An Application of Combining Results from Multiple Methods|Statistical Evaluation of Uncertainty for NIST SRM 1508a

Published

October 1, 2005

Author(s)

Adriana Hornikova, Nien F. Zhang, Michael J. Welch, Susan S. Tai

Abstract

NIST Standard Reference Materials (SRM) are certified reference materials that are developed at NIST and provided to laboratories (industry, government and academia) for assessment and improvement of measurement quality. SRM 1508a, is re-certified to update values was evaluated at the Statistical Engineering Division at NIST using several different statistical models. The recertification is based on combining of the measurement results in subsets at each level varying due to use of different measurement methods and/or years. For statistical analysis, we decided to treat levels separately and to consider each of five method/year combinations as individual subset. In this study, we considered several different statistical models and corresponding estimators for the certified value, its uncertainty and the corresponding 2 sigma coverage interval.

Proceedings Title

NCSL International Conference 2005

Conference Dates

August 7-11, 2005

Conference Title

Advances in Science and Technology -Their Impact on Metrology Proceedings

Pub Type

Conferences

Keywords

Bayesian estimators, factor, mixed linear model with one random facto, MLE

Citation

Hornikova, A. , Zhang, N. , Welch, M. and Tai, S. (2005), An Application of Combining Results from Multiple Methods|Statistical Evaluation of Uncertainty for NIST SRM 1508a, NCSL International Conference 2005 (Accessed April 23, 2024)