Sage Journals: Discover world-class research

Abstract

There are rising concerns that social network sites (SNS) facilitate the creation of echo chambers, in which attitude-consistent information becomes the norm while attitude-challenging information is avoided. This study aims to investigate theoretically derived predictors of attitude-consistent and attitude-challenging exposure on SNS. We theorize that three key sets of predictors may influence attitude-consistent and attitude-challenging exposure: ideology, cognitive, and behavioral indicators of political involvement, and network characteristics. In a two-wave panel study, we predict the frequency of attitude-consistent and attitude-challenging exposure as well as relative attitude-consistent exposure, measured as attitude-consistent exposure as a share of overall opinion exposure. The results demonstrate that extreme ideological positions, higher political knowledge, and low-effort political participation predicted an increase in (relative) attitude-consistent exposure. Cross-social class exposure predicted a decrease in (relative) attitude-consistent exposure. The findings challenge existing arguments that SNS may per se facilitate attitude-consistent exposure.

Keywords

opinion exposure selective exposure ideology social network sites

Social network sites (SNS) are providing new opportunities for individuals to seek and curate political news content. On SNS, individuals no longer need to rely on newsroom editorial boards and can instead self-select the political information they want to see and interact with. This form of self-selected personalization has the potential to expose individuals to more diverse opinions but can also limit exposure to attitude-challenging political information (Bakshy et al., 2015; Stroud, 2018). Such diminishing levels of exposure to attitude-challenging political information have raised concerns about homogeneous opinion networks, sometimes referred to as “echo chambers” (Vraga & Tully, 2021; Zuiderveen Borgesius et al., 2016). They are characterized by selective exposure which describes the selection of information that is consistent with our attitudes, and selective avoidance, that is, the avoidance of information that challenges our attitudes (Skoric et al., 2021; Zhu et al., 2017). As a consequence, homogeneous opinion networks could put a threat to the quality of democracy, as they may foster polarization and impair the heterogeneity of the public political discourse.

Previous research mostly investigated the (non-) existence of homogeneous networks in both the online and offline context (e.g., Brundidge, 2010; Garrett, 2009a; 2009b; Stroud, 2018). Most studies (e.g., Knobloch-Westerwick & Meng, 2009) found that, compared to offline news, online news consumption increases exposure to both attitude-consistent and attitude-challenging information. Only a few studies (e.g., Iyengar & Hahn, 2009; Messing & Westwood, 2014) found that SNS may contribute to homogeneous network formation.

Despite the progress made, three core research gaps remain: First, research on the antecedents of homogeneous networks on SNS is limited to a few predictors. Previous studies suggested that partisanship (e.g., Beam & Kosicki, 2014; Brundidge, 2010), ideology (e.g., Dahlgren et al., 2019; Skovsgaard et al., 2016), or political interest (e.g., Knobloch-Westerwick & Meng, 2009) may affect attitude-consistent exposure in online or offline contexts. However, we lack a systematic analysis of these antecedents in the SNS context. In other words, while previous scholarship aimed to clarify if SNS may contribute to homogeneous networks, we do not have sufficient knowledge about which predictors explain relative attitude-consistent exposure on SNS.

Second, the vast majority of previous examinations on SNS applied cross-sectional research designs (e.g., Fletcher & Nielsen, 2017; Garrett, 2009b). Cross-sectional designs are unable to clarify causal directions. Although there are some notable panel studies on the topic, they all examine traditional media outlets, or use a summary measure of “the Internet” (e.g., Dahlgren et al., 2019; Skovsgaard et al., 2016; Stroud, 2010), which cannot be generalized to SNS. Other studies explicitly refer to SNS (e.g., Beam et al., 2018; Weeks et al., 2017), yet they solely rely on partisanship but not on a systematic analysis of various individual predictors.

In what follows, we address these research gaps in multiple ways: First, we conduct a systematic empirical analysis of various individual predictors of attitude-consistent and attitude-challenging exposure, as well as relative attitude-consistent exposure. To measure relative attitude-consistent exposure, we calculated the share of attitude-consistent exposure in one’s total opinion exposure. We show that ideology, political involvement, and network characteristics are key individual predictors of attitude-consistent, attitude-challenging, and relative attitude-consistent exposure on SNS. Second, we applied a two-wave panel study during a National election campaign. This design affords advantages over cross-sectional data, because it allows to investigate stability and change on the individual level during different points in time. Panel designs also cancel out the problem of reversed causation and reduce problems associated with sample bias (Hsiao, 2014).

Personalization of Political Information Environments

Research on self-selected personalization has its origins in selective exposure theory.¹ Lazarsfeld et al. (1944) already observed selective exposure for partisans and generalized that “a positive relationship exists between people’s opinions and what they choose to listen to or read” (p. 164). Similarly, Festinger (1957) posited that people are more likely to select information confirming rather than contradicting their views. The advent of the Information Age added new impetus to selective exposure research. The Internet fostered a pluralization and fragmentation of the information environment, thus, allowing individuals a more autonomous news selection than offline (Iyengar & Hahn, 2009). SNS allow users to interact with each other (e.g., via liking, commenting, or sharing content) but also to receive comprehensive information from various sources. Hence, the different interaction dynamics that occur online and the unprecedented flow of obtainable information are likely to have diverse effects on information selection compared to offline environments (Bakshy et al., 2015; Messing & Westwood, 2014). Sunstein (2007) argued that the digital context encourages individuals to select attitude-consistent information over attitude-challenging information. Thus, people may no longer expose themselves to disagreeing views and therefore tend to stay in their own so-called “echo chambers”—a rather disputed concept that describes an information environment steadily reinforcing someone’s own views (see Zuiderveen Borgesius et al., 2016). As a consequence, these developments would diminish the heterogeneity of the public political discourse, increase political segregation, and have negative implications for the quality of democracy. Although a few studies (e.g., Iyengar et al., 2008; Jamieson & Cappella, 2010) have confirmed Sunstein’s assumptions, the major stock of literature did not find empirical evidence for the existence of echo chambers (Dahlgren et al., 2019; Y. Kim, 2011; Knobloch-Westerwick & Meng, 2009). For instance, Garrett (2009a, 2009b) showed that selective exposure and selective avoidance are two distinct concepts: Individuals may exhibit a preference for attitude-consistent information without systematically avoiding attitude-challenging information. Similarly, Knobloch-Westerwick and Meng (2009) found that the likelihood of selecting an online news article is 58% for attitude-consistent content and 43% for attitude-challenging content. Additionally, the online information environment can also increase incidental exposure to attitude-challenging information (Brundidge, 2010; Weeks et al., 2017). For SNS, Y. Kim (2011) demonstrated that SNS can positively affect exposure to cross-cutting information and also indirectly influence exposure to cross-cutting information via online political messaging. Situational motivations, such as accuracy, defense, or impression, can also affect selective exposure on SNS (Winter et al., 2016).

Regarding individual predictors, research has predominantly investigated the role of partisanship as a driver of selective exposure. The majority of findings showed that partisanship tends to increase exposure to attitude-consistent content and decrease exposure to attitude-challenging content (Iyengar et al., 2008; Stroud, 2018). For example, Long et al. (2019) observed a stronger effect for conservatives than for liberals and also for the degree of partisanship. Subsequently, partisan selective exposure is likely to increase political polarization (Stroud, 2010). Research has also investigated other individual predictors such as ideology or political interest. Dahlgren et al. (2019) found that ideological preferences can affect exposure to ideologically consistent online and offline news. Similarly, political interest, and also higher levels of habitual news consumption and attitude certainty, can increase both selective exposure and selective avoidance (Knobloch-Westerwick & Meng, 2009). Interestingly, Skovsgaard et al. (2016) showed that political interest can be a stronger predictor of selective exposure than ideological preferences.

