After a brief historical overview over various approaches to the foundations of statistics, the very general, simple and basic concept of (potential) surprises is introduced, which may be subjective or objective and goes beyond previous approaches by I.J. Good and by the author. The surprises are conditional on the background knowledge or belief of the person experiencing it; the updating of the so-called background, and the merging or, if not possible, the contrasting of different backgrounds by two or more persons (otherwise they talk past each other) are very important operations in practice. A number of examples from real life, in complement to two previous, more qualitative papers, are given.
The paper is available in the following formats:
Seminar fuer Statistik
Leonhardstrasse 27
ETH-Zentrum, LEO D2
CH-8092 Zurich
Switzerland
| Frank Hampel | hampel@stat.math.ethz.ch |
Send any remarks to isipta11@uibk.ac.at.