Rumors and rumor corrections on Twitter: Studying message characteristics and opinion leadership

As rumors often ripple across the cyberspace, posting rumor corrections on social media can bring about social good by spreading the truth. However, rumors and rumor corrections are not easily distinguishable from one another. Therefore, this paper investigates how three message characteristics, namely, the use of emotions, clarity and credible source attribution, can predict message veracity on social media. Message veracity denotes whether a message is a rumor or a rumor correction. In addition, the paper further examines the extent to which opinion leadership moderates the relation between message characteristics and message veracity. Set against the context of the death hoax of Singapore's first Prime Minister Lee Kuan Yew in March 2015, data for this paper came from Twitter. Analysis involved binary logistic regression. All the three message characteristics predicted veracity. Rumor corrections were characterized by lower use of emotions, higher clarity, and higher credible source attribution compared with rumors. Furthermore, opinion leadership moderated the relation between the use of emotions and message veracity as well as that between credible source attribution and message veracity.