合理进行均值比较——随机区组设计定量资料多元方差分析

Reasonably carry out mean value comparison:MANOVA of the quantitative data collected from the randomized block design

本文目的是介绍与随机区组设计定量资料多元方差分析有关的基本概念、计算方法、一个实例以及SAS实现。基本概念包括区组因素、如何选定区组因素、随机区组设计和不完全随机区组设计;计算方法涉及一般统计量和检验统计量;一个实例涉及“长期饲喂高锌日粮对断奶仔猪免疫机能影响的动物试验及其多元定量资料”。借助SAS实现随机区组设计定量资料的一元方差分析和多元方差分析。并讨论当区组因素对结果的影响无统计学意义时,合理的统计分析方法是不考虑区组因素,直接采用单因素多水平设计一元和多元定量资料方差分析。

The purpose of this article was to introduce the basic concepts,calculation methods,an example and SAS implementation related to the randomized block design quantitative data multivariate analysis of variance(MANOVA).Basic concepts included block factors,how to select block factors,randomized block design and incomplete randomized block design.Calculation methods involved the general statistics and test statistics.The example involved long-term feeding of high-zinc diets animal experiments and multivariate quantitative data on the effect on the immune function of weaned piglets.With the help of SAS software,the one-way analysis of variance(ANOVA)and MANOVA for the quantitative data in the randomized block design were realized.And it was discussed that when the influence of block factors on the results was not statistically significant,the reasonable statistical analysis method was to directly use single factor multilevel design quantitative data univariate and multivariate ANOVA without considering block factor.