Analysis of Messy Data, Volume I: Designed Experiments
Analysis of Messy Data, Volume I: Designed Experiments
复制标题
杂乱数据分析，第一卷：设计实验
DOI：
10.1080/00401706.1985.10488628
复制
发表时间：
1985
影响因子：
6.7
通讯作者：
J. Schmee
中科院分区：
文献类型：
--
作者：
J. Schmee
这卷书勇敢地试图解决一个非常复杂的问题，尽管统计软件广泛存在，但仍然充满了困难。到目前为止，大多数关于实验设计的教科书都回避了分析不平衡数据或涉及缺失或离群观察的数据所涉及的复杂性。这本书确实是开辟了新天地，通过犁通过无数的相当实际的设计情况，根本无法解释的方法，涉及漂亮的数学。材料是在一个可以理解的水平上提出的工人与坚实的，但不一定压倒性的背景分析线性模型。一门好的方差分析课程应该足以让你以令人振奋的速度阅读本书。粗略的浏览表明，读者可以通过获得SAS研究所（1982）的GLM程序来增加本书的益处。前四个涵盖了多重比较和设计结构的概念等基本技术。第5-8章继续介绍基础知识，但与具有类似呈现水平的书籍相比，人们会惊讶地发现分裂情节，重复测量设计和可评估性早期解释。在这里，作者明确了均值模型相对于效应模型的优势。它们涵盖了几个有趣的案例研究中的第一个案例中的主效应和交互效应的对比。这些案例研究是讨论中的各种情况的有指导意义的例子，并提供了简明的统计解决方案。不幸的是，作者仅限于介绍统计结果，很少将统计语言翻译成原始研究者的概念。
This volume makes a valiant attempt at tackling a very complex subject that, despite the wide availability of statistical software, remains fraught with difficulties. So far most textbooks on experimental design have shied away from the intricacies involved in analyzing unbalanced data or data involving missing or outlying observations. This book is indeed breaking new ground by plowing through a myriad of rather practical design situations that simply cannot be interpreted by methods involving pretty mathematics. The material is presented at a level understandable to workers with a solid but not necessarily overwhelming background in the analysis of linear models. A good course in the analysis of variance should be sufficient to proceed through this book at a motivating pace. A cursory browsing reveals that a reader might increase the book's benefits by having available the GLM procedure of the SAS Institute (1982).The book is separated into 32 brief chapters. The first four cover such basic techniques as multiple comparisons and the concept of a design structure. Chapters 5-8 continue to cover basics, but one is startled to find split-plot, repeated measurement designs, and estimability explained early when compared to books with a similar level of presentation. Here the authors make clear the advantages of the means model over the effects model. They cover contrasts of main and interaction effects in the first of several interesting case studies. These case studies are instructive examples of the various situations under discussion presented with a concise statistical solution. Unfortunately, the authors restrict themselves to a presentation of the statistical results and rarely translate statistical language into the concepts of the original investigator.