Which control chart should I use? A Six Sigma decision guide

Continuous (variables) data

Measurements such as length, time or weight:

• I-MR (individuals and moving range): one measurement at a time, subgroup size of 1.
• X-bar and R: small subgroups, typically 2 to about 9 measurements each.
• X-bar and S: larger subgroups, typically 10 or more, where the standard deviation is a better estimate of spread than the range.

Attribute (count) data

First decide whether you are counting defective units (each unit passes or fails) or defects (one unit can have several). Then check whether the sample size is constant.

• p chart: proportion of defective units; sample size may vary.
• np chart: number of defective units; constant sample size.
• c chart: number of defects; constant area of opportunity.
• u chart: defects per unit; area of opportunity may vary.

Control limits are not specification limits

Control limits are calculated from the process's own data, conventionally three standard deviations either side of the center line, and show what the process is doing. Specification limits come from the customer and show what it should do. A process can be in control and still make out-of-specification product, and the exam will test that distinction.

Reading the chart

A point beyond a control limit signals a special cause. Run rules catch subtler signals, such as a long run of points on one side of the center line, or a steady trend up or down. Investigate special causes; do not adjust the process in response to common-cause variation, which only adds variation.