Optimal rank-based tests for homogeneity of scatter
Hallin, Marc ; Paindaveine, Davy
Ann. Statist., Tome 36 (2008) no. 1, p. 1261-1298 / Harvested from Project Euclid
We propose a class of locally and asymptotically optimal tests, based on multivariate ranks and signs for the homogeneity of scatter matrices in m elliptical populations. Contrary to the existing parametric procedures, these tests remain valid without any moment assumptions, and thus are perfectly robust against heavy-tailed distributions (validity robustness). Nevertheless, they reach semiparametric efficiency bounds at correctly specified elliptical densities and maintain high powers under all (efficiency robustness). In particular, their normal-score version outperforms traditional Gaussian likelihood ratio tests and their pseudo-Gaussian robustifications under a very broad range of non-Gaussian densities including, for instance, all multivariate Student and power-exponential distributions.
Publié le : 2008-06-15
Classification:  Elliptic densities,  scatter matrix,  shape matrix,  local asymptotic normality,  semiparametric efficiency,  adaptivity,  62M15,  62G35
@article{1211819564,
     author = {Hallin, Marc and Paindaveine, Davy},
     title = {Optimal rank-based tests for homogeneity of scatter},
     journal = {Ann. Statist.},
     volume = {36},
     number = {1},
     year = {2008},
     pages = { 1261-1298},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1211819564}
}
Hallin, Marc; Paindaveine, Davy. Optimal rank-based tests for homogeneity of scatter. Ann. Statist., Tome 36 (2008) no. 1, pp.  1261-1298. http://gdmltest.u-ga.fr/item/1211819564/