English

Critical issues with the Pearson's chi-square test

Methodology 2025-05-13 v1

Abstract

Pearson's chi-square tests are among the most commonly applied statistical tools across a wide range of scientific disciplines, including medicine, engineering, biology, sociology, marketing and business. However, its usage in some areas is not correct. For example, the chi-square test for homogeneity of proportions (that is, comparing proportions across groups in a contingency table) is frequently used to verify if the rows of a given nonnegative m×nm \times n (contingency) matrix AA are proportional. The null-hypothesis H0H_0: ``mm rows are proportional'' (for the whole population) is rejected with confidence level 1α1 - \alpha if and only if χstat2>χcrit2\chi^2_{stat} > \chi^2_{crit}, where the first term is given by Pearson's formula, while the second one depends only on m,nm, n, and α\alpha, but not on the entries of AA. It is immediate to notice that the Pearson's formula is not invariant. More precisely, whenever we multiply all entries of AA by a constant cc, the value χstat2(A)\chi^2_{stat}(A) is multiplied by cc, too, χstat2(cA)=cχstat2(A)\chi^2_{stat}(cA) = c \chi^2_{stat} (A). Thus, if all rows of AA are exactly proportional then χstat2(cA)=cχstat2(A)=0\chi^2_{stat}(cA) = c \chi^2_{stat}(A) = 0 for any cc and any α\alpha. Otherwise, χstat2(cA)\chi^2_{stat} (cA) becomes arbitrary large or small, as positive cc is increasing or decreasing. Hence, at any fixed significance level α\alpha, the null hypothesis H0H_0 will be rejected with confidence 1α1 - \alpha, when cc is sufficiently large and not rejected when cc is sufficiently small, Yet, obviously, the rows of cAcA should be proportional or not for all cc simultaneously. Thus, any reasonable formula for the test statistic must be invariant, that is, take the same value for matrices cAcA for all real positive cc. KEY WORDS: Pearson chi-square test, difference between two proportions, goodness of fit, contingency tables.

Keywords

Cite

@article{arxiv.2505.06318,
  title  = {Critical issues with the Pearson's chi-square test},
  author = {Vladimir Gurvich and Mariya Naumova},
  journal= {arXiv preprint arXiv:2505.06318},
  year   = {2025}
}
R2 v1 2026-06-28T23:27:40.299Z