Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
481744 | European Journal of Operational Research | 2009 | 11 Pages |
Abstract
Bayesian techniques specify how to update beliefs about a variable given information on that variable or related variables. In many cases, statistical analyses also provide information about the relationship between variables, but the Borel Paradox prohibits many natural ways of updating beliefs conditioned on information about a relationship. This paper presents a method by which beliefs can be updated without violating the Borel Paradox under certain circumstances. We apply our approach to relationships specified by a statistical model (i.e., regression), and relationships described by statistical simulation.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science (General)
Authors
Robert Bordley, Vicki Bier,