期刊名称：Sankhya. Series A, mathematical statistics and probability
印刷版ISSN：0976-836X
电子版ISSN：0976-8378
出版年度：2009
卷号：71
期号：02
页码：284--297
出版社：Indian Statistical Institute
摘要：Given a parametric statistical model, with ∈ IRp, if we assume, for any suf-
ficiently smooth prior, asymptotic normality for the multivariate posterior
distribution, the derivation of the reference prior and the probability match-
ing prior only depend on the structure of the Fisher information matrix.
In this paper we show that the derivation of the above mentioned objec-
tive priors can be substantially simplified under some specific assumptions
on the structure of the Fisher matrix. In particular, we generalize known
results about diagonal and blockwise diagonal Fisher matrices. We also ar-
gue, through several examples, that the sufficient conditions we state in our
results are quite common to meet in practice.
关键词：Objective Bayesian analysis, probability matching
priors, parameter orderings, reference priors.
