A bayesian analysis of interrelated bernoulli processes with an application for clinical trials
作者:
Philip J. Smith,
期刊:
Communications in Statistics - Simulation and Computation
(Taylor Available online 1984)
卷期:
Volume 13,
issue 1
页码: 69-84
ISSN:0361-0918
年代: 1984
DOI:10.1080/03610918408812359
出版商: Marcel Dekker, Inc.
关键词: Bayes factor;Bayesian statistics;bivariate Bernoulli processes;interrelated Bernoulli processes;likelihood ratio;posterior probability;p-value;test of hypotheses
数据来源: Taylor
摘要:
Prior information regarding the interrelation of two Bernoulli processes may justify a clinical trial designed to corroborate this information. Antelman (1973) has studied the Dirichlet-beta which permits the expression of the prior knowledge of such interrelation. However, use of this prior distribution leads to complicated and intractable analyses. Alternately, such prior information regarding the interrelation of the processes may be adequately summarized by a simple Dirichlet distribution. Procedures for testing hypotheses regarding a priori interrelations of the success probabilities of the processes are given. Exact expressions for the posterior probabi1ities of these hypotheses are shown to be approximately equal to weighted p-values or 1ikelihood ratios.
点击下载:
PDF (386KB)
返 回