|
|
- Improvement of processes (and outcomes) requires statistical analysis of critical performance metrics
- A process/outcome can only be improved if it is “statistical control” – i.e., the performance metric varies over time in a random (but predictable) fashion around a mean (predictable = normal distribution)
- Variation within three standard deviations of the mean is usually “common variation” (exception: 7 or more consecutive data points above or below the mean = “special variation”)
Variation outside the confidence limit (mean+/- 3 S.D.) indicates “special cause” variation“
- Common Cause” variation
- Variation intrinsic to the process or system
- Interventions in response to “common cause” variations are usually futile, and usually make the process worse
- Individual performance is not the issue; performance is dictated by the process
- If you want to change the mean, or narrow the band of “common cause” variation (i.e., reduce the standard deviation), you need to change the process
- “Special Cause” variation
- Variation due to unusual circumstances
- Individual performance may (or may not) be the cause of such variation
- Special investigation warranted to identify and correct the cause
|
|
