SEQUENTIAL METHOD OF SAMPLING FOR DECIDING BETWEEN 2 COURSES OF ACTION
成果类型：
Article
署名作者：
WALD, A
刊物名称：
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
ISSN/ISSBN：
0162-1459
DOI：
10.2307/2965624
发表日期：
1945
页码：
277-306
关键词：
摘要：
The sequential plan provides a rule for continuing to accumulate cases until the total evidence is strong enough to decide between 2 possible courses of action. Since a decision is made on the basis of a sample, rather than the whole population, there will be some chance that a wrong choice will be made. The investigator is at liberty to decide what risks of wrong decisions he can afford. These 2 risks correspond to 2 chosen values of a parameter, such as an average value or the % of cases having a given characteristic. The commonest application is for deciding when to accept and when to reject a lot of articles. The basis of the test is the likelihood ratio, which measures how relatively greater a chance one chosen value of the parameter has of producing the actual sample, than has another chosen value. When this evidence is sufficiently strong a decision is made in favor of one of the values and the indicated action taken. The calculations involved in setting up a sequential plan are quite simple. The cases presented are (a) % of cases having a characteristic, (b) difference between 2 percents, and (c) the mean, for known population standard deviation. It would seem that many applications could be made in the biological sciences, since the plans provide for a minimum average amt. of testing or experimentation.