Final Report on Contract N00014-79-C-0522.
Abstract
The research performed on this contract addressed statistical methodologies that would be appropriate in censored survival analyses and competing risks models. Both general parametric and nonparametric models were considered. The problems of efficient estimation of parameters of the underlying failure time distribution were addressed when the data may only be partly observable due to the presence of censorship. The methodology developed permits costs associated with monitoring of experimentation and of recruitment of units to be incorporated into the basic infrastructure leading to the description and properties of optimal procedures that minimize overall expected cost. These optimal designs are inherently sequential in nature, in that the optimal stopping time or sample size is solely determined by the observed information gathered by experimeter.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1985
- Accession Number
- ADA162703
Entities
People
- Joseph C. Gardiner
Organizations
- Michigan State University