Смещение МП-оценки параметров бета-логистической модели c аддитивными стохастическими искажениями

Abstract

The paper is devoted to the robustness of parameter estimation for the beta-logistic model of clustered binary data in case of additive stochastic distortions of binary observations. Stochastic expansions for the bias of the maximum likelihood estimator (ML-estimator) under distortions at obtained. It is proved that the ML-estimator is inconsistent and biased under distortions. The bias corrected estimator in the case of known distortion levels is proposed. Computer simulation result are also given.