ESTIMATION OF CONTINUOUS SUBJECTIVE PROBABILITY DISTRIBUTIONS IN A SEQUENTIAL DECISION TASK,
Abstract
A range betting method (RBM) for measuring continuous subjective probability distributions is presented. Predictions derived from the method are tested in a computer-controlled sequential decision making task. In particular, twenty-two undergraduate students participated individually in the task. The subjects were asked occasionally ao place a wager upon the range of an unknown value of P, 0 < P < 1, generated by an LGP-30 computer. In addition, the subjects predicted the exact value of P by typing a number on a Flexowriter, and then estimated the value of P after obtaining information about its value. The information was given in the form of n values (1 < n < 20) of 0's or 1's generated by a binomial distribution with mean P and printed on the Flexowriter in front of the subject. Predicted decisions were derived both from the RBM method and from a relatively simple confidence limits' model. The subjects' estimates of the range boundaries were compared to the predicted values from both models. In both comparisons the results indicated a regression effect and a block-to-block tracking behavior. Attempts to compare two independent measures of the same subjective probability distribution were unsuccessful. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1964
- Accession Number
- AD0614684
Entities
People
- Amnon Rapoport
Organizations
- University of North Carolina at Chapel Hill