麻豆社区

Stella Victoria Laier Bernkilde from Political Science at the University of Southern Denmark received support from the DDC Thesis Grant for her thesis, “AI or Accountability? An Experimental Study of Political Scandals, Crisis Response Management, and Political Trust in Denmark.” Her project examines how different types of political scandals and crisis responses affect citizens’ trust, with particular attention to how politicians may use artificial intelligence to deny responsibility and challenge public accountability.

The grant enabled Stella to conduct a survey experiment with 1,155 Danish respondents through Norstat. Her findings show that political scandals reduce trust in politicians, but that the effectiveness of an apology or an AI based denial depends on the type of scandal. The study also identifies a clear partisan bias in how citizens evaluate politicians, while finding that broader trust in Denmark’s political system remains resilient.

Stella’s thesis received the highest grade and was subsequently named Best Political Science Thesis of the Year at the University of Southern Denmark.

You can read more about her project and findings in the abstract below.

Abstract:
Political scandals are an almost inevitable part of contemporary democratic life, yet they pose a severe threat by undermining citizens’ trust in political actors and institutions - the very foundation of our political system’s legitimacy. Consequently, effective crisis communication has become paramount for maintaining democratic stability. While previous research consistently demonstrates that scandals erode political trust, the literature remains divided on whether different types of scandals and crisis responses are evaluated differently by citizens. In particular, limited attention has been devoted to how emerging technologies, such as artificial intelligence (AI), provide politicians with new opportunities to deflect responsibility (i.e., The Liars Dividend) and what this means for digital democracy and public accountability. This thesis, “AI or Accountability? An Experimental Study of Political Scandals, Crisis Response Management, and Political Trust in Denmark”, addresses this gap by examining how different types of political scandals and response strategies affect political trust in Denmark, with particular attention to the growing role of AI in political crisis management. Additionally, the study investigates whether these effects are shaped by partisan ingroup/outgroup dynamics and mediated by emotional reactions such as anger. 

The study is based on a randomized survey experiment conducted among 1,155 Danish respondents through a professional panel (Norstat). Participants were exposed to one of eight experimental scenarios in a 2x2x2 factorial design varying by party affiliation (Social Democrats vs. Liberals), scandal type (policy scandal vs. norm scandal), and response strategy (AI-based denial vs. apology), alongside a control group. The analysis finds that political scandals significantly reduce trust in politicians, with policy scandals generating lower levels of trust than norm scandals. Moreover, the study reveals an unexpected paradox when it comes to the effectiveness of the response strategy; an accommodating strategy (apology) is more effective in policy scandals, whereas a defensive strategy (AI-based denial) performs better in norm scandals. The mediation analysis finds that anger partially explains this relationship. The study also finds strong evidence of partisan outgroup bias, as respondents evaluate opposing-party politicians more negatively regardless of scandal type, but it does not further amplify the scandal’s impact on trust. Overall, these findings suggest that citizens evaluate political crisis communication in a highly context-dependent manner. This has important implications for democratic accountability, as trust in individual politicians appears vulnerable to strategic communication. However, trust in the political system remained confined, showcasing strong institutional resilience in Denmark.