Об асимптотике распределения двухшаговых статистических оценок одномерного параметра

The authors' approach to study two-step estimators of one-dimensional unknown parameters is extended to a wider classes of the first- and second-step estimators which include well known M-estimators. Under general restrictions necessary and sufficient conditions are found for the normalized difference between the second-step estimator and the unknown parameter to converge weakly to an arbitrary distribution.