Gage R&R (Measurement System Analysis)
Validate your measurement system before you trust any process data. A Gage R&R study separates measurement variation into repeatability and reproducibility, compares it to total variation, and reports %GRR and the number of distinct categories.
Run Gage R&R →What is Gage R&R?
Gage Repeatability and Reproducibility (Gage R&R) is the core tool of Measurement System Analysis (MSA). It quantifies how much of the observed variation in your data comes from the measurement system itself, rather than from real differences between parts.
The measurement variation is split into two components. Repeatability (equipment variation, EV) is the variation when the same operator measures the same part repeatedly with the same gauge. Reproducibility (appraiser variation, AV) is the additional variation introduced when different operators measure the same parts.
The study compares this combined measurement variation to the total variation. If measurement error consumes too large a share, the gauge cannot reliably distinguish good parts from bad, and any capability study or control chart built on that data is untrustworthy. The modern method uses ANOVA, which also estimates the operator-by-part interaction.
In plain terms: Before blaming the process for variation, you have to make sure your ruler isn't the problem. Gage R&R checks whether the same person measuring the same part gets the same answer (repeatability) and whether different people agree (reproducibility). If the measurement noise is too big, your data is lying to you.
Variation Components
Repeatability (EV)
Equipment variation: the same operator, same part, same gauge, measured repeatedly. It reflects the gauge's inherent precision.
Reproducibility (AV)
Appraiser variation: differences between operators measuring the same parts. Large AV points to inconsistent technique or unclear procedures.
Part-to-Part & Total
Part variation is the real difference between parts. Total variation combines part variation and measurement variation; %GRR is measurement variation as a share of total.
Key Formulas
Acceptance Guidelines
The common MSA guideline: a %GRR under 10 percent indicates an acceptable measurement system; 10 to 30 percent may be acceptable depending on the application, cost and criticality; above 30 percent is unacceptable and the system needs improvement.
The number of distinct categories (ndc) should be at least 5. It represents how many separate groups of parts the measurement system can reliably tell apart. An ndc below 5 means the gauge lacks the resolution to support process analysis or control charting.
Assumptions & Validation
Representative Parts
The parts chosen should span the normal range of process variation.
If violated: Reselect parts to reflect the true spread; parts too similar understate part variation and inflate %GRR.
Randomized, Blind Measurement
Operators should measure parts in random order without knowing prior readings.
If violated: Enforce randomization and blinding to avoid bias.
Stable Gauge
The gauge should be calibrated and in stable condition throughout the study.
If violated: Calibrate before the study and confirm no drift during it.
Adequate Design
A typical study uses about 10 parts, 3 operators and 2 to 3 trials.
If violated: Increase parts, operators or trials for more reliable estimates.
⚠️ Check assumptions first
A misleading %GRR often comes from the parts, not the gauge. Because %GRR is measurement variation divided by total variation, choosing parts that are too similar shrinks part variation and makes the gauge look worse than it is. Always select parts spanning the real process range, and prefer the ANOVA method, which correctly separates the operator-by-part interaction that the older range method ignores.
When NOT to Use Gage R&R (MSA)
Attribute (Pass/Fail) Data
For go/no-go or visual pass/fail inspection, use an attribute agreement analysis (attribute MSA), not a variables Gage R&R.
Destructive Testing
When measuring destroys the part, the same part cannot be remeasured. Use a nested or destructive MSA design instead.
Single Operator, Automated Gauge
With no appraiser influence, reproducibility is not relevant; a repeatability-and-bias study may suffice.
Industry Applications
Pre-Capability Gatekeeping
Validate the gauge before any process capability or control-charting work, so decisions rest on trustworthy data.
Supplier & Audit Requirements
Provide MSA evidence required by automotive and aerospace quality systems before part approval.
Inspection Improvement
Diagnose whether inconsistent inspection comes from the gauge (repeatability) or the operators (reproducibility).
New Equipment Qualification
Confirm a new measurement device has adequate resolution and precision before deployment.
Frequently Asked Questions
What do repeatability and reproducibility mean?
Repeatability, also called equipment variation, is the variation seen when the same operator measures the same part multiple times with the same gauge; it reflects the instrument's inherent precision. Reproducibility, or appraiser variation, is the extra variation introduced when different operators measure the same parts. Together they make up the measurement system's contribution to total variation.
What is an acceptable %GRR?
A widely used guideline treats a %GRR below 10 percent as acceptable, 10 to 30 percent as marginal depending on the application's cost and criticality, and above 30 percent as unacceptable. The percentage expresses how much of the total variation is consumed by measurement error; the lower it is, the more of your data reflects real part differences.
What is the number of distinct categories (ndc)?
The ndc estimates how many distinct groups of parts the measurement system can reliably distinguish, calculated as about 1.41 times the ratio of part variation to measurement variation. A value of at least 5 is required for the gauge to support process analysis and control charting. A low ndc means the gauge cannot resolve real differences between parts.
Why should I use the ANOVA method over the range method?
The older average-and-range method estimates repeatability and reproducibility but cannot separate the operator-by-part interaction, the tendency for certain operators to measure certain parts differently. The ANOVA method estimates this interaction explicitly and gives more accurate variance components, which is why modern MSA practice prefers it.
How many parts, operators and trials do I need?
A standard Gage R&R study uses about 10 parts, 3 operators and 2 or 3 trials, giving enough data to estimate the variance components reliably. The parts must span the normal range of process variation; using parts that are too similar understates part variation and makes the measurement system appear worse than it is.
What if my %GRR is too high?
First confirm the parts genuinely span the process range, since unrepresentative parts inflate %GRR. Then separate the cause: high repeatability points to the gauge itself, calibration or fixturing, while high reproducibility points to operator technique or unclear procedures. Address the dominant component before repeating the study.
Validate Your Measurement System First
Separate gauge error from part variation and get %GRR and ndc. Free during Beta.
Run Gage R&R →