Managing Diverse Online Networks in the Context of Polarization: Understanding How We Grow Apart on and through Social Media

Abstract

Social media enable their users to be connected with a diverse group of people increasing their chances of coming across divergent viewpoints. Thus, network diversity is a key issue for understanding the potentials of social media for creating a cross-cutting communication space that is one of the premises of a functioning democracy. This article analyzes the strategies social media users adopt to manage their network diversity in the context of increasing polarization. The study is based on 29 semi-structured interviews with diverse social media users from Turkey and qualitative network maps. Furthermore, the study adopts a cross-platform approach comparing Facebook, Twitter, and WhatsApp in relation to the diversity of their users’ networks. The study shows that social media users adopt different strategies interchangeably in specific contexts. These strategies include visible (unfriending, blocking) and invisible (muting, unfollowing, and ignoring) forms of disconnection, debating, observing divergent opinions, and self-censorship. Political interest of social media users, political climate, issue sensitivity, and “imagined affordances” of social media platforms play a role in users’ choices about which strategy to choose when they are confronted with divergent viewpoints through their diverse online networks. Building on the unfriending literature that points out to rather partisan users, who unfriend, unfollow or block others, this article demonstrates that in peak moments of polarization, also the politically disengaged or moderate users disconnect from diverse others.

Introduction

Social media users are connected with a diverse set of weak and strong ties through social media. This diversity of online social networks also increases users’ chances of being exposed to cross-cutting opinions. Cross-cutting exposure is defined as one of the basic premises of democracy as it can help people to understand competing perspectives better, reflect on their own positions, and expand their horizons (Habermas, 1989; Mutz & Martin, 2001; Papacharissi, 2002). However, people also adopt online and offline strategies to avoid cross-cutting exposure, such as unfriending on social media.

Several studies have focused on unfriending and the creation of echo chambers among like-minded social media users in recent years (Dubois & Blank, 2018; Himelboim et al., 2013; Sunstein, 2002). Nevertheless, “engagement with difference” on social media does not always lead to “post hoc filtration and dissolution of social ties” (Skoric et al., 2018). Users also adopt strategies to actively avoid echo chambers. Thus, it is important to develop a contextualized understanding of how and why social media users adopt certain strategies over others when they face divergent opinions. Furthermore, polarization in different countries—the increasing political, emotional, and social distance among competing political views (Iyengar et al., 2012; McCoy et al., 2018)—also influences the way social media users deal with network diversity. Looking at the highly polarized context of Turkey, this article analyzes users’ strategies to manage the diversity of their networks in the context of polarization.

Studies on social media, polarization, and network diversity often focus on singular platforms—mostly Twitter (e.g., Colleoni et al., 2014; Conover et al., 2011; Himelboim et al., 2013; Kearney, 2019). However, users might be part of a diverse set of networks. Some are involved in like-minded and polarized networks on one platform, while also being exposed to cross-cutting opinions on another (Y. Kim & Chen, 2015). Comparing different platforms that users engage with is key for evaluating the effects of echo chambers and online network diversity (Dubois & Blank, 2018). Moreover, existing research often focuses on singular strategies, such as unfriending/blocking (John & Dvir-Gvirsman, 2015; John & Gal, 2018) or self-censorship (Beam et al., 2018; Thorson, 2014). This article analyzes these different strategies in relation to each other aiming to understand the motivations of users as they develop their strategies for managing network diversity.

The study combines qualitative interviews with ego-centered network maps. The sample consists of 29 people from Istanbul with very diverse backgrounds (see Supplemental Appendix). The data were analyzed through inductive and deductive codes that contributed to mapping of different strategies that users adopt to manage the diversity of their online networks and the factors that influenced which strategy they chose at particular moments.

The findings show that the prevailing political climate at any given time has a strong influence on people’s decisions on how to react to divergent opinions on social media. Existing research points out that politically engaged and partisan users are the users who most often unfriend others (Kearney, 2019; J. K. Lee et al., 2014). Building on these findings, this article also demonstrates how politically disengaged and moderate users disconnect from others in peak moments of polarization. Furthermore, the article makes a distinction between visible strategies (unfriending and blocking) and invisible strategies of disconnection (muting, unfollowing, and ignoring). Whereas visible strategies indicate to the unfriended person that they do not belong to the networks of the unfriending person any more, the invisible strategies show an implicit intention to somehow stay connected in the future. The article discusses debating with others, observing diverse opinions, and self-censorship as further strategies. The section about the strategies for managing diverse networks is followed by a discussion about the different factors in the management of network diversity. These include political interest and engagement of the users, political climate, issue sensitivity, and “imagined affordances” (Nagy & Neff, 2015) of the platforms.

Polarization and Social Media

Polarization in the broadest sense can be defined as the distance between competing political orientations (Kearney, 2019). Although some level of polarization can be found in all pluralist democracies and can even have a mobilizing effect by increasing political participation, severe polarization can also lead to distancing between social groups and can be very damaging to democracies (McCoy et al., 2018). Severe polarization can be defined as “a process whereby the normal multiplicity of differences in a society increasingly align along a single dimension and people increasingly perceive and describe politics and society in terms of ‘Us’ versus ‘Them’” (McCoy et al., 2018, p. 16). The “Other” is increasingly perceived as a threat or enemy rather than a political adversary and strong emotions of antipathy and distrust toward opposing social groups are developed (Iyengar et al., 2012; McCoy et al., 2018). The political distance between the two sides is increasingly based on belonging and social identities rather than ideological differences (Iyengar et al., 2012; McCoy et al., 2018). Thus, severe polarization has a strong affective dimension (McCoy et al., 2018, p. 18; Iyengar et al., 2012). Polarization in its most severe form becomes a serious threat to democracy by undermining social cohesion and political stability (McCoy et al., 2018).

