Abstract
Two types of estimates of process level, namely repeated median estimates (Siegel, Citation1982) and full online estimates (Gather et al., Citation2006) based on repeated median filters, are used to develop control charts. The distributional properties of the estimates are studied using simulation and these are found to closely follow normal distribution. The repeated median being robust against outliers with asymptotically 50% breakdown value and having small standard deviation is found to be useful as a basis for monitoring process averages. The control charts using repeated median estimates have been recommended for general use.
Mathematics Subject Classification:
Acknowledgment
The authors are indebted to the referee for his suggestions in revising and improving the article.