Files
Abstract
Different interpretations of the probability concept are considered. In a situation of repeated trials, indepedent in the frequentistic sense, the frequentistic theory appears to be adequate only if a sufficiently large number of observations is available. In those cases where only a limited number of observations is available, the subjectivistic approach towards probability might be of help. The underlying assumptions of this notion of probability are considered. From the fundamentals of normative probabilities, the ambiguity of a prior distribution for an "unknown" parameter is shown: Sometimes such a prior distribution can be considered as a reflection of (personal) beliefs, sometimes it must be considered as a derivative of the learning process that is chosen. When these two .interpretations overlap, the meaning of a prior distribution becomes rather vague.