How social media twists retracted COVID-19 papers, fueling mistrust in science

A recent study published in Public Understanding of Science has shed light on how retractions of scientific papers, intended to correct errors and maintain the integrity of research, can be twisted on social media to spread misinformation and erode public trust in science. Researchers found that when articles about COVID-19 were retracted, the way people interpreted these retractions online depended heavily on their pre-existing beliefs about the pandemic and science itself.

The COVID-19 pandemic saw an unprecedented amount of scientific research emerge at rapid speed. However, this urgency also led to a number of studies being published that contained errors or even fabricated data. When these flawed studies are discovered, the scientific community has a system of retraction, where journals publicly withdraw the problematic articles. This process is meant to ensure the scientific record remains accurate and reliable.

However, scientists became concerned that in a climate of already declining public trust in science and increasing political division, retractions might backfire. Instead of reassuring the public that science corrects its mistakes, retractions could be misinterpreted as evidence that science is unreliable, biased, or even corrupt. This concern is amplified by the widespread use of social media, where retracted articles and discussions around them can quickly spread and be taken out of context.

“I’m generally interested in how people and organizations use (and misuse) science in order to advance their agendas,” said study author Rod Abhari, a doctoral candidate in the School of Communication at Northwestern University. “There’s a long history of this in the United States with corporations and conservative politicians misrepresenting science in order to obfuscate scientific consensuses around issues like tobacco harms and climate change. I became interested in retractions specifically during COVID-19, after seeing many accounts on Twitter, including Fox News commentator Laura Ingraham, using conspiracy theories about retractions in order to claim that science should not be trusted.”

In their study, the researchers examined discussions on the social media platform Twitter (now known as X) surrounding two specific scientific articles about COVID-19 that were later retracted. They chose these two articles because they were among the most widely shared retracted COVID-19 studies online, ensuring there was enough public discussion to analyze.

The first article, referred to as Mehra20 by the researchers, had made claims about the harmful effects of hydroxychloroquine, a drug that was controversially promoted as a COVID-19 treatment, particularly by then-President Donald Trump. This article was retracted due to concerns about the reliability of its data. The second article, called Rose21, focused on potential heart-related side effects of COVID-19 vaccines and was also retracted, although the exact reasons for its retraction were less clearly stated publicly.

The researchers collected a large dataset of 1,723 tweets that contained links to either the Mehra20 or Rose21 articles. They looked at tweets posted both before and after each article was officially retracted, over a period stretching from early February 2020 to the beginning of June 2023. To analyze the content of these tweets, they used a method called manual content analysis. This involved reading each tweet and categorizing them based on the topics discussed.

The researchers developed a set of categories that emerged from reading the tweets themselves and considering existing research on how scientific issues are discussed in public debates. These categories included whether a tweet discussed the findings of the article, whether it mentioned flaws in the study’s methods, whether it speculated about the motivations of the authors or publishers in initially publishing the article, and whether it discussed the reasons behind the article’s retraction.

To ensure the analysis was consistent and reliable, two independent confederates coded each tweet, and their coding was compared. When there were disagreements, a senior researcher made the final decision. This process helped to ensure that the analysis was as objective as possible. The researchers then looked at the patterns in these categories to understand how the online conversation around these retracted articles evolved over time and how it differed between the two articles.

The analysis revealed striking differences in how the retractions of Mehra20 and Rose21 were discussed on social media. In the case of Mehra20, before its retraction, many tweets uncritically shared its findings that hydroxychloroquine was harmful. Some of these tweets even used the article as evidence to criticize Donald Trump’s promotion of the drug. However, even before the retraction, a significant number of tweets also pointed out flaws in the study’s methods and called for it to be retracted. This critical discussion gained momentum after prominent scientists publicly raised concerns about the article.

Once Mehra20 was officially retracted, the online conversation shifted. Very few tweets continued to defend the article’s findings. Instead, the focus turned to discussing the flaws of the study and speculating about the motivations behind its initial publication. Many tweets suggested that the article had been published with a political agenda to undermine hydroxychloroquine and, by extension, Donald Trump.

