Traditional statistical process control (SPC) and calculation of process performance indices rely on the textbook assumption that the critical to quality (CTQ) characteristic follows a normal distribution, or bell curve. If the processes on the factory floor do not cooperate with this assumption, control charts will not work properly while estimates of defects per million opportunities (DPMO) can be off by five or more orders of magnitude. That's when our customer might tell us, "Your 'Six Sigma' process is not even capable."
Textbook statistical process control (SPC) relies on the assumption that the process follows a bell curve or normal distribution. When the process does not cooperate with this assumption, the risk of false alarms (and consequent overadjustment) can be several times more than expected, and process performance indices can be grossly misleading.
This underscores the need to (1) test the assumption of a normal distribution and, if the distribution is not a bell curve, (2) identify the correct distribution and then calculate appropriate control limits and process performance indices. While we can never prove that a process follows any particular distribution, we can prove beyond a reasonable doubt (the Type I or alpha risk) that it does not follow a selected distribution.
Join this session by our expert speaker William A. Levinson to get a better understanding of statistical distribution and what to do when it’s not a bell curve. Learn about some qualitative and quantitative tests for normality (and other distributional assumptions).
Who should attend?
Quality engineers and technicians
William A. Levinson P.E.
William A. Levinson, P.E., is the principal of Levinson Productivity Systems, P.C. He is an ASQ Fellow, Certified Quality Engineer, Quality Auditor, Quality Manager, Reliability Engineer, and Six Sigma Black Belt. He is also the author of several books on quality, productivity, and management, of which the most recent ... More info