Problem Solving & Quality · Quality Tools

Gauge R&R Study

Have several operators measure the same parts several times to see how much of the observed variation comes from the measurement itself.

Time1 h 30
FormatSmall group
StageQuality Tools

Gauge R&R Study: what it is and why it works

A gauge repeatability and reproducibility study quantifies how much of the observed variation in your data comes from the measurement system rather than from the parts. Several appraisers measure the same set of parts several times, blind and in random order. Repeatability is the variation seen when one appraiser measures the same part repeatedly with the same gauge; reproducibility is the variation between appraisers. Combined, they give the gauge R&R, usually expressed as a percentage of total study variation or of the tolerance, together with the number of distinct categories the system can resolve.

Every other data-driven tool assumes the numbers are trustworthy. If the gauge contributes a large share of the variation, control charts will show false signals, capability indices will look worse than the process really is, and root-cause analysis will chase noise. A gauge study is cheap compared with weeks of investigation built on bad data. It also points to the fix: poor repeatability suggests the instrument, fixture or part positioning, while poor reproducibility suggests the method, training or appraiser technique. It is typically run before a control chart, before a capability study and in the Measure phase of DMAIC.

What you need

About ten parts or samples that span the real process range, identified but hidden from appraisers
Two or three appraisers who use the gauge in normal work
The gauge, fixture and written measurement method as used on the floor
The tolerance and an estimate of process variation for the percentage calculations
A randomized run order and a data sheet or software template (average and range or ANOVA method)

What you get

Repeatability and reproducibility components and the total gauge R&R
%GRR against study variation and against tolerance, with the number of distinct categories
A decision: accept, accept conditionally, or improve the measurement system
Specific leads for improvement: instrument, fixture, method or training

When to use it

When the lab and the line disagree on the same sample, or the chart moves when nothing changed.

How to do it, step by step

Select about ten parts or samples that cover the real process range, and two or three appraisers who normally measure.
Have each appraiser measure each part two or three times, in random order, without seeing previous results.
Calculate repeatability (same person, same part) and reproducibility (between people).
Express the gauge variation as a share of the process variation or tolerance: under 10 % is generally good, above 30 % is not acceptable.
If the gauge fails, fix the method, the fixture or the instrument before using its data to solve anything.

Worked example: Bore gauge disagreement at a pump manufacturer

Illustrative scenario — figures are realistic but not from a real company.

A pump manufacturer machines cast iron impellers with a hub bore toleranced at ±0.0015 in. The line used a handheld dial bore gauge, the inspection room used a coordinate measuring machine, and the two regularly disagreed on borderline parts. A control chart on the line showed frequent out-of-limit points with no identifiable cause.

The quality technician picked ten impellers covering the full production range and marked them on the underside, where appraisers could not see the numbers.
Three line inspectors measured each part three times in random order over two days, giving 90 readings.
An ANOVA analysis gave a gauge R&R of 38 % of tolerance, with repeatability dominating. The number of distinct categories was 2, meaning the system could barely sort parts into big and small.
Watching the inspectors revealed the cause: the gauge was rocked by hand to find the minimum reading, and each person rocked differently. The team added a simple centering fixture and a written technique with a photo.

Result. A repeat study gave 12 % of tolerance and six distinct categories. The out-of-limit points on the line control chart mostly disappeared, confirming they had been measurement noise. The team concluded that three weeks of machine adjustments had been spent reacting to the gauge, not the process.

Open it in the FieldKit Problem Solving app → Get the Problem Solving template pack — join the waitlist

Common pitfalls and how to avoid them

Choosing ten parts that are almost identical, taken from the same hour of production.Select parts that span the real process range; otherwise the study understates part variation and exaggerates %GRR against study variation.
Letting appraisers know which part they are measuring or see their previous readings.Code the parts, randomize the order and have someone else record results so that memory does not flatter repeatability.
Using lab engineers or the best inspector instead of the people who measure every day.Use the normal appraisers, instruments and method, so the result reflects the measurement system as it is really used.
Running a variable gauge R&R on a pass/fail check.For go/no-go gauges or visual inspection, run an attribute agreement analysis instead, with known reference parts.

Frequently asked questions

What is an acceptable gauge R&R percentage?

A widely used guideline is under 10 % as acceptable, 10 to 30 % as possibly acceptable depending on the importance of the characteristic and the cost of improvement, and above 30 % as unacceptable. Check the number of distinct categories too; five or more is the usual minimum. Always say whether the percentage is of study variation or of tolerance, because the two answer different questions.

What is the difference between repeatability and reproducibility?

Repeatability is the variation when the same appraiser measures the same part several times with the same gauge; it reflects the instrument and the setup and is often called equipment variation. Reproducibility is the variation between the averages of different appraisers measuring the same parts; it reflects differences in method and technique and is often called appraiser variation.

Can you do a gauge R&R on a laboratory analysis?

Yes, with adaptations. For destructive or chemical tests where the same sample cannot be measured twice, use homogeneous split samples from one well-mixed batch and treat each split as a repeat. The analysts play the role of appraisers. The logic is the same: separate variation within an analyst from variation between analysts and compare it with the process variation and the specification.

Origin

Measurement system analysis and Gauge R&R — AIAG Measurement Systems Analysis (MSA) reference manual, first edition 1990.

Used in these playbooks

Process stability review 2 weeks

Two weeks to find out whether a disputed characteristic is a measurement, stability or capability problem — and to lock the answer into the control plan.

Related methods

More in “Quality Tools”

The statistical and risk tools of industrial quality: Pareto, SPC, capability, measurement, FMEA.