Special cause is suggested in CC whenever a single point falls outside the CL or where multiple points follow a nonrandom pattern within the CL. The patterns depend on the type of chart (ie data) you are using and can results in a more sensitive interpretation of special cause.
As for why the 3 sigma limits are used, it is my understanding that it goes to the underlying assumption of normality and the corresponding Empirical Rule ( ie probability theory) , where app. 99.73% of your data will fall within plus or minus 3 sigmas of the mean. Thus, you would expect only about 3 points out of 1000 to occur outside the control limits naturally, making the chart a highly reliable instrument (if done right) for determining where to ¨debug¨ a process of special cause.
Make sure you are using the right chart the right way…they can be a little tricky…also look into run charting when working with a small number of individual values.
And double check my explanation. It has been awhile.
Good luck.
.
