Capability for Attribute Data

Capability for attribute data is the quality assessment approach for pass-fail or count-based outcomes where continuous-variable Cp and Cpk are not applicable - it uses defect-rate and binomial/Poisson metrics to evaluate process performance.

What Is Capability for attribute data?

• Definition: Capability evaluation for discrete outcomes such as defect present/absent or defects per unit.
• Core Metrics: DPMO, ppm defective, yield, sigma level equivalents, and confidence bounds.
• Data Models: Binomial for pass-fail and Poisson/negative-binomial for defect counts.
• Reporting Focus: Expected nonconformance rate under current process conditions.

Why Capability for attribute data Matters

• Method Correctness: Applying continuous capability indices to binary data gives misleading conclusions.
• Quality Governance: Attribute metrics align with inspection and escape-rate management workflows.
• Customer Alignment: Many contracts specify ppm or DPMO limits for acceptance.
• Improvement Tracking: Discrete metrics reveal effectiveness of defect-prevention actions over time.
• Cross-Process Comparability: Standardized attribute indices support benchmarking across product lines.

How It Is Used in Practice

• Data Stratification: Segment defects by mechanism, tool, lot, and opportunity count.
• Rate Estimation: Compute defect rates with confidence intervals using correct discrete models.
• Control Deployment: Use attribute control charts and targeted corrective actions for dominant defect categories.

Capability for attribute data is the correct statistical lens for binary quality outcomes - discrete defects require discrete metrics for honest process assessment.