Attribute Control Charts

Definition

Attribute Control Charts are statistical process control charts for count-based or classification-based data. They are used when a unit is counted as defective or not defective, or when defects are counted per unit, batch, area, time period, or opportunity set.

The most common attribute charts are p charts for proportion defective with varying sample sizes, np charts for number defective with constant sample size, c charts for number of defects with constant area or opportunity, and u charts for defects per unit when the number of units or opportunities varies. The goal is to distinguish routine common-cause variation from special-cause signals that require investigation.

History

Attribute control charts developed from the same statistical process control foundations as variable control charts. As quality control expanded beyond continuous dimensions and measurements, practitioners needed methods to monitor defect counts, rejected units, complaint rates, audit findings, injuries, errors, and other count data.

They remain widely used because many operational outcomes are naturally attribute-based: a shipment is late or on time, a claim is correct or incorrect, a unit has defects or does not, a record has missing fields, or a patient form is complete or incomplete. Attribute charts provide a practical way to monitor these outcomes over time.

When to Use

Use attribute charts when the response is a count, proportion, percentage, rate, or classification and the team wants to monitor process stability over time. Use them for defect rates, customer complaints, late orders, scrap units, audit findings, safety incidents, invoice errors, rework counts, missing documents, or inspection rejects.

Use variable control charts when a continuous measurement is available and meaningful, because variable data usually contains more information. Before using attribute charts, confirm that the classification system is reliable, the sampling logic is stable, and the opportunity definition is consistent.

Step-by-Step

1. Define the attribute clearly. Specify what counts as a defective unit, defect, complaint, error, incident, or opportunity.
2. Choose the chart type. Use a p chart for proportion defective with variable sample size, np chart for count defective with constant sample size, c chart for defects with constant opportunity, and u chart for defects per unit with variable opportunity.
3. Collect time-ordered data. Preserve the sequence of production, service, inspection, or reporting periods. Control charts depend on time order.
4. Check classification reliability. Use attribute agreement analysis where human judgment affects the data.
5. Calculate the center line and limits. Use the correct binomial or Poisson-based limit formulas for the chosen chart.
6. Plot the data. Show each subgroup, sample, day, shift, batch, or period against the center line and control limits.
7. Interpret special-cause signals. Look for points outside limits, runs, trends, shifts, cycles, or other nonrandom patterns based on the organization's SPC rules.
8. Investigate signals at the process. Connect signals to actual process changes, materials, people, equipment, suppliers, methods, or environment.
9. Update controls and sustain. Use findings to improve standard work, control plans, error proofing, training, or reaction plans.

Examples

• p chart: A quality team tracks the percent of rejected lots each week while weekly inspected volume changes. The p chart accounts for varying sample size.
• np chart: A line inspects exactly 200 units each shift and tracks the number of units with at least one defect. The constant sample size makes an np chart appropriate.
• c chart: A paint department tracks the number of surface defects on one standard panel size per shift. Because the opportunity area is constant, a c chart fits.
• u chart: A claims team tracks errors per 100 claims when the number of claims reviewed changes each day. The u chart normalizes defects per unit.
• Healthcare example: A clinic tracks missing fields per patient record. If the number of records reviewed varies by day, a u chart is generally more appropriate than a simple count trend.

Common Pitfalls

• Choosing the wrong chart. Defects and defectives are different. Constant and varying sample sizes require different chart logic.
• Ignoring changing opportunity. A raw defect count can rise simply because more units were inspected or more opportunities existed.
• Using percentages without sample size context. A 5 percent defect rate from 20 units has different uncertainty than 5 percent from 2,000 units.
• Poor operational definitions. Inconsistent defect classification creates false signals or hides real ones.
• Treating every high point as special cause. Control limits help distinguish expected variation from signals that need investigation.
• Recalculating limits too often. Frequent limit changes can hide process shifts. Recalculate when the process has genuinely changed and is stable.
• Using attribute data when variable data is available. Continuous measurements often support stronger analysis and earlier detection.

Related Tools

• Statistical Process Control (SPC)
• Control Charts
• p chart, np chart, c chart, and u chart selection logic
• Attribute Agreement Analysis
• Pareto Analysis
• DPMO (Defects Per Million Opportunities)
• Control Plan
• Control Limits Generator

Further Reading

• SPC Control Limits Generator
• ASQ CQE Section 6: Quantitative Methods and Tools
• Yield, Defect Rate, and DPMO Converter
• Quality Engineering Topic Hub
• Six Sigma Topic Hub