HE Plots for Repeated Measures Designs
HE Plots for Repeated Measures Designs
复制标题
重复测量设计的 HE 图
DOI：
10.18637/jss.v037.i04
复制
发表时间：
2010
影响因子：
5.8
通讯作者：
M. Friendly
中科院分区：
文献类型：
--
作者：
M. Friendly
Friendly（2007）中介绍的假设误差（HE）图提供了可视化多变量线性模型中假设检验的图形方法，通过将假设和误差协变显示为椭球体并提供效应大小和显著性的可视化表示。这些方法在R（Fox，Friendly和Monette 2009 a）和SAS（Friendly 2006）的heplots中实现，通常适用于具有固定效应因子（MANOVA），定量回归（多变量多元回归）和组合病例（MANCOVA）的设计。本文介绍了这些方法的重复测量设计，其中的多变量响应代表一个或多个“受试者内”因素的结果的扩展。这个扩展是用R的hepplots来说明的。示例描述了单样本特征分析、具有多个S间和S内因子的设计以及双多变量设计，其中在多个场合观察到多变量响应。
Hypothesis error (HE) plots, introduced in Friendly (2007), provide graphical methods to visualize hypothesis tests in multivariate linear models, by displaying hypothesis and error covariation as ellipsoids and providing visual representations of effect size and significance. These methods are implemented in the heplots for R (Fox, Friendly, and Monette 2009a) and SAS (Friendly 2006), and apply generally to designs with fixed-effect factors (MANOVA), quantitative regressors (multivariate multiple regression) and combined cases (MANCOVA). This paper describes the extension of these methods to repeated measures designs in which the multivariate responses represent the outcomes on one or more “within-subject” factors. This extension is illustrated using the heplots for R. Examples describe one- sample profile analysis, designs with multiple between-S and within-S factors, and doubly- multivariate designs, with multivariate responses observed on multiple occasions.