合理进行均值比较——析因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 factorial design

本文目的是介绍与析因设计定量资料多元方差分析有关的基本概念、计算方法、两个实例以及SAS实现。基本概念包括交互作用、析因设计、含区组因素的析因设计和分式析因设计;计算方法涉及一般统计量和检验统计量;两个实例分别涉及“制造塑料胶片过程中挤压度和添加剂剂量对其性能的影响”和“寻找挤压塑料胶卷的最优工艺条件”。借助SAS对两个实例中的定量资料进行多元方差分析,并对如何确定最终的方差分析模型进行讨论。

The purpose of the paper was to introduce the basic concepts,calculation methods,two examples and SAS implementation related to the factorial design quantitative data multivariate analysis of variance(MANOVA).Basic concepts included interaction,factorial design,factorial design with block factor and fractional factorial design.Calculation methods involved the general statistics and test statistics.The two examples were respectively related to the influence of extrusion degree and additive dosage on its performance in the process of manufacturing plastic film and searching for the optimal process conditions of extruded plastic film.With the help of SAS software,MANOVA was carried out on the quantitative data in two examples,and how to determine the final analysis of variance model was discussed.