RE: Short Run SPC

[John Dowd <jsdwd@netcom.com>]

There is a question of whether 'short runs' can be analyzed with only a 
few data points.  Of course, they can.  They can be analyzed (and 
frequently are) with NO data points.

The question of the adequacy of the analysis is a complex one and depends 
as much on how the data were gathered as on how much data there is.

One question is that of robustness.  The robustness of a procedure is the 
extent to which it will yield similar results when (among other 
considerations) there is 'inadequate' data.  The 'correct' amount of data 
needed is clearly a risk/benefit decision that must be made.  How much 
benefit can we get with out appreciably increasing the risk.

Statistical Control Charts have been shown to be robust.  See Wheeler for 
a discussion of this.

John Dowd
jsdwd@netcom.com
==========================================================================