Predictors of Attitude-Consistent Exposure

Although previous research has enhanced our knowledge about homogeneous opinion networks and selective exposure in the online and offline context, individual-level predictors are under-researched. While some work showed that partisanship may increase selective exposure (Iyengar et al., 2008; Long et al., 2019) or that attitude certainty and political interest increase both selective exposure and avoidance (Knobloch-Westerwick & Meng, 2009), no study has systematically tested a variety of relevant individual-level predictors. We will analyze three types: ideology, political involvement, and network characteristics.

The Role of Ideology

While there is some evidence that partisanship predicts attitude-consistent exposure (e.g., Beam & Kosicki, 2014; Matuszewski & Szabó, 2019; Messing & Westwood, 2014), we still lack research on the effects of political ideology on relative attitude-consistent exposure. We argue that ideology might be a more important predictor than single partisan preferences. The pluralized Austrian party system has undergone substantial transformations as a result of various political, economic, and social developments (Crotty, 2009; Mair, 2009). This pluralization also gave citizens a greater variety of political parties to potentially vote for and thus also makes the electorate rather volatile. Due to these circumstances, the degree of party affiliation in Austria is considered rather low (Grotz & Weber, 2012). At the same time, the Austrian party system was transformed by the upswing of right-wing populist parties, such as the Freedom Party of Austria (FPÖ) (Wagner & Meyer, 2017). This development paved the way for more extreme ideological views across citizens and has also reinforced this phenomenon on the ideological right which, in turn, also led to an increased mobilization of the ideological left (Lefkofridi et al., 2014). In fact, most parties in Austria can be described along the left-right ideology dimension and ideology may serve as a more important predictor of outcome variables than partisanship.

More importantly, people with extreme ideological views may be less open to diverging ideological positions than people with moderate political views (Rogowski & Sutherland, 2016). In fact, scholars have argued that individuals from the far left and the far right react in similar ways to attitude-challenging information. For instance, the findings of Rodriguez et al. (2017) suggest that ideology leads to a steeper trajectory in the increase of partisan selective exposure. According to these initial findings, one may predict that the process behind selective exposure should be similar for strong liberals and strong conservatives, or, in the European context, for far left and far right individuals. There are two theoretical explanations for this. First, the ideological rigidity hypothesis (Lammers et al., 2017) predicts a symmetric effect: “Strong conservatives and strong liberals will categorize political stimuli more sharply, compared to neutrals and moderates” (p. 613). In fact, Lammers et al. (2017) observed a positive U-shaped quadratic effect when explaining simplified categorization of information. Individuals with extreme ideological positions form simpler categories of political information compared to moderates. This simplified categorization is arguably one precondition to select attitude-consistent and to avoid attitude-challenging information. Second, according to the preference for certainty hypothesis, people on the ends of the political spectrum have a higher desire for clarity, certainty, and stability compared to the political moderate (Boutyline & Willer, 2017). This, in turn, should foster selective exposure and thus increase relative attitude-consistent exposure.

Surprisingly, however, there are hardly any studies that have tested the role of ideology in this context (Rodriguez et al., 2017). Most studies are U.S.-focused and simply test differences between liberals and conservatives to engage in selective exposure, independent of their ideological position (e.g., Bakshy et al., 2015). Nevertheless, based on research on the theoretical explanations outlined above, we expect extreme ideological views to increase the frequency of attitude-consistent and to decrease the frequency of attitude-challenging exposure. Therefore, we also expect such views to increase relative attitude-consistent exposure. Thus:

Ideological extremity (at both ends) is associated with (a) an increase in exposure to attitude-consistent content, (b) a decrease in exposure to attitude-challenging content, and (c) an increase in relative attitude-consistent exposure on SNS over time.

The Role of Political Involvement

Political involvement, which can be generally defined as being engaged into politically related activities (Scheufele et al., 2002), is a vital component of democracy. Previous research showed that political involvement can be a predictor of news media use and selective exposure in the offline and online context (e.g., Knobloch-Westerwick & Meng, 2009; Skovsgaard et al., 2016). Thus, it is relevant to also examine the relationship between political involvement and relative attitude-consistent exposure on SNS. The degree of political involvement can, among several aspects, be related to SNS activities. For instance, on SNS, information-seeking behaviors, political discussion characteristics, and political activities can affect participatory political behaviors in different ways (Matthes et al., 2020; Vaccari et al., 2016)

Although previous research has enhanced our knowledge about the effects of attitude-consistent and attitude-challenging information exposure on different forms of political involvement (see, for a meta-analysis, Matthes et al., 2019), it has not systematically investigated the role of different forms of political involvement as predictors of relative attitude-consistent exposure on SNS.

Drawing on research on public opinion formation (e.g., Donsbach & Traugott, 2008; Leeper & Slothuus, 2014), political involvement can have differential effects on exposure to attitude-consistent and attitude-challenging information: On the one hand, citizens with a high degree of political involvement may be very likely to be exposed to different political opinions, for example, because they engage in active news seeking and engage in political discussion (Lee et al., 2014). In other words, people who are highly engaged in politically related activities could be more likely to be exposed to attitude-challenging content. One reason for this argument may be that the politically involved seek to know all sides of an issue because they may expect to discuss their views with others. In order to defend one’s views, one needs to know the views and arguments of the opposing sides. On the other hand, people with high political involvement could also develop a biased political opinion. This is related to high self-identification, strong attitudes, and a strong political opinion of the politically involved (Leeper & Slothuus, 2014). Thus, they could be more likely to select attitude-consistent content. Taken together, we lack studies on different forms of political involvement as predictors of relative attitude-consistent exposure on SNS. The findings on political involvement and public opinion formation are also mixed. Therefore, we apply an explorative approach and pose a research question.

Furthermore, in order to assess political involvement comprehensively, we distinguish four different dimensions of political involvement spanning cognitive and behavioral components (Scheufele et al., 2002): political interest, political knowledge, high-effort participation, and low-effort participation. Interest and knowledge are classic dimensions signaling involvement at the cognitive level. Participation, by contrast, is a behavioral concept. In line with Knoll et al. (2020), we distinguish low-effort from high-effort participation. We do so, because we understand political participation as a goal-directed behavior which is contingent upon attainability assessments and the involved effort, rather than the online versus offline dichotomy (Knoll et al., 2020; Nanz et al., 2020; Valentino et al., 2011). Low-effort participation includes activities that require a relatively small amount of time and energy (e.g., sharing information and signing a petition). High-effort participation includes time- and energy-consuming activities (e.g., protesting and writing a blog).

RQ1: How are different forms of political involvement (political interest, political knowledge, and low- and high-effort political participation) associated with exposure to attitude-consistent and challenging content, as well as relative attitude-consistent exposure on SNS over time?

Network Characteristics Hypotheses

We argue that network characteristics are a substantial individual predictor of relative attitude-consistent exposure on SNS. We distinguish three aspects, (a) network size, (b) cross-class exposure, and (c) political exposure on SNS.

Network Size

Network size—the number of people a user is connected with on SNS—is likely to affect exposure to attitude-consistent and attitude-challenging content on SNS. Generally, there is reason to assume that people are more likely to befriend individuals similar in race, age, or general values (see Tang & Lee, 2013). However, if the network size becomes larger, the average tie to individuals in the network becomes weaker. There is reason to believe that individuals are only able to have closer ties with a limited number of individuals offline (e.g., Dunbar, 2016 suggests 150; see also Hill & Dunbar, 2003). Prior research in this context has also shown that network size does not have a significant influence on ambivalent political views (Mutz, 2002a, 2002b; Song & Eveland, 2015). On SNS, however, individuals are frequently connected to many more individuals compared to offline settings, which they connect with for other reasons than political affiliation or social status (Ellison et al., 2007; Y. Kim & Chen, 2016). Such weak ties arguably bring in more diversity to the network. Hampton et al. (2011) show that SNS can afford higher levels of network diversity than offline networks.

