Michael L. DeKay

Heinz School of Public Policy and Management
Carnegie Mellon University

"Reversals in Assessments of Outcome Utilities: The Result of Precautionary Decision Making?"

Abstract:

In a series of experiments, participants reported their utilities for outcomes of real-world binary decisions and their threshold probabilities for taking specified actions (e.g., responding to a dam-failure warning). Across several scenarios, many participants evaluated false positives (e.g., evacuating when there is no dam failure) as more desirable than true negatives (e.g., not evacuating when there is no dam failure). The most common rationale for such utility reversals was the precautionary maxim "it is better to be safe than sorry." Participants appeared to have evaluated these outcomes on the basis of the decisions that could lead to them, although such behavior is irrational from the perspective of expected utility theory (or any other consequentialist theory of decision making). Consistent with this explanation, the number of utility reversals was substantially reduced when the link between decisions and outcomes was weakened by taking participants out of the role of decision maker. A final study provides evidence that people's initial preferences for a course of action distort their evaluations of information in ways that support these preferences, and that this distorted information may be associated with utility reversals that imply that precautionary action is the dominant strategy. These results amplify the importance of information distortion by documenting its role in high-consequence decisions for which precautionary actions have been proposed or adopted. In this talk, we will describe some of these experiments and results, but we are particularly interested in obtaining reactions and learning about alternatives to the way in which we have framed the problem.

Joint work with Paul S. Fischbeck and Dalia Patino Echeverri

Back to Talks Page