The Impact of Fake News on Public Trust in Media

Abstract

This study evaluates how exposure to fake news shapes public trust in media using a mixed-methods design that combines an individual-level randomized feed experiment with a quasi-experimental panel around a mid-year platform labeling policy. A stratified online sample received verified, fake, or hybrid news streams with orthogonal prebunking and accuracy-prompt manipulations; digital traces and surveys captured immediate and four-week outcomes. Measurement modeling validated latent constructs for trust, credibility, literacy, and belief accuracy, and structural equation models decomposed pathways from exposure to trust. Difference-in-differences and event-study analyses assessed policy timing effects, while qualitative interviews illuminated interpretation strategies and repair talk. Results show that fake news reduces post-exposure trust relative to verified and hybrid streams; part of this effect is mediated by lower belief accuracy and perceived credibility. Prebunking and accuracy prompts attenuate trust losses, but mitigation is heterogeneous, stronger among higher-literacy participants and weaker among ideologically committed respondents. Platform labeling improves belief accuracy modestly and transiently, with limited direct effects on trust. Cross-country exposure diversity correlates with trust resilience, whereas ownership/source network centrality aligns with greater distrust, suggesting distributional and structural channels beyond content alone. Findings recommend layered interventions: durable prominence and transparency obligations for platforms, newsroom practices centered on verifiability and timely correction, and scalable prebunking and literacy programs to build robust accuracy heuristics.