United We Stand, Divided We Fall: Mainstream Party Agreement and Vote Switching to Competitors
P12-S300-1
Presented by: Luis Sattelmayer
Does mainstream party agreement on divisive issues reduce vote switching to competitors? The literature on party competition emphasizes mainstream parties’ incentives to converge toward the median voter. Recent findings suggest that convergence along the left-right ideological scale can repel voters. This paper challenges such assumptions by arguing that mainstream party agreement on salient and divisive issues, like immigration, stabilizes voter support by altering the dynamics of political competition. Specifically, agreement fosters valence-based competition, where parties compete on competence and credibility, while divergence intensifies conflict, polarizes voters, and increases defections to competitors. To test this argument, the paper constructs an innovative dataset combining panel survey data from the GLES with fine-grained measures of German parliamentarians' issue stances on Twitter. Using Natural Language Inference (NLI), it tracks shifts in immigration positions over time and introduces a novel measure of agreement between mainstream parties. The analysis reveals that when mainstream parties align on immigration stances, they not only reduce the likelihood of vote switching to far-right competitors but also attract disillusioned voters back to the mainstream. These effects are most pronounced when immigration salience is high, highlighting the strategic importance of message alignment. The findings have significant implications for understanding electoral dynamics in multiparty systems. Agreement on divisive issues can serve as a powerful, yet underexplored, tool for mainstream parties to stabilize support, counter voter defection, and challenge the rise of far-right competitors. This study underscores the need to move beyond simplistic convergence theories and explore how issue-based coordination can reshape political competition.
Keywords: Mainstream Party Agreement, Natural Language Processing, Vote Switching, Immigration, Valence-Based Competition