The Implied Truth Effect: Attaching Warnings to a Subset of Fake News Headlines Increases Perceived Accuracy of Headlines Without Warnings

Pennycook, G., Bear, A., Collins, E., & Rand, D. G. The implied truth effect: Attaching warnings to a subset of fake news headlines increases perceived accuracy of headlines without warnings. Management Science, Forthcoming

52 Pages Posted: 14 Sep 2017 Last revised: 16 Jan 2020

See all articles by Gordon Pennycook

Gordon Pennycook

University of Regina

Adam Bear

Yale University

Evan Collins

Department of Biomedical Engineering, Yale University

David G. Rand

Massachusetts Institute of Technology (MIT)

Date Written: August 7, 2019

Abstract

What can be done to combat political misinformation? One prominent intervention involves attaching warnings to headlines of news stories that have been disputed by third-party fact-checkers. Here we demonstrate a hitherto unappreciated potential consequence of such a warning: an “implied truth” effect whereby false headlines that fail to get tagged are considered validated and thus are seen as more accurate. With a formal model, we demonstrate that Bayesian belief updating can lead to such an implied truth effect. In Study 1 (N = 5,271 MTurkers), we find that while warnings do lead to a modest reduction in perceived accuracy of false headlines relative to a control condition (particularly for politically concordant headlines), we also observed the hypothesized implied truth effect: the presence of warnings caused untagged headlines to be seen as more accurate than in the control. In Study 2 (N = 1,568 MTurkers), we find the same effects in the context of decisions about which headlines to consider sharing on social media. We also find that attaching verifications to some true headlines – which removes the ambiguity about whether untagged headlines have not been checked or have been verified – eliminates, and in fact slightly reverses, the implied truth effect. Together, these results undermine theories of motivated reasoning while identifying a potential challenge for the policy of using warning tags to fight misinformation – a challenge that is particularly concerning given that it is much easier to produce misinformation than it is to debunk it.

Note: Note: A previous version of this paper was titled "Assessing the effect of “disputed” warnings and source salience on perceptions of fake news accuracy", and included an additional set of experiments on source salience which were ultimately published separately as Dias Pennycook Rand (2020) Emphasizing publishers does not effectively reduce susceptibility to misinformation on social media. The Harvard Kennedy School (HKS) Misinformation Review. Doi.org/10.37016/mr-2020-001

Keywords: fake news, news media, social media, fact-checking, misinformation, source credibility

Suggested Citation

Pennycook, Gordon and Bear, Adam and Collins, Evan and Rand, David G., The Implied Truth Effect: Attaching Warnings to a Subset of Fake News Headlines Increases Perceived Accuracy of Headlines Without Warnings (August 7, 2019). Pennycook, G., Bear, A., Collins, E., & Rand, D. G. The implied truth effect: Attaching warnings to a subset of fake news headlines increases perceived accuracy of headlines without warnings. Management Science, Forthcoming , Available at SSRN: https://ssrn.com/abstract=3035384 or http://dx.doi.org/10.2139/ssrn.3035384

Gordon Pennycook

University of Regina ( email )

3737 Wascana Parkway
Regina, Saskatchewan S4S OA2 S4S 0A1
Canada

Adam Bear

Yale University ( email )

2 Hillhouse Ave
New Haven, CT 06511
United States

HOME PAGE: http://campuspress.yale.edu/adambear/

Evan Collins

Department of Biomedical Engineering, Yale University ( email )

New Haven, CT 06511
United States

David G. Rand (Contact Author)

Massachusetts Institute of Technology (MIT) ( email )

77 Massachusetts Avenue
50 Memorial Drive
Cambridge, MA 02139-4307
United States

HOME PAGE: http://www.daverand.org

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
6,035
Abstract Views
49,064
Rank
2,682
PlumX Metrics