Description and analogy in the practice of statistics
Description and analogy in the practice of statistics
复制标题
统计实践中的描述和类比
DOI：
10.1093/biomet/66.2.195
复制
发表时间：
1979
期刊：
影响因子：
2.7
通讯作者：
P. D. Finch
中科院分区：
文献类型：
--
作者：
P. D. Finch
我们将讨论统计工作中描述和推理的作用之间有争议的对比。我们从 Kempthorne & Doerfler (1969) 使用的虚构实验开始，强调有时考虑“这个实验本身告诉我们什么？”这个问题是有用的。肯普索恩和多弗勒提出这个问题是为了形成一种观点，即当人们使用相同的实验单元时，在虚构的重复实验中会发生什么，然而，这些重复在实践中是无法实现的，因为实验本身“破坏”了相关单元。对于这些作者来说，实验的随机设计提供了适当的重复群体。假设我们希望检查介绍性统计文本 D 在统计入门理论课程的教学中是否优于竞争对手 D'。即将入学的 26 名学生随机分为两组，人数相等，由同一位教师在不同时间授课，一组使用文本 D，另一组使用文本 D'。课程结束时，两组进行共同考试并记录他们的分数。假设 26 个考试成绩按升序排列，没有并列，D 类的排名为
We shall be involved with controversial contrasts between the roles of description and inference in statistical work. We start with the fictitious experiment used by Kempthorne & Doerfler (1969) to emphasize that it is sometimes useful to consider the question 'What does this experiment, on its own, tell us?' Kempthorne & Doerfler posed this question as the formation of opinion about what would happen in fictitious repetitions of an experiment when one used the same experimental units, those repetitions being unrealizable in practice, however, because the experiment itself 'destroys' the units in question. For those authors an appropriate population of repetitions was provided by the randomized design of the experiment. Suppose that we wish to examine if an introductory statistical text D is superior to a competitor D' in the teaching of a beginning theory of statistics course. An incoming class of 26 students is divided at random into two groups of equal size which are taught by the same instructor at different times, one group using text D and the other text D'. At the end of the course the two groups take a common examination and their scores are recorded. Suppose that the 26 examination scores are ranked in order of increasing magnitude, that there are no ties and that the ranks in class D are