Re: Trendlines

[Paul Hollingworth <PH@paulholl.demon.co.uk>]

In message <74.1d02967.26bca702@aol.com>, DANSWART@aol.com writes
>I remember seeing (somewhere) a control chart whose control limits were 
>created using the trend information contained in the data.  It looked just 
>like any other control chart, the CL, UCL and LCL were all parallel, but 
>tilted.  

The basis for these 'tilted' limits (on individuals e.g. XmR charts) is
that the average moving range is still a valid measure of dispersion for
a trend. It may therefore be used to calculate the distance for the
control limits either side of the centre (trend) line, as for a 'level'
chart. Where you get your centre line from is matter for discussion. . .

>  It may have been a feature of an SPC package, 
>but not necessarily recommended as useful by Dr. Wheeler, et al.

On the contrary, Don explores the use of trended limits in his book
Building Continual Improvement, which also includes some excellent work
on dealing with seasonality.

Paul H

----------------------------------------------------------------
Paul Hollingworth                       4GM Consulting  
email: PH@4GM.com                       http://www.4GM.com
phone: +44(0) 1423 322225               fax: +44(0) 1423 322205 
----------------------------------------------------------------
================================================================