Working Paper: CEPR ID: DP14738
Authors: Emeric Henry; Ekaterina Zhuravskaya; Sergei Guriev
Abstract: Using an online randomized experiment in the context of the 2019 European elections campaign in France, we study how fact-checking affects sharing of false news on social media. We exposed a random sample of French voting-age Facebook users to statements on the role of the European Union made by the far-right populist party Rassemblement National. A randomly selected subgroup of participants was also presented with fact-checking of these statements; another subgroup was offered a choice whether to view the fact-checking information. Then, all participants could choose whether to share the false statements on their Facebook pages. We show that (i) both imposed and voluntaryfact-checking reduce sharing of false statements by about 45%; (ii) the size of the effect is similar between imposed and voluntary fact-checking; and (iii) each additional click required to share false statements substantially reduces sharing. These results inform the debate about policy proposals aimed at limiting propagation of false news on social media.
Keywords: alternative facts; fake news; sharing; social media; facebook
JEL Codes: D8; D91
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| exposure to fact-checking (C91) | reduced sharing of false statements (Z13) |
| imposed fact-checking (F38) | reduced sharing of false statements (Z13) |
| voluntary fact-checking (Y30) | reduced sharing of false statements (Z13) |
| each additional click (Y50) | decreased likelihood of sharing (D16) |
| requirement for a second click (C99) | reduced actual sharing rate (F62) |
| fact-checking information (Y10) | decreased intention to share (D16) |
| fact-checking information (Y10) | decreased actual sharing of false statements (Z13) |