| description abstract | This article presents an analysis of the risk warning messages sent via Twitter during Hurricane Irma, which struck Key West, Florida and the Florida mainland on September 10, 2017. The study analyzed 497 tweets sent by six emergency management organizations between September 1, 2017 and September 14, 2017, for the presence of five warning elements: hazard, guidance, location, time, and source. Using descriptive statistics and negative binomial regression models, this study confirmed that tweets can contain all five warning elements, and tweets that contain these elements have a positive effect on retweet counts, while controlling for message style, microstructures, account features, and time effect. In the context of Hurricane Irma, organizations managed to include all five warning elements in a tweet that contains a graphic image. The findings also indicate, however, that not all warning elements had a significant impact on retweet counts. Consequently, the five warning elements should be viewed as a guide that can help organizations design risk warning messages that reflect to their contextual needs. | |
