SOME COMPONENTS OF PROBABILISTIC INFERENCE.
Abstract
In most studies of probabilistic inference, subjects have behaved conservatively, i.e., they have revised their probability estimates less than the amount prescribed by Bayes's theorem. This study attempts to shed further light on the conservatism effect by analyzing the process of inductive inference into two components, and determining the effects on conservatism of these two components. One component is deduction: determining the probabilistic relationships between the datum observed and the hypotheses being considered. The second component is combination of the data: combining the effects of individual items of data in order to make an inference based on all the data observed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1966
- Accession Number
- AD0650223
Entities
People
- Lawrence D. Phillips
Organizations
- University of Michigan