Political information is now commonly consumed embedded in user-generated content and social media. Hence, peer users (as opposed to professional journalists) have become frequently encountered sources of such information. This experiment tested competing hypotheses on whether expo`sure to attitude-consistent versus -discrepant political messages (confirmation bias) depends on association with peer versus professional sources, through observational data and multi-level modeling. Results showed the confirmation bias was differentiated, as attitude importance fostered it only in the peer sources condition: When consuming user-generated posts on political issues, users showed a greater confirmation bias the more importance they attached to a specific political issue. Furthermore, exposure generally affected attitudes in line with message stance, as attitude-consistent exposure reinforced attitudes, while attitude-discrepant exposure weakened them (still detectable a day after exposure). Attitude impacts were mediated by opinion climate perceptions.
All Science Journal Classification (ASJC) codes