Exact algorithms for computing p-values of statistics-linear combination of 3-nomial variables

  • Authors:

  • Farid Beninel;Gérard Grélaud

  • Affiliations:

  • CREST, ENSAI and UMR CNRS 6086, campus de Ker Lann, 35170 Bruz, France;Université de Poitiers and UMR 6086, UFR de Mathématiques Bd. Marie et Pierre Curie, 86962 Futuroscope, France

  • Venue:
  • Computational Statistics & Data Analysis
  • Year:
  • 2007

Quantified Score

Hi-index 0.03

Visualization

Abstract

Algorithms for computing exact distribution values of statistics-linear combination of some multinomial variables called 3-nomial variables are derived. In the general case, there are no explicit formulae of the cumulative distribution function of these statistics that allow the direct computation of p-values. The exact distribution in the independent case and in some dependent ones is studied. For each of the two cases of dependence, an algorithm based on recursive relationships is proposed, giving the number of favorable cases and combinatorics to minimize computation time. For each of these NFC algorithms, concurrent PGF algorithms and some realized comparisons are presented.