As a consequence, larger network sizes may facilitate exposure to content from weak ties, which may be more likely to express opposing political views. However, as exposure to attitude-challenging views may increase, exposure to attitude-consistent views may increase, too. Taken together, it follows that SNS users with large network size are more likely to be exposed to both attitude-consistent and attitude-challenging content. However, since the proportion of exposure to challenging views may become larger with higher numbers of weak ties in larger networks, we also assume that relative attitude-consistent exposure may decrease, too. Thus:

A larger network size is associated with (a) an increase in attitude-consistent content, (b) an increase in attitude-challenging content, and (c) a decrease in relative attitude-consistent exposure on SNS over time.

Cross-Class Exposure

Cross-class exposure describes how frequently SNS users are exposed to content from individuals who belong to a different social class (Groshek & Koc–Michalska, 2017; Lee et al., 2014). Such differences may increase the likelihood of exposure to attitude-challenging content, because citizens from different social classes often share opposing viewpoints, for example in terms of income redistribution (Visser et al., 2014). We hence argue that cross-class exposure may decrease exposure to attitude-consistent content and increase exposure to attitude-challenging content. We also assume that it decreases relative attitude-consistent exposure on SNS.

High levels of cross-class exposure are associated with (a) a decrease in exposure to attitude-consistent content, (b) an increase in exposure to attitude-challenging content, and (c) a decrease in relative attitude-consistent exposure on SNS over time.

Political Exposure on SNS

Finally, we assume that more frequent exposure to political information in general may increase overall political opinion exposure. Individuals who are frequently exposed to political information in their network are more likely to encounter discussions and opinionated posts from their network acquaintances (Zhang et al., 2010). Thus, frequent exposure to political information may not necessarily increase the share of attitude-consistent over attitude-challenging exposure. This is, because if citizens have high political activity in their network, the opportunity for incidental encounters with cross-cutting political opinions increases (Knoll et al., 2020; Weeks et al., 2017). Following this reasoning, we assume that frequent political exposure will increase both attitude-consistent and attitude-challenging exposure, and decrease relative attitude-consistent exposure.

High levels of political exposure are associated with (a) an increase in attitude-consistent content, (b) an increase in attitude-challenging content, and (c) a decrease in relative attitude-consistent exposure to political opinions on SNS over time.

Method

We conducted a two-wave panel survey in the context of the Austrian parliamentary election campaign 2017 (N = 559). The first wave was conducted between 29 August and 2 September, the second wave between 5 October and 12 October, one week before the election. Participants were exposed to one month of intensive election campaign between the two waves. The company Survey Sampling International (SSI) recruited the sample. We defined quotas based on the distribution of age, gender, and education in Austria. Because the questionnaire was mainly concerned with SNS use, we sampled people from 16 to 65 (M = 44.49, SD = 12.61) and who reported to use either Facebook, Twitter, YouTube, or Instagram. Therefore, all participants in our sample used SNSs. We did not sample individuals above the age of 65, because the penetration rate of SNS in this age group is still very low.

Sample

We collected data from 764 individuals in the first wave and some 73% responded in the second wave. Our sample is composed of 50.27% female. Some 19% had college degrees, 27% had degrees from college-bound high schools, and 48% apprenticeship or vocational schools. The remaining participants held degrees from compulsory schools. Although the attrition rate was 27%, our sample is still fairly representative for the Austrian population. The original quotas (based on national population survey) were 18% college-bound high schools, 13% college degrees, 44% apprenticeship/vocational school, and 25% compulsory school only.

Measures

All survey items addressing SNS use referred to the four most widely used SNSs in Austria: Facebook, Twitter, Instagram, and YouTube. We measured the dependent variables (opinion exposure) at both time 1 and time 2. All other variables were measured at time 1.

Opinion Exposure

To measure attitude-consistent SNS exposure (Spearman-Brown coefficient ρ_(t1) = .97, M_(t1) = 3.69, SD_(t1) = 1.88; ρ_(t2) = .96, M_(t2) = 3.66, SD_(t2) = 1.81), we asked participants how often they see political posts from contacts on SNS who (a) agree with their political opinion and (b) have the same political attitudes. To measure attitude-challenging exposure (ρ_(t1) = .92, M_(t1) = 3.54, SD_(t1) = 1.83; ρ_(t2) = .95, M_(t2) = 3.51, SD_(t2) = 1.86), we asked them how often they see political posts from contacts in these social networks who (a) challenge their own political opinion and (b) who have a different political attitude. We also tested this two-factor solution using parallel analysis and principal axis factoring with oblimin rotation (see Worthington & Whittaker, 2006). The factor loadings for attitude-consistent exposure were .96 and .98 and for attitude-challenging exposure .94 and .91. The eigenvalues for the two factors were 1.88 and 1.71 (numbers are based on the wave 1 data, but we also replicated these results for wave 2). Since attitude-consistent and attitude-challenging exposure are highly correlated, we control for both variables in the first two models. Based on these scales, we constructed a relative attitude-consistent exposure measure, which is calculated as incongruent exposure as a proportion of total opinion exposure (M_(t1) = 0.51, SD_(t1) = 0.09; M_(t2) = 0.51, SD_(t2) = 0.10). This relative score was calculated by: Attitude-consistent exposure/(attitude-consistent exposure + attitude-challenging exposure). This measure assesses the relative share of attitude-consistent exposure in individuals’ total opinion exposure. This gives us an idea about asymmetric increases in attitude-consistent relative to attitude-challenging exposure.

Ideology

We measured ideology (M = 4.06, SD = 1.32) with a standard item with higher values indicating right-wing ideology (1 = left, 7 = right). To assess the effect of extreme ideological positions at both ends, we estimated the quadratic effect of ideology (Lammers et al., 2017). Multiple-item batteries are not required for concepts that tap one single characteristic, that is, the left-right dimension can only be expressed with the words “left” and “right” (Bergkvist & Rossiter, 2009).

Political Involvement

We measured political interest (ρ = 0.92, M = 4.90, SD = 1.81) with two items asking how strongly respondents were interested in a) politics and b) the current election (1 = not interested, 7 = very interested). We conceptualize political participation as how much effort it takes an individual to perform the behavior, and thus distinguish between high- and low-effort activities (Knoll et al., 2020). To measure high-effort participation (M = 0.52, SD = 1.02), we asked participants whether they performed the following activities: (1) writing a longer political comment online (e.g., Facebook message, E-mail, and blog entry) to convince others with their arguments, (2) contacting a politician or journalist via email or social media to increase awareness of political issues, (3) creating a political group online (e.g., WhatsApp and Facebook) in order to increase awareness of political issues, (4) taking part in a demonstration or protest related to a political issue, (5) taking part in a political assembly to discuss political topics (e.g., community or school assembly), and (6) working for a political organization (political party, NGO, and school organization). To assess low-effort participation (M = 1.39, SD = 1.63), we asked participants whether they performed the following six activities: (1) liking or sharing a political post on social media, (2) adding a short comment to a political post on social media, (3) signing an online petition related to a political issue, (4) reminding others of a political event or engagement opportunity (e.g., voting and signing a petition), (5) using a campaign sticker, pen, bag or similar of a political party, and (6) signing a petition in the street. We measured political knowledge (M = 4.03, SD = 2.25) by exposing participants to four correct and four incorrect statements on issue positions of four major parties. Participants evaluated the statements (true, false, or do not know). We summed up correct responses, resulting in an additive index reaching from 0 to 8.

