Reference priors in multiparameter nonregular cases

被引:18
|
作者
Ghosal, S
机构
[1] Indian Statistical Institute,Division of Theoretical Statistics and Mathematics
关键词
asymptotic expansion; Bayes risk; discontinuous densities; Kullback-Leibler number; multidimensional parameter; posterior distribution; reference prior;
D O I
10.1007/BF02564432
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
The reference prior in the sense of Bernardo is derived in some multiparameter nonregular cases. The family of densities we consider have discontinuities at some points which depend on one component of the parameter (say, theta) while, for fixed values of theta, the family is regular with respect to the other components (say, phi). We obtain the reference prior through an asymptotic expansion of Lindley's measure of information. The expansion is in itself of some importance. The results are illustrated using examples.
引用
收藏
页码:159 / 186
页数:28
相关论文
共 50 条