Welfare impact of policies to fight disinformation: stopping the false while preserving the true

Department d'Economie et Sciences de la Décision  
Intervenant : Emeric Henry (SciencesPo)
Salle T-007

Abstract:

"Using an online randomized experiment in the context of the 2022 US midterm elections, we compare the impact of different policies proposed to fight the circulation of false news. We exposed a random sample of Twitter users to a series of 4 Tweets, two true and two false, and allowed them to share at most one. The participants were randomly allocated to different policies, standard in the literature: a control group, a fact check treatment, a policy priming accuracy and a policy requiring an extra click. The reduced form evidence shows that priming accuracy performs best: it decreases sharing of false news and increases the circulation of true news. A structural model of sharing shows that the effect of the policies can be decomposed into three channels: priming accuracy decreases veracity estimates for the false tweets, increases the weight put on veracity in the sharing function and increases the cost of sharing. The structural model is used to guarantee external validity and to evaluate alternative policies."

Joint with Sergei Guriev, Theo Marquis and Ekaterina Zhuravskaya