Network Characteristics

For social media network size (M = 3.98, SD = 2.96), we asked participants about the estimated total number of people they are connected with on SNSs. We used 11 predefined categories (lowest = 0, highest = 1000+ contacts), resulting in a numeric scale reaching from 1 to 11. To measure cross-class exposure (M = 3.30, SD = 1.89), we asked individuals how frequently they encounter social media posts from people who come from a different social class (1 = never, 7 = often). We measured participants’ political exposure (α = .76, M = 3.37, SD = 1.59) as how often participants encounter political posts from (1) friends, (2) media organizations, and (3) political actors in their newsfeed (1 = never, 7 = very often).

Control Variables

Partisan strength (M = 5.19, SD = 1.96) was measured by asking about a strong preference for one party or one candidate (1 = no preference, 7 = very strong preference). For SNS platform use, we asked how frequently participants use Twitter (M = 1.60, SD = 1.22), Facebook (M = 3.86, SD = 2.05), Instagram (M = 1.82, SD = 1.47), and YouTube (M = 3.58, SD = 1.71) (1 = never, 7 = often). We also included age in years, gender (reference category = male), and education (high college degree; medium college-bound high school degree; reference category = compulsory school degree).

Results

We ran OLS regression models with lagged dependent variables. We also included the autoregressive paths (i.e., attitude-consistent exposure at T1 as a predictor of attitude-consistent exposure at T2) in our analysis. Such autoregressive models allow us to explain changes between the dependent variables at T1 to T2 which are not explained by individuals’ T1 scores. Even though we are still relying on observational and self-reported data, this procedure can decrease common problems with cross-sectional surveys, such as reversed causality or sample bias (see Prior, 2005). The full regression results are shown in Table 1. A correlation table with all variables (except demographics) is provided in the appendix (Table A1). We ran three separate models, one for attitude-consistent exposure (model 1), one for attitude-challenging exposure (model 2), and one for relative attitude-consistent exposure (model 3). The robustness test revealed no signs of multicollinearity. Hypothesis 1 assumed that ideological extremity (at both ends) is associated with (a) increasing exposure to attitude-consistent content, (b) decreasing exposure to attitude-challenging content, and (c) increasing relative attitude-consistent exposure. H1a was confirmed, as indicated by the significant quadratic effect of ideology in model 1 (b = 0.072, p = .002). No significant support was found for H1b. However, one could also argue that individuals at the ideological ends are exposed to more political opinions in general. Thus, we also looked at relative attitude-consistent exposure and found support for H1c. Our results showed that individuals at the ideological ends are exposed to a higher share of attitude-consistent opinions (b = 0.004, p = .014). The nature of the quadratic effects is depicted in Figure 1. As Figure 1 indicates, the exposure to attitude-consistent SNS information in wave 2 is slightly more pronounced for individuals at the right ideological end compared to individuals at the left end. This trend is also indicated by the weak and close to significant linear effect of ideology in model 2 (b = 0.078, p = 0.08). This weak linear effect persists also when excluding the quadratic term from the model. Looking at the control variables in Table 1, we also find that the coefficient of partisan strength is weak and insignificant in all models.

Table 1.

Predicting (Relative) Attitude-Consistent and Attitude-Challenging Political Exposure on SNS.

	Attitude-consistent Exposure W2	Attitude-challenging Exposure W2	Relative Attitude-consistent Exposure^a W2
	b (SE)	b (SE)	b (SE)
Independent variables W1
Attitude-challenging exposure	−0.046 (0.051)	0.422^*** (0.056)
Attitude-consistent exposure	0.365^*** (0.051)	−0.115^* (0.056)
Relative attitude-consistent exposure			0.406^*** (0.043)
Left-right ideology
Ideology^b	0.078⁺ (0.045)	0.039 (0.049)	0.003 (0.003)
Ideology^2	0.072^** (0.023)	0.022 (0.026)	0.004^* (0.002)
Pol. Involvement
Pol. Interest	0.012 (0.042)	0.005 (0.046)	−0.000 (0.003)
Pol. Knowledge	0.107^*** (0.029)	0.039 (0.032)	0.004^* (0.002)
HE Participation^c	0.009 (0.067)	0.026 (0.074)	−0.001 (0.004)
LE Participation^c	0.124^** (0.046)	0.030 (0.050)	0.005⁺ (0.003)
Network
Network size	0.068^** (0.021)	0.060^* (0.023)	0.001 (0.001)
Cross-class exposure	−0.079⁺ (0.047)	0.029 (0.052)	−0.008^** (0.003)
Political exposure	0.190^** (0.059)	0.235^*** (0.065)	−0.001 (0.004)
Control variables
Age	−0.001 (0.005)	0.004 (0.006)	−0.000 (0.000)
Female	−0.026 (0.121)	−0.268^* (0.133)	0.015⁺ (0.008)
Medium education^d	0.041 (0.139)	0.046 (0.153)	0.000 (0.009)
High education^d	−0.019 (0.155)	−0.213 (0.171)	0.011 (0.010)
Partisan strength	0.025 (0.035)	0.054 (0.038)	−0.002 (0.002)
Facebook use	0.157^*** (0.033)	0.089^* (0.037)	0.005^* (0.002)
Twitter use	0.119^* (0.054)	0.042 (0.059)	0.005 (0.004)
YouTube use	−0.032 (0.039)	0.026 (0.043)	−0.004 (0.003)
Instagram use	−0.041 (0.047)	−0.003 (0.051)	−0.003 (0.003)
Constant	0.367 (0.386)	0.241 (0.425)	0.311^*** (0.032)
Adjusted R ²	0.474	0.393	0.199
N	559	559	559

+ p < .1; * p <.05; ** p < .01; *** p < .001.

^aRelative measure = Attitude-consistent exposure/(Attitude-consistent exposure + Attitude-challenging exposure).

^bIdeology is mean-centered.

^cHE = High-effort, LE = low-effort.

^dReference category: Low education.

Figure 1.

Quadratic effect of ideology on attitude-consistent exposure (left graph) and on relative attitude-consistent exposure (attitude-consistent exposure as share of total opinion exposure, right graph). Graphs are based on Table 1. Numerical covariates are set to their median, categorical covariates to their mode values.

RQ1 asked how different types of political involvement (interest, knowledge, and low- and high-effort participation) may affect attitude-consistent and attitude-challenging exposure, as well as relative attitude-consistent exposure. We found no significant effects in all three models for political interest. Political knowledge was associated with an increase in attitude-consistent exposure (b = 0.107, p = .0002) and in relative attitude-consistent exposure (b = 0.004, p = .02). High-effort participation showed no significant association with the outcome variables in all three models. Finally, low-effort political participation had a significant positive effect on attitude-consistent exposure (b = 0.124, p = .006) and had an effect close to significance on relative attitude-consistent exposure (b = 0.005, p = .07).²

Hypothesis 2 assumed that SNS users with large network size increase their exposure to (a) attitude-consistent content and (b) attitude-challenging content, and (c) decrease relative attitude-consistent exposure. We found evidence for H2a (b = 0.068, p = .001) and H2b (b = 0.060, p = .01). H2c was not supported, because larger networks do not significantly predict asymmetric exposure to attitude-consistent or attitude-challenging content. Hypothesis 3 assumed that cross-class exposure (a) decreases exposure to attitude-consistent content, (b) increases exposure to attitude-challenging content, and (c) decreases relative attitude-consistent exposure. As indicated in model 1 and model 2, we do not find significant effects of cross-class exposure on attitude-consistent and attitude-challenging exposure. Thus, H3a and H3b were rejected. However, H3c was supported, because we do find that cross-class exposure decreases relative attitude-consistent exposure (b = −0.008, p = .003). Hypothesis 4 proposed that frequent overall political exposure on SNS may increase (a) exposure to attitude-consistent content and (b) exposure to attitude-challenging content and (c) decrease relative attitude-consistent exposure. We find evidence for H4a and H4b. Political exposure in the network predicted an increase in attitude-consistent exposure (b = 0.190, p = .001) and attitude-challenging exposure (b = 0.235, p = .0003). However, H4c was rejected.

