Variation - difference in the output of a process (or inputs to a process) over time. Variation consists of common cause variation, special cause variation and structural variation (and some include tampering).
Shewart provide the control chart as a tool to use when managing processes. The reduction of variation has profound impacts on the reduction of costs and increase in customer satisfaction.
The control chart helps identify what improvement stategy to use. This is done by helping identify when to try special cause thinking (identify the assignable cause) and when to apply common cause thinking (system improvements). It is possible to make up "assignable causes" for common cause variation. This is what we often do. The control chart is meant to help us avoid looking to that form of thinking and instead in cases where a special cause is not indication to use system improvement thinking instead.
Special cause thinking should be used when point is outside the control limits (more than 3 standard deviations away from the mean) of there is a pattern in the data that indicates a special cause. These patterns are often know as the western electric rules.
- common cause variation - variation resulting from the system. Every system will have some amount of variation of results. The way to improve common cause variation is to change the system.
- special cause variation - variation resulting from a assignable cause. Special causes should be addressed by finding the special cause and taking action. Special Causes can be beneficial, in which case identify the cause and seek to incorporate it in the standard practice.
- patterns indicating a special cause (often refered to as the Western Electric rules, though more rules have been added)
- 8 points in a row on one side of the mean.
- 4 of 5 points between 1 sigma and 3 sigma from the mean (none within 1 sigma of the mean).
- 2 of 3 points between 2 sigma and 3 sigma from the mean.
- Trend of 6 points in a row increasing or decreasing.
- 14 points in a row that alternate above and below the mean.
- 8 in a row not within 1 sigma of the mean.
- Structural Variation - trends within the data. They often take the form of seasonal variation and growth or decline (where the data has, for example, a underlying trend of increasing 4% annually).
- Tampering - taking action based on the belief that a common cause is a special cause. The tendancy to take action, often leads to action without reason which causes more problems than it fixes. Dr. Deming stated that most variation (97% plus) was common cause variation not due to special causes. Tampering can also be considered a form of variation.
- Out of Control Process (unstable process) - a process not in statistical control. Many Quality and Statistical tools require a stable process and therefore using them when the process is out of control will not provide accurate results.
- Stable Process - one in statistical control. Various methods exist to determine whether a process is in statistical control.
- Control Chart - chart of data to aid in determining wether a process is in control and to make it easier to spot data that may indicate special causes.
Find More Management Resources
Management Dictionary: Affinity Diagram, Cause and Effect Diagram, Control Chart, Cycle Time, Kanban, One piece flow (continuous flow), PDSA, Poka Yoke, Red Bead Experiment,
Root Cause Analysis, Run Chart, Statistical Process Control (SPC)