非虚假设综合渐近正态法

目的:提出非虚假设综合渐近正态法。方法:将渐近正态法加以扩展以覆盖非虚假设和分层设计,从而推导出单样本和两样本基本关系式,功效函数,样本量,及检验统计量。结果:当最小感知差别取零时,它还原对应经典方法,包括经典Cochran检验和Mantel-Haenszel检验。当层数取1时,它还原为非虚假设渐近正态法,包括Dunnett-Gent检验。当最小感知差别取零且层数取1时,它还原为经典渐近正态法,包括单样本和两样本比例检验。结论:这种方法可用于分层设计的有效对照临床试验,以建立临床优效性,非劣效性,或等效性。

A Summary Asymptotic Normal Procedure of the Non-Null Hypothesis

Objective:This paper proposes a summary asymptotic normal procedure of the non-null hypothesis.Methods:The asymptotic normal procedure is extended so as to cover the non-null hypothesis and stratified design.Hence we derive the one-sample and two-sample equations of basic relationship,power functions,sample sizes,and test statistics.Results:They reduce to their classical counterparts including the classical Cochran test and the Mantel-Haenszel test setting the minimal detectable difference equal to zero.They reduce to the asymptotic normal procedure of the non-null hypothesis including the Dunnett-Gent test when there is only one stratum.And they reduce to the classical asymptotic normal procedure including the one-and two-sample tests on proportions setting both the minimal detectable difference equal to zero and the number of strata equal to unity.Conclusion:It can be applied to the active control clinical trials with stratified design for establishing the clinical superiority,non-inferiority,or equivalence.