Control Charts

• Identify significant events (called "special cause" variation) that require immediate management action.

• Graphically display process stability and capability.

• Identify trends before they become "special causes".

Types of Variation

• Special Cause - An event or run that statistically differs from the rest of the data. This means that the probability of the event or run occurring by chance are vanishingly small (usually <1%). This type of variation is often called assignable cause because it is likely that only one of the myriad of "causes" impacting the process actually caused the event.

   

• Finding Special Cause

  + Any point outside the control limits is due to "special cause" variation and requires investigation.

   

  + Individual points within the limits are a part of the normal (common cause) variation in the process. Improving them requires examination of the entire process.

   

  + Run chart trend analysis is also applicable. For example runs up/down.

   

  + 2 of 3 consecutive point above or below the 2 standard deviation limits.

   

  + 4 of 5 consecutive points above or below the 1 standard deviation limits.

   

  Note: Using more tests increases the likelihood of generating a false alarm.

• Common Cause - Other than the special cause events or runs process data is random "noise" and the "signal" of no one individual cause is discernible. When the process is operating at this level it is stable and predictable. Any improvement efforts must address the entire causal structure.

Types of Data

o Variables Data - Numeric data that can (theoretically) assume any value.

o Attributes Data - Data regarding the state or condition rather than the actual numeric value. For example "Normal" or "Abnormal" for a lab result rather than the actual result.