How Life-like Digital Humans in Voting Advice Applications Can Stimulate Young Voters to Inform Themselves About Politics: The Role of Familiarity and Expertise

Conversational Agent Voting Advice Applications (CAVAAs) allow voters to ask questions about political topics in Voting Advice Applications (VAAs). Given their potential to inform young voters, designing effective CAVAAs is crucial. We conducted two experimental studies in run-up to the European Elections 2024 on the effects of different avatar types in CAVAAs. Study 1 (N = 81 theoretically educated students) examined the impact of disembodied versus embodied CAVAAs with a renowned or non-renowned digital human avatar. In study 2 (N = 199 practically educated students) the avatar's expertise was also considered. While both studies confirmed CAVAAs are perceived as relevant information resources among young voters, the tool's embodiment, familiarity, and expertise did not positively impact tool evaluation measures, political measures, and user behavior. However, study 2 showed that non-renowned avatars were perceived as more eerie than their renowned counterparts, confirming previous findings. We discuss the implications for theory and practice.