RE: Hillside Yellers?

["Ralph Stauffer" <rstauffer@mn.rr.com>]

Petter said,

"...most introductory books on statistical methods would probably have to be
rewritten and start with SPC diagrams before they introduce concepts like
population and sample.  Is this the way most people on the DEN think about
it?"

This is an interesting observation, and kind of goes with a theme we have
touched on several times over the past several years; namely, that the
theory of variation is different from statistical theory, as usually taught.

Don Wheeler makes an interesting case in his latest book, "Guide to Data
Analysis." It's not a new case. Tukey, Shewhart, Deming, Box and others have
made this case (or parts of it) over the years. Wheeler himself has made the
case in every book he's written. It was a principal discussion at the
ASA/ASQ Statistics Division Fall Technical conference a couple of years ago.
The case is that, lacking evidence of homogeneity (for process data,
evidence of a reasonable degree of statistical control), most assumptions we
make for most data sets are untested assumptions. 

This has, I think, profound implications. I have many statistics textbooks;
very few discuss this idea at all. Most discuss distributions, and many have
chi-square and other tests for testing a data set for congruence with
particular distributions; almost none suggest that the test may be
meaningless if the assumption of homogeneity doesn't hold. Almost none
discuss which tests might or might not be robust to that assumption, and
under what conditions. Almost none suggest a test for homogeneity. Most Six
Sigma DMAIC models don't mention SPC or control charts until you get into
the Control Phase. 

This is not universally taught, and there seems to be very little evidence
that there is much awareness of the concept. Maybe there is awareness, and
it's just not universally accepted, although I haven't yet seen a convincing
counter-argument. 

Best regards to all,

Rip Stauffer
Woodside Quality Solutions
623 Carver Bluffs Parkway
Carver, MN 55315          
612-916-0197
rstauffer@mn.rr.com