I have a simple and may be stupid question. Is six sigma +/- 3 sigma from the mean or +/- 6 sigma from the mean.

In other words the total area inside the normal curve contains 6 sigmas or 12 sigmas?

I always believed that it is +/- 3 sigma on either side of average.

Please help!

Harsh

The spec limits should be 12 standard deviations apart. Ideally that would be 6 standard deviations on either side of the mean, but standard "Six Sigma" methodologies allow a 1.5 standard deviation shift, so the mean could be just 4.5 sigma from one spec limit, and 7.5 from the other.

Tim F

To expand on Tim's explanation, the confusion stems from the fact that the control limits on Shewhart charts are not related to specification limits. The +/- 3 sigma limits are meant to capture nearly all (99.73%) of a normally distibutedpopulationand thus give evidence that something statistically unlikely has happened. As Tim said, the "Six Sigma" thing refers to the distance of specification limits from the mean.

Here is a graphic that might help you see what Tim is saying.

Dave

Here is a graphic that might help you see what Tim is saying.

Dave

Sorry Dave, but that graphic will only confuse things further, because it shows control limits when they should be identified as spec limits. See my earlier post.

For 6 sigma approach, divide total specification tolerance band into 12 equal intervals which corresponds to +/- 6 sigma and try to improve the process so that it uses only half of that band i.e. +/- 3 sigma.

Even in absence of special cause, a process will shift by 1.5 sigma in either side of the target. In above scenario, even with shift, you are going to have a very small number of defective parts. That is the process robustness which six sigma approach gives you even when mean shifts which is a way of life.

Sorry Dave, but that graphic will only confuse things further, because it shows control limits when they should be identified as spec limits. See my earlier post.
6 sigma relates to capability of the process. Control limits are the natural variation of the process. There is a relationship between the two. Because the sigma from the control chart is the same sigma for the 6-sigma calculation.

6 sigma relates to capability of the process. Control limits are the natural variation of the process. There is a relationship between the two. Because the sigma from the control chart is the same sigma for the 6-sigma calculation.
Sigma is sigma, yes. Tim Folkert's original response in this thread answered the OP's question concisely, I think.