What MSA actually is
MSA (Measurement System Analysis) is the set of studies that quantify how much of the variation you see in your data comes from the measurement system rather than the parts. It is one of the five automotive core tools, and its most-used study is Gauge R&R. The premise is uncomfortable but essential: every measurement is part variation plus measurement variation, and if the measurement variation is large, your inspection, your capability studies and your SPC are all built on sand.
MSA is distinct from calibration. Calibration proves a gauge reads true against a traceable standard; MSA proves the gauge can discriminate between parts in real use, in the hands of real operators. A gauge can be perfectly calibrated and still fail Gauge R&R.
The sources of measurement variation
MSA splits measurement error into two families — location errors (the system reads off-centre) and width errors (the system scatters):
| Property | Family | What it means |
|---|---|---|
| Bias | Location | The average reading differs from the true (reference) value |
| Linearity | Location | The bias changes across the operating range of the gauge |
| Stability | Location | The readings drift over time (day to day, week to week) |
| Repeatability (EV) | Width | Scatter when one operator measures the same part repeatedly — equipment variation |
| Reproducibility (AV) | Width | Difference between operators measuring the same parts — appraiser variation |
Repeatability and reproducibility
Gauge R&R combines the two width errors. Repeatability (equipment variation, EV) is the inherent scatter of the gauge itself — how much a single operator’s readings of one part vary when repeated. Reproducibility (appraiser variation, AV) is the disagreement between operators measuring the same parts, usually caused by differing technique, fixturing or interpretation. The combined Gauge R&R is the total measurement-system width, expressed relative to either the total study variation or the tolerance.
How a Gauge R&R study is run
The classic crossed study uses 10 parts, 3 operators and 3 trials — each operator measures every part three times, in random order, blind to the previous reading. The parts are chosen to span the expected process range. Two calculation methods are common: the Average and Range (X-bar & R) method, which is quick and hand-calculable, and ANOVA (analysis of variance), which is more rigorous because it also isolates the operator-by-part interaction. ANOVA is the preferred method in the AIAG MSA manual.
%GRR acceptance and ndc
Two numbers decide whether a measurement system is acceptable. The first is %GRR — the Gauge R&R expressed as a percentage of the total variation (or of the tolerance). The AIAG guidelines are:
| %GRR | Verdict | Action |
|---|---|---|
| Under 10% | Acceptable | The measurement system is good for the application |
| 10% to 30% | Conditional | May be acceptable depending on importance, cost of the gauge and application |
| Over 30% | Unacceptable | The system must be improved before use |
The second is the number of distinct categories (ndc) — how many separate groups the measurement system can reliably tell apart within the process spread. It is calculated as ndc = 1.41 × (PV / GRR), where PV is the part variation, and it should be 5 or greater for the system to be usable for variables data. An ndc below 5 means the gauge is effectively grading parts into too few buckets to control the process.
Attribute (go/no-go) MSA
Not every check is a variable reading — many are attribute decisions: pass/fail, go/no-go, present/absent. Attribute MSA (an attribute agreement analysis) tests whether appraisers agree with each other and with a known standard, usually via the Kappa statistic, where a value above roughly 0.75 indicates good agreement. It is the right study for visual inspection, gauging with plug/ring gauges, and any accept/reject judgement.
How Fast Quality Software runs MSA
Fast Quality Software records MSA and Gauge R&R per gauge, alongside the calibration register, so a gauge that cannot measure a characteristic reliably is caught before it is ever used on a control-plan check.
Because MSA studies sit in the same system as the gauge register and the control plan, the gauge attached to a special characteristic is one that has both a valid calibration and a passed R&R — and the study is retained as a PPAP element. See it applied for precision machining, or review pricing in INR.
Frequently asked questions
What is the difference between calibration and MSA?
Calibration proves that a gauge reads true against a traceable reference standard; MSA (Measurement System Analysis) proves that the gauge can actually discriminate between parts in real use, in the hands of real operators. A gauge can be perfectly calibrated and still fail Gauge R&R because of poor repeatability or operator disagreement, so both are required before a measurement can be trusted.
What is an acceptable Gauge R&R percentage?
Under the AIAG guidelines, a %GRR below 10% is acceptable, a %GRR between 10% and 30% is conditionally acceptable depending on the importance of the application and the cost of the gauge, and a %GRR above 30% is unacceptable and the measurement system must be improved before use. %GRR is the Gauge R&R expressed as a percentage of the total variation or of the tolerance.
What is ndc in Gauge R&R?
The number of distinct categories (ndc) is how many separate groups a measurement system can reliably tell apart within the process spread. It is calculated as ndc = 1.41 x (part variation / Gauge R&R) and should be 5 or greater for a variables measurement system to be usable. An ndc below 5 means the gauge cannot resolve the parts finely enough to control the process.
What is repeatability versus reproducibility?
Repeatability, also called equipment variation, is the scatter in readings when a single operator measures the same part repeatedly with the same gauge. Reproducibility, also called appraiser variation, is the disagreement between different operators measuring the same parts, usually caused by differing technique or interpretation. Gauge R&R combines both into the total measurement-system variation.
How many parts, operators and trials are used in a Gauge R&R study?
The classic crossed Gauge R&R study uses 10 parts, 3 operators and 3 trials, where each operator measures every part three times in random order and blind to previous readings. The parts are chosen to span the expected process range. Results can be calculated with the Average and Range method or, more rigorously, with ANOVA, which also isolates the operator-by-part interaction.