ENBIS: European Network for Business and Industrial Statistics

Mind the gap

21 September 2009, 15:45 – 16:05

Abstract

Submitted by

Antje Christensen

Authors

Antje Christensen, Finn Junager

Affiliation

Novo Nordisk

Abstract

When validating an analytical method, or transferring it to another laboratory, requirements to the method’s precision and accuracy are often based on what the method can be expected to deliver. That’s reasonable: There is no point in setting so low a limit for the bias that you would need an outrageous sample size to stay below it, given the method’s precision. But what if you actually need that low limit on the bias for your application?
This situation occurred in a Lean Six Sigma project on a tabletting process, which addressed a systematic difference between planned and actual strength. Tracking down the bias involved using a model for water sorption that is based on the physical properties of the ingredients. Thus theory and process measurements were brought together to sharpen our view of the issue.

View paper

Return to programme