Voting in the Echo Chamber? Patterns of Political Online Activities and Voting Behavior in Switzerland

Understanding the political consequences of digitalization is among key challenges for modern societies. A pressing issue question whether online activities make individuals more close-minded and less willing to consider alternative arguments. We examine this using a peculiarity Swiss electoral system – possibility split votes as behavioral outcome measure. argue that might either likely (“echo chamber”-argument) or spread their across parties (“deliberation”-argument). Empirically, we use data from Election Study Selects 2019 test these The results hierarchical logistic regression analysis do not support any conflicting Yet, additional analyses suggest interest moderates relationship between vote splitting: makes activists votes. Eine grosse Herausforderung für moderne Gesellschaften besteht darin, die politischen Folgen der Digitalisierung zu verstehen. Zentral ist hierbei Frage, ob ausgeübte politische Aktivitäten dazu führen, dass Menschen weniger offen andere Meinungen und Argumente sind. Wir nutzen eine Besonderheit des Schweizer Wahlsystems, um diese Frage untersuchen: Möglichkeit Stimmensplittings („Panaschieren“). Somit können wir untersuchen, im Netz das Verhalten an Wahlurne zusammenhängen. argumentieren, entweder Wahrscheinlichkeit verringern („Echokammer“-Argument) oder erhöhen („Deliberation“-Argument) können. Empirisch überprüfen anhand Daten Wahlstudie 2019. Die Ergebnisse einer hierarchischen logistischen Regressionsanalyse stützen keines gegensätzlichen Argumente. Zusätzliche Analysen deuten jedoch darauf hin, politisches Interesse Beziehung zwischen ausgeübten Stimmensplitting moderiert: Wer politisch interessiert gleichzeitig aktiv ist, eher geneigt, Stimmen splitten. Comprendre les conséquences politiques de la numérisation est un défi important pour sociétés modernes. Une urgente savoir si activités en ligne rendent gens moins ouverts à opinions et arguments divergents. Nous examinons cette nous appuyant sur une particularité du système électoral suisse - le panachage, c’est-à-dire répartition voix entre plusieurs partis. soutenons que l’activisme politique peut soit rendre enclins répartir leurs (argument « chambre d’écho »), plus au panachage délibération »). Les résultats d’une analyse statistique basée données l'étude électorale ne corroborent aucun deux Néanmoins, suggèrent l’intérêt modère lien l’activité panachage. Within past years, social media Internet generally have become increasingly relevant source information arena (Shin & Thorson, 2017). While almost everyone everyday matters on daily basis, it has also overtaken other (Latzer al., 2020). Particularly during election campaigns, voters platforms gain about candidates express own positions. It argued medium “that provides public with needs quicker, cheaper, in convenient form […] change patterns behavior” (Tolbert McNeal, 2003: 175). In times debates increasing polarization, what changes induced by relevance look like they strengthen rather undermine democratic processes. More precisely, how relate attitudes behavior? Do lead echo chambers create closed mindsets ultimately increase opinion polarization? Are no longer open considering standpoints, “hear side”? Or deliberation openness viewpoints? Finally, does all play out behavior at ballot box? address issues questions studying link splitting national elections If increased voters’ attitudinal segregation but those who consume news not, would be serious concern (Flaxman 2016). Knowing, processing, deliberating divergent views can considered core element liberal democracy (Downs, 1957; Huckfeldt 2004; Mutz, 2006). Accordingly, citizens build far selection specific, maybe biased, affects formation, long history research (e.g., Festinger, Fischer 2005; Zaller, 1992). was then expected encourage creation filter bubbles, fears accentuated recent shift towards (Allcott Gentzkow, On contrary, there hope will opposing accessible (Garrett 2013; Shaw Benkler, 2012). This so assumption could overcome divides. Even though numerous studies asked changed citizens’ diet formation worse better, existing knowledge still inconclusive (Dubois Blank, 2018). Against background, our study goes beyond previous four ways. First, while many articles focus consequences, box. Considering system, actively engage politics are different parties. unique factor (Selb Lutz, 2015), which seen maximum variant preferential voting (Lutz, 2011), novel voter may various reasons, e.g., strategically, because she undecided two knows likes specific personalities parties, expect likelihood related voter’s degree positions vein, wrapped up his preferred party’s chamber very cast another party. Conversely, been exposed perceive several least single eligible and, thus, splitting. Second, most evidence stems United States Great Britain. Due majoritarian systems, societies prone polarization. Thus, Switzerland prototype consensus indicate extent earlier findings generalized contexts. Third, typically focused Facebook Twitter (Bakshy 2015; Barberá 2017; Himelboim 2013). These vary concerning polarization contribute chambers. Moreover, supply-side perspective neglects receive sources Focusing engagement individual independently enables us general measurement activities. Lastly, phenomenon one’s under-researched topic case. From comparative perspective, addressed distinguishing horizontal (i.e., when equivalent offices contested) vertical the, regional level) (Burden Helmke, 2009).1 within same representative body, i.e., lower parliament, only discussed institutional descriptive (Bühlmann 2016; Linder Mueller, To best knowledge, exists examining determinants (in context).2 Hence, present Switzerland. assess splitting, panel module includes comprehensive measure respondents’ discussing online, sharing commenting information), identify activists.3 set camps. serve dependent variable. thereby assume camps polarized mindset. were systematically votes, indication Our conclusion. There overall non-finding covers group-specific pattern driven level interest. Whereas significantly low politics, interested, positively