Gage R&R (Measurement System Analysis)
Gage R&R quantifies how much observed variation comes from the measurement system versus genuine part differences. Answers whether the gauge can reliably distinguish good parts from bad.
Step 1 — Provide your data
Crossed design: multiple operators each measure the same parts multiple trials.
Typical: 3 operators x 10 parts x 3 trials = 90 measurements.
Supports Excel Import, Sample Data Generator, and Manual Entry.
Step 2 — Results
Variance components:
- Repeatability (EV) — same operator, same part variation
- Reproducibility (AV) — between-operator variation
- Gage R&R (GRR) — combined EV + AV
- Part-to-Part (PV) — genuine part variation
Percentage breakdowns (% study variation, % contribution)
ANOVA table — Operator, Part, Interaction, Repeatability
Number of Distinct Categories (NDC)
A UniversalChatBot is available for discussion.
Statistical methods used
ANOVA method — two-factor crossed design (operators x parts with replicates).
Variance components derived from mean squares.
Acceptance criteria (AIAG):
| % Gage R&R | Verdict |
|---|---|
| < 10% | Acceptable |
| 10% - 30% | Marginal |
| > 30% | Unacceptable |
NDC: floor(sqrt(2) * SD_part / SD_gageRR). Need NDC >= 5 for adequate discrimination.
Interpretation:
- High repeatability = gauge imprecise (fix/replace gauge)
- High reproducibility = operators differ (train/standardise)
- High part-to-part = good (real variation being detected)