Poisson Capability evaluates defect-count performance when opportunities occur over units, time, area, length, or volume and count data follow Poisson behavior.
Definition
Poisson Capability is capability analysis for count data where defects or events occur over a defined area of opportunity, such as defects per unit, leaks per meter, calls per hour, or errors per transaction batch. It estimates whether the process count rate is acceptable relative to requirements.
Unlike normal capability, Poisson methods are designed for discrete counts and event rates.
History
Poisson models come from probability theory and have long been used for counts of rare or independent events. Quality practitioners use them in capability, control charts, and defect-rate analysis when continuous measurement assumptions do not fit.
When to Use
Use Poisson Capability when the response is a count of defects or events and the opportunity area is known. It is common for surface defects, errors per form, service incidents, contamination counts, and calls per period.
Step-by-Step
- Define the event or defect clearly.
- Define the opportunity unit: item, area, time, length, or transaction.
- Confirm counts are collected consistently.
- Check stability using an appropriate count chart.
- Evaluate whether the Poisson assumption is reasonable.
- Estimate defect rate and capability relative to requirements.
- Stratify if rates differ by product, shift, location, or condition.
- Improve causes and verify the event rate changed.
Examples
- Coating: Pinholes are counted per square meter.
- Service: Data-entry errors are counted per 1,000 transactions.
- Maintenance: Stops are counted per operating hour.
Common Pitfalls
- Using Poisson methods for continuous measurements.
- Ignoring changing opportunity size.
- Combining different count-rate populations.
- Ignoring overdispersion or clustering.
- No stability check before capability.
- Confusing defect counts with defective-unit counts.