A control chart is a line chart that offers graphical representation of changes in processes and behaviors over time. They are also known as Shewhart charts. The components of a control chart are statistical points, the mean of this statistic, a center line for the mean, a standard deviation, and upper and lower limits usually represented at three errors from the center line. The chart may also include separate zones with associated rules; and annotations that describe significant events.
Control charts were developed by Walter A. Shewhart in the 1920s while working for Bell Labs. There was a strong need to manage product failures and repairs. In 1920, engineers realized that a reduction in manufacturing process variation was critical in this area. They also realized that modifications to manufacturing in response to product issues was not effective. Shewhart articulated the problem as common and special causes, and introduced the control chart as a means for separating them.
In a control chart, a process will be interpreted as “in control” if all points exist within the control limits. If any points exist outside those limits, they are interpreted as sources of variation, which are known as special-cause variations. Control charts are heavily employed in manufacturing, and variations indicate that quality costs will rise if the issue is not investigated.
Control charts offer valuable decision-making information, but they do not offer specific information related to the actual issue. They merely flag a process. This is both its strength and weakness. It can signal that something needs attention, but it cannot truly analyze the process as negative or positive; thus, it is difficult to use this analysis in processes that vary by default.