逆抽样条件下配对试验设计中优比的统计推断  

Statistical Inference for Odds Ratio in Matched Pairs Under Inverse Sampling

作　　者：江绍萍[1] 李慧敏[1] 缪清[1] Jiang Shaoping;Li Huimin;Miao Qing(School of Mathematics ＆ Computer Sciences,Yunnan Minzu University,Kunming 650500,China)

机构地区：[1]云南民族大学数学与计算机科学学院

出　　处：《统计与决策》2018年第17期5-8,共4页Statistics & Decision

基　　金：国家自然科学基金资助项目(11461083;31360277);云南省教育厅科学研究基金项目(2015Z106)

摘　　要：在现代医学研究中,优比通常被用于配对实验设计下两治疗方案等价性评价问题的研究。在研究中加入逆抽样过程,并采用Fisher-score方法求解感兴趣参数的方差,从而导出六个检验统计量：wald型检验统计量（基于样本方差和基于原假设下方差）、对数wald型检验统计量（基于样本方差和基于原假设下方差）、Score检验统计量、似然比检验统计量。文章通过对各种检验犯第一类错误的概率和实际功效的模拟研究,比较了各统计量渐进检验的统计性能。结果表明Score统计量是最优的,在相同参数设置下其既能保证犯第一类错误概率最接近于显著性水平,又能使得功效达到最大。In modern medical research, odds ratio is often used in matched pair design to assess the equivalence of two treatments. This paper adds inverse sampling process in the study, and uses fisher-score method to solve the variance of the interested parameters, thus deriving six test statistics： the wald-type test statistic（with sample variance and original hypothetical variance）,the logarithmic wald-type test statistic（with sample variance and original hypothetical variance）, score test statistic, score and likelihood ratio test statistic. Through simulation studies on the probability and actual efficacy of various types of test errors, the paper compares the statistical performance of the asymptotic test of each statistic. The results show that the score statistics is optimal, and under the same parameter setting, it can not only guarantee the probability of making the first kind of error is closest to the significance level, but also make the efficiency reach the maximum.

关 键 词：优比 逆抽样 检验统计量 配对试验 

分 类 号：O212[理学—概率论与数理统计]