Rumors are regular features of crisis events due to the extreme uncertainty and lack of information that often characterizes these settings. Despite recent research that explores rumoring during crisis events on social media platforms, limited work has focused explicitly on how individuals and groups express uncertainty. Here we develop and apply a flexible typology for types of expressed uncertainty. By applying our framework across six rumors from two crisis events we demonstrate the role of uncertainty in the collective sensemaking process that occurs during crisis events.

Widespread rumoring can hinder attempts to make sense of what is going on during disaster scenarios. Understanding how and why rumors spread in these contexts could assist in the design of systems that facilitate timely and accurate sensemaking. We address a basic question in this line: To what extent does rumor evolution occur (1) through reliance on a centralized information source, (2) in parallel information silos, or (3) through a web of complex informational interactions? We develop a conceptual model and associated analysis algorithms that allow us to distinguish between these possibilities. We analyze a case of rumoring on Twitter during the Boston Marathon Bombing. We find that rumor spreading was predominantly a parallel process in this case, which is consistent with a hypothesis that information silos may underlie the persistence of false rumors. Special attention towards detecting and resolving parallel information threads during collective sensemaking may hence be warranted.

Widespread rumoring can hinder attempts to make sense of what is going on during disaster scenarios. Understanding how and why rumors spread in thesecontexts could assist in the design of systems that facilitate timely and accurate sensemaking. We address a basic question in this line: To what extent does rumor evolution occur (1) through reliance on a centralized information source, (2) in parallel information silos, or (3) through a web of complex informational interactions? We develop a conceptual model and associated analysis algorithms that allow us to distinguish between these possibilities.