Variation
Variation is the degree to which the sequential outputs of a process or system differ. That is, it is a measure of how much the outputs of a process vary when the intent is to make them the same.
As an example, consider the 6 capital letter Cs shown in Figure 1. The intent of this effort was to draw 6 Cs that were, for all intents and purposes, the same, yet no two Cs are alike. Some are tall, while others are short. Some have smooth arcs, while others are jagged. All 6 efforts were an attempt to draw a capital letter C but no two of the Cs are the same.
Figure 1. Even simple efforts to draw multiple capital letter Cs reveal the influence of variation.
Granted, the example in Figure 1 is a little contrived but it illustrates the point nonetheless. Whether we like it or not and whether we can measure it or not, all processes (even simple ones) are influenced by variation. Thus, at its core, the task of manufacturing products and delivering services is the task of understanding variation. It is the task of reliably and repeatedly producing parts, assemblies, and products that are, for all intents and purposes, the same.
Two types of variation
Variation manifests in two forms: common causes of routine variation and assignable causes of exceptional variation.
When a process is influenced by only common causes of routine variation, future process behavior can be predicted within limits. This is because common causes, while many in number, are minimal in their effect.
While the influence of common causes can never be eliminated, they can be mitigated. However, mitigating the influence of common causes is not always economically feasible since it requires fundamental process changes (reengineering). These changes introduce new technology, new methods, new equipment, new materials, and new procedures into a process, making them costly to execute.
For details outlining how to determine if it is economically feasible to reengineer a process see Donald J. Wheeler’s Quality Digest article The Effective Cost of Production and Use and his book Reducing Production Costs.
When a process is influenced by both common causes of routine variation and assignable causes of exceptional variation, future process behavior cannot be predicted within limits. While few in number, assignable causes are dominant in their effect. They cause a process to bounce and shift in unpredictable ways. To improve a process influenced by assignable causes, the assignable causes must be eliminated.
Unlike a process influenced by only common causes, it is always economically feasible to eliminate the influence of assignable causes. Their dominant and unpredictable influence makes them a liability that cannot be ignored.
Figure 1. The two types of variation.