Two-stage estimators of the common mean of k normal populations are considered, and compared with the Graybill-Deal estimator in terms of asymptotic variance. Some of them are found to be superior to the estimator under regularity conditions. A comparison of the Graybill-Deal estimator, an alternative estimator and the maximum likelihood estimator is made when samples are of the same size. In addition, several simulation results concerning finite sample behaviour of proposed two-stage estimators are presented.
- Asymptotic relative efficiency
- Estimation of common mean
- Two-stage procedure
ASJC Scopus subject areas
- Statistics and Probability