On the equality of estimators under a general partitioned linear model with parameter restrictions
- 109 Downloads
Assume that a linear regression model is written as a partitioned form. In such a case, it is quite convenient to determine the role of each subset of regressors, and to derive estimators of unknown partial parameters in the partitioned model. In this paper, we consider the relationships between the well-known ordinary least-squares estimators (OLSEs) and the best linear unbiased estimators (BLUEs) of the whole and partial mean parameter vectors in a general partitioned linear model with parameter restrictions. We first review some known results on the OLSEs and the BLUEs and their properties under general linear models. We then present a variety of necessary and sufficient conditions for OLSEs to be BLUEs under a general partitioned linear model with parameter restrictions.
KeywordsPartitioned linear model Parametric functions Estimability OLSE BLUE Estimator equality
Mathematics Subject Classification62F12 62F30 62J10
- Alalouf IS, Styan GPH (1984) Characterizations of the conditions for the ordinary least squares estimator to be best linear unbiased. Chaubey YP, Dwivedi TD (Eds), Topics in Applied Statistics, Department of Mathematics, Concordia University, Montréal, pp 331–344Google Scholar
- Baksalary JK, Puntanen S, Styan GPH (1990) On T.W. Anderson’s contributions to solving the problem of when the ordinary least-squares estimator is best linear unbiased and to characterizing rank additivity of matrices. In: Styan GPH (ed) The Collected Papers of T.W. Anderson: 1943–1985. vol 2, pp 1579–1591. Wiley, New YorkGoogle Scholar
- Puntanen S, Styan GPH (1989) The equality of the ordinary least squares estimator and the best linear unbiased estimator, with comments by O. Kempthorne, S.R. Searle, and a reply by the authors. Am Stat 43:153–164Google Scholar