Severe polarization effects not only Turkey, but also many countries around the world, including the United States, Venezuela, Hungary, and India, among others (McCoy et al., 2018). The divides in Turkey are not new, rather, they have historical roots (Çelik et al., 2017). Social and political tensions exist along three main fault lines: ethnic (Kurds and Turks), sectarian (Alevites and Sunnis), and ideological (AKP supporters and AKP opponents; Çelik et al., 2017). However, AKP, the party in power since 2002, deepened these divides by utilizing polarizing discourses and continuously pointing out the “foes” of AKP/New Turkey in a simplified frame. Thus, the existing fault lines have been crystallized over the past few years (Erdogan & Uyan-Semerci, 2018; McCoy et al., 2018). Two recent studies on polarization in Turkey funded by the German Marshall Fund show that polarization in Turkey increasingly creates distancing between the supporters of different political parties (Erdogan & Uyan-Semerci, 2018; Erdogan, 2016). People are very reluctant to discuss contentious political matters in public spaces (e.g., neighborhood meetings) or even at family dinners (Erdogan & Uyan-Semerci, 2018, p. 4). Only 25% of people surveyed say that they would discuss critical issues on social media (Erdogan & Uyan-Semerci, 2018).

The relationship between social media and polarization has become a contested issue in recent studies. Several studies show that social media leads to increasingly selective exposure to like-minded views and the formation of echo chambers (Himelboim et al., 2013; Kearney, 2019; Sunstein, 2002). At the same time, there are other studies balancing these findings that argue social media reduces mass political polarization (Barberá, 2014) or that echo chambers are overstated and actually only apply to a small segment of the population (Dubois & Blank, 2018). These mixed findings imply that although social media might have the affordance to create or strengthen polarization, their “polarizing potential” is not always (fully) realized (F. L. Lee, 2016, p. 58).

As F. K. Lee (2016) argues, we need to understand the immediate political contexts to understand the polarizing effects of digital media. Polarization itself is not a pre-given and static situation, but is in itself an ongoing process in a particular context. As political polarization turns into social distance in severely polarized countries, it is important to understand if and how social media contribute to people’s growing apart.

Social Media and Network Diversity

Cross-cutting exposure is closely linked with the diversity of users’ social networks in online and offline contexts. Higher network diversity can increase one’s chances for being exposed to views of people of different backgrounds (Hampton et al., 2011). Network diversity or heterogeneity points to “a mix of divergent viewpoints within a given network” (Y. Kim & Chen, 2015, p. 2347). Although different definitions of diversity can be found in the literature, it is defined here as a divergence between the preference of a person and the others in a given network. This definition would not only include “oppositional viewpoints but also those with different demographic and socioeconomic backgrounds who could provide diverse viewpoints” (Y. Kim & Chen, 2015, p. 2347).

Network diversity is not necessarily a new phenomenon; neighborhoods, religious institutions, and public spaces in an urban context have always been marked by a certain level of diversity (Hampton et al., 2011). However, with the emergence of social media, we have more diverse mechanisms that facilitate networking with diverse others (Brundidge, 2010; Choi & Lee, 2015). According to Choi and Lee (2015), social media first makes the maintenance of larger communication networks possible at reduced costs. Second, weak ties are often also included in people’s social media networks alongside stronger ties. Third, people are exposed to diverse views in social media in serendipitous ways.

The research on network diversity, cross-cutting exposure, and social media has so far provided inconsistent results (J. K. Lee et al., 2014). On one hand, there is a line of research arguing that social media users increasingly interact with like-minded people and avoid dissimilar viewpoints leading to network homophily, fragmentation, and polarization in the society (Himelboim et al., 2013; Sunstein, 2002; Van Alstyne & Brynjolfsson, 2005). On the other hand, there are studies showing that users have more diverse networks on social media than they do in offline contexts and therefore are confronted with a more diverse range of differing views online (Papacharissi, 2002; Hampton et al., 2011; Y. Kim & Chen, 2015; J. K. Lee et al., 2014). Like in relation to polarization, developing a contextualized approach is key for understanding under which conditions social media increases or decreases the diversity of online networks.

A closer look at the studies on social media, network diversity, and homophily shows that there is an overrepresentation of studies on Twitter (e.g., Colleoni et al., 2014; Conover et al., 2011; Himelboim et al., 2013; Kearney, 2019). However, different platforms might be used differently by social media users and have diverging effects in relation to polarization (C. Kim & Lee, 2016). For understanding if people are connected with diverse others and experience cross-cutting exposure, we also need to look at the broader picture through a cross-platform approach (Dubois & Blank, 2018). Furthermore, there is a lack of qualitative studies on network diversity. Qualitative studies can give us a more detailed insight into users’ perspectives and behaviors and help us understand why and how people engage with or avoid interactions with diverse others on social media. They can also help us move beyond what is visible on social networks (such as likes, retweets, followers–followees) and understand what users avoid doing on social media (e.g., not reading, not connecting with others). Furthermore, qualitative studies can shed light on the changes in network diversity as social media users add diverse others to their networks, delete, mute, and ignore others or reconnect with people with whom they disconnected at some point.

