Hypothesis Testing and Confidence Intervals for a Multiplicative Poisson Model With Applications to Reliability and Bioassay.
Abstract
ATIVE INTEGERS AND LET (Y sub ij) be mutually independent Poisson random variables with parameters (n sub ij)(alpha sub i)(beta sub j) respectively. N = (n sub ij) is called the design matrix and (n sub ij) may be regarded as the number of observations on independent Poisson random variables with parameters (alpha sub i)(beta sub j). Let theta = the product from j=1 to m of beta sub j. This is a natural parameter of interest in models using increased severity testing in reliability theory and in bioassay problems. When certain conditions on the design matrix are satisfied, tests of hypotheses and confidence intervals for theta are obtained. The randomized forms of these tests and confidence intervals will be uniformly most powerful similar tests and uniformly most accurate unbiased confidence intervals respectively. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1973
- Accession Number
- AD0773605
Entities
People
- Andrew P. Soms
- Bernard Harris
Organizations
- University of Wisconsin–Madison