Dear Experts,
When we have to decide to create a new baseline average?
I have a process with 31 process values (31 months, from Jan05 until July07); process has been modified in June 2007, in this case can we create a new baseline average from June 2007 to evaluate the new changes?
Another question please, can we cerate a new baseline for out-off control process?
Thanks.

Dear Experts,
When we have to decide to create a new baseline average?
I have a process with 31 process values (31 months, from Jan05 until July07); process has been modified in June 2007, in this case can we create a new baseline average from June 2007 to evaluate the new changes?
Another question please, can we cerate a new baseline for out-off control process?
Thanks.

As of right now, it would not be possible to establish a new baseline average, since you only have two data points (June and July 07). My practice when modifying a process is to continue to plot data against the old baseline, until the process shows itself to be out of control. This shows that the process modification actually did change the process - hopefully in the good direction! Then calculate the new process baseline once sufficient data has been accumulated. Opinions vary on what is sufficient data. For myself, I prefer at 25 data points, but with only 1 point/month, that may be impractical.

Geoff Withnell

I have a writeup on the process at http://www.hanford.gov/rl/uploadfiles/VPP_20_Life_Cycle.ppt and "The Life Cycle of a Trend" may also be found here at the Cove.

A few important points:

1. Do not change the baseline unless you first have a statistically significant trend signal on the control chart. Davis Ballestracci coined the phrase "a baseline is innocent until proven guilty".

2. I have found with empirical experience that once you have identified a trend, if you wait for three changes of direction in the data (increasing to decreasing to increasing to decreasing or vice versa) then you will have enough data to make a sufficient baseline for future prediction.