I'm looking for a mathematical model which represents the R&R of an attribute defined product. Other than the representation offered in the AIAG manual, do you know of any statistical convention for analyzing the R&R of a measurement system which qualifies a product by attribute?

I don't off hand, but Howard Atkins (one of the forum moderators) is pretty statistical. You might try shooting him an e-mail. I don't know that he stops by this forum much if at all. I might be wrong, though.

I do not know of an R&R study for attribute data. What I have used in the past, rather, was the model for process capability for attributes. You should be able to define the "process capability" of your gage using this methodology.

Process capability for attributes is defined as:

Capability % = 100*(1-((F+0.7)/N)

Where N is the number of tests and F is the number of failures (gage fails to detect nonconforming product or incorrectly identifies conforming as nonconforming).

Under the assumption the gage always functions correctly, then process capability can be determined by:

Capability % = 100*(0.5)^((1/(N+1))

Hope this helps, or was I completely off base?

Best Regards,

Don

Thanx for the detailed response, Don. Not my specialty... http://www.qs9000.com/ubb/wink.gif

Not a problem, Marc. Anything I can do to help. Before I started ISO stuff, I was a quality engineer in Florida for 10 years. If I do not know it, I may be (may not be) able to find it. I still have a lot of that stuff laying around and will keep an eye on this forum for ya. All for now. Gotta go.