Re: Noise .... slightly off topic

[KINLEY_ROBERT@Lilly.com]

This is slightly to one side of this topic

Steve Prevette writes :-

>
>- I would add the criteria that no other "rule" that is in use for 
control
>charting (such as 7 points in a row above average) is violated.  There 
are
>many variations on these "rules", so one does need to establish up front 
an
>operational definition for your own control chart rules. 
> 

The 'rule' of 7 points above average is in quite comon use but 
can be a bit 'trigger happy',  as you will discover if you do a 
simulation where you generate, say, 20 points from a Normal 
distribution, calculate upper & lower limits, and then plot,say, 
a further 20 points on a control chart using these limits.

Many textbooks recommend runs of 8, and the JMP(SAS) stats 
package uses 9 . 

It all depends how many 'special causes' you have time to deal with  :-)

                regards Bob
 

Bob Kinley
Site Statistician
Eli Lilly & co
Speke, UK