Trendlines

[Helen Ganley <HGanley@doh.health.nsw.gov.au>]

In a message dated 08/04/2000 5:10:15 AM Pacific Daylight Time, 
ferrell1@home.com writes:

<< Is there anytime that a 'trendline' provides value to a time series
 chart?
 Or more specifically a control chart? >>


Davis Balestracci's book " Practical Applications for Medical Group Practice"
has some good examples of inappropriate use of trendlines.

I also quote from his 8 common statistical traps: Fitting inappropriate "trend" lines to a time series data sequence;

-another form of "tampering" attributing a specific type of special cause (linear trend) to a set of data which contains only common cause.
Attributing an inappropriate  specific special cause (linear trend)  to a data time series that contains a different kind of special cause.

Comment:  Typically occurs when people always use the "trend line" option in spread sheet software to fit a line to data with no statistical trends.

Improvement often takes place in "steps' where a stable process moves to a new level and remains stable there.  However, a regression line will show statistical significance, implying that the process swill continually improve over time"

Hope this helps

Helen Ganley
Quality Manager
Royal North Shore Hospital & Community Health Services
Sunny Sydney
Australia
======================================================================