The spread of misinformation and fake news on social media platforms has become a global concern, with substantial risks associated with its dissemination. While efforts have been made to fact-check and flag individual pieces of content, these often fall short due to the overwhelming volume of posts shared across platforms like X and Meta. In response to this challenge, a research study conducted by by Verena Schoenmueller (Esade), Simon J. Blanchard (McDonough School of Business, Georgetown University) and Gita Johar (Columbia Business School, Columbia University), sheds light on the significant information contained in social media users' post histories that can help identify those most likely to spread fake news. These users can then be targeted with misinformation mitigation efforts.
The Role of Post Histories in Predicting Fake News Sharing
Rather than focusing solely on the content of shared articles, the researchers delved into the post histories of social media users to uncover patterns in language that could help predict whether an individual is likely to share fake news. This shift in focus offers a new approach to the issue, emphasizing the behavioral cues in users' previous posts rather than relying solely on the misinformation content itself. By analyzing the last 3,200 tweets from users sharing fake news and multiple comparison groups, the team was able to identify distinct linguistic markers and emotional patterns in the posts of individuals who shared fake news.
The study found that users who shared fake news tended to use language associated with power, anger, and existential concerns such as death and religion. They were also more likely to display high levels of neuroticism and openness, while being less agreeable, extroverted, and conscientious. These findings suggest that certain emotional states and personality traits could be reliable indicators of a user's propensity to share misinformation.
The Emotional Landscape of Fake News Sharers
The study reveals a heightened use of emotional language by fake news sharers. Words associated with high-arousal emotions like anger and anxiety were significantly more prevalent in their posts compared to those of random social media users. The research also revealed a higher frequency of power -related terms, suggesting that fake news sharers tend to use more language related to power and control. A subsequent exploratory experiment further indicates that Twitter users who use more power-related language actually report a lower sense of personal power.
Interestingly, the emotional state of users did not only correlate with fake news sharing but also with the willingness to share fact-checked articles. Both groups exhibited similar emotional triggers, suggesting that those who share fake news and those who share fact-checks are driven by similar emotional responses, particularly anger.
A Predictive Framework for Identifying Fake News Sharers
By applying machine learning techniques to the textual cues found in users' post histories, the researchers were able to build a predictive model to identify individuals who are likely to share fake news. This framework holds potential for social media platforms looking to improve their ability to target fact-checking efforts and mitigate the spread of misinformation. Rather than attempting to manually fact-check every post, platforms could focus on users who exhibit the linguistic and emotional characteristics associated with fake news sharing.
Testing Interventions to Reduce Fake News Sharing
To highlight how interventions can be derived from the findings, the researchers conducted exploratory experiments to test the effectiveness of various strategies in reducing fake news sharing. One such intervention involved manipulating anger. However, the authors found that attempting to mitigate anger in the moment had no significant impact on users' sharing behavior. This finding suggests that anger might be a chronic trait among fake news sharers, rather than a temporary emotional reaction. In a second exploratory experiment, the authors found that an ad for a fact-checking browser extension was more effective - both in terms of click-through and downloading - when it used language that emphasized users’ sense of power and control.
Conclusion: A New Approach to Combatting Fake News
This research emphasizes the importance of post histories as a valuable source of data for predicting and preventing the spread of fake news. As social media platforms continue to grapple with the challenge of misinformation, the insights gained from this research provide a novel and actionable approach to identifying and potentially influencing the behaviors of fake news sharers.
Article details
Who Shares Fake News? Uncovering Insights from Social Media Users’ Post Histories
Verena Schoenmueller, Simon J. Blanchard, Gita Venkataramani Johar
First Published: September 1, 2024
DOI: 10.1177/00222437241281873
Journal of Marketing Research
