Variation is the enemy! Variation can introduce waste and errors into a process. The more variation, the more errors and the more waste!
The more you adjust a process to nominal (the center), the more variation you will create. (Also see Quincunx.)
Six elements contribute to variation in a process. Those six elements – 5 Ms and on P (or 6Ms) influence variation in all processes – manufacturing or not.
Process spread vs centering
Types of Variation
Walter Shewhart developed the idea of the control chart in the 1920’s to help decide when the output of a process was part of “a stable system of chance causes”, or whether there was an “assignable cause.”
Common Cause Variation
Common cause variation are the usual, historical, quantifiable variation in a system. The best treatment is to look at all of the data available for the process and try to gain a better understanding of the system. You could then make basic, impactful changes to your process that would improve the whole.
While it would be a bad idea to try to determine what was different between each data point, your could investigate the overall trends and then adjust your process so that the range falls within the specification limits.
Some of those changes could be a great opportunity for a DMAIC project.
Special Cause Variation
Special cause variation are unusual, not previously observed, non-quantifiable variation. Best practices for fighting special cause variation include obtaining process information quickly, taking immediate action on them, and develop a quick solution to prevent recurrence.
Common Cause vs Special Cause Variation
If you were to treat common cause like special cause, you would miss the trend of variation that is in your process and thus miss the opportunity to improve it over the long run. For example, a recurring problem of late deliveries all treated like a special case would cause you to miss out on the root cause of the problem.
If you were to treat special cause like common cause, you would be over-deploying resources to fight an problem that isn’t as likely to happen. For example, if shipping was delayed due to a once-in-a-generation storm and you pivoted all of your attention to fixing the process for when that storm hit, you would de-prioritize other more likely, and more meaningful process improvements.
Why Businesses Should Measure Variation
It costs them money.
Fun game illustrating variation.
ASQ Six Sigma Black Belt Exam Variation Questions
Question: Legal requirements specify that a bottled product must contain at least the volume printed on the label. A bottling company wants to reduce the amount of overfilled bottles.
On the basis of the data above, what is the most effective strategy to accomplish this task?
(A) Decrease the target fill volume only
(B) Decrease the target fill variation only
(C) First decrease the target fill volume, then decrease the target fill variation
(D) First decrease the target fill variation, then decrease the target fill volume.
Answer: (D) Similar to the bullet diagrams above, the best strategy is to reduce variation in your process before trying to make an improvement. Think of the quincunx example. If you try to make an improvement, there change the place you drop the puck, then you actually get worse results because you never shrunk the dispersion (ie reduced variation.)