Some authors recommend the use of an additional test for detecting increases in the spread, when using a control chart for individual observations. We examine this recommendation both in a practical situation and theoretically. Both studies show that the additional test gives somewhat more power for detecting a 25% increase of the process variation. For nearly all other deviations from the in-control state the test is more likely to cause confusion. From a practical viewpoint we therefore advise against its use.
|Status||Published - 2003|