Abstract
The median rankit control chart produced by a quantile approach has been presented and discussed in some quality literature to place emphasis on usefulness for a process with non-normal observations. Nevertheless, the ability of median rankit charts to detect the process change has not been examined now. To fit this gap, this paper presents the performance of the median rankit chart in terms of the average run length that is widely used to assess the statistical efficiency of a control chart. Under the assumption of process observations from the normal distributions or from some non-normal distributions (heavytailed symmetric and skewed distributions), above results are also compared to that for the individual Shewhart chart and EWMA chart.
Keywords and phrases: