Social Network, Activity Space, Sentiment, and Evacuation: What Can Social Media Tell Us?

Hurricanes are one of the most common natural hazards in the United States. To reduce fatalities and economic losses, coastal states and counties take protective actions, including sheltering in place and evacuation away from the coast. Not everyone adheres to hurricane evacuation warnings or orders. In reality, evacuation rates are far less than 100 percent and are estimated using posthurricane questionnaire surveys to residents in the affected area. To overcome limitations of traditional data collection methods that are costly in time and resources, an increasing number of natural hazards studies have used social media data as a data source. To better understand social media users’ evacuation behaviors, this article investigates whether activity space, social network, and long-term sentiment trends are associated with individuals’ evacuation decisions by measuring and comparing Twitter users’ evacuation decisions during Hurricane Matthew in 2016. We find that (1) evacuated people have larger long-term activity spaces than nonevacuated people, (2) people in the same social network tend to make the same evacuation decision, and (3) evacuated people have smaller long-term sentimental variances than nonevacuated people. These results are consistent with previous studies based on questionnaire and survey data and thus provide researchers a new method to study human behavior during disasters. Key Words: big data, disaster management, evacuation, hurricane, social media.