UNSOLVED PROBLEMS OF EXPERIMENTAL STATISTICS
成果类型：
Article
署名作者：
TUKEY, JW
刊物名称：
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
ISSN/ISSBN：
0162-1459
DOI：
10.2307/2281535
发表日期：
1954
页码：
706-731
关键词：
摘要：
The author shows how these problems are generated naturally as a consequence of the following hypergeneral principles: (1) Different ends require different means and different logical structures. (2) In each area, statistical method must and does evolve, mainly by adding both immediate ends and considerations. (3) While techniques are important in experimental statistics, knowing when to use them and why to use them are more important. (4) In the long run, it does not pay a statistician to fool either himself or his clients. Each of these is then broken down into 4 or 5 general principles which are illustrated by the discussion of particular problems. Finally, 51 specific questions are stated which imply many, but not all, of the existing unsolved problems.