Discussion

We found that citizens with extreme ideological positions increased exposure to attitude-consistent information but not to attitude-challenging information. As a consequence, they also increased relative attitude-consistent exposure. These findings can be explained by the ideological rigidity (Lammers et al., 2017) and the preference for certainty (Boutyline & Willer, 2017) of individuals with extreme ideological views. These individuals may form simpler and more clustered categories of political information and thus engage in more black-and-white thinking than moderates. People at the ideological ends may also have a higher desire for clarity, certainty, and stability of worldviews compared to the political moderate. This particularly holds for the intensive phase of an election campaign which is characterized by the densification of political information. Political parties may aggressively advertise their campaigns, supporters express their opinions, and the media covers the campaigns more intensely (Iyengar & Simon, 2000). Individuals at the ideological ends may not only selectively expose themselves to political congruent information, they also engage in confirmation bias processes in which they reinforce and further strengthen their initial attitudes. Even though there is some evidence that individuals may still see a fair amount of attitude-challenging information on the aggregate level (see Bakshy et al., 2015), some groups of individuals may use these opportunities to avoid attitude-challenging information and engage primarily with attitude-consistent information. Considering the fact that people increasingly switch their political information activities to SNS, the content they are exposed to on these sites is key for their political attitudes and behavior.

We controlled the linear effect of ideology, and we can say that not left versus right, but extreme ideology at both ends explained relative attitude-consistent exposure. Yet, one may argue that the effects of ideology are driven by other related factors such as partisan strength (Brundidge, 2010; Messing & Westwood, 2014). However, we also controlled for partisan strength in our model. The variable did not yield any significant effects. These findings also indicate that ideology can be a more important predictor of relative attitude-consistent exposure on SNS than partisanship. A potential explanation for these results could be rooted in the multi-party system in Austria. The Austrian party system is very pluralized and the entry of new parties into the parliament due to political, economic, or social transformations has been frequently observed (Müller, 2000; Wagner & Kritzinger, 2012). As such, the electorate is very volatile and prone to strong swing-voting which arguably can decrease the degree of partisanship and also lessen the impact of partisanship on outcome variables. Thus, we may conclude that selective exposure is primarily driven by ideology, which is often confounded with partisan attitudes in two-party systems. Moreover, we found some first evidence that political involvement may affect individuals’ opinion exposure on SNS. For example, we found that political knowledge significantly increased both overall exposure to attitude-consistent content and relative attitude-consistent exposure. This is somewhat surprising since one could argue that individuals with higher political knowledge would have a stronger need to learn about different political views. At the same time, however, individuals with higher knowledge may have formed stable issue-specific attitudes and are more convinced about their mentally stored arguments. A more contextual explanation for this finding could be the transition from low-choice to high-choice media environments. In low-choice media environments with strong public broadcasting structures such as Austria (Hallin & Mancini, 2004), individuals with high political knowledge are widely exposed to rather balanced and pluralistic political news content (Skovsgaard et al., 2016). In a high-choice media environment such as SNS, these individuals are better equipped to find and appraise confirming information and selectively avoid attitude-challenging political information.

We also found evidence that low-effort (but not high-effort) political participation may increase relative attitude-consistent exposure. Low-effort activities are comprised of liking or quickly commenting a political post, or engage in low-effort offline activities. Such low-effort activities are based on little cognitive involvement and heuristic processes may come into play (see Knoll et al., 2020). For example, individuals may simply like a post or sign a petition, because they feel it is consistent with their personal views. In other words, they use attitude-conformity as a heuristic to engage in low-effort behavior. As a consequence, due to follow up processes (e.g., they may re-encounter the post they have previously liked or they may follow the process of the petition), they may re-engage with the attitude-consistent content they previously engaged with. At the same time, there is no need for low-effort engagement with attitude-challenging content. By contrast, if individuals engage in high-effort participation, such as by writing longer comments or contacting a politician, they may engage in more in-depth political thinking and may hence also encounter opposing political arguments. However, these findings are exploratory and need to be replicated.

Our findings showed that citizens’ opinion exposure is also determined by their social media network characteristics. Citizens who have a large network size and were more frequently exposed to political information increased their exposure to attitude-consistent and attitude-challenging information over time. The reason for this might be that these network characteristics encourage opinion exposure in general. Most weak ties may have attitude-consistent attitudes, leading to a rather proportional increase of exposure to attitude-consistent and challenging views. The same is true for political exposure in the network: A high frequency of political exposure may also involve incidental encounters, coming from diverse sources (Weeks et al., 2017). Furthermore, we observed that cross-class exposure decreased relative attitude-consistent exposure. The reason might be that individuals from different social classes have different political views, for example when it comes to economic redistribution (Van der Waal et al., 2007). Exposure to content shared by people with different social backgrounds may be important so that individuals become more tolerant towards opinions of different social groups.

Limitations and Future Research

Two-wave panel data afford advantages over cross-sectional data, for they cancel out the problem of reversed causation and also reduce problems associated with sample bias. However, also panel data are observational data and therefore limited. Experimental studies are needed to provide a more robust test for the causal relationships we are theorizing in this study. We rely on self-reported measures of opinion exposure. We calculate relative attitude-consistent exposure as the share of attitude-consistent opinion exposure within total opinion exposure. Since opinion encounters are not always stored in memory, our findings need to be replicated using in situ data (i.e., mobile experience sampling). Our key measures referred to SNS in general. However, political exposure and engagement may vary across different SNS platforms (Halpern et al., 2017; C. Kim & Lee, 2016). Thus, our findings need to be replicated for specific SNS platforms. We chose a time frame of 1 month, so replications with a larger time frame are warranted. However, the study was conducted during the hot phase of a national election campaign, so we can expect that political newsfeed curation is intense in this period, such as that social media users follow or de-follow political candidates, weak tie fellow citizens, or media organizations due to political posts they encounter in their newsfeed. Future research could also benefit from complementing survey designs with social network analysis (SNA). For instance, a panel survey could investigate different sets of individual predictors for attitude-consistent and attitude-challenging exposure. A SNA could examine how these individual predictors shape actual selective exposure clusters of the participants from the panel study on SNS, i.e., how participants with extreme ideological views or high political knowledge interact with others SNS users or how they are exposed to attitude-consistent and attitude-challenging content from other SNS users (see Himelboim et al., 2013). Finally, cross-country replications between the U.S. and European countries with different party systems, such as Great Britain, France, Italy, or Spain, would be in order (Dimitrova & Matthes, 2018). Such studies could not only reveal potential differences between two-party and multi-party systems, but also differences within subtypes of multi-party systems.

