This article is cited in 21 papers

Decomposable statistics and hypothesis testing. The case of small samples

G. I. Ivčenko, Yu. I. Medvedev

Moscow

Abstract: Questions of statistical hypotheses testing are considered for polynomial outcome probabilities in the scheme of series under the conditions that the number of outcomes $N$ and the number of tests $n$ increase simultaneously to infinity so that $n/N$ remains bounded. A class of tests is studied which is based on decomposable statistics of outcomes, frequencies. In particular, some results are obtained concerning the classical $\chi^2$-test and maximum likelihood ratio test.

Received: 31.05.1976

 English version: Theory of Probability and its Applications, 1979, 23:4, 764–775

Bibliographic databases:

• 
• 
•