Abt K
Department of Biomathematics, Medical School, University of Frankfurt, Germany.
Stat Med. 1991 May;10(5):777-95. doi: 10.1002/sim.4780100513.
In controlled clinical trials the problem of multiplicity of desired inferential statements finds attention at an increasing rate. In this paper the previously proposed concept of Descriptive Data Analysis (DDA), situated between Confirmatory and Exploratory Data Analysis, is applied to the planning aspects of controlled trials for which the problem of multiplicity exists. The (non-Bayesian) DDA planning concept should provide the investigator with tools to draw final conclusions from data of several variables possibly observed at several time points in possibly several groups of subjects by combining his pre-trial medical experience with descriptive inferential statements (confidence intervals and test results) at nominal significance levels. DDA also provides for confirmatory statements concerning individual null hypotheses and partially global hypotheses.
在对照临床试验中，所需推断性陈述的多重性问题正日益受到关注。本文将先前提出的描述性数据分析（DDA）概念应用于存在多重性问题的对照试验的规划方面，该概念介于验证性数据分析和探索性数据分析之间。（非贝叶斯）DDA规划概念应为研究者提供工具，使其能够通过将试验前的医学经验与名义显著性水平下的描述性推断性陈述（置信区间和检验结果）相结合，从可能在几组受试者中的几个时间点观察到的多个变量的数据中得出最终结论。DDA还提供了关于单个原假设和部分全局假设的验证性陈述。
