What quality engineers should know about control charts based on weighted averages

Abstract

The exponentially weighted moving average (EWMA) and moving average (MA) charts use a weighted average of current and past data values. These have the advantage of detecting process shifts more quickly than the Shewhart chart which uses only the current datum to make an inference about whether the process is in control. There has been a proliferation of control chart statistics, many obtained by taking the statistic from one charting procedure, like the EWMA, and using it as input to another procedure, like the cumulative sum chart. These charts, called compound charts, are sometimes more efficient at detecting process shifts, but the improvement is almost always slight. We review many of the compound charts and conclude that these compound charts offer at best only slight improvement and are more complicated to apply than the simpler EWMA.