Selecting the right chart is fundamental to control chart accuracy. If the wrong control chart is selected, control limits will not be correct for the data. Knowing this, facing the possibility of an array of control charts may engender confusion, if not panic. Should I create charts, charts, np-charts, n-charts, p-charts, c-charts, or u-charts? Who’s to know?
Actually, determining the control chart to use is a straightforward, simple process that depends only on the format of data to be plotted and the sample or subgroup size in which it is collected. Data collected is either in variables or attributes format, and the amount of data contained in each sample, or subgroup, is provided. Got it?
Variables data is expressed as a measurement. Examples are height, weight, time, or length. If you use an instrument—a measuring tape, a scale, a gauge, or a clock, for example—to gather the data, it is most likely variables data. This kind of data, unlike attributes data, may have decimal places. The charts that are commonly used for variables data are charts and charts.
Attributes data, on the other hand, is a count of an occurrence, such as the number of employees, the number of errors, the number of customer calls, or the number of inoculations. Notice that the data can potentially be larger than the subgroup. One patient, for example, may have several different inoculations, or one purchase order might have many errors. Attributes data is always expressed in whole numbers (no decimals).
The amount of data contained in each sample offers another clue to selecting a control chart. Examples of sample size:
- When assessing the temperature in a vat of liquid, the reading is measured once per time period (such as hourly); therefore the sample size is one.
- When measuring the height of parts, a sample of five parts is taken and measured every 15 minutes; therefore the sample size is five.
- When checking 10 invoices per day for errors, the sample size is 10.
Note: “Sample size” refers to the amount of data in each sampling; this is also known as “subgroup size.” This is to be distinguished from the “sample.” For example, 10 samples, each with a sample size, or subgroup size, of 5, might be examined.
A flow chart demonstrates the ways in which type of data and sample (subgroup) size determine the control chart to be used:
© Cleary, M., and Graham, J. Practical Tools for Continuous Improvement (Dayton, OH: PQ Systems, Inc. 2000)
In the diamond above that asks whether the data being collected can be larger than the subgroup (sample) size, you will notice that the decision paths lead directly to the selection of particular types of charts, from c-charts across to p-charts. (Page numbers refer to references in Practical Tools for Continuous Improvement.)