Thus, there is a need for more in-depth studies on network diversity that focus on how network diversity is being (re-)constructed and the dynamics behind it. The literature on audience and privacy management on social media offers a guiding framework for such a perspective focusing on everyday strategies of social media users.

Audience Management on Social Media

People’s everyday lives, in which social media use takes place, are marked by complex decisions including both the avoidance of cross-cutting exposure and the intentional search for diverse perspectives at different times. Social media users continuously “juggle multiple layers and kinds of audiences” (Baym & boyd, 2012, pp. 321–322). This nuance and complexity of everyday life calls for “strategic management” by the users (Baym & boyd, 2012, p. 320).

Bringing a diverse group of social ties together, social media are argued to flatten multiple audiences into one, leading to a “context collapse,” a term coined by danah boyd (2010). While dealing with their networked audiences, users navigate different social relationships (Marwick & boyd, 2011) and need to manage the dynamics of invisible audiences (boyd, 2010; Davis & Jurgenson, 2014). To be able to manage their invisible audiences, users monitor their audiences, take their feedback, and imagine them like the imagined audiences of a writer. Cognizant of these imaginaries, users adopt different strategies to manage their online social networks.

The strategies of social media users are facilitated by both social processes and technical possibilities that the platforms offer (Hayes et al., 2015). Self-censorship—especially in relation to sensitive content, for example, politics, news, religion—is one of the strategies that many users adopt to avoid conflicts on social media (Beam et al., 2018; Duggan & Smith, 2016; Thorson, 2014). Some users change their privacy settings using the technical possibilities of the social media platforms to maintain privacy on their public profiles, but interact with sub-networks (Chen, 2018). Unfriending/unfollowing is another strategy that users adopt in an attempt to exercise sovereignty over the personal public sphere that social media offers (John & Dvir-Gvirsman, 2015; John & Gal, 2018; Skoric et al., 2018).

Similar to Hayes et al. (2015), this study looks at “online management behaviors” including both users’ disclosure of and exposure to information because network diversity is relevant for both. Unfriending, for example, is often not only about not seeing a person’s posts, but also hindering him or her from seeing your own posts. Defining network diversity as a mix of divergent viewpoints from the perspective of the ego (Y. Kim & Chen, 2015), managing diverse online social networks includes the different strategies users adopt when they are confronted with these differences on social media.

Research Methods

For analyzing the relationship between social media, network diversity, and polarization, this research focuses on Turkey, a highly polarized country with a quite large group of active social media users (Yanatma, 2018). 75% of Turkey’s population comprises internet users (TUIK, 2019). Among these, 87% uses Facebook and one third uses Twitter actively (Erdogan & Uyan-Semerci, 2018, p. 5). Turkey’s mass media landscape is also highly polarized and there are strong parallels between media, political parties, and media use (Erdogan & Uyan-Semerci, 2018; Panayırcı et al., 2016). This makes Turkey a very interesting case for studying the following research questions:

• What strategies do users adopt to manage the diversity of their social networks on social media when they are confronted with divergent viewpoints?

• What are the factors that influence how users manage the diversity of their social networks on social media?

• How does polarization influence the ways in which users manage the diversity of their online social networks?

For answering these questions, the presented research adopts a qualitative approach combining qualitative semi-structured interviews and ego-centered network maps. The data collection took place between March 2016 and August 2016. On 15 July 2016, a coup attempt took place in Turkey, which dominated the political agenda of Turkey and had a major influence on the content of the 15 interviews that were carried out in its aftermath. At the same time, it presented a very interesting case for discussion as an example of a peak moment of polarization that will be discussed later on.

The data collection took place in Istanbul, a large metropole representing a microcosm of the diversity of the population in Turkey. With its qualitative design, the study did not aim for representation but rather for purposive sampling and reaching a diverse sample of social media users (Patton, 2002) in relation to how they perceive and manage divergent views on social media. 29 people—15 women and 14 men—with different educational levels and within the age range of 20–57, were interviewed. Among these were Alevites, Sunnis, and non-believers; Kurds and Turks; as well as supporters of the biggest five parties (AKP, CHP, MHP, HDP, and SP) in the Turkish parliament at the time of the data collection. The participants’ professions also varied (see Supplemental Appendix for an overview).

Semi-structured interviews were preferred for developing an in-depth understanding of interviewees’ perspectives on network diversity. The interview guidelines included questions about daily social media use, network diversity (online and offline), political interest and orientation, participation in online political debates, boundary-making strategies on social media, and experiences with political conflict online. Adopting a cross-platform approach as discussed earlier, the interviewees were openly asked about their daily social media use. According to their own emphasis, more specific questions were asked about each platform including Facebook, Twitter, Instagram, WhatsApp, and possible others.

In addition to interviews, ego-centered network maps were used for data collection (Berg & Hepp, 2018). The interviewees were asked to put themselves in the middle of two pieces of blank paper and freely draw their friendship networks on Facebook and the follower/followee networks on Twitter. The network maps did not only demonstrate the different layers in the interviewees’ networks, but also how the interviewees categorize and imagine these networks. The interviewees were asked to explain their network maps after drawing them and in this regard, the maps also served as a very good prompt for asking further questions about the diversity of each group in their maps (see, for example, Figure 1).

Figure 1.

Facebook network map—Nevin.

