When to use attributes versus variables controls charts

You have attributes data when:

·    Your data can take only one of two values, such as pass/fail, go/no-go, or present/absent. For example, in a service call center incoming calls are either answered (a success) or not (a failure). This is defectives data, which takes the form of the binomial distribution.

·    Your data are counts, such as number of surface scratches, number of accidents, or number of typographical errors. For example, a fabric manufacturer charts the number of blemishes per 10 square yards of fabric. This is defects data, which takes the form of the Poisson distribution.

You have variables data when:

·    Your data are measurements, such as length or cylinders, fill weight of tubes, or growth rate of plants.

It is possible, however, to convert variables data to attributes data by classifying whether or not the product meets specifications. For example, if you are interested in the diameter of ball bearings, you can

·    Measure the ball-bearing diameters precisely and record the exact measurements. In this case you would be using variables data.

·    Measure the ball bearings and then classify the measurements according to whether or not they fall within the specification limits. If you record the number of ball bearings that do not meet the specifications, you would be using attributes data.

The following lists show the features of variables control charts and attributes control charts. Use information about the charts and your process to determine which chart best meets your needs.