RE: stable system?

["DeRoeck, Rich" <RDeRoeck@alphaind.com>]

>>Control charting is not rocket science, at its best, you could have a 6th,

7th or 8th grade student preparing the charts and drawing conclusions (about

stability, common cause variation, special cause variation, and
prediction).<<

Most control charts that I have seen in a manufacturing environment are
poorly designed. Why, because they were not set up to ask the right
questions (just thrown out there). Little attention was paid to the
following:

1. Is this a critical process characteristic?
2  Does controlling this parameter affect product quality?
3. Will it make a difference to the customer?
4. How shall we subgroup the data? What makes sense for this process?
5. Who shall take the measurements? Have they been properly trained?
6. What is a sampling plan (frequency) that makes sense?
7. Is the measurement process in control? Is it accurate?
8. Are there documented procedures for out-of-control conditions?
9. What detection rules shall we use? What makes sense for this process?

As someone once said: Think first, then think statistically.

Rich DeRoeck