Experimental discussion of the (x2, P) test for goodness of fit.
成果类型：
Article
署名作者：
Pearson, K
刊物名称：
BIOMETRIKA
ISSN/ISSBN：
0006-3444
DOI：
10.2307/2331972
发表日期：
1932
页码：
351381
关键词：
摘要：
When the sample or any class frequency is small the Incomplete B- or r-Function Tables will give a better value of P than Elderton''s Table. The main use of x 2 was intended to be the comparison of a considerable graduated sample of a parent population with further samples in order to test whether such samples were likely to be samples from the parent population, only known through this graduated sample. Experimental sampling shows that the x''2''s and the Ps from the graduated sample are highly correlated with the %2''s and P''s from the parent population. Even for small samples a far better P is obtained from a graduated than from a raw sample. The x 2 comparison of 2 samples can not properly be regarded as a contingency table. A sample graduated by aid of moments is a possible sample from a population; the degrees of freedom are not limited.