Estimation in semiparametric models. Some recent developments (Q1188637)

In practical statistical problems one is often confronted with the fact that something is known about the unknown underlying probability measure, but less than parametric theory takes for granted. Estimators are constructed also for such intermediate models, but often the question of their optimality remains open. In the present book, the author gives a survey of a general (i.e. not necessarily parametric) asymptotic theory for such problems. The central point of this approach is the derivation of asymptotic bounds for the concentration of estimator-sequences in general nonparametric models. The development of the basic ideas of the general theory in the first sections is followed by the investigation of more special semiparametric models where a finite-dimensional parameter is to be estimated in the presence of a general nuisance parameter. Since even in this situation a general procedure for the construction of asymptotically optimal estimator-sequences is not in sight, more special cases are considered and the general concept is applied to families of distributions admitting a sufficient statistic for the nuisance parameter. The text is clearly written. The book should be recommended to statisticians in research which are dealing with asymptotic methods.