Knocks the final few nails into the coffin which contains the remains of the notion that the theory and practice of control charting depend on assumptions of normality. The subject’s creator, Dr Walter Shewhart, denied this as long ago as 1939! His most famous student, Dr W. Edwards Deming, denied it repeatedly thereafter. There appear to be two most crucial arguments as to why the “orthodox” statistician claims that normality is necessary. One is to enable probability interpretations of control limits. The other is to justify the conversion factors which are in common use in control‐chart calculations. The truth is that, even under normality, the usual probability interpretations are meaningless in practice and that, in the latter case, the behaviour of the conventional conversion factors is not at all dependent on normality but is in fact very similar over a wide range of differently‐shaped probability distributions.
MCB UP Ltd
Copyright © 1997, MCB UP Limited