Statistical Tests Based on Empty Cells

Abstract

In professional literature on statistics, a constantly growing interest is observed in non-parametric methods. Commonly these methods are based on counting, rank or position statistics and on a number or length of series. In this paper, the least popular tests, namely tests based on a number of empty cells, are presented. David-Hellwig test and a two-sample consistency test are considered. Empirical power of the tests is presented in comparison to classic tests: Kolmogorov and Shapio-Wilk test for testing normality of a distribution and t-Student’s and Wilcoxon tests for testing consistency of two distributions.