期刊名称：Brazilian Journal of Probability and Statistics
印刷版ISSN：0103-0752
出版年度：2003
卷号：17
期号：1
页码：91-105
出版社：Brazilian Statistical Association
摘要：Bayesian inference in factor analytic models has received re-newed attention in recent years, partly due to computational advances butalso partly to applied focuses generating factor structures as exemplified byrecent work in financial time series modeling. The focus of our current workis to investigate the commonly overlooked problem of prior specification andsensitivity in factor models. We accomplish that by implementing P′erez andBerger's (1999). Expected Posterior (EP) prior distributions. As opposedto alternative objective priors, such as Je.reys' prior and Bernardo's prior,EP prior has several important theoretical and practical properties, with itsstraightforward computation through MCMC methods and coherence whencomparing multiple mo dels perhaps the most important ones
关键词：Bayes' factors; Bayesian inference; expected posterior prior;latent factor models; model selection criteria; model uncertainty
