Statistical Engineering Division
Seminar Series
Using Bayesian Methods to Account for Measurement Error
Michael Hamada
Los Alamos National Laboratory
Room 145, NIST North
July 29, 2003, 2:00pm
When information about a measurement system is available (e.g., from
a designed experiment), a Bayesian approach provides a
straightforward way to account for measurement error in making
statistical inferences about an individual or population. Examples
of such experiments include calibration and gauge R & R experiments.
Statistical inferences may take the form of calculating calibration,
tolerance and prediction intervals or setting specification limits.
Both the measurement system data model and the individual/population
model may be nonstandard. The desired inferences involve functions of
these model distributions and parameters and require that uncertainty
in the model parameters be characterized. In this talk, I will
illustrate the Bayesian approach with a number of examples.
Mike Hamada, Ph.D., is a technical staff member in the Statistical
Sciences Group at Los Alamos National Laboratory in New Mexico. He
has published more than 30 papers and has won the Technometrics'
Wilcoxon Prize and the ASQC Brunbaugh Award.
NIST Contact:
Ivelisee Aviles, x-2849.
