OeNB Jubiläumsfonds Project Final Presentation at WU  

The final presentation of the OeNB-funded project “An Economic Model of Fake News” took place on November 10 2025 at the WU Vienna D4.2.008 from 12:00-13:30.

Melis Kartal (principal investigator) presented results from two studies of the project in the presentation.

Study 1: Sorting fact from fiction when reasoning is motivated (joint with Edoardo Cefala, Sylvia Kritzinger, and Jean-Robert Tyran)

Abstract: We investigate how sorting fact from fiction and updating from new information are shaped by cognitive ability, motivated reasoning, and overconfidence in societally important but politicized topics, such as climate change and public health. We predict and find evidence that the ability to make correct news assessments that counter one's existing issue opinions increases in both IQ and education. However, when we disaggregate data by news topic, we find that cognitive ability sometimes boosts motivated decision making. Our findings suggest that institutions matter. For example, higher trust in institutions reduces the magnitude of motivated reasoning, which may help limit opinion polarization in the longer term.

Study 2: Social (mis)influence

Abstract: We examine whether allowing individuals to share their confidence alongside factual information improves social learning online. In an experiment with over 3,600 U.S. participants, senders either shared only their answers to quiz questions or could additionally report their confidence. Confidence sharing substantially improves information transmission: receiver accuracy more than doubles relative to baseline, with especially large gains in neutral questions and in misleading items where baseline accuracy is below chance. Answer-only sharing yields weaker effects and none in political questions. However, the benefits of confidence sharing are highly heterogeneous: women, lower-income, and less-educated respondents benefit exclusively from confidence sharing, while men and higher-income or higher-educated respondents benefit similarly across both sharing institutions. The results highlight confidence reporting as a scalable design tool for improving online information quality especially among populations vulnerable to misinformation.