Attribute Process Capability: Capability for Count Data
What is Attribute Process Capability?
Variables capability is measured in σ; attribute capability is measured by the nonconforming fraction or defect rate — a binomial model for fraction nonconforming p and a Poisson model for defects per unit λ — then converted to a capability Z value (Z = Φ⁻¹(1−p)), sigma level and DPMO. Higher Z means better capability, and the tool performs the conversion from raw counts automatically, reducing manual table-lookup errors; count data only needs to be organized into batch counts.
When to Use It
Use it when only go/no-go or defect-count data are available: services, assembly operations, visual inspection, shipping errors and other cases where no continuous measurement exists, and for baseline assessment and before/after comparison in Six Sigma projects. Attribute capability quantifies count data and complements variables CpK to cover every data type.
How to Use It (Step by Step)
Enter the inspected quantity and the number of nonconforming items (or defect counts per unit) per batch, and the tool estimates p̄ or λ with the corresponding Z value, sigma level and DPMO, together with confidence intervals. Confirm the process is in control with a P or U chart before reading the capability, and for zero-defect batches report the confidence upper limit rather than a single zero.
Key Formulas / Example
Z = Φ⁻¹(1−p); DPMO = 1,000,000 × p on a per-opportunity basis; the long-term sigma level is commonly quoted as Z + 1.5 under the shift convention. Z can be compared with variables capability through Cpk ≈ Z/3, but the assumptions differ, so state them. Confidence intervals matter: capability conclusions depend on total sample size, and very low rates need very large samples.