Conclusion

SNS have created high-choice media environments, in which users personalize their news experience by curating their newsfeed according to their preferences. Such personalization can be beneficial on many aspects but has also drawn several public concerns. Among them is the selection of attitude-consistent political information and the avoidance of attitude-challenging political information (Skoric et al., 2021; Stroud, 2018). Although previous research has enhanced our knowledge about echo chambers on SNS, we lack a systematic analysis of individual predictors of exposure to attitude-challenging and attitude-consistent information on SNS. In this study, we addressed this research gap by investigating how ideology, cognitive, and behavioral forms of political involvement, and different network characteristics are associated with exposure to attitude-challenging and attitude-consistent information, as well as relative attitude-consistent exposure on SNS. Our findings demonstrated that extreme ideology, political knowledge, and low-effort political participation increased attitude-consistent and relative attitude-consistent exposure over time. Only cross-class exposure predicted a decrease in attitude-consistent and relative attitude-consistent exposure.

Our study contributes to previous research by clearly indicating that individual predictors do matter. Such individual predictors may not just be related to partisanship, as most previous U.S.-focused research suggested, but may also encompass ideology, cognitive and behavioral forms of political involvement, and network characteristics. In fact, our findings demonstrate that partisanship may not be related to (relative) attitude-consistent and attitude-challenging exposure. This could be explained by the pluralization of the Austrian party system since the late 1980s which made voting decisions more volatile and thus lessened the degree of partisanship (Grotz & Weber, 2012). In turn, the Austrian party system was transformed by the strong rise of right-wing populist parties which increased the degree of extreme ideological positions (Wagner & Meyer, 2017). This development would also explain why ideological extremity at both ends rather than partisanship predicts (relative) attitude-consistent exposure. Similarly, our findings highlight the importance of contextual factors, such as different party systems, which can influence (relative) attitude-consistent and attitude-challenging exposure on SNS. Certain predictors which may be relevant in a two-party system, such as partisanship in the U.S., may not matter in a multi-party system. To gain further evidence, comparative studies between countries with different party systems are warranted (Dimitrova & Matthes, 2018).

Our research also contributes to the previous research by using panel data, which are more appropriate than cross-sectional data to understand how individual characteristics induce changes in (relative) attitude-consistent and attitude-challenging exposure on SNS over time (Hsiao, 2014). This particularly holds for studies in a campaign context. Our findings indicate that over the course of a campaign and with the election date approaching, certain individual predictors can fuel, while other individual predictors can diminish a relative attitude-consistent information environment on SNS. Although more evidence is needed, we may speculate that those factors that can foster participatory democracy, such as political knowledge and low-effort participation, and factors that can increase ideological and affective polarization, such as ideological extremity, may necessarily contribute to a relative attitude-consistent information environment on SNS. In contrast, predictors such as cross-class exposure, which can reduce ideological and affective polarization, may attenuate a relative attitude-consistent environment on SNS. Overall, our findings clearly demonstrate that SNS do not per se contribute to a relative attitude-consistent information environment but that individual predictors in this context matter.

Data Availability

Data are available under OSF: https://osf.io/pjdk7/?view_only=55f7e13f27714ea49d7b741cd30ed792.

Appendix ATable A1. Correlation Table.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1 Partisan strength
2 Facebook	0.073
3 Twitter	0.003	0.168^***
4 YouTube	−0.113^**	0.270^***	0.354^***
5 Instagram	−0.071	0.305^***	0.424^***	0.364^***
6 Relative attitude-consistent exposure	−0.033	−0.024	0.000	−0.046	−0.044
7 Attitude-challenging exposure	0.248^***	0.281^***	0.155^***	0.119^**	0.126^**	−0.453^***
8 Attitude-consistent exposure	0.313^***	0.372^***	0.180^***	0.153^***	0.124^**	−0.382^***	0.701^***
9 Ideology	0.109^*	0.131^**	−0.051	−0.033	−0.085^*	0.093^*	−0.071	0.041
10 Political interest	0.493^***	−0.001	0.063	−0.032	−0.031	−0.034	0.334^***	0.302^***	0.027
11 Political knowledge	0.260^***	−0.114^**	0.014	−0.058	−0.126^**	0.063	0.177^***	0.137^**	−0.038	0.438^***
12 High-effort participation	0.155^***	0.101^*	0.155^***	0.100^*	0.133^**	−0.070	0.236^***	0.301^***	0.028	0.210^***	0.036
13 Low-effort participation	0.263^***	0.178^***	0.193^***	0.116^**	0.126^**	−0.104^*	0.382^***	0.460^***	0.004	0.319^***	0.145^***	0.548^***
14 Network size	0.103^*	0.303^***	0.088^*	0.106^*	0.217^***	−0.029	0.297^***	0.308^***	0.045	0.110^**	−0.041	0.226^***	0.242^***
15 Cross-class exposure	0.222^***	0.314^***	0.181^***	0.129^**	0.147^***	−0.335^***	0.723^***	0.670^***	0.015	0.301^***	0.125^**	0.277^***	0.391^***	0.310^***
16 Political exposure	0.307^***	0.382^***	0.168^***	0.151^***	0.151^***	−0.223^***	0.651^***	0.724^***	0.090^*	0.389^***	0.145^***	0.300^***	0.459^***	0.354^***	0.668^***

*p <0.05; **p < 0.01; ***p < 0.001.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jörg Matthes

Notes

Appendix

Author Biographies

Franz Reiter is a Ph.D., candidate at the Department of Communication at the University of Vienna. His research interests include media contents and effects, political communication, and dirty campaigning.

Raffael Heiss (Ph.D., University of Vienna) is lecturer and postdoctoral researcher at the MCI Center for Social and Health Innovation. His research interests include public health, communication, digital media, and civic engagement.

Jörg Matthes (Ph.D., University of Zurich) is a professor in communication science and director of the Department of Communication at the University of Vienna. His research interests include political communication, digital media, advertising, and quantitative methods.

References

Bakshy

Messing

Adamic

L. A.

(2015). Exposure to ideologically diverse news and opinion on Facebook. Science, 348(6239), 1130–1132. https://doi.org/10.1126/science.aaa1160.

Beam

M. A.

Hutchens

M. J.

Hmielowski

J. D.

(2018). Facebook news and (de)polarization: Reinforcing spirals in the 2016 US election. Information, Communication & Society, 21(7), 940–958. https://doi.org/10.1080/1369118X.2018.1444783.

Beam

M. A.

Kosicki

G. M.

(2014). Personalized news portals. Journalism & Mass Communication Quarterly, 91(1), 59–77. https://doi.org/10.1177/1077699013514411.

Bergkvist

Rossiter

J. R.

(2009). Tailor-made single-item measures of doubly concrete constructs. International Journal of Advertising, 28(4), 607–621. https://doi.org/10.2501/S0265048709200783.

Boutyline

Willer

(2017). The social structure of political echo chambers: Variation in ideological homophily in online networks. Political Psychology, 38(3), 551–569. https://doi.org/10.1111/pops.12337.

Brundidge

(2010). Encountering “difference” in the contemporary public sphere: The contribution of the internet to the heterogeneity of political discussion networks. Journal of Communication, 60(4), 680–700. https://doi.org/10.1111/j.1460-2466.2010.01509.x.

Crotty

(2009). Party transformations: The United States and Western Europe. In Katz

R. S.

Crotty

(Eds.), Handbook of party politics (pp. 499–514). SAGE.

Dahlgren

P. M.

Shehata

Strömbäck

(2019). Reinforcing spirals at work? Mutual influences between selective news exposure and ideological leaning. European Journal of Communication, 34(2), 159–174. https://doi.org/10.1177/0267323119830056.

DeVito

M. A.

(2017). From editors to algorithms: A values-based approach to understanding story selection in the Facebook news feed. Digital Journalism, 5(6), 753–773. https://doi.org/10.1080/21670811.2016.1178592.