The collected data were analyzed through inductive and deductive categories. These included the strategies users adopt to manage their diverse networks, such as unfriending, blocking, changing privacy settings, adding behavior, debating, ignoring, observing, and self-censorship. These strategies that are also discussed in the literature, and as discussed in earlier sections, served as deductive categories for the analysis. The analysis also included inductive categories that pointed to factors that influence these strategies like political positions, perceptions of political climate, issue sensitivity, and perception of specific social media platforms.

This study is limited by its small sample of people living in the urban context of Istanbul and it is hard to generalize the findings of this contextualized study. Nevertheless, the strength of such an in-depth study lies in the insights that it provides in relation to the complexity of people’s orientations, behaviors, and motivations.

Managing Diverse Networks in the Context of Polarization

The composition of people’s online networks varies depending on family relations, education level, profession, neighborhood, and hobbies, among other factors. While drawing and discussing their network maps, several people mentioned close friends as a group of like-minded people and family members as a partly diverse group. Weak ties are quite diverse groups including “people, whom [interviewees] met only once,” “not very close friends,” elementary, high school or university friends, military pals, or professional ties (pet owners for a vet; other taxi drivers for a taxi driver). Hobbies and interests also added new groups to some interviewees’ networks as can be seen in the case of Nevin’s Facebook network map below (see Figure 1).

The context of the social media use is socially constructed through the interplay between the users’ imaginations, practices, and the platforms’ technical affordances (boyd, 2014). Nagy and Neff (2015) refer to users’ imaginations about the action possibilities of certain technologies as “imagined affordances.” The interviewees attributed different characteristics to different platforms and this influenced their reactions to network diversity on these platforms. For example, those who use Twitter saw it as a “much more political space” (Cagla, F, 31, journalist). Nilgün (F, 22, university student) argued that “Facebook felt more useless” after getting to know Twitter, where users “have a higher educational level.” They are “people like [her], who have an ideology, opinions.” Facebook, on the other hand, was often associated as a “family space” (Kemal, M, 30, teacher) “where even grandparents are active” (Nilgün). WhatsApp, on the other hand, was used by all interviewees for communicating with close ties and partly to communicate with colleagues. One of the interviewees said that WhatsApp discussions are like “Facebook carried on there” (Leman, F, 36, lecturer/filmmaker). Others state that it is much less conflictual than Facebook “because there are more people like [oneself] there” (Melike, F, 21, student).

Interviewees perceived the social media platforms also as a communication space that can be partially controlled and needs to be “managed” (John & Gal, 2018). For example, Leman, (F, 36, lecturer/filmmaker) described Facebook as “a neighborhood,” where you would find “all kinds of people. Good or bad, it depends on how you manage that neighborhood.” Another interviewee said: “the wall is my wall; you cannot share everything there” (Nalan, F, 31, banker). This perception comes with a certain responsibility for monitoring and controlling the content on “the wall” and the newsfeed. The interviewees developed different strategies to manage what they are being exposed to on their timelines and which audiences are exposed to their own posts considering the diversity of their networks. These strategies included visible forms of disconnection (unfriending and blocking), invisible forms of disconnection (muting, unfollowing, and ignoring); actively debating, observing, and self-censorship. Visible forms of disconnection are used for seeing and not being seen by divergent others. With invisible forms of disconnection, the strategy is more about being or not being exposed to others’ posts, whereas debate and self-censorship are more related to exposing one’s opinions to others. Furthermore, most interviewees kept their accounts restricted to their friends, but did not adopt any further strategies of changing privacy settings. Therefore, privacy settings will not be discussed here as a separate strategy.

Visible Forms of Disconnection: Unfriending, Blocking

Unfriending is one of the users’ strategies for selective avoidance of divergent opinions on social media (John & Dvir-Gvirsman, 2015; John & Gal, 2018; Skoric et al., 2018). Although unfriending and muting is often considered to be in the same category in existing research (e.g., Duggan & Smith, 2016; Skoric et al., 2018), they have different implications for network diversity (John & Gal, 2018). Unfriending (Facebook) and blocking (Twitter) are visible to the disconnected person and indicate an explicit statement about the wish to disconnect. Muting (Facebook, Twitter), unfollowing (Twitter), and ignoring (Facebook and Twitter) are rather invisible to the counterpart and imply a wish not to completely break up the relationship.

Interviewees’ unfriending behavior reflects the different axes of polarization in Turkey as demonstrated by Erdogan and Uyan-Semerci (2018). For example, the supporters of conservative-nationalist MHP and conservative-religious AKP often unfriend HDP (the left-liberal Kurdish party) voters, which they feel most distanced from. In particular, the more partisan interviewees identifying strongly with a particular ideology did not tolerate people supporting parties/groups that they feel most distanced to at any given moment. For example, Yasin (M, 26, waiter), a member of the conservative-nationalist MHP, openly said that “[he] does not like Kurds, [he] does not like Alevites” and does not have them in his online and offline networks. Another interviewee, who defines himself as a secularist, says that he directly blocks AKP voters without even looking at the closeness level of their relationship:

If there is [anything that disturbs me], block them. Especially when there are AKP supporters, I don’t look at the closeness of our relationship, I remove them. I don’t see what I don’t want to see there. (Mustafa, M, 32, technical draftsman)

As the quote from Mustafa shows, one reason for unfriending is “to not see things that you don’t want to see” on your timeline. Mustafa also implies that if someone is an AKP-supporter, even “their degree of closeness” does not matter. In a sense, this other person does not deserve to be part of his social network as he or she is supporting an unacceptable political position.

