sample reported a monthly income between $4000 and $6000 and just over 10% reported over $6000. The largest share of Korean participants resided in Seoul (19.3%) and Kyungki-do (22.8%), the capital of South Korea and its suburb; 27.6% resided in large cities in different provinces, such as Busan or Incheon; and 30.3% in other more rural areas such as Gangwon, Chungbuk, and Kyoungbuk. Overall, the Korean sample consisted of slightly more educated and higher-income participants than the U.S. sample, but gender and age distributions were fairly compatible between the two samples.

Measures

Personal Network Size

Three categories of relationship networks sizes were measured: strong, weak, and mediated latent ties. For strong ties, the numbers of immediate family members and close friends were included (interitem r = .47 for the U.S. sample; r = .38 for the Korean sample); for weak ties, the numbers of neighbors, coworkers, and voluntary organizational members were combined (Cronbach’s α = .88 for the U.S. sample; Cronbach’s α = .58 for the Korean sample); for mediated latent ties, the number of contacts in mobile phones and the most frequently used SNS were added and the numbers of strong and weak ties were subtracted from those sums (interitem r = .56 for the U.S. sample; r = .38 for the Korean sample; S. Lee et al., 2019). As is typical for network analysis, all network size variables were heavily skewed in their distributions, so square root and natural log transformations were performed for their normalization.

Political Network Attributes

For political network diversity, how often participants socialized with people of various political orientations was measured with a 7-point scale (1 = not at all to 7 = very frequently). To measure strong ties’ political homogeneity, two questions were asked: “How likely is it that the people in your network of strong personal ties share your political views and how likely is it that they supported the same presidential candidate as you?” (H. Lee et al., 2015, p. 578). The responses ranged from 1 = very unlikely to 7 = very likely and were summed to form an additive index (interitem r = .86 for the U.S. sample; r = .80 for the Korean sample). Table 1 shows descriptive statistics of each network size and attribute variable.

Cross-Cutting Discussion

The frequency of participants’ engagement in the following two types of discussion was asked: talking to others who do not support the candidate they favor and having a conversation about politics or social issues that involves disagreement (H. Lee et al., 2015). The frequency was reported on a 7-point scale (1 = never to 7 = daily). An additive index score was calculated (interitem r = .76 for the U.S.; r = .83 for Korean sample).

Political Participation

Four types of political involvement were measured (H. Lee et al., 2015): attending a political meeting, rally, or speech; working for a candidate or a party; contacting a public official or a political party; and contributing money to a candidate or a political party. The frequency in each type was reported on a 7-point scale (1= never to 7 = daily). The responses were averaged to form a composite index (M = 1.90, SD = 1.55, Range = 1–7, Cronbach’s α = .93 for the U.S. sample; M = 1.30, SD = 0.77, Range = 1–6.5, Cronbach’s α = .91 for the Korean sample). Log transformation was used to meet the normality assumption.

News Access and Controls

The amount of news accessed through various media outlets was measured in days per week from 1 = 0 day to 8 = 7 days. Data regarding two traditional news media (i.e., television and newspaper) and three social media (i.e., Facebook, Twitter, YouTube) outlets were considered. Political news interest was measured as a proxy for political interest with a 7-point scale (1 = very little to 7 = very much). Media literacy was measured in two dimensions: knowledge and self-efficacy. Items such as “News consumers can find news that reflects their political opinions” and “News is designed to gain attention from users” were used to measure knowledge of the news industry and its mechanism on a 7-point scale (1 = strongly disagree to 7 = strongly agree). An average score was created from 10 items (M = 5.46, SD = 1.04, Cronbach’s α = .93 for the U.S.; M = 5.30, SD = 0.83, Cronbach’s α = .92 for the Korean sample). Items such as “I can understand the diverse perspectives related to important news” and “I can express my opinions well related to important news through the Internet and social network sites” were used to measure the self-efficacy of media literacy with the 7-point scale, and an average score was created from six items (M = 5.25, SD = 1.12, Cronbach’s α = .90 for the U.S. sample; M = 4.75, SD = 0.88, Cronbach’s α = .86 for the Korean sample).

Interaction Terms

Three interaction terms were created: one between CCD and news access via Facebook, another between CCD and news access via Twitter, and the third between CCD and strong ties’ political homogeneity. To reduce the likelihood of collinearity between interaction terms and their components, all the component variables were mean centered (Tabachnick & Fidell, 2013).

