Lessons from social media history for the war on fake news
Analyzing users' past social media posts can help predict who is likely to share fake news and inform interventions to potentially curb its spread.
Fake news on social media has emerged as one of the most critical problems of the digital era. With the exponential growth in social media use, the flow of misinformation has increased significantly, leading to the mass spread of misleading content.
What is fake news on social media?
Fake news refers to content designed to manipulate public perception, distorting the truth and affecting important decisions. Understanding what is the percentage of fake news on social media helps contextualize the scope of this challenge. Recent studies suggest that misinformation can account for a substantial portion of viral content, though exact percentages vary by platform and context.
New research proposes a new way to identify people who are likely to share fake news — and shows that the post histories of social media users can help to better predict people that are more likely to spread fake news. To do this, it is crucial to know how to identify fake news on social media, as the behaviour patterns of users who share it can provide clues about their propensity to spread misinformation.
The study, by Verena Schoenmueller (Esade), Simon J. Blanchard (McDonough School of Business, Georgetown University) and Gita Johar (Columbia Business School, Columbia University), was published in the Journal of Marketing Research. Its authors say they hope the methods they have developed will aid marketers and misinformation researchers in the ongoing battle against fake news.
Fake news on social media: Unreliable sources
The spread of fake news on social media, such as misinformation, disinformation, conspiracy theories and false claims, is an ever-present threat. The responsibility of social media platforms with regard to fake news in Spain and other countries has become a crucial issue, as these platforms have the power to rapidly amplify misinformation. With the World Economic Forum ranking it among the top five global risks for 2025.
Attempts to mitigate the spread of this fake news have typically focused on identifying its sources, with websites such as Hoaxy, Media Bias/Fact Check and NewsGuard labeling outlets as legitimate or otherwise.
But when news can spread across the globe in seconds, simply labeling its source does little to stem the flow. The spread of fake news on social media is amplified by the speed with which users share content without verifying it, facilitating its large-scale propagation. Few users pause to check the origins of the information they share, and mainstream media has been known to publish stories lacking veracity.
Manual fact-checking websites like Snopes and FactCheck.org offer insight into individual articles, but by the time they’re in the public domain the damage is done. Facts about fake news on social media show that a large proportion of users do not verify the source of news stories before sharing them, which contributes to the cycle of misinformation.
Researchers have recently turned their attention to the consumers, rather than creators, who share misleading content. However, many of these studies are limited. Controlled experiments using recruited participants may not accurately reflect those who are most likely to encounter and share fake news. In field experiments, researchers cannot know the content that users have seen but decided not to share.
Examples of fake news on Social MediaAlthough specific examples of fake news on social media may vary depending on the context and platform, common patterns can be identified in the nature of these stories. Fake news is generally characterised by:
These general patterns are useful for identifying fake news on social media and understanding how it spreads. Data about fake news on social media shows that misinformation on the most popular platforms often focuses on hot-button and emotionally charged topics, which are more likely to go viral. |
The importance of language in the spread of fake news
An important yet under-explored aspect of social media that can shed light on the promulgation of fake news is the post histories of known sharers. Comparing the language they use with others who share similar demographics and online behaviors can help to identify the patterns of fake news sharers. This, in turn, can predict future behavior and inform theories to influence effective interventions.
Post histories are a valuable source of data to formulate evidence-based approaches to combat fake news
Schoenmueller and her co-authors analyzed the post histories of selected Twitter users over a series of studies. In the first, they identified 66 articles that the factchecking website Snopes had flagged as containing misinformation. The Twitter usernames, location and gender of those who had shared the original articles were collected.
Three comparison groups were then formed: one containing users who had shared the fact-checked pages from Snopes; a random sample of Twitter users; and a sample of Twitter users matched based on the sociodemographics of the group that shared the fake news. A second dataset was gathered from users who had shared at least one article from publishers that had been identified by the website Hoaxy as more likely to publish inaccurate claims. The last 3,200 posts of each user were analyzed.
Why is fake news on social media dangerous?The danger of fake news on social media lies in the speed at which it spreads and the lack of effective control over the content that circulates. Understanding the effect of fake news on social media is critical for developing effective countermeasures. Misinformation can lead to dangerous behaviour, influence elections, undermine public health initiatives, and erode trust in legitimate institutions. Shining light on the 'dark side' of social media How does fake news affect social media users?Research shows that fake news affects social media platforms by creating echo chambers where users are repeatedly exposed to the same viewpoints. This limits open discussion and deepens divisions, leading to increased polarization within communities. During critical events, fake news can also spread panic and misinformation, making it harder for people to access accurate and reliable information. |
Spreading negativity
Several insights gained by the researchers aligned with existing literature. Fake news sharers tended to be older, more active on social media, and have conservative leanings. Links were also found between intense emotional states: those who shared fake news were more likely to use emotional language associated with negative emotions such as anger and anxiety.
The personality traits of sharers were also consistent with previous findings, revealing higher levels of neuroticism and openness, with lower levels of extroversion, agreeableness and conscientiousness. There was also a higher incidence of power- and death-related words from fake news sharers.
A predictive framework using machine learning classifiers was then developed by the researchers to quantify the extent to which post histories could accurately predict fake news sharers. The findings confirmed that incorporating textual cues from post histories did improve the ability to classify those likely to share fake news.
Exploring interventions to combat fake news
The final two exploratory experiments recruited Twitter users from research platforms to examine how textual cues from post histories related to future sharing intentions. First, a state of anger was induced by asking 398 participants to describe an article they had read that made them feel extremely angry. Half of the participants were then randomly assigned a mitigating condition to induce calm.
After being shown a series of news headlines, they were asked to rate their willingness to share each article, whether they believed the articles to be accurate, and the importance they placed on believing the articles to be accurate, surprising, interesting, aligned with their politics, and funny. The results suggest that anger is correlated with higher rates of sharing both fake and genuine news. The interventions to mitigate anger had no impact.
Finally, 481 Twitter users were shown a timeline that included an ad for a fact-checking browser extension. Half of the group were shown an advertising message that had been modified to strengthen the emphasis on control and power. They were then asked to indicate whether they were likely to click on the ad or download the extension. They were also asked to complete scales that measured their ‘desire for control’ and ‘personal sense of power’.
Valuable data in the fight against fake news on social media
When the survey results were analyzed in line with textual cues from post histories collected by the researchers, they indicated that the use of empowering language did increase the frequency of both clicking on the ad and downloading the browser.
The researchers say the study highlights the use of post histories as a valuable source of data to formulate evidence-based approaches that aim to combat fake news. However, they note that the Twitter API used in their analysis has since been removed following the rebranding of Twitter to X, making studies of this type unfeasible for researchers with limited budgets.
In this sense, the study also serves to remind funding bodies and social media platforms of the importance of facilitating reasonable access to APIs.