En anglais, mais ça peut nous dépanner. Envoyez ces études aux équipes de communication. :)
Study finds warnings about false claims can “prebunk” fraud talk.
John Carey and Brendan Nyhan
Government professors John Carey, left, and Brendan Nyhan conducted studies of election perceptions in the United States and Brazil.
Bright Line Watch Stands Sentry on Democracy
Body
With control of Congress and a check on the Trump administration at issue in the 2026 midterm elections, the upcoming election cycle may again see claims of voter fraud.
But warning voters beforehand that there may be false claims about the election, and providing them with information on election security measures through “prebunking,” can increase confidence in the results and decrease beliefs in voter fraud, according to a new study published in Science Advances.
“Prebunking is effective because it provides people with novel facts about how elections are secured,” says co-author Brendan Nyhan, the James O. Freedman Presidential Professor of Government at Dartmouth.
To examine if corrective information can change false beliefs in widespread election fraud, the research team conducted a series of studies in the United States and Brazil to evaluate perceptions about past and future elections.
The two countries were selected because of the prevalence of misperceptions about voter and election fraud after President Donald Trump lost to Joe Biden in 2020 and Brazilian President Jair Bolsonaro lost to Luiz Inácio Lula da Silva in 2022. Both incumbents promoted claims of fraud, and supporters stormed their capitols in protest.
Prebunking is effective because it provides people with novel facts about how elections are secured.
The studies were conducted online in the U.S. before the 2022 midterm elections and in Brazil after its 2022 presidential elections. Each study contrasted the effect of prebunking with a credible sources treatment, which tests the effect of hearing from sources who might be the most persuasive to people who are inclined to be skeptical.
More than 5,500 participants from the two countries were surveyed online about their views on the credibility of elections via YouGov in the U.S. and Qualtrics in Brazil. Participants were randomly assigned to one of two treatment conditions or to a control group:
The first group received prebunking information with facts on election security measures in place in their country with a warning that others may make false claims to mislead them about an upcoming election. Prior studies have shown that exposure to corrective information can be effective at discerning fact from fiction. In the U.S., for instance, participants received information about how elections are secured via steps like testing voting machines, validating mail ballots, and using secure dropboxes to collect ballots.
The second group was assigned to a “credible sources” treatment in which they received information about allies of either Trump in the U.S. in 2020 or Bolsonaro in Brazil in 2022 or credible neutral sources who affirmed the previous election result as legitimate (such as Republican judges and officials, Bolsonaro’s son, and the Brazilian election agency).
The first study was conducted in the U.S., the second in Brazil, and the third was also administered in the U.S.
Broadly speaking, in the U.S., both the prebunking and credible sources proved effective.
For the prebunking treatment about the upcoming 2022 election, beliefs that Biden won in 2020 and confidence in the 2020 election increased while the number of House seats people thought were won by fraud in 2020 decreased. Results were similar for the credible sources treatment. The results showed that among Republicans, belief that Biden was the rightful winner of the 2020 election increased from 33% in the control group to 44% for credible sources and 39% for prebunking.
In Brazil, prebunking seemed to be more effective than information from credible sources, increasing confidence in the 2022 and 2026 elections and decreasing beliefs about the prevalence of election fraud in both.
For the third study, which was conducted in the U.S. following the 2022 midterm election, the authors again found that participants expressed more confidence in elections and less belief in fraud if they received the prebunking. Notably, these effects were only observed when they omitted the part of prebunking treatment which warns people that they might encounter false information, suggesting that the novel information provided to participants in the treatment is the most important factor, rather than how that information is framed.
“These results are encouraging,” says co-author John Carey, the John Wentworth Professor in Social Sciences and interim dean of the Faculty for Arts and Sciences. “You often hear that people aren’t responsive to facts or interested in details, but our experiments increased the accuracy of beliefs, even among those who started out most committed to ideas that aren’t supported by evidence.”
The co-authors say election officials and local and state governments should share information about how elections are secured to help preempt false fraud claims that can diminish faith in the integrity of elections.
