Some new estimators for Cox regression (Q1317247)

The author presents asymptotic results associated with the behavior of proportional hazard estimators in the Cox regression model. A class of estimators which generalizes maximum weighted partial likelihood estimators is considered, and sufficient conditions for the functionals associated with these estimators to be well defined are given. The asymptotic properties of these estimators are considered, and also the problem of constructing an estimator with minimum variance subject to a bound on the bias on local neighborhoods of the Cox model is discussed. Further, illustration by medical data is given.