What You Always Wanted to Know About Censoring But Never Dared to Ask Parameter Estimation for Censored Random Vectors

This article considers a wide class of censoring problems and presents a construction rule for an objective function. This objective function generalises the ordinary likelihood as well as particular "likelihoods" used for estimation in several censoring models. Under regularity conditions the maximiser of this generalised likelihood has all the properties of a maximum likelihood estimator: it is consistent and the respective root-n estimator is asymptotically efficient and normally distributed.