Search

Highlights • We find an inverse S-shape in stated subjective probabilities in a novel experiment. • Prelec's (1998) conditional invariant model provides the best fit. • We find a large degree of individual heterogeneity. Abstract We examine the nature of stated subjective probabilities in a complex, evolving context in which participants are not told what the actual probability is: we collect information on subjective expectations in a computerized car race game wherein participants must bet on a particular car but cannot influence the odds of winning once the race begins. In our setup, the actual probability of the good outcome (a win) can be determined based on computer simulations from any point in the process. We compare this actual probability to the subjective probability stated by participants at three different points in each of six races. In line with previous research in which participants have direct access to actual probabilities, we find that the inverse S-shaped curve relating subjective to actual probabilities is also evident in our far more complex situation, and that there is only a limited degree of learning through repeated play. We show that the model in the inverse S-shaped function family that provides the best fit to our data is Prelec's 1998 conditional invariant model. We also find that individuals who report a greater degree of ambiguity are more pessimistic and less responsive to actual changes in real probabilities. [ABSTRACT FROM AUTHOR]

Bayes Rule specifies how probabilities over parameters should be updated given any kind of information. But in some cases, the kind of information provided by both simulation and physical experiments is information on how certain output parameters may change when other input parameters are changed. There are three different approaches to this problem, one of which leads to the Garbage-In/Garbage-Out Paradox, the second of which (Bayesian synthesis) violates the Borel Paradox, and the third of which (Bayesian melding) is a supra-Bayesian heuristic. This paper shows how to derive a fully Bayesian formula which avoids the Garbage-In/Garbage-Out and Borel Paradoxes. We also compare a Laplacian approximation of this formula with Bayesian synthesis and Bayesian melding and find that the Bayesian formula sometimes coincides with the Bayesian melding solution. [ABSTRACT FROM AUTHOR]