Conduct Gage R&R studies and validate measurement systems per AIAG MSA manual. Covers variable and attribute studies, acceptance criteria, and calculation methods. USE WHEN user says 'MSA', 'gage R&R', 'GR&R', 'measurement system', 'repeatability', 'reproducibility', 'attribute agreement', or 'ndc'. Integrates with ControlPlan, SPC, and AutomotiveManufacturing skills.
MSA determines how much of the observed process variation is due to the measurement system rather than the actual process. Before making decisions based on measurement data, we must verify the measurement system is adequate.
Without MSA:
With MSA:
For measurements that produce numerical data (dimensions, weight, temperature, etc.)
| Study Type | Purpose | Method |
|---|---|---|
| Repeatability | Same operator, same gage, same part, multiple measurements | Single operator, 10+ measurements |
| Reproducibility | Different operators, same gage, same parts | Multiple operators measure same parts |
| Gage R&R | Combined repeatability and reproducibility | Standard study |
| Bias | Difference between measured and true value | Compare to master |
| Linearity | Bias across measurement range | Multiple references |
| Stability | Variation over time | Control chart on master |
For measurements that produce pass/fail, good/bad, or categorical results
| Study Type | Purpose | Method |
|---|---|---|
| Attribute Agreement | Operator consistency and accuracy | Multiple operators, multiple trials |
| Kappa | Agreement beyond chance | Statistical calculation |
| Effectiveness | Correct decisions vs. actual status | Reference evaluation |
| Parameter | Minimum | Preferred | Notes |
|---|---|---|---|
| Operators | 2 | 3 | Include typical operators |
| Parts | 5 | 10 | Represent process variation |
| Trials | 2 | 3 | Repeat measurements |
| Total readings | 20 | 30-90 | More = better discrimination |
| Metric | Acceptable | Marginal | Unacceptable |
|---|---|---|---|
| %GR&R (vs Process) | <10% | 10-30% | >30% |
| %GR&R (vs Tolerance) | <10% | 10-30% | >30% |
| ndc (Number of Distinct Categories) | ≥5 | 3-4 | <3 |
%GR&R <10%:
%GR&R 10-30%:
%GR&R >30%:
ndc (Number of Distinct Categories):
Analysis of Variance separates total variation into:
Simpler calculation, widely used:
Repeatability (EV) = R̄ × K₁
Where: R̄ = average range across all operators
K₁ = factor based on number of trials
Reproducibility (AV) = √[(X̄diff × K₂)² - (EV²/nr)]
Where: X̄diff = range of operator averages
K₂ = factor based on number of operators
n = number of parts
r = number of trials
GR&R = √(EV² + AV²)
%GR&R = (GR&R / TV) × 100
Where: TV = Total Variation = √(GR&R² + PV²)
PV = Part Variation
ndc = 1.41 × (PV / GR&R)
| Parameter | Minimum | Preferred |
|---|---|---|
| Appraisers | 2 | 3 |
| Samples | 20 | 30-50 |
| Trials | 2 | 3 |
| Sample mix | Include borderline | 50% good, 50% bad, include borderline |
| Metric | Description | Target |
|---|---|---|
| Within Appraiser Agreement | Self-consistency | ≥90% |
| Between Appraiser Agreement | Appraiser vs. Appraiser | ≥90% |
| Appraiser vs. Standard | Appraiser vs. Reference | ≥90% |
| Kappa | Agreement beyond chance | ≥0.75 |
| Kappa Value | Interpretation |
|---|---|
| <0.20 | Poor agreement |
| 0.21-0.40 | Fair agreement |
| 0.41-0.60 | Moderate agreement |
| 0.61-0.80 | Substantial agreement |
| 0.81-1.00 | Almost perfect agreement |
Measures systematic error (difference from true value)
Method:
Acceptance: Bias ≈ 0 or within calibration tolerance
Measures bias across the measurement range
Method:
Acceptance: Linearity (slope × Process Variation) <5%
Measures variation over time
Method:
Acceptance: Stable control chart, no trends
| Characteristic | Required MSA | Criteria |
|---|---|---|
| Critical (CC) | Gage R&R (variable) or Attribute Agreement | %GR&R <10% |
| Significant (SC) | Gage R&R (variable) or Attribute Agreement | %GR&R <30% |
| Standard | Gage R&R recommended | %GR&R <30% |
| SPC-monitored | Gage R&R required | ndc ≥5 |
| Issue | Likely Cause | Solution |
|---|---|---|
| High repeatability | Gage resolution, condition | Better gage, calibrate, repair |
| High reproducibility | Training, technique | Standardize method, train |
| High interaction | Operator-dependent method | Simplify method, fixture |
| Poor ndc | Gage can't see variation | More sensitive gage |
| Low Kappa | Ambiguous criteria | Define clearer standards |
| Bias | Calibration, wear | Recalibrate, adjust |
When generating MSA content:
# MSA Study Report
## Study Information
| Field | Value |
|-------|-------|
| **Study Type** | Gage R&R / Attribute Agreement |
| **Gage ID** | [ID] |
| **Gage Description** | [Type, range, resolution] |
| **Characteristic** | [What is measured] |
| **Specification** | [Tolerance] |
| **Study Date** | [Date] |
| **Conducted By** | [Name] |
## Study Parameters
| Parameter | Value |
|-----------|-------|
| Operators | [Number and names] |
| Parts | [Number] |
| Trials | [Number] |
| Total measurements | [Count] |
## Results
| Metric | Value | Acceptance | Status |
|--------|-------|------------|--------|
| %GR&R | [X]% | <10% / <30% | PASS/FAIL |
| ndc | [X] | ≥5 | PASS/FAIL |
| Repeatability | [X]% | - | - |
| Reproducibility | [X]% | - | - |
## Conclusion
[ACCEPTABLE / MARGINAL / UNACCEPTABLE]
## Actions (if required)
- [Action items]
All gages in Control Plan require MSA:
Load: read ~/.claude/skills/Controlplan/SKILL.md
SPC validity depends on MSA:
Load: read ~/.claude/skills/Spc/SKILL.md
MSA supports work instruction development:
Load: read ~/.claude/skills/Automotivemanufacturing/SKILL.md
For detailed guidance:
read ~/.claude/skills/Msa/CLAUDE.md
For study templates:
ls ~/.claude/skills/Msa/templates/
For acceptance criteria:
read ~/.claude/skills/Msa/reference/acceptance-criteria.md
For calculation formulas:
read ~/.claude/skills/Msa/reference/calculation-formulas.md