Like-minded and cross-cutting talk, network characteristics, and political participation online and offline: A panel study

Cross-cutting talk, like-minded talk, and political participation

Since the early work of Lazarsfeld and colleagues (1944), public opinion scholarship has been concerned with the impact that interpersonal discussion has on individuals’ willingness and motivation to participate in activities that are relevant to society. Ideally, citizens should be confronted with both political views they agree with and opposing opinions, as this enables them to be as comprehensively informed as possible. However, confrontation with opposing attitudes and information may lead people to become less politically active, which is why homogenous networks, in theory, are better for the functioning of democracy (Matthes, 2012; Mutz, 2002). In line with this, Mutz (2002) argues that network-heterogeneity dampens political participation because disagreement within a close circle of family, friends, and acquaintances leads people to experience conflict and negative feelings. This, in turn, suppresses political action because individuals do not wish to offend their network. At the same time, cross-cutting in contrast to like-minded views offer new information, making people feel ambivalent (i. e., insecure) about their own positions, further dampening participation. However, other studies have found no support for that claim (Bello, 2012; Pattie and Johnston, 2008)

In contrast to disagreement, opinion congruence can be assumed to encourage participation (e. g., Matthes and Marquart, 2015). However, most of the available research has focused on the effects of opinion disagreement, rather than of congruence. As two exceptions, Eveland and Hively (2009) as well as Valenzuela, Kim and Gil De Zuñiga (2012) found evidence for the notion that participation is positively affected by like-minded interpersonal discussion. Possibly, opinion-friendly talk reminds people of the benefits of their party alignment (McGhee and Sides, 2011), therefore prompting political engagement. As another explanation, spiral of silence theory (Noelle-Neumann, 1974) suggests that opinion-friendly talk is processed as a cue that the majority of citizens hold similar views, which may strengthen one’s viewpoint or be a signal for normative pressure to do the right thing (Matthes and Marquart, 2015).

Strong vs. weak network ties

When looking at the impact of interpersonal discussion, it matters to whom individuals talk when they agree or disagree on a given issue. Accordingly, social network ties’ strength plays an important role. Roch and colleagues (2000, p. 779), in reference to the seminal work by Granovetter (1973), define tie strength by the frequency of interaction between two individuals: “Strong ties generally reflect repeated interaction within a small group of family, relatives, or close friends [while] weak ties characterize the less frequent contacts that occur within more extensive, specialized networks”. The latter entail relationships with co-workers, students at schools, or fellow members of other associations. Strong ties can be further distinguished from comparably weak connections by closer emotional bonds and reciprocity of service (Morey, Eveland, and Hutchens, 2012). These characteristics have consequences for the way in which citizens talk about politics with their network ties: Not only are individuals confronted with political disagreement more often when talking to their strong social network members (Morey et al., 2012), the effects of such interpersonal discussions also vary when compared to weak-tie cross-cutting talk.

When it comes to disagreement, there are two competing theoretical accounts: On the one hand, strong-tie connections are better able to withstand pressure from disagreement since such individuals most likely agree on other more important issues – mere disagreement on political topics does not constitute a threat to strong relationships per se. In addition, “spouses and family members contribute far more to exposure to sustained [emphasis added] disagreement” (Bello and Rolfe, 2014, p. 145), which is to say that close relationships offer a safe place for interpersonal political discussions and remain stable over time; whether one agrees on political topics or not, family members are bound to stay.

On the other hand, citizens may avoid disagreement out of fear of social exclusion (e. g., Lu et al., 2016; Matthes, 2013; Mutz, 2002), which may be stronger when disagreement is faced with strong as opposed to weak ties. Torcal and Maldonado (2014), for instance, found cross-cutting discussions to affect political interest in different ways, depending on the strength of the respective personal relationship: Individuals become more apathetic about politics when their strong network ties (i. e., family and close friends) disagree with them. This effect is less pronounced for cross-cutting talk with weak ties because “it is more difficult to be interested in something that places the individual in conflict with family and friends” (Torcal and Maldonado, 2014, p. 696). While Torcal and Maldonado focused on political interest and did not investigate effects for participation, it may well be argued that political interest is an important prerequisite for any participatory behavior. Taken together, the role of strong versus weak ties in predicting the effects of cross-cutting interpersonal discussion is far from being entirely understood. When it comes to like-minded talk, we completely lack insights about the role of different network ties.

Online vs. offline participation

Offline participation refers to rather traditional forms of participation in political activities, such as voting, working in a party organization, attending speeches, or displaying campaign buttons. Typical forms of online participation are signing an online petition, sending a message to a public office holder, or sharing political messages on social networks (Bernhard and Dohle, 2018). When it comes to the effects of cross-cutting and like-minded discussions, we lack studies that compare online versus offline participation, and extant findings are far from being consistent (see Bello, 2012; Ikeda and Boase, 2011; Matthes et al., 2019).

On the one hand, cross-cutting talk’s effects on participation may be weaker in online compared to offline participation contexts: The online environment allows for a huge variety of opinions, and individuals experiencing cross-pressures in their interpersonal network can easily find alternative opinion support online and participate there. When online settings are anonymous, unpopular opinions can be easily voiced without putting personal relationships at stake. For instance, so called “click speech” – “liking” and sharing content without having to produce one’s own content (Pang et al., 2016) – does not necessitate much cognitive effort and commitment, and may not depend on interpersonal cross-pressures. On the other hand, the boundaries between online and offline communication are becoming increasingly blurred (Bernhard and Dohle, 2018; Pang et al., 2016). For instance, Facebook contacts are often based on real-world social relationships, and online political participation may be visible to one’s interpersonal network. It follows that cross-cutting or like-minded talk may determine whether citizens participate online or not.

