Orchestrating the Beat of Discord: How Citizens’ Understandings of Democracy Shape the Linkage Between Party System Polarization and Electoral Behavior

The impact of party system polarization on democracy has received much academic attention. Yet, whereas the conventional account suggests that heightened party competition improves turnout (i.e., the more voters receive strong and clear cues stressing parties’ ideological differences, the more they can find a party expressing their preferences and see the utility of participation), the mechanisms that may combine with spatial polarization to shape electoral behavior have garnered less scrutiny. This paper contends that the linkage between party system polarization and voting behavior is mediated by citizens’ understandings of democracy. More specifically, drawing on the 'group-based democracy' model proposed by Achen and Bartels (i.e., the notion that democratic claims are rooted in identity aspirations specific to social and political groups) (2016), it is argued that party polarization on the issue of popular sovereignty has a differential impact on voting depending on the model of democracy citizens support. Within the rapidly developing literature on layman’s understandings of democracy, empirical evidence points to a widespread support of the populist model (emphasizing popular sovereignty and direct democracy), especially among citizens dissatisfied with the liberal model (emphasizing rule of law and protection for individual liberty). Populist challenger parties have strong incentives to attract those voters disgruntled with the way democracy works by positioning themselves as entrepreneurs on democratic issues. This is likely to result in heightened party system polarization, possibly leading to increased turnout and significant electoral gains for populist parties (H1). It can also be anticipated that party polarization on the issue of popular sovereignty will differentially appeal to voters endorsing liberal and populist models of democracy. In particular, citizens supportive of the liberal (populist) model are expected to be more inclined to vote for non-populist (populist) parties when there is a high level of polarization on popular sovereignty (H2). The following methodological approach is devised to address these assumptions. First, Chapel Hill Expert Survey data is used to capture party stances on whether 'the people' should have the final say on the most important issues. Dalton’s canonical formula (2008) is then applied to measure party polarization on popular sovereignty at the country level. Second, based on opinion data from the European Social Survey (round 10) and expert data from the PopuList, it is confirmed that Europeans endorsing liberal and populist models of democracy exhibit different voting patterns. Finally, logistic multilevel models lend empirical credence to the assumption that the probability of voting for populist parties increases with party polarization on popular sovereignty (H1). Still, this tendency is significantly more pronounced among citizens who hold strong populist views of democracy, while their counterparts supportive of the liberal model of democracy are more inclined to vote, as well as more inclined to cast a ballot for a non-populist party (H2). Findings have ambivalent implications for democracy, as party system polarization on popular sovereignty depresses the propensity to abstain from voting but also antagonizes citizens supportive of the liberal and populist models by elevating their propensity to vote for opposing parties.