“Polarization Versus Democracy“: Informing the Hungarian Electorate to Reduce Affective Polarization and Foster Democratic Attitudes

This project aims to clarify the - currently somewhat contested - relationship between affective polarization and anti-democratic attitudes among the public by developing an experimental intervention that tackles both phenomena. Various reports such as Freedom in the World assert the continuing global decline of democracy, despite a broad public support for democratic principles in most countries (Svolik 2019; Ferrin & Kriesi 2025). Several studies propose affective polarization as the variable that might explain this puzzle, as emotionally highly polarized voters are more likely to trade off democratic principles for partisan interests (Graham & Svolik 2020; Orhan 2021; Kingzette et al. 2021; Gidengil et al. 2022). Svolik (2019) has framed this dilemma as “polarization versus democracy”. This implies that reducing affective polarization among voters could also improve their vigilance against democratic transgressions. However, several studies that have succeeded in mitigating affective polarization with experimental interventions could not simultaneously curtail anti-democratic attitudes, raising the question of whether these two phenomena are causally linked at all (Broockman et al. 2023; Voelkel et al 2023). With this project, we particularly aim at simultaneously reducing both affective polarization and anti-democratic attitudes. Hungary provides a highly suitable context for such aims, as a regime divide over democracy itself has become a salient dimension of partisan conflict. As such, democratic backsliding could both result from and fuel affective polarization (Gessler & Wunsch 2025), constituting a vicious circle which suggests that these two issues should be addressed in conjunction. To unravel this puzzle, we implement several innovations as compared to the existing studies. We will develop an experimental treatment – a conversational chatbot correcting various misperceptions among the Hungarian electorate – that puts particular focus on (misperceptions about) the state of Hungarian democracy and adherence to democratic principles among supporters of different parties. Moreover, we create different versions of the chatbot based on the party preference of the respondents, which should result in more well-aimed treatment. We also plan to improve the measurement of anti-democratic attitudes, designing more nuanced survey items to tackle the “floor effect” that similar studies have often faced, as support for explicitly anti-democratic behavior is already low, leaving little room for further reduction. We will conduct a two-wave survey among the Hungarian electorate in spring 2026 to develop and test this chatbot treatment. The first (pre-election) survey will map the central misperceptions among/about the supporters of different parties, and the baseline levels of affective polarization and anti-democratic attitudes, serving as an input for the chatbot(s). The second (post-election, experimental) survey will test the effects of the chatbot on both affective polarization and (anti)-democratic attitudes. Our findings will make an important contribution to the affective (de)polarization literature, while also carrying social relevance. This project is inspired by the idea of “constructive polarization” that combines clearly drawn lines between different parties/leaders with a unifying tone towards the population (Somer 2025). Our chatbot becomes freely available for public use, hopefully fostering a less polarized and more collaborative society that will step up for the Hungarian democracy.