Logistic响应分布分位数的鞍点近似置信区间  

Saddlepoint Approximate Confidence Intervals for the p-th Quantile of Logistic Response Distribution

作　　者：田玉斌[1] 张英[1] 李国英[2] 

机构地区：[1]北京理工大学理学院数学系,北京100081 [2]中国科学院数学与系统科学研究院,北京100080

出　　处：《数学物理学报（A辑）》2005年第1期110-118,共9页Acta Mathematica Scientia

基　　金：国家自然科学基金(10371126)资助

摘　　要：论文基于响应数据,应用鞍点近似方法,给出构造Logistic响应分布分位数的近似置信区间的方法.论文还对这种置信区间进行了模拟,并将该方法应用于QD-8电雷管.模拟和实例结果表明,当样本量较小时,该方法能够较好地推断Logistic响应分布的分位数.For the p-th quantile of Logistic response distribution, by using the saddlepoint approach, the method to construct the approximate confidence intervals is presented based on response data. A simulation for calculating these confidence intervals is conducted. The proposed approximation is also applied to a real data set of QD-8 electric detonators. Results obtained from the simulation and application show that the method provides more precise inference for the p-th quantile of Logistic response distribution with small sample sizes.

关 键 词：响应分布 鞍点近似 置信区间 极大似然估计 

分 类 号：O212.1[理学—概率论与数理统计]