Statistical process control for nonnormally distributed variables through the use of transformations

The use of normalizing transformations that allow the use of standard SPC (statistical process control) procedures is discussed. A case study is presented to illustrate the use of tests for normality, of normalizing transformations, and the subsequent calculation of the control limits and construction of the SPC graph. It is shown that using normalizing transformations provides a viable method to monitor and control nonnormally distributed variables using standard SPC methods. This approach reduces the rate of false alarms, allows the use of the AT&T rules to detect trends in the data, increases confidence in the ability of SPC to detect points out of control, and uses standard SPC methods, so that learning new procedures is not necessary. One drawback of this method may be that the ordinate of the SPC chart is in the units of the transformed variable