(iPoster) What Drives Negative Campaigning against Far-Right Parties? Evidence from EP2024

• In Event: Poster Sessions: Political Communication II & The Political Power of Film and Reality TV
  In Poster Session: Poster Session: Political Communication II

Abstract

Far-right parties (Pirro, 2023) around the globe are gaining ground, challenging liberal democracies and mainstream parties. The latter respond with different strategies (Heinze, 2018) and to varying degrees of success (Krause et al., 2023). One prominent strategy is negative campaigning and delegitimization (Schwörer and Fernández-García, 2021; van Heerden and van der Brug, 2017). Although attacking may appear the most obvious response, mainstream parties in European multiparty systems use it to very different degrees (Daur, 2025).
This comparative study aims to explain the variation in negative campaigning against far-right competitors, focusing on online campaigns during the 2024 European Parliament elections. Against the backdrop of far-right’s self-portrayal as victims of the establishment (Marcks and Pawelz, 2022), we further ask whether the drivers of negative campaigning against far-right parties differ from those of negative campaigning in general. Prior research provided solid evidence on the general drivers of negative campaigning (e.g., Dolezal et al., 2016; Maier and Nai, 2022; Walter et al., 2014). A separate strand of research examined strategies toward far-right parties (Heinze, 2018; Schwörer and Fernández-García, 2021) and their effectiveness (Krause et al., 2023; van Heerden and van der Brug, 2017). What is missing, however, is a systematic analysis of the drivers of negative campaigning specifically directed at far-right parties.
To close this gap, we link these strands of research and argue that the extent of negative campaigning against the far right hinges on strategic considerations of mainstream parties. Based on a review of the negative campaigning literature, we derive and test several hypotheses. First, we expect ideological distance to have a U-shaped effect, with both distant and proximate parties being more negative toward each other. Second, we expect parties trailing challengers in the polls to be more likely to resort to negativity (Haselmayer, 2019). Third, we expect such campaign contexts and past coalitions to moderate the first effect. Fourth, we expect negativity to be driven by retaliation, that is, a responses to previous attacks(Dolezal et al., 2016). While we expect no significant differences between the drivers of negative campaigning in general and against far-right parties, we do expect cross-national differences between countries where far-right parties have pariah status and those where they are politically integrated (Daur, 2025). We control for incumbency effects and test whether the findings are specific to the far right or also apply to the far left.
These hypotheses are tested against a dataset of N = 47,563 social media posts (Facebook and Instagram) from 349 profiles—party accounts and politicians—representing 53 political parties across seven European countries (AT, DE, FI, FR, IT, PL, and SE), which vary considerably in the status of far-right parties. Our focus is on social media communication, which is particularly well suited for observing dynamics of attacks and retaliation. The analyzed time frame covers three months, two months before the election held on 6-9 June 2024 and one month after the election, which allows us to isolate campaign effects and to examine short-term dynamics. Parties are classified as far right following the PopuList (Rooduijn et al., 2024). Fine-grained ideological positions are drawn from the Comparative Manifesto Project (Lehmann et al., 2025), and polling data from the aggregator Europe Elects. We operationalize attacks using a computational targeted sentiment analysis, validated against human-annotated posts (n=1,447). The study introduces a novel R package providing flexible and valid methods for named entity recognition (F1=0.82), entity linking, and targeted sentiment classification (F1 = 0.90). Finally, the data are aggregated into day-wise dyadic observations on party-level and analyzed using multilevel Bayesian regression models.
We discuss our findings in light of growing challenges to liberal democracies posed by far-right parties, affective polarization, and communicative counterstrategies within networked public spheres.

Authors

• Daniel Thiele, Freie Universität Berlin

• Annett Heft, Freie Universität Berlin and Weizenbaum Institute for the Networked Society, Berlin

• Baoning Gong, Weizenbaum-Institute

• Barbara Pfetsch, Universität Hohenheim