Researchers analyzed a large corpus of COVID-19 tweets using a domain-specific NLI model to classify 264,737 posts as supporting or opposing false claims. The study compared 23 user- and text-level features across these two groups to understand the dynamics of the counter-misinformation ecosystem.
- Anti-misinformation posts exhibit higher levels of anger, disgust, and sadness compared to pro-misinformation posts, contradicting the assumption that negative emotion is a signature of falsehood.
- Posts opposing misinformation tend to originate from more established users, characterized by older account ages, higher follower counts, and greater listed counts.
These findings challenge dominant assumptions about emotional signatures in misinformation and highlight the role of established users in contesting false claims.