基于共轭先验分布的两小样本Bayesian检验及实例分析  

Two Small Samples Bayesian Test Based on Conjugate Prior Distribution and the Instance Analysis

作　　者：赵丽华[1,2] 刘桂芬[2] 刘志永[2] 卢准炜[1] 

机构地区：[1]太原理工大学理学院,山西太原030024 [2]山西医科大学卫生统计教研室,山西太原030001

出　　处：《中北大学学报（自然科学版）》2011年第1期108-110,共3页Journal of North University of China(Natural Science Edition)

基　　金：山西省自然科学基金资助项目(200911005-2);太原理工大学校青年基金资助项目

摘　　要：介绍基于共轭先验分布的两小样本Bayesian检验原理及方法,并与经典方法作对比研究.当两组资料确实存在差别且为小样本时,基于共轭先验分布的两小样本Bayesian检验方法比传统的两样本t检验法得到的置信区间范围要短,精度要高.本文通过实例分析验证了此结论.The theory and method of Bayesian test for two small samples based on conjugate prior distribution were introduced and compared with the classical method.While there exists surely difference between two parameters which are of small samples,conclusions drawn from Bayesian are more steadily than that from classical t test and can be interpreted reasonably.By comparing the Bayesian test with the classical t-test for two samples,the confidence interval of the former is shorter,and the accuracy is higher.The conclusion was verified by the instance analysis.

关 键 词：共轭先验分布 Bayesian检验 两样本t检验 高血脂 低密度脂蛋白 

分 类 号：O212.8[理学—概率论与数理统计]