Examining the Dynamic Ways That People Evaluate and Respond to Evolving Hurricane Risks

Hurricanes are dynamic, evolving in space and time. Forecasts of hurricanes are now available days before a storm makes landfall, and they are updated and refined as the threat evolves. Public officials use this weather risk information to inform recommended protective actions (e.g., evacuation orders) and for other preparations (e.g., road closures, public transportation interruptions). An unprecedented ability now exists to communicate this dynamic hurricane risk information via multiple sources and channels, including with social media. Just as hurricane risks and communication about them are dynamic, so too are the ways that people evaluate and respond to them. This presentation will discuss a study of these dynamic processes through content analysis of Twitter data from people who were in the Far Rockaway neighborhood of New York City leading up to Hurricane Sandy’s landfall. The Twitter data offer a quasi-real-time record of these at-risk Twitter users’ perspectives as Sandy approached and as evolving information about it was communicated. Drawing on relevant risk theories, including the Protective Action Decision Model, we deductively and inductively investigate the risk information people obtained and used, their risk perceptions, and their protective behaviors and other responses, along with influencing factors, such as past experience and situational barriers and motivations. In doing so, we identify how risk information, risk perceptions, and responses differed across Twitterers and how these changed as the Hurricane Sandy threat evolved. The analysis reveals that the predominant risk information--i.e., forecasts, evacuation orders, social cues, and environmental cues--referred to by the Twitterers shifted as Sandy approached and made landfall. The analysis also reveals that several specific emotions emerged (e.g., fear, worry, anger, excitement) at different times and with different implications. Finally, the Twitterers displayed a range of behavioral responses that reflect their perceptions of the evolving risk; examples include evacuating a day or more before landfall based on the evacuation order, preparing on the day of landfall for loss of utilities (power and water), and moving to a different location based on environmental cues as the storm approached. This presentation will further discuss these results and their implications for improving understanding about how, when, and why people respond to evolving hurricane risks in the ways that they do.