Contributed Poster Presentations: Quality and Productivity Section

Online statistical monitoring of a process to detect a shift in the conforming percentage involves periodically selecting a unit or units to examine. Based on some protocol or decision rule a judgment is made about whether the conforming percentage for the process has shifted from an acceptable percentage to one that is now unacceptable, i.e. the process is now out of control. It is possible that an item is misclassified and so making a final judgment would be improved by doing more than making one classification on a single item. A variety of protocols have been studied using different metrics involving among other things the probability of shift detection if it has occurred, the probability of declaring that a shift has occurred when no shift has occurred, the distribution and mean of the number of classifications needed to reach a judgment at any inspection period, the run length both when the process is in control and when it is out of control, and the distribution of the number of inspection periods until the process is declared out of control. In this paper, we explore some modifications of previously used protocols.