Following the outlined literature, we propose two hypotheses and two research questions describing the relationship between political discussion and political participation. First, we assume that:

H1: Exposure to cross-cutting talk dampens participation.

 

H2: Exposure to like-minded talk increases participation.

Since these relationships may depend on network characteristics as well as the type of political participation (i. e., online vs. offline), and competing theoretical arguments prevail in the literature, we also ask:

RQ1: What are the differences between strong and weak network ties when it comes to the influence of political talk on participation?

 

RQ2: Are online and offline participation affected to different extents when it comes to the influence of cross-cutting and like-minded political talk?

Measurements

All measurements are listed in the Appendix. Our main independent variables asked participants about the frequency with which they were exposed to various interpersonal discussion partners that were in line with, or opposing, their own political position during the past four weeks (ranging from 1 = never to 5 = very often; adapted from Eveland and Hively [2009] and Valenzuela et al. [2012]).^[2] It is important to note that these are formative measures, not reflexive measures (Kline, 2011). Items were summed up to create the following indices: cross-cutting talk with weak ties in W1 (M = 8.44, SD = 9.34) and W2 (M = 8.74, SD = 4.44), cross-cutting talk with strong ties in W1 (M = 6.54, SD = 2.24) and W2 (M = 6.44, SD = 2.33), like-minded talk with weak ties (W1: M = 9.98, SD = 4.48, W2: M = 10.53, SD = 4.89) and strong ties (W1: M = 10.23, SD = 2.76; W2: M = 10.32, SD = 2.83).^[3]

In order to assess participatory activities, we asked respondents in both waves whether they had, first, engaged (no/yes) in any of nine offline activities which were summarized to one index measuring participatory activities offline (W1: α = .77, M = 1.86, SD = 2.08, range 0–8; W2: α = .76, M = 1.76, SD = 2.04, range 0–8; adapted from Vissers and Stolle, 2014). Participants’ online participation activities were assessed by asking whether they had, during the past four weeks, engaged (no/yes) in any of four activities that were summed up to measure online participatory behavior (W1: α = .73, M = 1.32, SD = 1.35; W2: α = .69, M = 1.28, SD = 1.30; range 0–4, respectively). Online and offline political participation in wave 2 are our main dependent variables, and the panel analysis allows us to control for the respective autoregressive effect of former participatory activities in wave 1.

In addition to sociodemographics, we included a number of control variables from W1: newspaper (M = 3.40, SD = 1.25), television (M = 3.60, SD = 1.21), online news (M = 3.71, SD = 1.33), and social media use (M = 2.88, SD = 1.49; all ranging from 1 = not at all, to 5 = very often), political orientation (M = 4.24, SD = 2.04; range 1 = very left to 10 = very right), and general political interest (M = 3.35, SD = 1.28, range 1 = hardly interested to 5 = very interested) (Foos and de Rooj, 2016).

Data analysis

Data were analyzed using path analysis in Mplus. Since the dependent variables are count variables with clear signs of over-dispersion, OLS regression is not appropriate. We thus used negative binominal regression, controlling for the autoregressive states of offline and online participation (see Table 2).

Limitations and future research

Clearly, panel studies with more waves are needed, preferably including comparative data from multiple countries. Similar to previous research (e. g., Klofstad et al., 2013), we used a time gap of four weeks between the waves which is fully sufficient to detect cross-lagged relationships. Furthermore, we measured disagreement within the interpersonal social network by perceived disagreement. Such a question does not allow us to assess actual opposition; however, it has repeatedly been argued that perceived disagreement is, in fact, more decisive in this context (e. g., Klofstad et al., 2013; Lu et al., 2016). While this has not been the focus here, we agree that inconsistencies also stem from operationalizing central concepts differently (e. g., Klofstad et al., 2013), and strongly urge future work to put more emphasis on uniform measurements. In addition, our questionnaire was limited in asking about the frequency of political discussion encounters only; as such, we are not able to assess how often respondents talk to other people in their network in general. Future work should include respective measures in order to account for a) general network sizes (and the amount of weak and strong ties) and b) the possibility that more recent conversations might bias respondents’ memory. When doing so, scholars are also encouraged to take a closer look at study respondents’ conversation partners: As we argued here, the ‘who’ can play an important role when it comes to assessing the influence of talk to strong and weak network ties on participatory behavior. While we considered the frequency with which respondents in our sample were exposed to opinion-friendly and cross-cutting talk with predefined groups of conversation partners (e. g., spouses, relatives, teachers, or distant online contacts), individual perceptions of the importance of these groups might differ (see Eveland, Song, Hutchens, and Levitan, 2019). Social Network Analysis and the use of name generators in research can be beneficial here as well in order to identify central actors within networks (e. g., Miller, Bobkowski, Maliniak, and Rapoport, 2015; Nir, 2005). Moreover, we encourage data collection beyond the Austrian context of this study, and particularly recommend comparative designs to address questions related to cultural differences in political discussion setups as well. Research may also move beyond the simple online-offline distinction when it comes to political participation, but rather analyze high versus low effort forms (Knoll, Matthes, and Heiss, 2020).

Finally, scholars have stressed the importance of including moderating influences both at the individual and contextual level, arguing that interpersonal discussion heterogeneity may positively affect political behavior for some individuals while hindering participation for others. It was beyond the scope of this paper to further investigate the influence of such moderators, but we strongly encourage future work to take these into account in longitudinal data as well.