Attribute Control Charts (p, np, c, u)

When to use them

Attribute charts are for results you count rather than measure: parts that pass or fail a check, or flaws found on a unit. They show whether the defect rate is stable over time and flag the subgroups where something changed.

• p chart: the proportion of defective units, when the sample size can change from one subgroup to the next.
• np chart: the number of defective units, when every sample has the same size.
• c chart: the number of defects, when the same amount is inspected each time (one unit, one roll, one batch).
• u chart: defects per unit, when the amount inspected changes.

Enter the subgroups in production order. Twenty to 25 subgroups are a sensible minimum, and each sample should be large enough to find a few defectives or defects on average.

Method

Centre lines, control limits and the tests for special causes follow the AIAG SPC manual and Montgomery. When sample sizes differ, each subgroup gets its own limits, as specialised statistical software does. The results are tested against reference data sets for all four charts.

Worked example

The p chart data behind “Try an example”: 25 subgroups of 200 units, 213 defectives in total.

The rate shifted up around subgroup 14 and subgroup 18 is above the upper limit, so the process is not stable and 4.26% describes the past period only. Find what changed at subgroup 14 and what happened in subgroup 18 before using the rate as a baseline.