Besides the interviewees who completely excluded particular viewpoints from their social networks, there were also politically disengaged or moderate interviewees, who indicated that they “normally” tolerate diverse political positions on their social networks. These also referred to instances where they disconnected from divergent others because of heated debates and conflicts on social media. These conflicts often emerged in “acutely polarizing” (Kearney, 2019), peak moments of polarization such as Gezi protests (May–June 2013), pre-election periods, the terror attacks (several attacks in 2015–2016, among others), and the coup attempt (July 2016). Nilgün (F, 22, university student), for example, explains how she “grew apart from [her friends]” after the terror attacks in Ankara (10 October 2015). She adds that “she feels very ‘tired’ and ‘sad’ and she has ‘no tolerance towards others any more’.” Another interviewee says:

I also have so many Kurdish friends, we were like blood brothers. I even know many Kurdish songs. For example, after this last terrorist attack, after the coup [attempt], I see that some are still on the wrong path. Since they were sharing things like that, I removed them. (Sezgin, M, 29, student, vocational college)

Sezgin’s emphasis on “still being on the wrong path” shows that unfriending in his case was not due to the “collapse of the imagined homogeneity” (Schwarz & Shani, 2016, p. 406), but due to the strengthening of already existing fault lines between people who came together despite their ethnic and political differences.

Unfriending is also strongly linked to in- and out-group identities (Iyengar et al., 2012), which are strengthened through people’s emotional states during crisis moments. For example, during the Gezi protests, Leman (F, 36, lecturer/filmmaker) was unfriended by “friends from the other community,” supporters of the protests, “because of [her] posts.” Another interviewee unfriended someone during the Gezi protests who posted about his support of the AKP on Facebook while “[(we) protestors] were experiencing such intense things.” She unwillingly re-friended this person as he sent a request after 2 year, but she actually did not “want someone so conservative to see [her] posts, [her] pictures, family, and network.” Several interviewees stated that “what happens on social media does not only stay there” and online conflicts were “carried into real life” (Yasin, M, 26, waiter). In most cases, it was hard to repair these social ties that were damaged in the peak moments of polarization.

Invisible Forms of Disconnection: Muting, Unfollowing, Ignoring

In comparison with unfriending or blocking, there are also more subtle forms of disconnection that are not immediately visible to the other side. In this case, the interviewees seem interested in avoiding political conflicts and dissonance, but do not want to completely burn bridges or hurt the other side. Instead, they “have the person on [their] list but unfollow” (Nalan, F, 31, banker). It is about “not seeing things coming from this person,” but “remain[ing] friends” (Figen, F, 38, PR expert). One interviewee explains:

Muting or unfollowing [other users] is one of my favorite functions on Facebook because it would be impolite if I unfriend them and they realize this. (Halil, M, 28, university student)

Like Halil, the interviewees mute others because they do not want to completely break their ties with these people and try to avoid being “impolite.” They still deactivate their connection with them to not be confronted with their posts and political views. As in this case, the users adopt the material affordances of the platforms (or “the functions” as Halil refers to it) in different ways to avoid being exposed to posts that they do not wish to see. Whereas on Facebook users can unfollow their friends (to hide their posts on their newsfeeds) or unfriend them, on Twitter they can choose to mute, unfollow, or block others. The settings of the platforms set the framework for the possible actions that users can take in relation to being exposed to others or, on the other hand, avoiding their content. There are also some interviewees who stated that they muted others not necessarily because of political disagreements but for other reasons such as oversharing, “sharing everything [. . .] like the babblers in everyday life” (Figen). But divergent views seem to be the most common reason that interviewees muted or unfollowed others on their online social networks. For example, Rafet (M, 24, receptionist) unfollows extreme conservatives because he feels “increasingly backed into a corner.” He says he “respect[s] them, but [he doesn’t] have to follow them or read what they write.”

Besides muting and unfollowing, ignoring someone’s posts (Duggan & Smith, 2016; J. K. Lee et al., 2014) can also be a form of invisible disconnection. For example, Leman (F, 36, lecturer/filmmaker) says she does not read people’s posts or comment on them if they have “a harsh tone.” Another interviewee only follows what he “find[s] interesting and skip[s] what he doesn’t without reading.” He adds later that he likes the posts of people, like his cousin, “with whom he has a similar point of view” (Yasin, M, waiter).

Debating with Others

Most interviewees preferred not to share or comment on political posts on social media to avoid conflicts, among other reasons. This is in line with previous studies on social media use in Turkey and elsewhere (Duggan & Smith, 2016; Erdogan & Uyan-Semerci, 2018). However, there are also politically engaged interviewees who regularly post about politics and debate with others on social media (see also (Bode, 2016; Lu & Lee, 2020). These users aim to “share [their opinions] in some way” (Nilgün, F, 22, university student)”; tell the others “about the truth” (Kadir, M, 40, taxi driver); “reach out to people”; and “make an impact” (Cagla, F, 31, journalist). Others like Leman (F, 36, lecturer/filmmaker) want to “show where [they] stand” on certain issues. Leman adds that she first “observes the news and filters the information flow” and thinks a lot before posting something because it creates an “image” of who she is and can “lead people in a wrong direction.” Similarly, Rafet (M, 35, receptionist) explains that he posts very carefully and not “with the impulse of the moment” because he has “friends from within and outside of the country with diverse opinions.”