Results

Communicative Paths Toward Political Participation

To answer RQ1, examining the interactions between social media news access and political network diversity and their cross-national variation, two multiple hierarchical regressions were performed with political participation as a criterion variable and personal network composition, social media news access, and political discussion as predictors. First, age, gender, income, education, and political news interest significantly predicted political participation in the U.S. sample (see Model I in Table 2). Specifically, younger, higher-income, and highly educated U.S. males with higher interest in political news were more participatory. Second, frequency of newspaper and YouTube usage for news was positively associated while frequency of TV usage was negatively related with political participation (see Model II). Among social network variables, only weak ties’ size was positively related with political participation as Americans with larger weak-tie networks tended to participate more.

The interaction between social media news access and political network diversity (RQ1a) was significant for Twitter usage in the U.S. sample after controlling for the effects of demographics, news access, and weak-tie size on political participation (see Model III and IV in Table 2). U.S. participants with more diverse political networks who used Twitter for news access more frequently tended to participate in political activities more often.

For Korean participants, gender, education, income, political news interest, and media literacy (both knowledge and efficacy) significantly predicted their political participation (see Model I in Table 3). Consistent with the U.S. sample, higher-income and highly educated males with more interest in political news tended to participate in political activities more often. While media literacy did not have a significant impact in the U.S. sample, both knowledge and efficacy of media literacy had significant relationships with Koreans’ political participation; notably, those who reported lower levels of knowledge and higher levels of self-efficacy participated in political activities more. Similar to the U.S. sample, newspaper and YouTube usage for news access was positively associated with the political participation of Korean participants (see Model II in Table 3). Larger weak-tie networks were also associated with Koreans’ political participation, but adding the social network variables as a block did not significantly increase the amount of explained variance in political participation (see Model III).

After controlling for these effects of demographics, news access, and social network composition, the interaction between political network diversity and social media news access was significant for Twitter usage and not for Facebook usage for news access in the Korean sample (see Model IV in Table 3). This result was also similar with the case of the U.S. sample.

In relation to RQ2, examining the interaction between CCD and strong ties’ political homogeneity (STPH), it was positively significant in both samples (see Models V and VI in Table 2 and 3). CCD was positively associated with political participation in both groups, but much more strongly in the Korean sample (increased 60% of variances vs. 10%); and when the interaction with STPH was considered in the last step of the regressions, CCD became insignificant in the U.S. sample, but remained significant in the Korean sample. The results seem to indicate that when both Americans and Koreans engage in cross-cutting political discussion, being surrounded by politically homogenous close ties tend to facilitate greater participation.

Discussion

This study examined communicative paths shaping citizens’ engagement in politics—personal networks, social media news access, and political discussion—in two national samples with distinct cultures, political contexts, and media environments. In both samples, a gap in political participation by gender, education, and income level seemed to persist as participants who were females with lower education and income levels participated in political activities less frequently. As this would be more of a social structural issue than an individual matter, more effort by the government and legislature can be put in place by both countries to facilitate political participation from these groups.

Media literacy, both in terms of knowledge and efficacy, was significantly related with political participation only in the Korean sample. It was notable that the two dimensions of media literacy had an opposite effect on political participation; Korean participants who reported being more knowledgeable about the news media industry and its mechanism participated in political activities less frequently, whereas those who reported higher levels of self-efficacy related to media literacy (e.g., being confident about using various media sources to confirm facts and differentiate important from nonsignificant issues) participated more often. It might be the case that Koreans who think they are knowledgeable about how news media work in general are also skeptical of the media and politics given the measurement items comprised questions regarding agreement with news media owners’ influence on the news content and participants encountering news that reflects their own political opinions (Pariser, 2011). Still, more research needs to verify this counterintuitive result of media literacy with a representative sample of the South Korean population.

The association with personal network characteristics was similar in the two samples. The large weak-tie networks were positively related to political participation, which resonated with previous findings (e.g., S. Lee et al., 2019); but the size of strong ties or mediated latent ties was not related to political participation in either sample. Large weak-tie networks bring more diversity into political networks since one will be likely to encounter more people of varying political orientations and perspectives (Huckfeldt & Sprague, 1987). Such diversity can enhance political knowledge and deliberation through interactions, leading to more participation (Moy & Gastil, 2006; Nir, 2011).

