A point estimator is a tool which assigns to each sample (data set) an estimated value of the aspect of interest of the unknown distribution of the data. An estimator is considered qualitatively robust when small changes in the distribution of the data result only in small changes of the distribution of the estimator.

After an introduction into statistical models and point estimators I will recall the classical notion of qualitative robustness originated by F. Hampel in the 1970th. Hampel considers qualitative robustness as a
global property. This generates a sharp division of estimators into those that are called „robust“‚ and others that are called „not robust“‚. In the last part of the talk I will discuss a recent refinement of

Hampel’s terminology where qualitative robustness is considered as a local property. I will explain why it can be beneficial to know the sets of relevant distributions on which qualitative robustness holds, and I will present some criteria and examples. This final part is based on joint work with V. Krätschmer (U Duisburg-Essen) and A. Schied (U Waterloo).

Referent: Prof. Dr. Henryk Zähle, Universität des Saarlandes

Die Vorträge finden jeweils um 17.15 Uhr im Raum 210 des Mathematik-Gebäudes 48 statt. Zuvor gibt es ab 16.45 Uhr die Gelegenheit, den Sprecher beim Kolloquiumstee in Raum 580 zu treffen.