The composition of interviewees’ networks on specific platforms influences how they post and debate with others on them. On Facebook, the interviewees are mostly interacting with people that they know from their offline environments. This leads to more caution in the content and tone of their posts and in some cases leads to self-censorship. On Twitter, in contrast to Facebook, the interviewees often follow and interact with people whom they do not know personally. Kemal (M, 30, teacher) indicates that “[he feels] more free on Twitter” and that “[he does] not feel as free on Facebook” because he is anonymous on Twitter. Furthermore, Cagla believes that when she posts about politics, she “can reach more people through Twitter.” It was indeed the more politically engaged interviewees, like Kemal and Cagla, who are more active on Twitter.

The politically engaged interviewees even moderate the discussions on their walls and delete “comments that are too extreme, too harsh, or even seem like insult” because they have a diverse group on their list (Orhan, M, 29, lecturer). Orhan explained that he believes in the “culture of social media” and that it is important to “keep this culture,” meaning that he believes “everyone can share something. They should be free to do this, but should be respectful while sharing.” Figen (F, 38, PR expert) also sees the debates under her posts as a chance for her friends with divergent opinions to “explain their viewpoints to each other.” But she intervenes to keep the “good debates” going when someone is “being rude.”

Most interviewees who regularly post about politics and are very careful about what they post in “normal” times, often start posting more recklessly during “abnormal events” and are less likely to care about “whoever feels bothered” by their posts, as Kadir describes:

I have quite an extensive network. For example, there are all kinds of people in our cabstand—they are all my friends. I post things considering everyone. I am careful so that nobody is upset or bothered. But sometimes during very abnormal events, when you feel blinded by anger, you say “I don’t care whoever feels bothered.” (Kadir, M, 40, taxi driver/driver)

Not only politically engaged users like Kadir, but also the politically moderate or less engaged users, who usually avoid politics on social media, indicated that they post more in these “abnormal” times. The most common examples given for such abnormal events in the context of Turkey were the Gezi protests (May–June 2013), terror attacks (mostly referring to the large-scale terror attacks in 2015–2016), pre-election, and referendum periods and the coup attempt (15 July 2016). The coup attempt was the most recent of these events and happened in the middle of the presented field research. Some interviewees indicated that they “felt the need to express themselves” after the coup attempt and “take a position” “because you either support this or are against it” (Orhan, M, 29, lecturer). One interviewee said that she does not usually post anything political but the situation has changed now and is “not what we were used to” and “therefore [she] is posting now” (Nevin, F, 35, specialist in an animal hospital). Another interviewee “felt the need” to share his views on social media as a “normal patriot”:

I particularly do not engage with politics because [. . .] I have friends supporting all different political views. I know their opinions and they do mine. Therefore, I try not to engage in quarrels and don’t comment. But I shared a lot of political posts on the day of the coup [attempt]. I felt the need to do this as a normal patriot (Sezgin, M, 29, student in vocational college)

Several interviewees reported that they posted more than usual on social media during/after events that they felt were “abnormal” or “unusual.” However, these peak moments of polarization affected the interviewees’ posting behaviors differently depending on their political positions. For example, it was primarily the conservative-religious and (conservative-)nationalist people, who expressed “the need to share their opinions” during and after the coup attempt. There were also others who were critical of the government and afraid to express any thoughts after the coup attempt. One interviewee sees it as a turning point for herself as “she felt worried about what she wrote on social media for the first time” (Figen).

Observing Others

Although the majority of social media users find discussing politics on social media with people of divergent views as frustrating and stressful, there is still a considerable group of users who find these discussions to be interesting and informative (Duggan & Smith, 2016). In particular, politically engaged interviewees who wanted “to understand political situations in detail and understand alternative perspectives” (Dubois & Blank, 2018, p. 735), tried to avoid echo chambers as they saw value in being exposed to different ideas. Some said that they seek to keep their online social network as heterogeneous as possible because it allows them to observe others “whose lives are very different than their own” (Figen, F, 38, PR expert). For example, Kemal (M, 30, teacher) said that he never unfriends anyone because of divergent views as he is “even more curious about his/her thoughts.” Leman (F, 38, lecturer/filmmaker) said that she consciously added “people from every faction” to her list to follow their posts and opinions especially “when there is a topic that is not yet very clear.” She reads “the two sides” to understand them and develop “an opinion of [her] own.” Another interviewee (Meryem, F, 34, secretary) mapped the diverse Facebook accounts on her network map showing what she finds “inspiring” and follows intensely (Figure 2).

Figure 2.

Network map—Meryem.

Nevertheless, even the users who “read the opinions of the other side,” point to exceptional issues such as “violence against women” that they cannot be tolerant about (Nalan, F, 31, banker). Issue sensitivity also plays a role for tolerance toward others in this regard. The issues vary depending on the interviewees’ political interests. For example, Nilgün (F, 22) is a university student who is interested in “observing different opinions,” so she normally “leave[s] people on [her] page so that [she] can see [them].” But she says if “there is someone, who intervenes in the values [. . .] that [she] mentioned before, [she] do[es] not follow them.” Nilgün mentioned earlier in the interview that she is raised in a right-wing family and things like “the flag” (symbolizing the Turkish nation) cannot be questioned for her. Another interviewee, who defines herself as left-liberal, says that she likes to read other people’s opinions’ but she cannot stand it “when they insult other people’s beliefs,” especially to Alevites (Fatos, F, 57, housewife).

