There is nothing common about common cause variation! When a process shows only common cause variation, it is predictable into the near future, and that is truly an uncommon advantage.
Overview: What is common cause variation?
A control chart can show two different types of variation, special cause variation (points out of limits or a non-random pattern of variation) and common cause variation.
Common cause variation is present when the control chart of a process measure shows a random pattern of variation with all points within the control limits. When a control chart shows common cause variation, a process measure is said to be in statistical control or stable.
Fundamental changes to your critical process components, such as machinery, material, personnel, environment, or measurement are necessary if you want to reduce common cause variation.
3 benefits of common cause variation
When your process consists of common cause variation, it is time to “Shrink, shrink, shrink that variation!” — W. Edwards Deming
When the process is in statistical control:
1. The process outcome is predictable in the short term
This is a huge benefit! When a process is predictable, you can expect the outcome will be between the control limits.
2. You can evaluate process capability
Process capability compares the common cause variation from your process measure to the customer or engineer’s specifications and target for that measure.
A process is considered capable if the process is stable AND the common cause variation is within the specification limits and centered on the target.
3. You can improve the process
A stable process is a prerequisite to a capable process. Once a process measure is stable, you can shrink its variation by making fundamental changes to its critical process components. For example, we conducted an experiment that showed certain machine settings could reduce the process measure’s common cause variation.
Why is common cause variation important to understand?
Treating the variation that is from common causes as though it is from a special cause of variation is a mistake called tampering. Tampering should be avoided. It wastes time and resources and will increase the variation of the process.
Common cause variation will present as the process measure varying around its center line within the calculated control limits. If this variation is unacceptable, you must make fundamental changes to the process components to reduce the variation.
For example, consider a stable control chart for the number of scratches on a casing, with an upper limit of 6 scratches and a lower limit of zero scratches. Sample averages moved from 3 to 2 to 5 scratches over three samples.
If you were to go to your workers and ask, “what happened?” when the error rate rose to 5, they would probably respond with “nothing new!” And they would be right because you are tampering — an error rate of 5 scratches normal because it is within the control limits.
If instead you were to improve the polishing process so that it was more effective, you might reduce the variation so that new process is in control with an upper limit of 4 scratches, thus shrinking the variation and avoiding the waste that comes from tampering.
An industry example of common cause variation
You can learn a lot about how to improve customer satisfaction by studying common cause variation.
In one example, a control chart was used to monitor cycle time for review of a loan application. Each day, a sample of applications was reviewed. The average time for review was 3 hours. The upper control limit was 5 hours, and the lower control limit was 1 hour. The process was in control. The customer required that the application be reviewed in 4 hours or less.
Today’s sample came in at 4.25 hours. The process is still in control. What can we conclude about the sample?
The manager should not search for a special cause, because a 4.25-hour review is in the expected range of process measure results.
- The customer will not be satisfied with the 4.25-hour cycle time.
- The process is stable, but not capable of meeting customer requirements.
- The manager must make fundamental changes to process components to shrink variation of the application review time.
The manager worked with the customer to streamline the application, making it shorter. This fundamental change reduced variation, making the review process more capable of meeting customer specifications.
3 best practices when thinking about common cause variation
Common cause variation indicates a process that’s in statistical control, or stable. Use common cause variation to:
1. Predict the range of values you can expect from your process measure
For example, a control chart of yield is operating in control at an average of 90% with an upper control limit of 92% and a lower control limit of 88%.
2. Implement fundamental changes to critical process components if the process requirements are not being met
Our manager wanted a better yield from her process, so she created a more aggressive mechanical maintenance schedule to ensure the critical machines were operating as desired.
3. Continue to plot the control chart after the change
The control chart indicated our manager’s change worked by showing a sudden shift upward of the yield after the change (our yield improved to 95%) and/or a reduction in variation (our yield variation stayed at +/- 2%).
Frequently Asked Questions (FAQ) about common cause variation
Can I put specification limits on a control chart that is stable?
No. Common cause variation should only be evaluated against control limits calculated from the data.
Once my process shows stability (common cause variation only), am I done improving my process?
Not always. Your process measure can be in control and still not meet your company’s or your customer’s requirements.
If it’s worth it to you or your customer, continue to improve critical process components, and monitor the control chart to see if you have stabilized the process measure with less variation.
Final thoughts on common cause variation
A process in statistical control will show common cause variation. It takes some work to achieve a stable process, but it is worth it — if only for the predictability that common cause variation buys you.