Uncertainty Forecasts Improve Decision Making Among Nonexperts

Abstract

The objective of this research was to evaluate the impact of weather uncertainty information on decision making in naturalistic settings. Traditional research often reveals deficits in human decision making under uncertainty as compared with normative models of rational choice. However, little research has addressed the question of whether people in naturalistic settings make better decisions when they have uncertainty information as compared with when they have only a deterministic forecast. Two studies investigated the effect of several types of weather uncertainty information on the quality of decisions to protect roads against icing and on temperature predictions and compared them with a control condition that provided deterministic forecast only. Experiment 1 was a Web-based questionnaire that included a single trial. Experiment 2, conducted in lab, included 120 trials and provided outcome feedback and a reward based on performance. Both studies indicated enhanced performance with uncertainty information. The best kind of uncertainty information tested here was the one that provided the probability at the threshold for the task at hand. We conclude that uncertainty information can be used advantageously, even when it does not result in perfectly rational performance, and that uncertainty can be communicated effectively to nonexpert end users, resulting in improved decision making.