Self-Censorship

In line with existing research (Chen, 2018; Duggan & Smith, 2016; Thorson, 2014), several of the interviewed social media users in Turkey avoid political posts on social media because they “did not see any benefit [in doing that]” and “it only pushes people away” (Kemal, M, 30, teacher). They also “feel upset if [they] write something” and debate with others (Efe, M, 41, self-employed). Therefore, many interviewees “prefer to remain silent” (Kemal and Leman, F, 36, lecturer/filmmaker) and “protect [themselves] with auto-control” (Efe). Selami explains this as follows:

Social media is bigger than it should be, and in Turkey especially, people use it carelessly. Since people do not respect opposing opinions, even when you write something correct, [the discussion] turns into to insults, profanity, and ends up being a waste of time. And it makes you angry. That’s why I don’t comment anything. (Selami, M, 28, veterinarian)

Selami completely avoids commenting on social media because of “disrespect to opposing opinions” in Turkey. Another interviewee openly indicated that he only comments others’ posts “if he likes what they post” and “never [comments] if he disagrees” (Kemal). Kemal added: “everyone is on Facebook now” and “we are a polarized society. No matter what you say in such a society, you will be excluded.” He stressed that Facebook is especially inadequate for political debates and he still shares political posts on Twitter with an anonymous account. Melike (F, 21, university student) also believes that “you cannot really debate through Facebook.” She recently came across a friend’s comments about the refugees in Turkey. She first thought about “unfriending him/her, but then thought why should [she]? Maybe [they] will talk to each other again one day.” She felt like she “should say something about this post,” but then based on “[her] previous experiences,” she “did not think that it would lead somewhere.” When she commented on posts before, “nothing changed” and she only lost “the chance to talk to people that [she] would have come to terms with if [she] sat down and talked to them” (Melike). Melike’s case shows that self-censorship is also employed to remain connected to diverse others. Similarly, Nilgün (F, 22, university student) explained that after experiencing several political conflicts in her university WhatsApp group, they agreed “not to share their political opinions” here to avoid conflict.

In the context of Turkey, self-censorship also has a political dimension as social media users try to protect themselves from the pressures of the government and the increasing restrictions on freedom of expression (Yeşil et al., 2017). In particular, since the Gezi protests in May-June 2013, the AKP government started to adopt more comprehensive strategies for controlling online content including new legislation that could increase censorship online, social media bans, content removals, throttling, and domain name server (DNS) poisoning during major events and surveillance and prosecution of users (Yeşil et al., 2017). Accordingly, Turkey’s status was changed from “partially free” to “not free” by the Press Freedom House in 2018 (Freedom House, 2019). Several internet users have been prosecuted in recent years because of their social media posts with justification coming from anti-terror laws or different articles of the penal code. In particular, article 299 of the penal code, against insulting the president, was used to prosecute 17,406 people (mostly, but not only based on their social media posts) between 2014 and 2018. Of these cases, 5,683 were settled with penalties (Dogruluk Payi, 2019). Against the background of the increased surveillance and pressure on social media users, the fear of sharing critical points of view was raised by several interviewees: Kemal, for example, started using an anonymous Twitter account instead of a personal account for this reason. Fulya (F, 38, teacher) is “reluctant in her posts” because she has “a family to take care of” implying that something could happen to her if she posted critically on social media. Meryem (F, 34, secretary) said her friends recommended she delete her posts on Facebook on the night of the coup attempt to avoid getting into trouble. Our recent qualitative study shows that fears of being socially/economically pressured and/or arrested because of critical social media posts have become even more significant in 2019 (see Kocer & Bozdağ, 2020).

Discussion: Factors Influencing Management of Diverse Networks on Social Media

The presented study maps social media users’ different strategies for managing the diversity of their online social networks. Although some users tend to adopt one strategy more often than others, most use multiple strategies to deal with the complex communication situations that emerge through network diversity and increasing polarization. Four issues emerge as key for understanding when and why people adopt specific strategies in certain times; namely, political interest and engagement of the users, the general political climate, issue sensitivity, and imagined affordances of different social media platforms.

The political interest and engagement of social media users influence their attitudes toward others on social media. Existing research shows that users who strongly identify themselves with a political party or ideology; are very interested in politics; and/or are politically active, tend to engage more in political debates on social media (Bode, 2016; Kearney, 2019; Lu & Lee, 2020). Therefore, they are more likely to face conflicts and disconnect with others because of these debates (Bode, 2016; Kearney, 2019; Lu & Lee, 2020). Supporting these findings, the interviewees who said that they would never include certain people (e.g., “AKP voters”; “Kurds and Alevites”) in their online social networks were those with highly partisan orientations. However, this study complements these findings by also showing that the politically moderate or disengaged users start to disconnect from others with divergent viewpoints in peak moments of polarization.

The general political climate also has an influence on the way people react to diverse others on social media. This study shows that even the politically moderate users, who “normally” tolerate others’ posts and define themselves as “relaxed in normal times” (Nevin, F, 35, specialist in animal hospital), become more “angry” and “sensitive” in “abnormal periods” (Kadir, M, 40 taxi driver/driver). During these peak moments of polarization, they either posted more impulsively without regarding “who might be disturbed by these posts” (Kadir), or they were more sensitive about the tone of others’ posts and unfriended them. The existing fault lines in the polarized societies are crystallized during these moments and social media increases the chances of being confronted with more impulsive and emotional reactions from others. This leads to a decrease in online network diversity.

