Fixing the Fact Checks: Maximizing Impact of Fact Checker Related Stories

• In Event: Attitudes Towards News

Abstract

Amid concerns about the prominence and effects of misinformation, fact checking has emerged as a much-heralded solution. Fact checking globally is on the rise (Amazeen, 2017), and underpins many of the efforts to combat misinformation. For example, fact checking reduces beliefs in misinformation among the public (Amazeen, Thorson, Muddiman, & Graves, 2016; Pingree, Watson, Sui, Searles, Kalmoe, Darr, Santia, & Bryanov, 2018; Young, Jamieson, Poulsen, & Goldring, 2017) and deters politicians from making false statements (Nyhan & Reifler, 2015).
One notable effort has been a partnership between the International Fact Checking Network (IFCN) and Facebook. The IFCN provides fact checks to Facebook that are featured in related stories below links containing identified misinformation. Research bears out this approach (Bode & Vraga, 2015; 2018), but the impact of the fact checks is likely contextual – the effect of seeing the fact check should depend upon how the fact check is presented. We test three related components of this presentation: The fact check’s headline, its summary (the blurb that appears under the headline), and any visual representation or cue regarding the results of the fact check.
Headlines are incredibly powerful conduits of information (Tewksbury & Althaus, 2000), and because people rarely click through to a story (Gabielkov, Ramachandran, Chaintreau, & Legout, 2016), the information contained in headlines is particularly persuasive and memorable, as compared to the information contained in the article itself (Tewksbury, Weaver, & Maddex, 2001). We consider whether the headline of the fact check is posed as a question (ie Did a study show that aluminum from vaccines killed Alfie Evans?) or as a statement of fact (ie Aluminum from vaccines did not kill Alfie Evans). As many people will only see the headline, we would expect that a statement would be more effective at correcting than a question. Statements were used in the existing research on the corrective impact of related stories (Bode & Vraga, 2015; 2018), but a cursory review of fact checks suggest that questions are the more common phrasing.
The information under the headline (which we refer to as the ‘blurb’) should be similarly persuasive. Again, because most people do not click through to the full story, they are likely receiving only the fact check information that is available on the Facebook platform. The blurb may provide useful additional corrective information beyond the headline. Whether that blurb includes a declarative statement about the veracity of the information (explicitly debunking the misinformation) or not should therefore impact the degree to which misinformation beliefs are corrected, potentially providing additional context to the headline.
Finally, multimedia elements tend to be particularly persuasive types of communication (Grabe & Bucy, 2009), and can affect whether people believe misinformation (Newman, et al., 2015). Many fact checks include some sort of visual representation of their conclusion of veracity (Fact Checker’s Pinnochios, Politifact’s truth-o-meter, etc.), but this image is not generally included as part of the fact check on the Facebook platform. Previous research has suggested that truth scales as part of a full article can enhance the corrective impact of fact checks (Amazeen, Thorson, Muddiman, & Graves, 2016), but has not examined this within the context of the headline only, nor on Facebook. We explore whether including a reinforcing multimedia graphic as part of the fact check response to misinformation would amplify the corrective effect of the fact check.
Because all three elements – headline, blurb, and visual – are likely interacting with one another, we employ a 2 (headline as a question versus a statement) by 2 (blurb includes the fact check conclusion or does not) by 2 (reinforcing visual is present or absent) experiment (plus a control where participants see no corrective information and a control where they see no misinformation at all) to test them simultaneously. We test these factors against four important outcomes: (1) reducing misperceptions generated by exposure to the misinformation, (2) perceptions of credibility of the fact-checking organization, (3) likelihood to click on the link or seek more information on the topic, and (4) likelihood to share or respond to the inaccurate post.
While we test these fact check formats in the context of the Facebook Related Stories function, we expect these results will apply to other spaces where people see fact checks, such as Google searches or news feeds. Results have important implications for understanding how technology companies can mitigate the effects of misinformation shared on social media platforms, and for developing empirical evidence of how fact checks are perceived in new media environments.

Authors

• Leticia Bode, Georgetown University

• Emily K. Vraga, George Mason University