A Spatial Analysis of Decisions Made in Response to Simulated Tornado Warnings in the United States

Although decision-making in response to tornado warnings is well researched, most studies do not examine whether individual responses to these warnings vary across different geographical locations and demographic groups. This gap is addressed by using data from a decision experiment that places participants virtually in a simulated tornado warning and asks them to minimize the costs of their decisions. The authors examine the following: 1) what demographic attributes may contribute to choices to minimize costs to protect assets at a specific location in a tornado warning, 2) whether there is a spatial component to how these attributes influence decision-making, and 3) if there are specific U.S. regions where individuals do not make protective decisions that minimize their overall cost. Multilevel regression analysis and poststratification are applied to data from the simulated decision experiment to estimate which demographic attributes and National Weather Service County Warning Areas are most associated with the costliest protective decisions. The results are then analyzed using spatial autocorrelation to identify spatial patterns. Results indicate that sex, race, and ethnicity are important factors that influence protection decisions. Findings also show that people across the southern portions of the United States tend to make more costly protective decisions, as defined in this work.SIGNIFICANCE STATEMENT: Tornadoes, although rare, threaten both life and property. Studies have shown that certain demographic groups are more negatively impacted by disasters than others and are at higher risk of severe weather hazards. We ask if there are demographic characteristics or geographic locations in common among people who are more prone to making protection decisions during tornado warnings to minimize economic costs. Results can help warning providers, such as the National Weather Service, direct resources and education to specific types of decision-makers or locations to improve sheltering decisions.