ABSTRACT
Experience demonstrates that he was correct to place an emphasis on this subject early in his book. Some things need no calculations because they are obvious when pattern and shape are examined.
Figure 11.2 shows a plot of points for the weight of pharmaceutical product extracted by a centrifuge, batch by batch. If examined closely, it reveals an almost perfect saw-tooth pattern. But what is driving this pattern? The regular up and down adjustments to a process found during over-control can cause such a pattern. If each point represented one shift, it could be shift-to-shift variation, where day and night shifts managed the process differently. Equally, it could be two machines operating at different levels. In this case, it was the latter. Where two or more machines or instruments exist, it is safe to start with the assumption that they are different. It is important to know how different, and in what characteristics. Six Sigma levels of performance will be difficult to achieve if these differences are significant. When the data were stratified and plotted as separate charts for each centrifuge, the pattern disappeared, as noted in Figure 11.2. The saw-tooth pattern
was caused by the difference in performance of the two machines. This pattern would have been invisible in a frequency distribution, and in many other statistical techniques.
