ROBUST TESTS BASED ON MINIMUM DENSITY POWER DIVERGENCE ESTIMATORS AND SADDLEPOINT APPROXIMATIONS

The nonrobustness of classical tests for parametric models is a well known problem and various robust alternatives have been proposed in literature. Usually, the robust tests are based on first order asymptotic theory and their accuracy in small samples is often an open question. In this paper we propose tests which have both robustness properties, as well as good accuracy in small samples. These tests are based on robust minimum density power divergence estimators and saddlepoint approximations.