Fake news may go viral more quickly when it uses words associated with anger.
Jichang Zhao at the Beijing Advanced Innovation Center for Big Data and Brain Computing and Yuwei Chuai at Beihang University in China have analysed the spread of both fake and real news across Weibo, one of China’s biggest social media platforms.
They analysed a data set of 22,479 posts from 20,532 users between 2011 and 2016, which Weibo had officially flagged as . They also looked at 10,000 real news posts from 1527 users.
Advertisement
The duo tasked a team of nine people with manually identifying the emotional content of posts, by scouring them for words that appear in a list of more than 6000 emotional terms. The reference list included 1323 terms associated with anger, 2066 with joy, 1243 with sadness, as well as terms relating to disgust and fear.
Each post was given an emotion rating – for example, if five angry words appeared in a 10-word post, it would be assigned an anger rating of 50 per cent. “Compared with real news, fake news carries much more anger and less joy,” says Zhao.
The proportion of anger in was three times greater than that in real news with few shares. Across all posts, was nearly 6 per cent angrier than real news and 17 per cent less joyful.
The pair then surveyed 1316 active Weibo users to identify people’s motivations for sharing fake news. They found that angry posts evoked greater feelings of anxiety, which created an incentive for sharing.
Flagging such angry posts could give social media users pause to critically analyse their content before reposting, say the pair. Labelling posts rated as more than 20 per cent angry would flag about 46 per cent of highly reposted fake news on Weibo, they found – but the catch is that it would also flag an estimated 22 per cent of popular real news.
To determine whether the findings also apply to fake news in English, the researchers used a similar approach to analyse a data set of more than 20,000 fake Twitter posts and 20,000 real news tweets. They found that fake news posts were on average angrier than real news posts. A comparison between fake news tweets that were highly shared and those that weren’t also showed that highly shared tweets were on average angrier.
The finding isn’t entirely unexpected, says Rowan Zellers at the University of Washington, Seattle. “Much disinformation is designed to play on people’s emotions so that they share it.”
Reference:
