Misinformation Prevention and Self-Correction

• In Event: Is There a Cure for Misinformation?

Abstract

How to stop the diffusion of misinformation has instigated tremendous interest recently. Much research has studied factors that lead people to believe or not believe in misinformation. Fewer studies have looked into how to stop the spread at source — by changing the behavior of original spreaders. This study addresses the question of what are the conditions that lead people to remove misinformation posts or even self-correct on social media.

We draw on data from two survey experiments in Taiwan and in the US. The design of the survey experiment is a sequential one where respondents are exposed to three consecutive scenarios. We leverage the context of “infodemic” in the COVID-19 pandemic. In the first scenario, respondents read an article on vaccine information. In the second scenario, respondents then read an article refuting the original source, but the source of refutation varies (four conditions: mass media, famous YouTuber, YouTuber, or public health expert). In the third scenario, respondents read further responses on social media, where comments on social media either support the original article or the refuting article (three conditions: supportive, opposing, mixed voices). In each scenario, we ask respondents to rate their likelihoods of engaging in two outcome behaviors that would stop the spread of the information, i.e., to remove the shared post and post a correction. In short, we have two outcomes with a 4X4 condition design.

The results indicate that it is much harder to drive people to make a correction post as compared to retracting the shared post. A refutation by the credible source of a public health expert could motivate the respondents to delete their original posts, but could not push them to make correction posts. YouTubers, whether famous or not, did not affect any actions of the respondents. Social media responses also influenced the respondents’ willingness to keep the original post. If opposing voices are present (i.e., opposing or mixed voices), the respondents are more likely to delete their original posts. However, social media reactions did not affect the respondents’ willingness to make correction posts. Overall, our results have implications on how dialogue processes of alternative arguments and social media reactions relate to the diffusion of misinformation.

Authors

• Yuan Hsiao, University of Washington

• Greg Chih-Hsin Sheen, National Cheng Kung University

• Ching-Hsing Wang, National Cheng Kung University