On Use of Bayesian Statistics to Make Guide to the Expression of Uncertainty in Measurement Consistent

Published

August 1, 2003

Author(s)

Raghu N. Kacker, Albert T. Jones

Abstract

The Guide to the Expression of Uncertainty in Measurement recommends a standardized way of expressing uncertainty in measurement and provides a comprehensive approach for combining information to evaluate uncertainty in all kinds of measurements in science, engineering, commerce, industry, and regulation. The Guide is being increasingly recognized from national metrology institutes to industrial laboratories. However the Guide is not fully consistent, which may impede its influence. The Guide supports uncertainties evaluated from statistical methods, referred to as Type A, and those determined by other means, referred to as Type B. The Guide recommends classical (frequentist) statistics for evaluating the Type A components of uncertainty but it interprets the combined uncertainty from Bayesian viewpoint. We suggest that the Guide can be made consistent by requiring that all Type A uncertainties be evaluated from Bayesian statistics or interpreted as their approximations. In applications of interest to metrologists, the Type A uncertainties evaluated from classical statistics may be used as approximations provided they are interpreted from Bayesian viewpoint.

Citation

NIST Interagency/Internal Report (NISTIR) - 6995

Report Number

6995

NIST Pub Series

NIST Interagency/Internal Report (NISTIR)

Pub Type

NIST Pubs

Keywords

Bayesian statistics, bias, error, frequentist statistics, uncertainty, uncertainty in measurement

Citation

Kacker, R. and Jones, A. (2003), On Use of Bayesian Statistics to Make Guide to the Expression of Uncertainty in Measurement Consistent, NIST Interagency/Internal Report (NISTIR), National Institute of Standards and Technology, Gaithersburg, MD (Accessed January 15, 2025)