Two analyses of Twitter posts showed that liberals produced more tweets about important social events, but that conservatives were much more likely to share rumors. While rumor-spreading decreased among liberals after official correction, it often increased among conservatives. The study was published in the Personality and Social Psychology Bulletin.
Psychologists have long understood that correcting misinformation is not easy. Even long after valid information is provided, falsehoods continue to exist creating a phenomenon known as the “continued influence effect”. In extreme cases, people may even exhibit a “backfire effect”, redoubling their commitment to a belief that has been debunked.
Properties of modern social media such as the organizations of users into “echo chambers,” groups of like-minded people who “echo” each other’s opinions, may worsen the “continued influence effect” by preventing the information that is not consistent with beliefs of people within an “echo chamber” to reach them. And even when people are exposed to information meant to correct their false beliefs, they may reject it through engaging a psychological mechanism known as “cognitive dissonance reduction.”
Previous studies have also shown that some political conservatives, such as the supporters of the QAnon Movement, are more likely to engage in conspiratorial thinking, are more susceptible to false information about dangerous events, and are also more receptive to pseudo-profound meaningless statements. But what about spreading rumors?
“There is a growing body of research suggesting that there may be some psychological asymmetries with respect to how liberals and conservatives interact with information online,” said study author Matthew DeVerna, a PhD candidate at Indiana University. “Building on this work, we were interested in seeing if there might be differences with respect to their rumor-spreading behavior on social media platforms.”
To study if and how political ideology is associated with spreading rumors on Twitter, DeVerna and his colleagues analyzed tweets related to the 2013 Boston Marathon bombing and the 2020 death of Jeffrey Epstein, jailed for sex trafficking at the time. They used software that collected tweets in the period between April 15 and April 25, 2013 (for rumors related to the Boston Marathon bombing) and between July 9 and December 31, 2019 (for rumors related to the death of Jeffrey Epstein).
“The use of Twitter data allowed us to conduct unique natural experiments to study this question in a really exciting way,” DeVerna said. “Furthermore, as the world is becoming increasingly interested in the spread of misinformation online, we feel it is important to understand the psychological underpinnings of what drives the spread of that content.”
The rumors they followed were those that a certain Saudi man that was cleared of accusations was in fact the bomber, that the Boston marathon bombing was a US government false flag operation, that Jeffrey Epstein’s death was not a suicide, that the Clintons were involved in Epstein’s death, and that Trump was involved in Epstein’s death. Although information exposing these rumors as false became available soon after they surfaced, their spread continued long after that information became available.
Researchers used a methodology based on analyzing political accounts the person followed to assess the political standing along the liberal-conservative continuum of each Twitter user whose tweets were included in the study. The first study included nearly half a million users who tweeted about the Boston Marathon, 67,4% of which were liberal.
The “Saudi bomber” rumor was referenced by 77,147 tweets and the false flag rumor by 19,756. Researchers were able to classify political standing of around half of the users who generated these tweets and reported that the majority of them (close to 60% in both cases) were conservatives.
Somewhat more than 410,000 tweets were identified as spreading rumors about the Jeffrey Epstein death and 67.8% of these were made by conservatives. The authors also compared fake news sharing habits of liberals and conservatives by registering how often users shared content from fake news websites such as Infowars.com or dailystormer.com. They used a list of 505 fake news websites.
Result showed that, although they generate less tweets in general, conservative users generated much more tweets about the studied rumors than liberal ones. In all cases, the spreading of rumor decreased sharply immediately after the official correction of the information contained in the rumor was published, only to more or less sharply increase again days after the publication of the correction.
In the case of the Saudi bomber rumor, tweets about the rumor reached a new peak among conservative users only 4 days after the official correction, while the rumor mostly died out with the liberal users. The false flag operation rumor followed similar patterns for the 5 days after the official correction for both groups, but showed decrease in the liberal group, and a new peak with conservatives after that. The 2020 rumors mostly died out with the liberal users after the official correction, but some months after the official correction started spreading intensely again among the conservative users.
“In our study, we observe evidence that suggests there was an ideological difference in rumor-spreading behavior before these official corrections were made — and these differences were exacerbated after those corrections,” DeVerna told PsyPost. “In both studies, conservatives — but not liberals — appeared to share rumors even more after they were debunked.”
The study highlights an important link between ideology and online news related behavior. However, it should be noted that for the 2013 analysis ideological standing of around half of the users could not be ascertained and results might have looked differently if these were properly identified. It is also possible that these patterns are influenced by certain specificities of Twitter and news related behavior might be different in other environments.
“We do not intend to imply that liberals would never display the type of rumor-spreading behavior we observe from conservatives in our studies,” DeVerna noted. “Our study does not rule out the possibility that we could see the opposite results if we were to track a different set of rumors that liberals were more motivated to spread.”
“We attempt to address this criticism in study two (about Jeffrey Epstein) with the rumors we track: #ClintonBodyCount, #TrumpBodyCount, and #EpsteinDidntKillHimself. These rumors were chosen specifically because the first two can be compared to one another (i.e., the motivation to spread these may theoretically “cancel out” for each political group) and the last rumor does not have any clear political affiliation (so neither group should be more inclined to spread it).”
“However, we still find that conservatives spread the rumor about the Clintons more than liberals spread the rumor about the Trumps and conservatives spread the politically neutral #EpsteinDidntKillHimself rumor more than liberals,”DeVerna explained.
The study, “Rumors in Retweet: Ideological Asymmetry in the Failure to Correct Misinformation”, was authored by Matthew R. DeVerna, Andrew M. Guess, Adam J. Berinsky, Joshua A. Tucker, and John T. Jost.