10.

Dimitrova

D. V.

Matthes

(2018). Social media in political campaigning around the world: Theoretical and methodological challenges. Journalism & Mass Communication Quarterly, 95(2), 333–342. https://doi.org/10.1177/1077699018770437.

11.

Donsbach

Traugott

M. W.

(2008). The SAGE handbook of public opinion research. SAGE.

12.

Dunbar

R. I. M.

(2016). Do online social media cut through the constraints that limit the size of offline social networks? Royal Society Open Science, 3(1), 150292. https://doi.org/10.1098/rsos.150292.

13.

Ellison

N. B.

Steinfield

Lampe

(2007). The benefits of Facebook “Friends:” Social capital and college students’ use of online social network sites. Journal of Computer-Mediated Communication, 12(4), 1143–1168. https://doi.org/10.1111/j.1083-6101.2007.00367.x.

14.

Festinger

(1957). A theory of cognitive dissonance. Stanford University Press.

15.

Fletcher

Nielsen

R. K.

(2017). Are news audiences increasingly fragmented? A cross-national comparative analysis of cross-platform news audience fragmentation and duplication. Journal of Communication, 67(4), 476–498. https://doi.org/10.1111/jcom.12315.

16.

Garrett

R. K.

(2009a). Echo chambers online?: Politically motivated selective exposure among Internet news users. Journal of Computer-Mediated Communication, 14(2), 265–285. https://doi.org/10.1111/j.1083-6101.2009.01440.x.

17.

Garrett

R. K.

(2009b). Politically motivated reinforcement seeking: Reframing the selective exposure debate. Journal of Communication, 59(4), 676–699. https://doi.org/10.1111/j.1460-2466.2009.01452.x.

18.

Groshek

Koc-Michalska

(2017). Helping populism win? Social media use, filter bubbles, and support for populist presidential candidates in the 2016 US election campaign. Information, Communication & Society, 20(9), 1389–1407. https://doi.org/10.1080/1369118X.2017.1329334.

19.

Grotz

Weber

(2012). Party systems and government stability in Central and Eastern Europe. World Politics, 64(4), 699–740. https://doi.org/10.1017/S0043887112000196.

20.

Hallin

D. C.

Mancini

(2004). Comparing media systems: Three models of media and politics. Cambridge University Press. https://doi.org/10.1017/CBO9780511790867.

21.

Halpern

Valenzuela

Katz

J. E.

(2017). We face, I tweet: How different social media influence political participation through collective and internal efficacy. Journal of Computer-Mediated Communication, 22(6), 320–336. https://doi.org/10.1111/jcc4.12198.

22.

Hampton

K. N.

Lee

C.-j.

Her

E. J.

(2011). How new media affords network diversity: Direct and mediated access to social capital through participation in local social settings. New Media & Society, 13(7), 1031–1049. https://doi.org/10.1177/1461444810390342.

23.

Hill

R. A.

Dunbar

R. I. M.

(2003). Social network size in humans. Human Nature, 14(1), 53–72. https://doi.org/10.1007/s12110-003-1016-y.

24.

Himelboim

Smith

Shneiderman

(2013). Tweeting apart: Applying network analysis to detect selective exposure clusters on twitter. Communication Methods and Measures, 7(3), 169–197. https://doi.org/10.1080/19312458.2013.813922.

25.

Hsiao

(2014). Analysis of panel data (3rd ed.). Cambridge University Press.

26.

Iyengar

Hahn

K. S.

(2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication, 59(1), 19–39. https://doi.org/10.1111/j.1460-2466.2008.01402.x.

27.

Iyengar

Hahn

K. S.

Krosnick

J. A.

Walker

(2008). Selective exposure to campaign communication: The role of anticipated agreement and issue public membership. The Journal of Politics, 70(1), 186–200. https://doi.org/10.1017/S0022381607080139.

28.

Iyengar

Simon

A. F.

(2000). New perspectives and evidence on political communication and campaign effects. Annual Review of Psychology, 51(1), 149–169. https://doi.org/10.1146/annurev.psych.51.1.149.

29.

Jamieson

K. H.

Cappella

J. N.

(2010). Echo chamber: Rush Limbaugh and the conservative media establishment. Oxford University Press.

30.

Kim

(2011). The contribution of social network sites to exposure to political difference: The relationships among SNSs, online political messaging, and exposure to cross-cutting perspectives. Computers in Human Behavior, 27(2), 971–977. https://doi.org/10.1016/j.chb.2010.12.001.

31.

Kim

Chen

H.-T.

(2016). Social media and online political participation: The mediating role of exposure to cross-cutting and like-minded perspectives. Telematics and Informatics, 33(2), 320–330. https://doi.org/10.1016/j.tele.2015.08.008.

32.

Kim

Lee

J. K.

(2016). Social media type matters: Investigating the relationship between motivation and online social network heterogeneity. Journal of Broadcasting & Electronic Media, 60(4), 676–693. https://doi.org/10.1080/08838151.2016.1234481.

33.

Knobloch-Westerwick

Jingbo Meng

(2009). Looking the other way. Communication Research, 36(3), 426–448. https://doi.org/10.1177/0093650209333030.

34.

Knoll

Matthes

Heiss

(2020). The social media political participation model: A goal systems theory perspective. Convergence, 26(1), 135–156. https://doi.org/10.1177/1354856517750366.

35.

Lammers

Koch

Conway

Brandt

M. J.

(2017). The political domain appears simpler to the politically extreme than to political moderates. Social Psychological and Personality Science, 8(6), 612–622. https://doi.org/10.1177/1948550616678456.

36.

Lazarsfeld

P. F.

Berelson

Gaudet

(1944). The people’s choice: How the voter makes up his mind in a presidential campaign. Columbia University Press.

37.

Lee

J. K.

Choi

Kim

(2014). Social media, network heterogeneity, and opinion polarization. Journal of Communication, 64(4), 702–722. https://doi.org/10.1111/jcom.12077.

38.

Leeper

T. J.

Slothuus

(2014). Political parties, motivated reasoning, and public opinion formation. Political Psychology, 35(1–3), 129–156. https://doi.org/10.1111/pops.12164.

39.

Lefkofridi

Wagner

Willmann

J. E.

(2014). Left-authoritarians and policy representation in Western Europe: Electoral choice across ideological dimensions. West European Politics, 37(1), 65–90. https://doi.org/10.1080/01402382.2013.818354.

40.

Long

J. A.

Eveland

W. P.

Slater

M. D.

(2019). Partisan media selectivity and partisan identity threat: The role of social and geographic context. Mass Communication and Society, 22(2), 145–170. https://doi.org/10.1080/15205436.2018.1504302.

41.

Mair

(2009). Cleavages. In Katz

R. S.

(Ed.), Handbook of party politics (pp. 371–375). SAGE.

42.

Matthes

Knoll

Valenzuela

Hopmann

D. N.

von Sikorski

(2019). A meta-analysis of the effects of cross-cutting exposure on political participation. Political Communication, 36(4), 523–542. https://doi.org/10.1080/10584609.2019.1619638.

43.

Matthes

Marquart

von Sikorski,

(2020). Like-minded and cross-cutting talk, network characteristics, and political participation online and offline: A panel study. Communications, 46(1), 113–126. https://doi.org/10.1515/commun-2020-2080.

44.

Matuszewski

Szabó

(2019). Are Echo Chambers Based on Partisanship? Twitter and Political Polarity in Poland and Hungary. Social Media + Society. https://doi.org/10.1177/2056305119837671.

45.

