Resumen de Loglinear models: an approach based on divergences

L. Pardo

• In this paper we present a review of some results about inference based on o-divergence measures, under assumptions of multinomial sampling and loglinear models. The minimum o-divergence estimator, which is seen to be a generalization of the maximum likelihood estimator is considered. This estimator is used in a o-divergence measure which is the basis of new statistics for solving three important problems of testing regarding loglinear models: Goodness-of-fit, nested sequence of loglinear models and nonadditivity in loglinear models.