Premium
Nonparametric weighted symmetry tests
Author(s) -
Abdous Belkacem,
Ghoudi Kilani,
Rémillard Bruno
Publication year - 2003
Publication title -
canadian journal of statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.804
H-Index - 51
eISSN - 1708-945X
pISSN - 0319-5724
DOI - 10.2307/3315851
Subject(s) - nonparametric statistics , mathematics , symmetry (geometry) , asymptotic distribution , estimator , generalization , wilcoxon signed rank test , scaling , statistical hypothesis testing , rank (graph theory) , parametric statistics , statistics , statistical physics , mathematical analysis , combinatorics , physics , mann–whitney u test , geometry
Weighted symmetry is an extension of the classical notion of symmetry in which the tails of a distribution are similar, up to a scaling factor. The authors develop test statistics of weighted symmetry based on empirical processes. The finite‐dimensional distributions of the proposed statistics are either non‐parametric or conditionally nonparametric, according as the parameters of weighted symmetry are known or estimated. Asymptotically, the distributions of the processes behave like Brownian bridges or motions, leading to familiar distributions for the proposed test statistics. The authors also establish the asymptotic normality of Hodges‐Lehmann type estimators based on a generalization of the Wilcoxon signed rank test. Furthermore, they propose density estimators in mat setting.