Here's the deal: for attributes MSA, you need to demonstrate that the measurement system as it's used in production will always discriminate between conforming and nonconforming parts. The only statistically reliable way to do this is to have operators check parts that are known to be "good" and "bad" in sufficient numbers. Your problem, it seems to me, is not having enough parts to be able to arrive at a statistically sound conclusion. You need to approach this parsimoniously--don't make it more complicated than it needs to be. You need to do it with as many parts as you can, and you need to make sure that your customer understands the situation and approves.
I am not really looking to satisfy a specific customer with this really just trying to establish good practice within the Lab.
Back to my original question - Is it OK to use variable data to define the repeatability and reproducability of an attribut gage - AND Why or Why Not?
Last edited by a moderator: