Fact-checking initiatives should consider labeling content as “true” more frequently, because social media users tend to share fact-checked information confirmed and labeled as “true” more than they share refuted fact-checked information labeled as “false,” according to a recently published study carried out by five Latin American researchers.
The study, “Truth be told: How ‘true’ and ‘false’ labels influence user engagement with fact-checks,” is based on a survey with Twitter users in Argentina in the context of the 2019 elections, won by the current president, Alberto Fernández. The study, published in the academic journal New Media & Society, was written by researchers Natalia Aruguete (National University of Quilmes, Argentina), Ingrid Bachmann (Pontifical Catholic University of Chile), Ernesto Calvo (University of Maryland, USA), Tiago Ventura (Georgetown University, USA) and Sebastián Valenzuela (Pontifical Catholic University of Chile).
Valenzuela explained to LatAm Journalism Review (LJR) that the authors of the study, who have spent years analyzing disinformation, wanted to contribute by studying not only the problem, but also its solutions. And one of the most widespread solutions against misinformation has been fact-checking.
“In Latin America, there were no empirical assessments of whether fact-checking works or not,” Valenzuela said. “And it seemed to us that it wasn't so obvious to say 'well, if this works in the United States, it will also work in Brazil, Chile or anywhere else.' In our countries, the journalism crisis and widespread distrust in public institutions are much deeper than in other places.”
In the study, 2,041 people took part in a survey experiment where they were asked whether they would engage with a tweet with a negative statement about Mauricio Macri, who ran for re-election in 2019 and was the main opponent of the current president, and a tweet with a negative statement about Fernández. Respondents then were exposed to fact-checked tweets labeled as either "true" or "false," and asked if they would share the verified statements on their Twitter profiles.
The researchers found that people prefer to share fact-checked information that favors their political preferences – a trend that has already been mapped by previous studies, the authors wrote. Furthermore, research participants tended to share more fact-checked content that reinforced their political preferences and that had a “true” label. In other words, confirmations, rather than fact-checked content with a “false” label, or refutations, were shared more often, even if the refutations were also favorable to their preferred candidates. And even among fact-checks that were unfavorable toward their political preferences, respondents still were more likely to share confirmations (“true” label) than refutations (“false” label).
For researchers, this can be explained by two reasons: refutations with a false label “carry a heavier cognitive burden,” or take more to be processed cognitively, than confirmations. Furthermore, the words “true” and “false” have distinct social connotations – “telling the truth is good, and telling a falsehood is bad,” the study authors wrote.
“Our results have a significant practical implication: fact-checkers may wish to frame more claims using the label ‘true’ (...) instead of the equivalent statement labeled ‘false,’” the researchers said. “While semantically equivalent, a confirmation frame increases sharing and engagement of factual content, while a refutation frame reduces sharing and engagement.”
The researchers noted, however, that it is important for fact checkers to continue to use the “false” label. According to the study, not only is the “fake” label able to “correct misperceptions, but it can also decrease the likelihood of retransmission of false stories, since ‘false’ labels have proven here to decrease engagement and, therefore, the spread of misinformation.”
“In other words, confirmation and refutation labels have practical implications, and while fact-checkers cannot always use the 'true' label – it depends on the original content they are verifying, after all – sometimes they do have the flexibility to focus on the true information rather than the misinformation, and in so doing, increase the spread of verifications and true content,” the authors wrote.
Valenzuela said that the results of the study have been shared with other disinformation researchers and with Latin American fact-checking organizations, who have said that the research conclusions “make a lot of sense.”
“The fact-checkers we have spoken to, from different parts of Latin America, generally recognize that there’s a difference; that it’s not the same thing to use ‘true’ or ‘false;’ that there’s a difference in terms of the engagement they generate on social media,” he said.
Valenzuela also added that the group of researchers is about to publish another study that brings together similar experiments carried out in Brazil, Chile and Colombia that reached the same conclusions as the research carried out in Argentina.
“We found exactly the same thing: confirmations, when you say ‘this is true,’ are shared much more than when you write the same correction but in the form of a refutation, when you say ‘this is false.’ So the evidence we have from four countries is consistent with the idea that this is a phenomenon that is not specific to just Chile or Argentina, but that it occurs across multiple countries,” he said.