APPROXIMATE RANDOMIZATION TESTS
APPROXIMATE RANDOMIZATION TESTS
复制标题
DOI：
10.1080/00223980.1969.10543491
复制
发表时间：
1969-01-01
影响因子：
2.7
通讯作者：
EDGINGTO.ES
中科院分区：
文献类型：
--
作者：
EDGINGTO.ES
在一篇关于随机化检验的文章中，简要讨论了近似随机化检验(1)。本文将更详细地考虑近似随机化检验。然而，首先有必要纠正前面文章中的一句话。在随机化检验中，获得的统计量的意义是抽样分布中与获得的统计量一样大的统计量的比例，而不是如错误地陈述的那样，超过获得的统计量的比例。需要对所计算的重要性值进行校正，以使其成为在零假设下获得这种极端结果的概率。
In an article on randomization tests, approximate randomization tests were briefly discussed (1). This paper will consider approximate randomization tests in further detail. First, however, it is necessary to correct a statement in the earlier article. In a randomization test, the significance of an obtained statistic is the proportion of the statistics in the sampling distribution that are as large as the obtained statistic, not, as was erroneously stated, the proportion that exceed the obtained statistic. The correction is required for the computed significance value to be the probability, under the null hypothesis, of obtaining such extreme results.