Notably, the interaction between social media news access and political network diversity was identified in the case of Twitter in both samples, but not for Facebook. For those who have politically diverse networks, consuming news through Twitter had a positive association with political participation. News access through Twitter might bring more diverse political perspectives as interactions on Twitter involve more weak ties (Valenzuela et al., 2018), facilitating more CCD. The reason why Facebook usage for news access did not interact with political network diversity to influence political participation might be found in Yarchi et al. (2021); when comparing the political communication across Twitter, WhatsApp, and Facebook, they found Facebook had the least homophilic and polarized structure. Combining this with Valenzuela et al.'s (2018) finding that Facebook was a place where people communicated more with their strong ties, frequent news access via Facebook does not seem to increase the benefit of exposure to politically diverse others. Although Facebook is already composed of people with various political orientations globally (Yarchi et al., 2021), its members may not engage in political discussion with them in their local, close networks.

Finally, despite the cultural and historical differences between the two countries, the interaction between CCD and strong ties’ political homogeneity worked positively for political participation in both samples, confirming previous findings identified with the U.S. samples (e.g., H. Lee et al., 2015; S. Lee et al., 2019). Although Americans engaged in CCD substantially more than Koreans, possibly due to their more diverse culture valuing individual freedom of expression and longer history of democracy, the positive association between CCD and political participation was much stronger (more than twice the effect size) in the Korean sample.

This finding of distinct effect size resonates with Eveland, Song, and Beck’s findings (2015) from cross-national comparisons that the association between political agreement and voting (and political discussion frequency) was stronger in collectivistic countries. Although the two findings may seem contradictory on the surface, Eveland, Song, and Beck measured political agreement only with strong ties (e.g., spouse and other discussion partners for important matters). Thus, the researchers found that CCD could both facilitate and hamper political participation depending on the discussion partners (e.g., strong vs. weak ties). Political disagreement with strong ties, such as family and friends, may hurt one’s feelings and discourage participation (Torcal & Maldonado, 2014), resulting in the stronger association between political agreement and voter turnout in collectivistic cultures (Eveland et al., 2015). Disagreement with weak ties may not have such a negative impact but increase one’s political knowledge and tolerance (Moy & Gastil, 2006; Nir, 2011; Price et al., 2002); this may have resulted in a stronger association between CCD and political participation in the Korean sample. However, this interpretation needs to be considered with the caveat that our samples for this study are not representative of each country; participants are members of voluntary panels of survey companies.

Nevertheless, when citizens are surrounded by close ties having the same political orientation, they are less afraid of political disagreement with weak ties and benefit from increased knowledge and deliberation, which engages them further in politics (H. Lee et al., 2015). The fear of being socially isolated and the pressure of conformity would be larger in a collectivistic culture, explaining Koreans’ less active engagement in CCD overall; however, when Koreans do engage in discussions involving disagreement while overcoming the social and cultural barriers and being supported by their close ties, the impact may be much more powerful in facilitating citizens’ political participation. It would be a challenging task for many South Koreans as the political polarization between the conservative and the liberal is still intense, reflected in the 2022 presidential election result. The candidate representing the conservative party won the election by a 0.7% point difference. This increasing political party-led ideological polarization in South Korea, along with other contributing factors such as social media, regional rift, and generational gap, was also pointed out by S. J. Lee (2022). An online space where people can discuss political issues safely and civilly with weak ties needs to be established replacing the existing structure of the echo chamber that reinforces political polarization (Keijzer et al., 2018; Yarchi et al., 2021)

Limitations and Directions for Future Research

Since the data for this study were collected based on online panels which were not representative samples of either country, we cannot make any generalizations of our findings. We also acknowledge potential errors in measurement of the weak tie size especially for the Korean sample. The score of internal scale reliability was less than ideal; thus, a future study should examine this issue further and find the reason.

Some potentially influential variables were not included as controls in the regressions. Our survey asked about participants’ political news interest and media efficacy, but not their political knowledge, party identification, or political orientation. Thus, we could not control for the effects of such relevant components on political participation. Future research should also include those controls for a more comprehensive analysis.

Conclusion

An examination of communicative paths toward political participation was conducted with a relatively large samples of Americans and South Koreans. Hierarchical regressions involving demographic variables, and those of personal network, news access, and political discussion and participation demonstrated mostly commonalities between the two samples except distinct effect sizes. We attribute the differences to the distinct cultures and histories of democracy of the two countries, but more research is needed to deepen the understanding of communicative paths toward political participation from international perspectives. While following the theoretical logic of O-S-R-O-R model of campaign communication mediation (Cho et al., 2009), the current study contributed to expanding the model by considering interactions between political network characteristics and political communication.