Some personal views with regard to the functions of statistics

Abstract

Descriptive, inductive and decision theory approaches to statistics are discussed in this paper. According to the author, the contrast between descriptive and inductive statistics is artificial since any description of the reality constitutes in fact an induction. In the words of Boldrini (1942) "every statistic, even one that seems documentary, has an investigative preoccupation and therefore, above all, an hypothesis". However, the author is doubtful also about inductive statistics conceived as a methodology of knowledge. Concerning the philosophy or pragmatism, it is worth while noticing that the interpretation due to Smith (1984) does not fit for any pragmatist in general and even less for its founders. In the last part of the paper, the functions of statistics are once more discussed from a neo-rationalistic viewpoint and a distinction between the process of analysis bringing forth to discovery and the operative process of decision is posed. The same application of probability in statistics is contested. Indeed, probability is an analytic tool we may use to explore hypotheses. Therefore, it cannot be reserved only for inference or decision, as we may see at present time in the text-books.