Abstract
Fans of the National Basketball Association (NBA) assigned probability judgments to the outcomes of upcoming NBA games, and rated the strength of each team involved. The probability judgments obtained from these “expert” subjects exhibited high intersubject agreement and also corresponded closely to the eventual game outcomes. A simple model that associates a single strength value with each team accurately accounted for the probability judgments and their relationship to the ratings of team strength. The results show that, in this domain at least, probability judgments can be derived from direct assessments of strength which make no reference to chance or uncertainty.
Highlights
To P(A ú B) and to P(B ú C), and a low value to P(A ú C)
The results of this study demonstrate that probability judgments for tournaments can be represented in terms of the normalized strength of the two teams involved in a given game
A strength model based on an extension of support theory (Tversky & Koehler, 1994) accurately accounted for the probability judgments of basketball fans predicting game outcomes; more general models imposing fewer constraints did not improve on the quantitative fit achieved by the strength model
Summary
To P(A ú B) and to P(B ú C), and a low value to P(A ú C). Neither standard probability theory nor support theory constrains the relationship among these three estimates. It is suggested that in many situations people’s judgments about the outcomes of a tournament may satisfy the following simple model. Assume that for each team in the tournament, the judge has a value s(A), interpreted as the strength of team A. The judged probability that team A will beat team B, is given by the following strength model: P(A ú B) Å s(A). In a recent article, Tversky and Koehler (1994) proposed a new model of subjective probability, called support theory, in which the judged probability of a hypothesis is given by the support (or strength of evidence) of that hypothesis normalized relative to the support of its alternative. According to this model, the judged probabilities associated with the results of a tournament depend only on the strengths of the respective teams. The model assumes no interactions; no team is expected to play especially well or especially poorly against any specific opponent
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
