A CONCEPT OF DEPARTURE-FROM-UNIFORMITY MEASURE AND ITS IMPACT ON THE TESTING FOR INDEPENDENCE IN TWO-WAY CONTINGENCY TABLES OF MEDIUM DIMENSIONS Cover Image

POJĘCIE MIARY ODEJŚCIA OD RÓWNOMIERNOŚCI ORAZ JEJ WPŁYW NA TESTOWANIE NIEZALEŻNOŚCI W TABLICACH DWUDZIELCZYCH ŚREDNICH ROZMIARÓW
A CONCEPT OF DEPARTURE-FROM-UNIFORMITY MEASURE AND ITS IMPACT ON THE TESTING FOR INDEPENDENCE IN TWO-WAY CONTINGENCY TABLES OF MEDIUM DIMENSIONS

Author(s): Piotr Sulewski
Subject(s): Economy, Methodology and research technology
Published by: Szkoła Główna Gospodarstwa Wiejskiego w Warszawie
Keywords: two-way contingency table; test of independence; untruthfulness measure; bar method; Monte Carlo method;

Summary/Abstract: Even when null hypothesis H0 is true, test statistics may not follow the chi-square distribution. It takes place when the contingency table is filled with a small sample. The relevant quantile of the chi-square distribution is no longer a proper critical value. Again and again, also in this case, the Monte Carlo method turns out to be irreplaceable. Modeling H0 means generating such tables in which values ascribed to rows are independent of values ascribed to columns. In paper a departure-from-uniformity measure mn was defined. When H0 is true measure mn has a strong impact on distribution of the test statistics. So, determining test critical values one has to take mn into account.

  • Issue Year: XVII/2016
  • Issue No: 2
  • Page Range: 111-122
  • Page Count: 12
  • Language: Polish
Toggle Accessibility Mode