Who needs capability analysis?

Confusion about whether you need to apply standard methods of calculating capability sometimes means that this powerful statistical application is either applied when it isn’t needed, or not used when it would provide useful information. When is capability analysis called for, anyway? Some guidelines will help.

If it’s important to understand how a system performs compared to specification limits, capability analysis is useful. Of course, the requirements or specifications must be available, and the system must be measured in the same way as the specifications are measured so a comparison will be useful. If both upper and lower specifications are designated, standard methods for capability analysis apply. One-sided specifications require a slightly different approach.

A system must be rendered stable in order for capability analysis to be accurate and useful. This means that there are no special causes of variation, determined by creating a control chart before undertaking capability analysis.

Data must be in variables form if the standard method for capability analysis is to be used. That means that it is measured data, such as weight, time, volume, or length. A histogram should be constructed to check for normal distribution. Something known as non-normal capability analysis can apply for data that does not reflect a normal distribution.

Data collected over a period of time using control charts offers the best opportunity to use the standard method of applying capability analysis.

To learn more about using capability analysis when one-sided specifications, attributes data, or non-normal distribution is used, see Practical Tools for Continuous Improvement.

Originally published in the January 2011 edition of Quality eLine, our free monthly newletter.

Subscribe now!