Johnson W D, Grender J M
Department of Biometry and Genetics, Louisiana State University Medical Center, New Orleans 70112-1393.
J Biopharm Stat. 1993 Mar;3(1):1-12. doi: 10.1080/10543409308835045.
Nonparametric methods are presented for the analysis of the two-treatment, two-period crossover design with multivariate response. After forming within-subject sums and differences, the usual tests, including those for carry-over effects and direct treatment effects, can be constructed using a multivariate analysis of variance. When the assumptions relevant to traditional analysis are in question, multivariate nonparametric tests based on ranks provide a realistic alternative. However, multivariate response gives rise to a wider class of hypotheses than is seen in crossover designs with univariate response. We discuss these hypotheses and formulate nonparametric tests for associated analyses.
本文介绍了用于分析具有多变量响应的两治疗、两阶段交叉设计的非参数方法。在形成受试者内的和与差之后，可以使用多变量方差分析构建常规检验，包括用于残留效应和直接治疗效应的检验。当与传统分析相关的假设存在疑问时，基于秩的多变量非参数检验提供了一种切实可行的替代方法。然而，多变量响应产生的假设类别比单变量响应的交叉设计中所见的更广泛。我们讨论这些假设并为相关分析制定非参数检验。
