SED navigation bar go to SED home page go to SED publications page go to NIST home page SED Home Page SED Contacts SED Projects SED Products and Publications Search SED Pages

contents     previous     next

4.2 Easy and Not-So-Easy Methods of Uncertainty Analysis

Carroll Croarkin, Mark Levenson
Statistical Engineering Division, ITL

Theodore Doiron
Precision Engineering Division, MEL

Concepts of uncertainty analysis for metrology were outlined in a two-day workshop presented in conjunction with the Measurement Science Conference (MSC) in Anaheim, CA on Jan. 25-26, 1999. As the subject of uncertainty has taken on new importance in testing and metrology laboratories because of requirements in the ISO 9000 series, engineers in those environments find themselves in need of guidance on data collection and analysis methods that are consistent with the procedures outlined in the ISO Guide to the Expression of Uncertainty in Measurement.

This workshop was initially presented in conjunction with the 1998 MSC in Pasadena, CA and was revised considerably this year to emphasize methodology appropriate to the industrial environment. Also, a hands-on experiment was added to the workshop this year with attendees estimating the value of Pi from measurements on a brass ring. Three separate apparatuses, each consisting of a ring, calipers for measuring diameter, and length standard for measuring circumference, afforded attendees and presenters the opportunity to test for components of variance and to debate the merits of various analytical techniques.

The lecture portion prepared the attendees for Type A and B analyses of uncertainty, including type A treatment of bias, and included discussions of: random and systematic components of error; statistical intervals; propagation of error.

Doiron discussed his experience and reliance on check standards for computing uncertainties for dimensional measurements in the NIST dimensional calibration programs.

The attendees worked through the analysis of variance on the data from the Pi experiment where they estimated individual variance components, decided on the appropriate form for the standard deviation of the reported value, estimated effective degrees of freedom and announced an expanded uncertainty for the value of Pi reported from this experiment.

contents     previous     next

Date created: 7/20/2001
Last updated: 7/20/2001
Please email comments on this WWW page to