Community Research and Development Information Service - CORDIS

Abstract

Parametric and nonparametric methods. It is usual, in dealing with an inference problem, to make a clear distinction between the parametric approach and the nonparametric one. A parametric method "is based on a specific assumption about the population sampled"; a nonparametric method is a method in which "specific distribution assumptions are replaced by very general assumptions". This distinction gives too much importance to a difference which undoubtedly exists. Moreover it becomes completely arbitrary when, as sometimes happens, the term "nonparametric" is taken as being synonymous with "assumptionless". In solving any statistical problem one must make some assumptions. The judgement about the strength of these assumptions is relative to the context in which one is working. Sometimes it can be reduced to a matter of taste. What is really important is to state clearly the assumptions which one must make in order to solve a specific problem. This is the aim of our paper.

Additional information

Authors: COLOMBO A G JRC ISPRA ESTAB. (ITALY) COSTANTINI D UNIVERSIT`A DI BOLOGNA (ITALY) , JRC ISPRA ESTAB. (ITALY);UNIVERSIT`A DI BOLOGNA (ITALY)
Bibliographic Reference: STATISTICA, VOL. XLV (1985), NO. 1, PP. 85-100
Record Number: 1989124007600 / Last updated on: 1987-01-01
Category: PUBLICATION
Available languages: en