Remark to an unambiguity of estimators in the universal linear statistical models with the type ii constraints

Abstract

Listen

Translate article

Abstract In the universal linear statistical model with the type II constraints, estimates of the unbiasedly estimable linear functions of the parameters of the mean value vector are given by special types of generalized matrix inverse. Since there exists many versions of such matrix inverse it is of some interest to check the unambiguity of obtained estimators. The aim of the paper is to find the best unbiased linear estimators of unbiasedly estimable functions and to show that they do not depend on the choice of the used generalized inverse of the matrices.