Bless you and thank you, John, for mentioning enumerative and analytic studies. These two terms have been all but lost since Deming's demise. Unfortunately, many statisticians have ended up taking most of their classes from mathematicians who have not been exposed to analytic studies themselves, using textbooks that deal only with probability theory, correlation, regression and tests of hypotheses. They often treat capability studies as though they are enumerative studies, with no understanding of what "within" means, because they don't understand rational subgrouping, they don't understand the difference between a sample and a subgroup, and they don't understand that a capability study is actually an analytic study, a function that is only valid within the purview of SPC.

For all readers: if you don't understand rational subgrouping (or have any questions), a great start to understanding may be found in Wheeler and Chambers' "Understanding Statistical Process Control," the Ball Socket Data example, starting on p 99 in the second edtion, and on p 100 in the third edition. On his deathbed, David Chambers told Don Wheeler that if he had it to do over again, he would teach the Ball Socket Data on every day in every SPC seminar they did (Don told me that they actually did include it in the required reading for each night).  If you don't understand it, read it again...and if you still don't understand it, call or write me, or Don, or John, and ask questions until you do understand it. Then you will be qualified to start using SPC and running capability studies.

Rip - I am one those people who lives and breathes this.  I teach it and use it extensively.  It has transformed my organization.  I too am frustrated by the use of precise enumerative statistics fro analytic studies.  (not trying to promote anything but I am speaking on this topic at the ASQ Lean Six Sigma conference in Phoenix this year so if you are attending I'd love to chat)

I'll add my support and thanks to John, and to the previous commentators, about the need to discuss analytic versus enumerative studies. It's a critical distinction that has gone missing in discussions about SPC, performance measurement & reporting, as well as management thinking and behavior. I'm adding my two-bits worth in an upcoming Significance magazine in an article entitled 'Significance, statistical and otherwise.'

Thanks for your clarification. This should help many people devling into capability analysis. SJM

"The proper understanding of rational subgrouping is fundamental to statistical process control, and if you think I am exaggerating, then I suggest you read ... Donald Wheeler on this subject."

I suggest that the author, John Flaig should read Wheeler. He obviously doesn't have a clue what he's talking about.  Ch6 "Advanced Topics in SPC".

I thought enumerative and analytic studies were different since they have different purposes. But for calculating process capability indices, the same data (k subgroups of size n) are used. The way it is described here, it seems the only difference between an enumerative and analytic study is whether one calculates Pp or Cp. That's not much to fuss about. So what am I not understanding?