Re: Noise ...

[GrantBlair@aol.com]

In a message dated 4/8/2001 1:20:03 AM Eastern Daylight Time, 
cascadia.systems@home.com writes:

<< First, would it be correct to say that the noise is the datum points within
 the UCL and LCL ?>>
This is correct- you are defining "noise" as Chance cause or Common cause 
variation which is within the UCL and LCL and system. The opposing term used 
is "signal", which is also know as Assignable cause or Special cause.  Points 
outside UCL and LCL indicate its presence. There are two types of Signals,
a shift in level and an increase in variation. In both cases, it is the 
responsibility of the local workforce to find and correct both sources of 
variation. ( It may also be appropriate for the workforce to adjust the 
process for changes in Level.)
It is important that Signals/Special causes are eliminated before attempting 
a further reduction in variation

 
<< Second, if required, and understanding the systemic causes of variability, 
I
 take it that it would be appropriate to work on reducing the variability of
 the process, and work if necessary to shift the process average ? >>
This is correct. I have used this illustration before, but you may not
have seen it.
Let' s suppose you are competing with an extablished product and are producing
the same product with the same process average and Common Cause variation 
(noise). It will be possible for you to take 25% of your competitor's 
business using a simple property of the normal distribution--- 1/2 of the 
variation will be above the process average, 1/2 below. When your customer 
compares your product to
the competitor, he will get:
1. Your product and competitors both below average- 1/4 of the time.. No sale
2. Your product and competitors both above average- 1/4 of the time..  No sale
3. Your product below average, competitors above- 1/4 of the time...     No 
sale
4. Your product above average, competitors below- 1/4 of the time.... BINGO! 
You
     have the business.
Now, suppose you can reduce the variation... ideally, to the point where 
nearly all of your product is exactly on target. (This is another important 
property of the normal
distribution). In this ideal case, you will now find:
1. Your product below aberage, competitors above- 1/2 of the time.. no sale
2. Your product above average, competitors below- 1/2 of the time..  BINGO 
AGAIN!
Our conclusion is that we can increase market share from 25% toward 50% as we 
reduce variation. Further increases are possible if we have learned enough 
about
our process along the way to adjust the target in an appropriate manner. 
Note also that this second case is MANAGEMENT's responsibility, not the 
workforce.


GrantBlair@aol.com
Ninety Six SC