How emergency managers engage Twitter users during disasters

Purpose Natural disasters are increasingly more frequent and intense, which makes it critical for emergency managers to engage social media users during crises. This study examined emergency official accounts' social media engagement at each disaster stage based on Fink's four-stage model of crisis and disaster: prodromal, acute, chronic and termination stages and linked topics and sentiments to engagement. Design/methodology/approach Using text mining and sentiment analysis, 1,226 original tweets posted by 66 major emergency official Twitter accounts and more than 15,000 retweets elicited across the life cycle of Hurricane Irma were analyzed. Findings Results identified the most engaging official accounts and tweets. Most tweets and the most engaging tweets were posted in the prodromal stage. Tweets related to certain topics were significantly more engaging than others. The most frequently tweeted topics by official accounts were less engaging than some seldom tweeted topics. Negative sentiment words increased the engagingness of the tweet. Sadness was the strongest predictor of tweet engagement. Tweets that contained fewer sadness words were more engaging. Fear was stronger in positively predicting tweet engagement than anger. Results also demonstrated that words for fear and anger were critical in engaging social media discussions in the prodromal stage. Words for sadness made the tweets less engaging in the chronic stage. Originality/value This study provided detailed instructions on how to increase the engagingness of emergency management official accounts during disasters using computational methods. Findings have practical implications for both emergency managers and crisis researchers.