associated section, theoretical considerations arguing differ choose party lists (“panaschieren”) compared online. following, mechanisms explain such difference: First literature (see, Spierings Jacobs, 2014), difference choosing result varying campaign effects behave differently different, namely resources attitudes. second particularly define (“online activists”). passively consuming sufficient. Rather, interested share, post comment content. Therefore, integrate activism. elaborate role campaigns outcomes intensively studied (Rady Johnston, 2006; Schmitt-Beck Farrell, 2002). Campaigns supposed influence mainly ways (Brady Dermont Stadelmann-Steffen, 2018; Holbrook McClurg, 2005): unfold persuasive effect, new based potentially opinion. mobilizing sensitizing importance (for party). both depending campaign.4 background growing Internet, effect communication received scholarly (and public) attention over last decades. context, some scholars warned early that, exposure strongly depend ideology-based selectivity confirmation bias where congruent initial view (Galston, 2003; Sunstein, 2001). point expectation environment previously identified structural motivational leading selective motivated reasoning (Huckfeldt Taber Lodge, years further One reason way information. beginning era, needed seek media, literally brought them pre-selected. Several authors emphasize consumption depends friends content see algorithms 2015). contrast, reading discrepant means searched explicitly, something minority chooses latter becomes especially unlikely, given often aware (Gillani Furthermore, structure (online) technologies, mostly biased much easily few persons reach broader partisan audience fact-checking effective tool reduce misinformation (Fridkin Wood Porter, 2019), real-world scenario, little help, since well. Shin Thorson (2017), show “partisan sharing” small highly active group users potential polarize even trust fact-checking. However, idea stronger uncontested. Some opportunities environments foster deliberation. particular, supply (diverse) counter-attitudinal just larger context 2018: 730), facilitate cross-cutting dialog (Shaw 2012), well-functioning 2004). case, according strand literature, greater choice necessarily but, conversely, assumed pro- correlated 2013), For US Garrett co-authors (2013) indeed find digital millennium led “turn avoidance”, explicit one-sided tend affecting voters, motivating actually (Holbrook 2005). When comes mobilization differs traditional offline activities, one prominent expectations recently, offer personalized cost-effective (Aldrich scope effectiveness mixed. personal contact, door-to-door mobilization, “gold standard” mobilize reaching smaller than methods others due costs, (Vaccari, Magalhães al. (2020) suggests effective, increases turnout 2003). Interestingly, however, boost greatest mobilized offline. supports notion Vaccari (2017) happen vacuum. already engaged. above focuses broadly, article role. people participatory non-activists. analyze participation (Anduiza 2010; Best Krueger, Feezell, 2016) (Krueger, 2006) education, income, age directly mobilization. classical interest, access, skills, turn strong predictors nourish inequalities 2017), require somewhat skills course former computer-based 2010: 364). Related, attributes likely. Kaye Johnson (2002) document trust, efficacy, partisanship search Swigger concludes basic values younger cohorts. conclusive differential “symptom” attitudes, engaged reached fundamental values. What discussion shown “offline counterparts” affected effects, homogeneous persuasion particular involved candidates. groups ex-ante, (or less) yet differences preferences eventually, behavior. example, Strandberg Finnish channels, its impact lower. suitable investigate Switzerland, elected proportional rule, adapt (party) ways.5 They candidate list, put list. concurrent, 2009). conceived proxy open-mindedness captures willingness ideological candidate-related demonstrates deprive proceed applying elaborated formulate hypotheses summarizing importantly, world, reasons believe occur via channels newspapers, TV, radio. One-sided (Chong Druckman, 2007; Matthes, information, conviction position “true” one. As result, should next context. respects. Parties sure ballots first step concentrate step. Additionally, possibilities incentivizes launch campaigns. features regard effects. Hitherto, documented Britain, systems countries system. Candidates establishing links and/or making themselves visible chances base Tresch 2020: 53). Using tools seems attractive cheaper campaigning (Magalhães 2020; Vaccari, through personalized, chamber”-argument). H1: Political H2: “analog” counterparts raise predispositions affect purpose. moderate Two aspects person’s stance need distinguished. hand, enjoy opinions, benefit “high-choice environment” 731) offers. Dubois Blank (2018) decreases users’ end chamber. involve contrasting mechanism. supporters, purpose partisans known (Bartels, 2000; 2007). Citizens extreme ideologies higher follow sites blogs preexisting viewpoints. H3a: levels H3b: negatively extremity ideology. H3c: strength partisanship. (2020). Starting May (wave 1: 20th 8th July respondents surveyed three survey. Wave 2 fielded 2: 2nd September 17th October 2019) wave 3 launched after day 3: 21st 9th December 2019). aims observe campaign. Its sample random drawn Federal Statistical Office sampling frame SRPH. 7’939 interviews completed represents response rate 31 percent. third wave, 5’125 (65 percent wave) (Tresch analysis, combine 1 3, sociodemographic variables included 1, 3. add macro cantonal survey data. include number seats National Council, allocated cantons main language spoken canton. Since variable, possible seats, exclude seat (Uri, Obwalden, Nidwalden, Glarus, Appenzell Ausserrhoden, Innerrhoden). deletion case-wise observations missing leaves 2’948 20 analysis. variable dichotomous capturing modified list Figure depicts share per average, every vote. variation implies model accounting nested reasonable (see below) mere function canton’s size thus practical Vote Note: Dashed line illustrates mean; calculations measuring citiz