Sources of Bias in Naturalistic Decision Making Under Risk

Abstract

Severe weather situations such as tornadoes and droughts require people to take protective action even when the probability of the severe weather is low because the consequence of not protecting is very serious. In naturalistic decision experiments based on these situations, people are risk-seeking such that they often do not take protective actions when it is economically rational to do so. This project studied this phenomenon from a signal detection theory perspective. A random likelihood model was introduced to estimate the subjective criterion which is the likelihood above which one takes protection actions. This model separates the subjective criterion from subjective likelihood, which is participants’ perception of the probability of the weather event. Three experiments manipulated the gain-loss framing and the economically rational criterion (the criterion based on expected value theory) to examine their effect on the subjective criterion. When the gain-loss framing was manipulated, the subjective criterion was higher in a loss frame than a gain frame. When the economically rational criterion was manipulated, the subjective criterion was between the economically rational criterion and the center of the possible likelihood range (50%). Neither manipulation affected subjective likelihood. In addition, participants showed an overestimation in subjective likelihood in all conditions. The shifted subjective criterion overcame this overestimation and resulted in risk-seeking decisions in some conditions. Thus, the random likelihood model analysis suggests that shift of the subjective criterion is the source of risk-seeking decisions in naturalistic decision tasks. Potential interventions are discussed with the aim to improve the placement of the subjective criterion.