The Impact of Twitter Labels on Misinformation Spread and User Engagement: Lessons from Trump’s Election Tweets

Published in WWW' 22. DOI:https://doi.org/10.1145/3485447.3512126

Rutgers Law School Research Paper

11 Pages Posted: 26 Feb 2022 Last revised: 13 May 2022

See all articles by Orestis Papakyriakopoulos

Orestis Papakyriakopoulos

Center for Information Technology Policy, Princeton University

Ellen P. Goodman

Rutgers Law

Date Written: February 22, 2022

Abstract

Digital platforms have turned to “soft moderation” to defang misinformation, and as an alternative to de-platforming or other hard moderation approaches. Fact checks and warning labels are soft moderation techniques designed to deter user engagement with misinformation. As counter speech, labels advance credible information without trenching on free speech. What remains to be understood, however, is what effect they actually have on users. Twitter labeled many of former-President Trump’s tweets which falsely claimed election fraud around the 2020 US federal elections; this body of data concerning the labeled and unlabeled tweets sheds light on how soft moderation might have influenced people's engagement with the misinformation. This study uses statistical analyses and various other empirical methods to answer these questions: how are fact checks and warning labels of different kinds associated with distinct user behavior (likes, retweets, quote tweets, replies) and with the toxicity of the platform's political discourse around the false tweets.

What we found, among other things, is that:
1. The soft moderation of fact checks and warning labels overall neither increased nor decreased engagement with the false tweets.
2. However, specific types of labels were associated with distinct patterns of user behavior, suggesting that label design makes a difference.
3. Tweets with more directly rebutting labels generated less toxic replies; tweets with more linguistically overlapping labels garnered less engagement.

These findings reveal complex relationships between labels and user reactions to misinformation, and suggest that policymakers and platforms consider label design (and not just the label binary) when encouraging or adopting soft moderation interventions.

Keywords: digital platforms, social media, misinformation, disinformation, content moderation, soft moderation, fact checks, warning labels, transparency, user engagement

Suggested Citation

Papakyriakopoulos, Orestis and Goodman, Ellen P., The Impact of Twitter Labels on Misinformation Spread and User Engagement: Lessons from Trump’s Election Tweets (February 22, 2022). Published in WWW' 22. DOI:https://doi.org/10.1145/3485447.3512126, Rutgers Law School Research Paper, Available at SSRN: https://ssrn.com/abstract=4036042 or http://dx.doi.org/10.2139/ssrn.4036042

Orestis Papakyriakopoulos

Center for Information Technology Policy, Princeton University ( email )

Ellen P. Goodman (Contact Author)

Rutgers Law ( email )

217 N. 5th Street
Camden, NJ 08102
United States
856-225-6393 (Phone)
856-225-6516 (Fax)

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
340
Abstract Views
3,586
Rank
159,553
PlumX Metrics