Essays in decision making: Risk, reward, and opportunity costs.

摘要

Do decision makers ignore pieces of information when making risky decisions? Brandstätter, Gigerenzer & Hertwig (2006) presented a fast and frugal heuristic of decision under risk called the priority heuristic (PH), where choices between two risky prospects are made according to a priority rule – first compare the minimum outcomes, then compare the probabilities of the minimum outcomes, then compare the maximum outcomes, and finally compare the probabilities of the maximum outcomes. We tested the validity of the PH by constructing risky choice "ladders," in which one dimension of the risky prospects – either probability or outcome – is systematically varied within each ladder. Participants in our study chose between two options within these ladders, which tested whether participants were sensitive to information when the PH predicts insensitivity. In the first demonstration, the PH predicts insensitivity to common differences in probabilities, but we found that participants were indeed sensitive to these common differences. In the second demonstration, the PH predicts insensitivity to a lower priority dimension based on priority dominance—the priority of a higher priority dimension over the lower priority dimension. However, we found that participants were indeed sensitive to lower priority dimensions the PH ignores. In the final demonstration, the PH predicts that choices involving prospects with three possible outcomes ignore the magnitude of the intermediate valued outcomes, but we found that participants were indeed sensitive to the magnitude of intermediate outcomes. In summary, we found that participants systematically choose between two risky prospects using information that the priority heuristic ignores. Our findings add to a growing body of evidence against the descriptive validity of the priority heuristic, and support algebraic models such as prospect theory.;1This research was conducted in collaboration with George Wu and Alex Markle. Special thanks to Reid Hastie, who assisted during the planning stages of this research.