TESTING AND MERGING OF CONFIDENCE AREAS.
Abstract
It is assumed that an entity can be described by two parameters (e.g., its position in a plane) and that the maximum likelihood estimates of these parameters with associated fiducial probabilities constitute an estimate. Two problems are addressed: (1) what is the probability that two estimates pertain to the same entity, and (2) if the two estimates pertain to the same entity, what is the optimum new estimate and associated confidence regions. Although the first problem can not be solved in its entirety without knowledge of a priori probabilities, a conditional probability is derived under the assumption that the given estimates can be described by a bivariate normal distribution. The second problem is solved under the same assumption. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 08, 1968
- Accession Number
- AD0673402
Entities
People
- M. E. B. Owens
- M. J. Romer
Organizations
- United States Naval Research Laboratory