A recent study published in the Harvard Kennedy School Misinformation Review explored the effectiveness of Twitter’s “disputed” tags in combating misinformation about election fraud during the 2020 U.S. presidential election. The researchers found that these tags, which were designed to flag false information, had little impact on changing people’s beliefs about the truthfulness of Donald Trump’s false claims regarding election fraud. In fact, some Trump supporters were more likely to believe misinformation when it included these disputed tags.
The researchers conducted this study in response to the growing concern about how misinformation on social media can influence public opinion, especially during elections. The 2020 U.S. election saw widespread claims of election fraud, mainly fueled by then-president Donald Trump. Social media platforms like Twitter attempted to counter these false claims by attaching warning tags to posts that made such allegations.
However, little was known about whether these tags were effective in reducing belief in misinformation, particularly among those who already supported Trump or had existing doubts about the election’s legitimacy. The study aimed to understand whether these disputed tags could alter perceptions of election fraud and to what extent political knowledge influenced these perceptions.
The study involved a sample of 1,078 U.S. adults who were recruited through CloudResearch, a participant-sourcing platform. Participants were randomly assigned to one of two conditions. In the first condition, participants were shown four tweets from Donald Trump, each falsely claiming election fraud. These tweets had disputed tags that indicated the claims were unverified. In the second condition (the control group), participants saw the same tweets without any warning tags.
Participants were then asked to rate how truthful they believed each tweet was on a seven-point scale ranging from “extremely false” to “extremely true.” In addition to these truthfulness ratings, participants were asked about their views on election fraud and the fairness of the 2020 election.
To understand the role of political knowledge, the researchers also assessed each participant’s knowledge of basic facts about American politics using a set of 10 factual questions. They additionally measured participants’ verbal ability with a vocabulary test to control for general cognitive ability. The researchers wanted to see if people with higher political knowledge would be more resistant to false information or, paradoxically, more likely to believe it when it aligned with their political beliefs.
The results showed that the disputed tags did not significantly reduce the belief in election fraud among Trump supporters. In fact, Trump voters with higher political knowledge were more likely to perceive the misinformation as truthful when the disputed tags were present compared to when they were absent. This finding suggests that for some individuals, especially those with strong political identities, the disputed tags might have backfired. Rather than reducing belief in misinformation, the tags may have triggered a sense of defensiveness or skepticism, leading these individuals to double down on their pre-existing beliefs.
Interestingly, Trump voters who were initially skeptical about election fraud also showed an increase in their belief in misinformation when the disputed tags were present, suggesting that the tags might have inadvertently reinforced doubts rather than dispelled them.
The study also found that Biden voters and non-voters were largely unaffected by the disputed tags. Biden supporters, in particular, showed consistently low levels of belief in Trump’s false claims, regardless of whether the tweets had warning tags or not. Third-party voters and non-voters showed a slight decrease in belief in misinformation when exposed to the disputed tags, but this effect was marginal and not as pronounced as the findings for Trump voters.
One of the major limitations of the study was its reliance on a specific sample of participants recruited through an online platform, which may not fully represent the broader U.S. population. Additionally, the context of the 2020 election, with its unique political climate and widespread claims of fraud, may have influenced the results. It’s unclear whether the findings would apply to other situations where misinformation is less politically charged or less pervasive.
Another limitation was the study’s inability to pinpoint the exact mechanism behind why the disputed tags seemed to backfire for some Trump voters. The researchers speculated that psychological factors like cognitive dissonance—where people experience discomfort when confronted with information that contradicts their beliefs—could explain the results.
Alternatively, the disputed tags might have triggered reactance, a psychological response where people resist being told what to think, leading them to strengthen their original beliefs instead of questioning them. However, the study did not directly test these mechanisms, leaving room for further investigation.
The researchers suggested several directions for future research. One possibility is to explore whether more direct or detailed corrections—such as tagging misinformation as outright “false” rather than simply “disputed”—would be more effective in reducing belief in misinformation. They also recommended testing these interventions in different contexts to see if the findings hold true outside the highly polarized environment of the 2020 U.S. election.
The study, “Trump, Twitter, and truth judgments: The effects of ‘disputed’ tags and political knowledge on the judged truthfulness of election misinformation,” was authored by John C. Blanchar and Catherine J. Norris.