Furthermore, users might be quite moderate and tolerant in relation to certain topics and more sensitive when confronted with others. For example, issue sensitivity regarding women rights; animal rights; rights of ethnic and religious minorities; and environmental issues, among others also plays a role in the ways people deal with political differences articulated on social media. Several interviewees stated that they can read and tolerate divergent views on social media as long as they “do not interfere with values” that are beyond the pale for them (Nilgün, F, 22, university student). The issues that “push the buttons” (Nalan, F, 31, banker) of the users are also issues that they post more about and more likely to lead to conflicts and disconnection on different occasions.

While developing strategies about managing the diversity of their networks, users also consider the imagined affordances of the platforms (Nagy & Neff, 2015). These are not necessarily pre-given or fixed affordances on the platforms, but more “affordances in practice” (Costa, 2018) shaped through social and cultural context and are in constant construction (Nagy & Neff, 2015). The interviewees’ perception of different platforms and patterns of use vary. For example, Twitter is more often used by the politically engaged interviewees, who see it as a “more political space.” Facebook, on the other hand, is defined as a “family space” and seen as “too mixed” by many users. Because of this diversity, users are either very cautious about what they post on Facebook in “normal times” or they refrain from posting anything political at all on this platform. WhatsApp has an important role for keeping contact with close ties and interviewees also occasionally discuss political issues with their contacts there. However, most interviewees are mostly connected with like-minded people on WhatsApp and see it as a less conflictual environment.

Conclusion

By mapping social media users’ strategies for dealing with network diversity and the factors influencing these strategies through an in-depth and a cross-platform study, this article contributes to the body of research on social media and polarization. The findings demonstrate how social media users in Turkey grow apart from their divergent social networks in times of polarization. However, this does not mean that specific social media platforms inherently bring along the affordances of polarization (F. L. Lee, 2016). The potential influence of specific social media platforms on polarization or on cross-cutting exposure should always be considered in their specific contexts. In certain times, specific social media platforms contribute to the emergence of a “public sphere” for a certain group of people as they inform themselves about the opinions of others and/or debate with them. In other times, for example, during peak moments of polarization as discussed above, social media reveal or amplify political differences and create conflicts. In this regard, social media should not be understood as public sphere versus echo chambers (Colleoni et al., 2014), but they might serve as both depending on the specific context.

People’s online social networks are also under constant construction and they may become less or more diverse as people add or unfriend others. Looking at the present context of Turkey, we can see that increasing polarization in the country leads to a decrease in the diversity of people’s online social networks. Most interviewees still have more diverse online networks than their day-to-day offline networks (see also Barberá, 2014; Brundidge, 2010; Choi & Lee, 2015), but they also report changes (at different levels) in the diversity of their networks in recent years because they disconnected from others with divergent viewpoints that were articulated through social media posts or comments.

The potential of different social media platforms for network diversity also varies depending on their “imagined affordances” (Nagy & Neff, 2015). For example, Twitter is increasingly used by a more partisan and politically engaged group of users, who are especially interested in disseminating their own viewpoints. Facebook, on the other hand, is also used by politically disengaged or moderate users who are more open toward cross-cutting discussions. Furthermore, Facebook still has the potential for unintended cross-cutting communication owing to the fact that Facebook users’ networks bring together a diverse group of weak and strong ties, as can be seen on the interviewees’ network maps. Several users also turn to WhatsApp for sharing their political views in more homogeneous groups, which is perceived as less risky in terms of surveillance by the government during social conflicts.

As people disconnect from each other, especially in peak moments of polarization, their chances of being confronted with different opinions and perspectives at other times are also decreasing. The interviewees themselves reflected on the effects of polarization and reported being concerned about “only speak[ing] among [them]selves” (Fatos, F, 57, housewife) and losing the “chance to talk to people that [they] would have come to terms with if [they] sat down and talked to them” (Melike, F, 21, student). One of the interviewees asks “we are all defending the same opinions, so what’s the point? [. . .] How do we progress?” (Fatos). This reflection points to a wish to depolarize. Some interviewees intentionally engaged with divergent others on social media for this purpose.

The study contributes to ongoing academic debates first by offering a contextualized qualitative analysis on polarization and social media pointing to the role of political context and climate. Second, the article makes a distinction between visible (unfriending and blocking) and invisible forms of disconnection (muting, unfollowing, and ignoring). Third, the article also demonstrates that users’ attitudes toward divergent viewpoints also depend on issue sensitivity and the “imagined affordances” of specific platforms.

The presented findings mainly relate to Facebook, Twitter, and WhatsApp, which constituted the most popular platforms and applications in Turkey during the field research. Future research can look at the potentials of Instagram for cross-cutting exposure as its prevalence increased and the platform became more relevant for political issues (see Kocer & Bozdağ, 2020). Although the study is also limited by its sample size and context, it offers key insights about polarization and social media that can guide future studies of other polarized contexts. Furthermore, the findings of this study are limited by self-report. Nevertheless, understanding users’ own discourses about social media and network diversity is important in evaluating social media’s potential for spurring democratic debates and contributions to democracy. Last but not least, this study focuses more on polarization and disconnection. Future research should look at moments of depolarization among people with divergent views, for example, the coronavirus pandemic reunifying old friends online.

Supplemental Material

sj-docx-1-sms-10.1177_2056305120975713 – Supplemental material for Managing Diverse Online Networks in the Context of Polarization: Understanding How We Grow Apart on and through Social Media

Supplemental material, sj-docx-1-sms-10.1177_2056305120975713 for Managing Diverse Online Networks in the Context of Polarization: Understanding How We Grow Apart on and through Social Media by Cigdem Bozdag in Social Media + Society