Some even promoted conspiracy theories, suggesting that pharmaceutical companies had orchestrated the article to discredit a cheap drug and promote more profitable treatments. Conservative commentators also amplified this narrative, claiming the media had eagerly promoted the flawed article initially to attack Trump but were now downplaying its retraction.

The response to the retraction of Rose21 was very different. Before its retraction, almost all tweets sharing Rose21 presented its findings – suggesting vaccine harms – uncritically. Many users highlighted the fact that the article had been peer-reviewed as proof of its validity and used it to argue against vaccine policies or personal vaccination.

In stark contrast to Mehra20, very few tweets before retraction mentioned any flaws in Rose21. When Rose21 was retracted, the reaction from those who had shared it was largely to ignore the retraction or to frame it as an act of censorship. Many users claimed that the retraction was evidence of a conspiracy to suppress the “truth” about vaccine harms, often blaming pharmaceutical companies or a corrupt “establishment.”

The findings show that “that retractions are sometimes misinterpreted, including by prominent conservative pundits like Robert F. Kennedy and Senator Ron Johnson, as proof that scientific publishers are colluding with the Democratic Party, mainstream media, and the pharmaceutical industry,” Abhari told PsyPost. “In reality, retractions are an important and necessary part of the scientific process, and we should be more worried about scientific credibility if articles were not being retracted.”

“However, inconsistencies in the retraction process, including a lack of transparency in retraction notices, have helped these theories spread. To combat the spread of conspiracy theories, scientific publishers need to do better about issuing clear and transparent retraction notices in accordance with established retraction guidelines.”

The authors of Rose21 themselves contributed to this narrative, with one author publicly describing the retraction as political censorship. This framing allowed those who supported the article’s initial claims to dismiss the retraction as illegitimate and to continue using the article’s findings to support their anti-vaccine stance.

“I was surprised that some of the conspiracy theories around specific retractions were actually spread by scientists who had their research retracted, including Didier Raoult and Peter McCullough,” Abhari said. “Rather than accepting their own errors or misconduct, these scientists instead presented their retractions as censorship on social media in order to situate themselves as courageous truth tellers, expand their online followings, and sow deeper scientific distrust.”

The researchers acknowledge that their study has some limitations. They only looked at discussions on Twitter, so the patterns observed might not be representative of all social media platforms or online discussions more broadly. They also focused on only two highly publicized retracted articles, and most retractions likely receive far less public attention. Furthermore, their analysis only included tweets that contained direct links to the articles, potentially missing some relevant discussions that did not include links.

Nevertheless, the study highlights a serious challenge for science communication. In a politicized environment, retractions, which are intended to be a sign of scientific rigor and self-correction, can be twisted to fuel misinformation and deepen distrust in science. The research underscores the importance of transparent and clear communication from journals and publishers when issuing retractions.

“Despite how unlikely retractions conspiracies are, retractions, like other scientific publishing decisions, are not democratic decisions,” Abhari told PsyPost. “They rely on the expert judgment of editorial staff and typically happen in closed door meetings. Usually the public is only privy to retraction notices, which range from detailed descriptions of the incident to no description at all. Thus, it’s unsurprising that, at a time of widespread scientific distrust, those who consume science skeptical media might also question the motivations for retractions.”

The long-term goal is to provide insights that can provide insights for better science policy and institutional management.

“I am interested in researching the effects of scientific governance on public trust in science,” Abhari said. “Public trust in science is essential for science to have a public impact, from promoting healthy behaviors to passing policies to protect our futures. My current studies are looking at how science credibility is discussed in news media coverage of retractions and the cognitive effects of exposure to retractions on scientific trust. My ultimate goal is to inform better decisions at all levels of scientific governance.”

The study, ““They Only Silence the Truth”: COVID-19 retractions and the politicization of science,” was authored by Rod Abhari and Emőke-Ágnes Horvát.