Messing

Westwood

S. J.

(2014). Selective exposure in the age of social media. Communication Research, 41(8), 1042–1063. https://doi.org/10.1177/0093650212466406.

46.

Möller

Trilling

Helberger

van Es

(2018). Do not blame it on the algorithm: an empirical assessment of multiple recommender systems and their impact on content diversity. Information, Communication & Society, 21(7), 959–977. https://doi.org/10.1080/1369118X.2018.1444076.

47.

Müller

W. C.

(2000). The Austrian election of October 1999: A shift to the right. West European Politics, 23(3), 191–200. https://doi.org/10.1080/01402380008425391.

48.

Mutz

D. C.

(2002a). Cross-cutting social networks: Testing democratic theory in practice. American Political Science Review, 96(1), 111–126. https://doi.org/10.1017/S0003055402004264.

49.

Mutz

D. C.

(2002b). The consequences of cross-cutting networks for political participation. American Journal of Political Science, 46(4), 838. https://doi.org/10.2307/3088437.

50.

Nanz

Heiss

Matthes

(2020). Antecedents of intentional and incidental exposure modes on social media and consequences for political participation: A panel study. Acta Politica. https://doi.org/10.1057/s41269-020-00182-4.

51.

Prior

(2005). News vs. Entertainment: How increasing media choice widens gaps in political knowledge and turnout. American Journal of Political Science, 49(3), 577–592. https://doi.org/10.2307/3647733.

52.

Rodriguez

C. G.

Moskowitz

J. P.

Salem

R. M.

Ditto

P. H.

(2017). Partisan selective exposure: The role of party, ideology and ideological extremity over time. Translational Issues in Psychological Science, 3(3), 254–271. https://doi.org/10.1037/tps0000121.

53.

Rogowski

J. C.

Sutherland

J. L.

(2016). How ideology fuels affective polarization. Political Behavior, 38(2), 485–508. https://doi.org/10.1007/s11109-015-9323-7.

54.

Scheufele

D. A.

Shanahan

Kim

S.-H.

(2002). Who cares about local politics? Media influences on local political involvement, issue awareness, and attitude strength. Journalism & Mass Communication Quarterly, 79(2), 427–444. https://doi.org/10.1177/107769900207900211.

55.

Skoric

M. M.

Zhu

Koc-Michalska

Boulianne

Bimber

(2021). Selective avoidance on social media: A comparative study of western democracies. Social Science Computer Review, 1–18. https://doi.org/10.1177/08944393211005468.

56.

Skovsgaard

Shehata

Strömbäck

(2016). Opportunity structures for selective exposure. The International Journal of Press/Politics, 21(4), 527–546. https://doi.org/10.1177/1940161216658157.

57.

Song

Eveland

W. P.

(2015). The structure of communication networks matters: How network diversity, centrality, and context influence political ambivalence, participation, and knowledge. Political Communication, 32(1), 83–108. https://doi.org/10.1080/10584609.2014.882462.

58.

Stroud

N. J.

(2010). Polarization and partisan selective exposure. Journal of Communication, 60(3), 556–576. https://doi.org/10.1111/j.1460-2466.2010.01497.x.

59.

Stroud

N. J.

(2018). Selective exposure theories. In Kenski

Jamieson

K. H.

Stroud

N. J.

(Eds.), Selective exposure theories - Oxford handbooks (pp. 1–19). Oxford University Press.

60.

Sunstein

C. R.

(2007). Republic.com 2.0. Princeton University Press.

61.

Tang

Lee

F. L. F.

(2013). Facebook Use and Political Participation: The Impact of Exposure to Shared Political Information, Connections With Public Political Actors, and Network Structural Heterogeneity. Social Science Computer Review, 31(6), 763–773. https://doi.org/10.1177/0894439313490625.

62.

Vaccari

Valeriani

Barberá

Jost

J. T.

Nagler

Tucker

J. A.

(2016). Of echo chambers and contrarian clubs: Exposure to political disagreement among German and Italian users of Twitter. Social Media + Society, 2(3), 1–24. https://doi.org/10.1177/2056305116664221.

63.

Valentino

N. A.

Brader

Groenendyk

E. W.

Gregorowicz

Hutchings

V. L.

(2011). Election night’s alright for fighting: The role of emotions in political participation. The Journal of Politics, 73(1), 156–170. https://doi.org/10.1017/S0022381610000939.

64.

Van der Waal

Achterberg

Houtman

(2007). Class is not dead-it has been buried alive: Class voting and cultural voting in postwar western societies (1956-1990). Politics & Society, 35(3), 403–426. https://doi.org/10.1177/0032329207304314.

65.

Visser

Lubbers

Kraaykamp

Jaspers

(2014). Support for radical left ideologies in Europe. European Journal of Political Research, 53(3), 541–558. https://doi.org/10.1111/1475-6765.12048.

66.

Vraga

E. K.

Tully

(2021). News literacy, social media behaviors, and skepticism toward information on social media. Information, Communication & Society, 24(2), 150–166. https://doi.org/10.1080/1369118X.2019.1637445.

67.

Wagner

Kritzinger

(2012). Ideological dimensions and vote choice: Age group differences in Austria. Electoral Studies, 31(2), 285–296. https://doi.org/10.1016/j.electstud.2011.11.008.

68.

Wagner

Meyer

T. M.

(2017). The radical right as niche parties? The Ideological landscape of party systems in Western Europe, 1980--2014. Political Studies, 65(1), 84–107. https://doi.org/10.1177/0032321716639065.

69.

Weeks

B. E.

Lane

D. S.

Kim

D. H.

Lee

S. S.

Kwak

(2017). Incidental exposure, selective exposure, and political information sharing: Integrating online exposure patterns and expression on social media. Journal of Computer-Mediated Communication, 22(6), 363–379. https://doi.org/10.1111/jcc4.12199.

70.

Winter

Metzger

M. J.

Flanagin

A. J.

(2016). Selective use of news cues: A multiple-motive perspective on information selection in social media environments. Journal of Communication, 66(4), 669–693. https://doi.org/10.1111/jcom.12241.

71.

Worthington

R. L.

Whittaker

T. A.

(2006). Scale Development Research: A Content Analysis and Recommendations for Best Practices. The Counseling Psychologist, 34(6), 806–838. https://doi.org/10.1177/0011000006288127.

72.

Zhang

Johnson

T. J.

Seltzer

Bichard

S. L.

(2010). The Revolution Will be Networked: The Influence of Social Networking Sites on Political Attitudes and Behavior. Social Science Computer Review, 28(1), 75–92. https://doi.org/10.1177/0894439309335162.

73.

Zhu

Skoric

Shen

(2017). I shield myself from thee: Selective avoidance on social media during political protests. Political Communication, 34(1), 112–131. https://doi.org/10.1080/10584609.2016.1222471.

74.

Zuiderveen Borgesius

F. J.

Trilling

Möller

Bodó

de Vreese

C. H.

Helberger

(2016). Should we worry about filter bubbles? Internet Policy Review, 5(1), 1–16. https://doi.org/10.14763/2016.1.401.

Explaining Attitude-Consistent Exposure on Social Network Sites: The Role of Ideology,Political Involvement,and Network Characteristics

Abstract

Keywords

Personalization of Political Information Environments

Predictors of Attitude-Consistent Exposure

The Role of Ideology

The Role of Political Involvement

Network Characteristics Hypotheses

Network Size

Cross-Class Exposure

Political Exposure on SNS

Method

Sample

Measures

Opinion Exposure

Ideology

Political Involvement

Network Characteristics

Control Variables

Results

Discussion

Limitations and Future Research

Conclusion

Data Availability

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iD

Notes

Appendix

Author Biographies

References