On some properties of ML and REML estimators in mixed normal models with two variance components

Gnot, Stanisław ; Michalski, Andrzej ; Urbańska-Motyka, Agnieszka

Discussiones Mathematicae. Probability and Statistics, Tome 24 (2004) no. 1, pp. 109-126.

Résumé

In the paper, the problem of estimation of variance components σ₁² and σ₂² by using the ML-method and REML-method in a normal mixed linear model Y,E(Y) = Xβ, Cov(Y) = σ₁²V + σ₂²Iₙ is considered. This paper deal with properties of estimators of variance components, particularly when an explicit form of these estimators is unknown. The conditions when the ML and REML estimators can be expressed in explicit forms are given, too. The simulation study for one-way classification unbalanced random model together with a new proposition of approximation of expectation and variances of ML and REML estimators are shown. Numerical calculations with reference to the generalized Fisher's information are also given.

Détail

Keywords: mixed linear models, likelihood-based inference, ML- and REML- estimation, variance components, Fisher's information

Bibliographie

[1] D. Birkes and S.S. Wulff, Existence of maximum likelihood estimates in normal variance-components models, J. Statist. Plann. Inference 113 (2003), 35-47.

[2] T. Caliński and S. Kageyama, Block Designs: A Randomization Approach, Vol. I: Analysis, Lecture Notes in Statistics, Springer 2000.

[3] M.Y. Elbassouni, On the existency of explicit restricted maximim likelihood estimators in multivariate normal models, Sankhyā Series B 45 (1983), 303-305.

[4] S. Gnot, D. Stemann, G. Trenkler and A. Urbańska-Motyka, On maximum likelihood estimators in multivariate normal mixed models with two variance components, Tatra Mountains Mathematical Publications 17 (1999), 111-119.

[5] S. Gnot, D. Stemann, G. Trenkler and A. Urbańska-Motyka, Maximum likelihood estimators in mixed normal models with two variance components, Statistics 36 (4) (2002), 283-302.

[6] E.L. Lehman, Theory of Point Estimation, Wiley, New York 1983.

[7] A. Olsen, J. Seely and D. Birkes, Invariant quadratic unbiased estimation for two variance components, Ann. Statist. 4 (1976), 878-890.

[8] H.D. Patterson and R. Thompson, Recovery of intra-block information when block sizes are unequal, Biometrika 58 (1971), 545-554.

[9] H.D. Patterson and R. Thompson, Maximum likelihood estimation of components of variance, Proceedings of 8th Biometric Conference (1975), 197-207.

[10] C.R. Rao, MINQUE theory and its relation to ML and MML estimation of variance components, Sankhyā Series B 41 (1979), 138-153.

[11] C.R. Rao and J. Kleffe, Estimation of Variance Components and Applications, North Holland, Amsterdam 1988.

[12] Poduri S.R.S. Rao, Variance Components Estimation, Mixed Models, Methodologies and Applications, Chapman and Hall, London, Weinheim, New York, Tokyo, Melbourne, Madras 1997.

[13] S.R. Searle, G. Casella and Ch.E. McCulloch, Variance Components, Wiley, New York 1992.

[14] J. Seely, Minimal sufficient statistics and completeness for multivariate normal families, Sankhyā Ser.A 39 (1977), 170-185.

[15] G.W. Snedecor and W.G. Cochran, Statistical Methods (8th ed.), Ames, IA: Iowa State University Press 1989.

[16] S.E. Stern and A.H. Welsh, Likelihood inference for small variance components, The Canadian Journal of Statistics 28 (2000), 517-532.

[17] W.H. Swallow and J.F. Monahan, Monte Carlo comparison of ANOVA, MIVQUE, REML, and ML estimators of variance components, Technometrics 26 (1984), 4, 7-57.

[18] T.H. Szatrowski, Necessary and sufficient conditions for explicit solutions in the multivariate normal estimation problem for patterned means and covariances, Ann. Statist. 8 (1980), 803-810.

[19] T.H. Szatrowski and J.J. Miller, Explicit maximum likelihood estimates from balanced data in the mixed model of the analysis of variance, Ann. Statist. 8 (1980), 811-819.