Re: control charts and non-normality

[DANSWART@aol.com]

In a message dated 05/11/2000 7:10:02 AM Pacific Daylight Time, 
GrantBlair@aol.com writes:

<< Are you considering control charts as enumerative statistics?  >>

No.  My understanding was that probabilities should not be assigned to future 
events based on control chart limits.  The previous posters indicated that 
you should expect to see an average number of signals per X many 
observations, and if you don't you should investigate your methodology.  
Since (to me) such a ratio is a form of probability statement, I was guessing 
that the assigning of probabilities to the number of expected out-of-control 
signals per number of observations was an enumerative technique.

I also am wary of using an average when looking for actual events.  I expect 
one could go a very long time before observing a sequence of coin tosses with 
exactly one-half heads, though the theoretical average would call for that.  
Or I might see it tomorrow.  I don't think I should I consider the coin 
suspect if either happens.

Have I gone off to the Milky Way?

:-)

Thanks for considering my questions.

Dan
danswart@aol.com
========================================================================