Conditional Nearly Best Linear Invariant Estimation of the Scale and Location Parameters of the First Extreme Value Distribution.
Abstract
Conditional, nearly best, linear invariant estimators of the scale and location parameters of the first extreme value distributions are developed using order statistics. A previously untried approximate covariance is used in place of the approximation of Gunnar Blom. Coefficients of the order statistics are calculated and tabled for sample sizes of 1 to 40. The tables include coefficients for estimation using complete samples and samples censored from above or symmetrically censored from both ends. A technique for simultaneous estimation using the conditional coefficients is described. Mean squared errors are listed for all estimators, and a comparison between these mean squared errors and the mean squared errors using the approximation of Blom is made. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1972
- Accession Number
- AD0743617
Entities
People
- Alfred A. Boyd Jr
Organizations
- Air Force Institute of Technology