Eliciting Public Risk Preferences in Emergency Situations.

We investigate decisions in the face of emergency situations where potential consequences include loss of life, for example, the choice of methods to rescue a group of people who are trapped in a fire. In particular, we conduct two sets of experiments to depict public risk preferences for potential fatalities in a rescue mission. The first set of experiments asks binary questions on a collection of hypothetical emergency scenarios such as fires, air crashes, and marine accidents. Our results show that people are inclined to choose risky options in emergency situations, indicating risk-seeking behavior. However, we observe no significant decline in the subjects' willingness to save one more life as the base number of people in danger increases, which contradicts the risk-seeking utility function implied by expected utility theory. We then conduct the second set of experiments to elicit possible distortions of probabilities in addition to the value function of fatalities under prospect theory. The value functions of most subjects are concave with increasing marginal effect of losing one more life, and most subjects deflate the probability of an adverse event no matter whether the probability is small or large. We compare our results with prospect theory studies involving money and time in the loss domain, and provide possible explanations for the observed discrepancies. [ABSTRACT FROM AUTHOR]