People's thresholds of decision-making against a tornado threat using dynamic probabilistic hazard information

With pervasive use of smartphones for acquiring weather information and efforts of National Oceanic and Atmospheric Administration (NOAA) in developing a tool under the banner of the Forecasting a Continuum of Environmental Threats (FACETs) to disseminate probabilistic information about a tornado event, it is important to investigate people's probabilistic thresholds for taking protective action when presented with dynamic visual information about a tornado threat on a smartphone. We presented dynamic displays of probabilistic information of five hypothetical tornado scenarios to 109 college students on their smartphones and asked them to report at what moment they would take protective action (if any). After conducting Cox proportional hazard regression and Poisson regression, we found that proximity to the tornado, likelihood of the threat occurrence, and being inside vs. outside of the risk area played a significant role in people's decision-making. This study illustrated that regardless of tornado trajectory, as a moving probabilistic swath showing probabilistic forecast of the tornado becomes closer to the information recipients, around 12% of the participants would take protective action prior to being impacted by the probabilistic zone with more than 0% chance of the tornado occurrence. Almost half of them would take protective action before being impacted by the probabilistic zone with more than 20% chance, and around 88% of the participants would take protective action before being impacted by the probabilistic zone with more than 40% chance of the tornado occurrence. Our study corroborates previous relevant research that providing probabilistic hazard information to the public could enhance the warnings' effectiveness.