期刊名称：Sankhya. Series A, mathematical statistics and probability
印刷版ISSN：0976-836X
电子版ISSN：0976-8378
出版年度：2010
卷号：72
期号：02
页码：376--406
出版社：Indian Statistical Institute
摘要：Inference procedures based on the Hellinger distance and other disparities
provide attractive alternatives to likelihood based methods for the statisti-
cian. The minimum disparity estimators are asymptotically ecient under
the model. Several members of this family also have strong robustness prop-
erties under model misspeci cation. Similarly, the disparity di
erence tests
have the same null distribution as the likelihood ratio test but are often
superior than the latter in terms of robustness properties. However, many
disparities including the Hellinger distance put large weights on the empty
cells which appears to be responsible for a somewhat poor eciency of the
corresponding methods in small samples. An arti cial empty cell penalty has
been shown to greatly improve the small sample properties of these proce-
dures. However all studies involving the empty cell penalty have so far been
empirical, and there are no results on the asymptotic properties of the mini-
mum penalized disparity estimators and the corresponding tests. In view of
the usefulness of these procedures this is a major gap in theory, which we
try to ll through the present work.
