Year: 1987
Type: article
Abstract: That portion of applied statistics which is used for the treatment of experimental data is essentially founded on unbiased estimators - notwithstanding the fact that any real measurement process implies so-called unknown systematic errors. Until now, those perturbations giving rise to unknown deviations from the true values have been formally randomized, i.e. they are treated as if they were of random origin. Biased estimators thus again become fictitiously unbiased. However, the associated procedures suffer from two disadvantages: they are too complicated and, in the author's opinion, they generate uncertainties which are not sufficiently reliable. The submitted paper attempts to establish "Metrological Statistics" which, being based on biased estimators, is much simpler to handle and which, in particular, yields safe estimates for measurement uncertainties.
Source: Metrologia
Author: M Grabe
Language: English
Subfield: Statistics, Probability and Uncertainty
Field: Decision Sciences
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W2006431632