Median Control Charts for Skewed Populations

Conventional median control charts with control limits determined by the means of subgroup medians and H-spreads are .based on the assumption that the distribution of the quality characteristic is normal or at least symmetric. In many situations, however, the normality assumption is not valid. This paper investigates the effects of skewness on the median charts and proposes a simple heuristic method of constructing median charts for skewed populations. The proposed method decomposes H-spread into upper and lower spreads, which are then used to set up asymmetric control limits. For a symmetric population, however, the chart reduces to the conventional median chart. The performance of the proposed median chart is investigated using Monte Carlo simulation and the results show that considerable improvement over the conventional median chart can be achieved for skewed populations.