Some comments on the decision approach to statistical inference

Abstract

The decision approach to statistical inference is discussed from the point of view of the aptitude of its underlying principles to be useful in the real problem solving area. Conditional and non conditional statistical decision inference are also discussed. Moreover a set of inferential problems for which the decision approach fails are examined. So that if a statistician wants to solve them he needs to have at his disposal a set of different inferential approaches, based on different underlying principles and/or axioms, useful in different situations and for different tasks. Therefore, in a complex problem, he could need many approaches at the same time, especially if this problem can be put into a set of inferential subproblems.