|
Abstract:
|
Shewhart control charts are sensitive to estimation effects and deviations from normality. Improvements have been derived, which are often adequate, but not always. In this talk, the remaining complication is attacked: What to do if a nonparametric approach is indicated, but too few Phase I observations are available? It is shown that grouping the observations during Phase II works well. Surprisingly, instead of using the group averages, it is preferable to compare the minimum for each group to a suitably chosen upper limit (in the two-sided case, also the maximum to an analogous lower limit). This 'minimum control chart' is demonstrated to be attractive; it is easy to explain and implement. Moreover, while it is truly nonparametric, its power of detection is comparable to that of the customary, normality assuming charts based on averages.
|
