Présentation de Théo MARQUIS, Sciences Po Paris
**Curtailing False News, Amplifying Truth: Evaluating Anti-Misinformation Measures on Social Media** _with Sergei Guriev, Emeric Henry and Ekaterina Zhuravskaya_ We develop a uniﬁed framework to evaluate policy measures aimed at reducing the circulation of false news on social media, such as raising awareness of the dangers of misinformation, oﬀering fact-checking, requiring conﬁrmation clicks, and asking users to think about the content they are about the share. We evaluate the impact of these interventions on the sharing of false and true political news on Twitter using a randomized experiment during the 2022 mid-term U.S. elections. We ﬁnd that priming users about the reputational costs of sharing misinformation is the most eﬀective policy in changing the balance of shared news towards true content. We build a model of sharing decisions, where a sharer derives utility from persuading the audience, signaling their partisan aﬃliation, and from maintaining reputation as a credible source. In equilibrium, sharing depends on the perceived veracity and partisanship of content, the salience of reputational concerns, and sharing frictions. Structural estimation of this relationship allows us to shed light on the mechanisms through which diﬀerent policy interventions aﬀect circulation of false and true content on social media. The impact of all considered policies comes mostly from their eﬀects on the salience of reputation and on the sharing frictions. The eﬀect policies have on the perceived veracity of content plays a negligible role in their impact on sharing, even when it comes to fact-checking. Priming intervention performs the best because it is designed to aﬀect salience with minimal added frictions.