Abstract
The aim of this article is to emphasize the fact, not observed previously in the literature, that many discrepancy measures used in tests related to different stochastic orders can be expressed as expectations of order statistics. In this way, we provide a new meaning to the corresponding test statistics which allows us to understand better, and potentially improve, the testing procedures. As illustration, we consider tests to detect overdispersion with respect to a specific probability model. In this setting, a test for the Weibull distribution is discussed in detail.
Mathematics Subject Classification:
Acknowledgments
The authors would like to thank the Editor and the reviewers for their comments which help to improve the first version of the article.
This research was supported by the Spanish MCyT, grants MTM2007-66632, MTM2010-17366, and MTM2011-27248, and Comunidad de Madrid-UAM grant CCG10-UAM/ESP-5494.