|Plan to examine process stability as part of your experiment||
Experimental runs should have control runs that are made at the
`standard' process setpoints, or at least at some standard operating
recipe. The experiment should start and end with such runs. A plot of
the outcomes of these control runs will indicate if the underlying
process itself has drifted or shifted during the experiment.
It is desirable to experiment on a stable process. However, if this cannot be achieved, then the process instability must be accounted for in the analysis of the experiment. For example, if the mean is shifting with time (or experimental trial run), then it will be necessary to include a trend term in the experimental model (i.e., include a time variable or a run number variable).