RE: Question regarding stable systems and correlations

[Steven_S_Prevette@rl.gov]

Here is the question: When the process is stable, the data from the response
variable will not correlate with any of
the data representing a particular common causes?

Here is a potential answer:

The theory, as put forth by Dr. Shewhart, is that if the response variable
looks like it's variation is random, then we assume it is random and there
will not be worthwhile correlations with other variables to explain the
variation.

Note that I have to have a method to determine if it appears to be behaving
randomly, that method being the control chart.

If the data are declared to be random (subject to common cause variation),
the assumption is that it is unlikely that we will find a correlation with
any other variable.  I should also point out that Shewhart's theory is
economic based - if you were able to find a correlation, it is likely that
the cost (dollars, time, effort) of finding the correlation will be greater
than the benefit we would gain from knowing the correlation exists.  Rather
than searching for a specific correlation within the data, we need to focus
on the entirety of the data and find ways to improve the process by looking
at the entire system.

As an example, if injury rates are too high, but are exhibiting common cause
variation, we don't ask why in a certain month we got X number of injuries
and then try to figure out why we got Y injuries the next month by
correlating to some other variable.  We instead collapse all the injury data
together in one pot, and look for common reasons for all injuries.  Pareto
charts and histograms will help here.

On the other hand, if we have special cause variation (a trend, a
statistical signal on a control chart) it is likely that we will find a
correlation that will prove economically viable.  We should ask why we used
to get X, but now get Y.  What other variable has changed?

Steve Prevette
Site Technical Authority for Statistical Trending
Environment, Safety and Health
Fluor Hanford, A Fluor Global Services Company
ASQ Certified Quality Engineer
steven_s_prevette@rl.gov
509-373-9371