Interpretation of SPC Control Chart Patterns

When interpreting statistical process control chart patterns consider the following. When a process is in control, the control chart pattern should exhibit natural characteristics as if it were from random data. Unnatural patterns involve the absence of one or more of natural -pattern characteristics. Some examples of unnatural patterns are: mixture, stratification, instability, data transformation need, and/or a process shift.

For identified unnatural patterns when using statistical business performance charts (i.e., 30,000-foot-level control chart), a reason for the occurrence should be, when possible, determined and documented on the chart. Identified process shifts need staging so that data from each staged region determines the time-period’s control limits.

For each stability regions, a process performance statement (e.g., percent non-conformance rate) can be made. Whenever there is a recent region of stability, the process can be said to be predictable and the data from this region can be considered a random sample of the future, which presumes that the process inputs and implementation procedures remain unchanged from the most recent stability region.

More information about this form of business performance scorecard charting can be found at “Creation of Effective Organizational Predictive Metrics that Lead to the 3 Rs of Business”