3

In proportional systems, parties generally signal their coalition preferences to voters during election campaigns. Breaking coalition promises after the election entails normatively undesirable repercussions for government identifiability and accountability. This study is the first to investigate voters’ short-term reactions to post-electoral coalition choices of parties in conflict with pre-electoral coalition signals. Firstly, I performed a quasi-experimental investigation that considers an instance of inconsistent coalition choice following the 1996 New Zealand election. Secondly, I conducted an innovative survey experiment during the government formation period following the 2020 Irish general election. I find that breaking coalition promises is costly not only for the inconsistent party but also for democracy in general. Inconsistencies concerning coalition choices make voters less likely to support the inconsistent parties and erode voters’ satisfaction with the way democracy works. The findings provide important implications for government identifiability and accountability in proportional systems, democratic stability, and the effectiveness of cordons sanitaires vis-à-vis radical parties.

Ruling out forming coalition governments with parties critical of liberal democracy, i.e., establishing cordons sanitaires vis-à-vis these parties, is often seen as a crucial contribution to safeguarding liberal democracy. However, little is known about whether cordons sanitaires are effective in reducing the vote share of parties isolated in this way. Specifically, the effects of cordons sanitaires on voting induced by changes in voters’ expectations of post-electoral government formation remain unclear. I address this research gap by conceptualizing cordons sanitaires as a specific class of pre-electoral coalition signals and drawing on theoretical knowledge about the electoral expectation-induced consequences of coalition signals. I integrate these theoretical insights into a formal agent-based model of dynamic party competition and perform counterfactual simulations in an artificial democracy calibrated to resemble the 1930s Weimar Republic. The results show that the vote share of an artificial NSDAP increases when a cordon sanitaire is erected against it. By illustrating the theoretical possibility (not inevitability) of these unintended expectation-induced consequences, the paper provides important implications for research on the mainstream parties’ response to radical parties.

This study formulates and tests an agent-based model of opinion dynamics which claims to explain the evolution of political preferences by means of social interaction effects. The approach incorporates a majority and a momentum mechanism claiming that individuals are affected by perceived opinion levels as well as by opinion changes. This theoretical model is empirically tested by estimating its parameters on government satisfaction in Germany. The results support the empirical validity of the approach, as a significant momentum mechanism can be identified while other significant parameters display meaningful estimates. Additionally, almost every week’s level of government satisfaction is a likely realization of the process given the data of the previous week. Beyond that, the findings suggest that individuals are rather affected by opinion changes than by opinion levels and that nonconformity plays a more important role in the evolution of the considered preference than conformity.