Abstract
The problem under consideration is the Γ-minimax estimation, under L2loss, of a multivariate normal mean when the covariance matrix is known. The family Γ of priors is induced by mixing zero mean multivariate normals with covariance matrix τI by nonnegative random variables τ, whose distributions belong to a suitable family G. For a fixed family G, the linear (affine) Γ-minimax rule is compared with the usual Γ-minimax rule in terms of corresponding Γ-minimax risks. It is shown that the linear rule is "good", i.e., the ratio of the risks is close to 1, irrespective of the dimension of the model. We also generalize the above model to the case of nonidentity covariance matrices and show that independence of the dimensionality is lost in this case. Several examples illustrate the behavior of the linear Γ-minimax rule.
Similar content being viewed by others
REFERENCES
Albert, J. H. (1984). Estimation of a normal mean using mixtures of normal priors, Tech. Report, Bowling Green State University, Ohio.
Berger, J. O. (1985). Statistical Decision Theory and Bayesian Analysis, Springer, Berlin.
Bickel, P. J. (1981). Minimax estimation of the mean of a normal distribution when the parameter space is restricted, Ann. Statist., 9, 1301-1309.
Brown, L. (1971). Admissible estimators, recurrent diffusions and insoluble boundary value problems, Ann. Math. Statist., 37, 1087-1136.
Brown, L. (1986). Fundamentals of Statistical Exponential Families, with Applications in Statistical Decision Theory, Lecture Notes vol. 9, Institute of Mathematical Statistics.
DasGupta, A. (1992). Distributions which are Gaussian convolutions, Tech. Report, No. 92-34, Department of Statistics, Purdue University, West Lafayette, Indiana.
DasGupta, A., Ghosh, J. K. and Zen, M. M. (1990). Bayesian analysis of multivariate normal mean with flat tailed priors, Tech. Report, No. 90-22, Department of Statistics, Purdue University, West Lafayette, Indiana.
Donoho, D., Liu, R. C. and McGibbon, B. (1990). Minimax risk over hyperrectangles and implications, Ann. Statist., 18, 1416-1437.
Efron, B. and Morris, C. (1971). Limiting the risk of Bayes and empirical Bayes estimators—part I: the Bayes case, J. Amer. Statist. Assoc., 66, 807-815.
Efron, B. and Olshen, R. (1978). How broad is the class of normal scale mixtures?, Ann. Statist., 5, 1159-1164.
Faith, E. R. (1978). Minimax Bayes estimators of multivariate normal mean, J. Multivariate Anal., 8, 372-379.
Ibragimov, I. A. and Has'minskii, R. Z. (1984). Nonparametric estimation of the values of a linear functional in Gaussian white noise, Theory Probab. Appl., 29, 1-32.
Kudō, H. (1967). On partial prior information and the property of parametric sufficiency, Proc. Fifth Berkeley Symp. on Math. Statist. Prob., 251-265, University of California Press, Berkeley.
Robert, C. (1990). Hidden mixtures and Bayesian sampling, Rapport technique 115, LSTA Université Paris 6.
Skibinsky, M. and Cote, L. (1962). On the inadmissibility of some standard estimates in the presence of prior information, Ann. Math. Statist., 34, 539-548.
Solomon, D. L. (1972). A-minimax estimation of multivariate location parameter, J. Amer. Statist. Assoc., 67; 641-646.
Vidakovic, B. and DasGupta, A. (1996). Efficiency of linear rules for estimating a bounded normal mean, Sankhyā Ser. A, 58, 81-100.
Watson, S. R. (1974). On Bayesian inference with incompletely specified prior distributions, Biometrika, 61, 193-196.
West, M. (1987). On scale mixtures of normal distributions, Biometrika, 74, 646-648.
Author information
Authors and Affiliations
About this article
Cite this article
Vidakovic, B. Linear Versus Nonlinear Rules for Mixture Normal Priors. Annals of the Institute of Statistical Mathematics 51, 111–124 (1999). https://doi.org/10.1023/A:1003835319163
Issue Date:
DOI: https://doi.org/10.1023/A:1003835319163