2X2 TABLES - THE POWER FUNCTION OF THE TEST ON A RANDOMIZED EXPERIMENT
成果类型：
Article
署名作者：
PEARSON, ES; MERRINGTON, M
刊物名称：
BIOMETRIKA
ISSN/ISSBN：
0006-3444
DOI：
10.2307/2332355
发表日期：
1948
页码：
331345
关键词：
摘要：
Various problems lead to analysis by 2 x 2 tables. The one considered here is that in which two treatments, the old (or no) treatment A and the improved treatment B are each given to half of a sample of N individuals such that X of the N would react favorably if given either treatment, Y if given treatment B, and Z would not react to A or to B. The hypothesis Y = 0 is to be tested against alternatives Y>0. The data are observed in the form of a 2x2 table, [image] The appropriate test is whether Xi is as small as a critical value given by Fisher''s exact or hypergeometric distribution based on the observed marginal totals. The power function of this test is obtained for nominal significance levies of 5% and 15% for N = 20, 30 and 50 various combinations of X, Y and Z. An appendix reconciles differences of notation with an earlier paper by Finney (Biometrika, 1948).