Seems Legit: An Investigation of the Assessing and Sharing of Unverifiable Messages on Online Social Networks

Abstract

Unverifiable messages abound on the Internet. As policymakers and social media platforms grapple with the spread of misleading, false, or otherwise harmful messages, it is important they better understand why users share messages they cannot verify. This article reports on two studies that shed light on such issues. In the first study, the authors leverage secondary data collected from Twitter to show that true and false unverifiable messages have different characteristics and that those characteristics are predictive of retweeting. In the second study, they conduct a controlled experiment to explain why such characteristics influence resharing. Jointly, these studies show that leaks (i.e., true-but-unverifiable) tend to be more plausible, more vivid, and are sent by more credible senders than rumors (i.e., false-but-unverifiable). Further, the relationships among these variables is multiplicative such that the effects of vividness and sender credibility are strengthened for plausible and weakened for implausible messages. Finally, message recipients use these characteristics to determine whether an unverifiable message is novel and/or helpful. In sum, the authors find that social media users are likely to reshare unverifiable messages when they exhibit characteristics of plausibility, vividness, and sender credibility, which signal the novelty of helpfulness of the message.