Attribute Gage R&R calculator
Enter each appraiser's pass/fail calls against a known reference and get effectiveness, miss and false-alarm rates, within and between-appraiser agreement, and a kappa statistic, with a traffic-light verdict against the AIAG guidelines. The go/no-go companion to variable Gage R&R. Everything runs in your browser.
Study setup
Enter P for pass (good, accept) and F for fail (bad, reject) in every cell, including the reference column. The tool also reads pass/fail, good/bad, 1/0, and yes/no. The reference is the known correct decision for each part.
Judgments
One P or F per appraiser, part, and trial, plus the reference call for each part. Results update live.
Paste a block from Excel starting at the reference cell of Part 1 (columns: Reference, then each appraiser's trials in order), or import a CSV with the downloadable template. Use the arrow keys or Enter to move between cells.
Result
Method: AIAG MSA 4th ed. attribute agreement analysis. Effectiveness, miss rate, and false-alarm rate follow the AIAG effectiveness table (miss ≤ 2% / ≤ 5%; false alarm ≤ 5% / ≤ 10%). Kappa is Cohen's kappa for each appraiser against the reference, pooled over every trial. The overall verdict uses the weakest appraiser.
Attribute agreement analysis, in plain terms
What an attribute study tells you
When your gage gives a number, you run a variable Gage R&R. When the decision is pass or fail, go or no-go, good or bad, there is no number to work with, so you study whether the people making the call agree. An attribute agreement study hands a set of parts to a few appraisers, has each of them judge every part more than once, and compares the calls against a known reference decision for each part. It answers three separate questions: does an appraiser repeat their own call, do the appraisers agree with each other, and do they agree with the truth.
Effectiveness, misses, and false alarms
Effectiveness is accuracy: the share of parts an appraiser gets right on every trial, judged against the reference. AIAG reads it against a traffic light, 90% or better acceptable, 80% to 90% marginal, below 80% unacceptable, and asks that every appraiser clear the bar, so this tool grades on the weakest appraiser rather than the average. Two errors sit underneath that number. A miss is passing a part that is truly bad, so a nonconforming part escapes. A false alarm is failing a part that is truly good, so you scrap or rework something fine. Because a miss is usually costlier, the AIAG bands are tighter on misses (2% and 5%) than on false alarms (5% and 10%).
Within, between, and versus the reference
Within-appraiser agreement is repeatability: does one person give the same call on the same part every trial. Between-appraiser agreement is reproducibility: do all the appraisers land on the same call for a part. Agreement versus the reference is effectiveness: do the calls match the known correct answer. They are not the same thing. An appraiser can be perfectly consistent with themselves and consistently wrong, and a whole team can agree with each other while all missing the same borderline parts, so the reference comparison is the one that decides whether the inspection can be trusted.
Kappa and the AIAG thresholds
A raw agreement percentage can look high just because most parts are obviously good. Kappa corrects for the agreement you would expect by chance given how many good and bad parts were in the study, on a scale where 1 is perfect and 0 is no better than a coin toss. This tool reports Cohen's kappa for each appraiser against the reference, pooled over every trial as one opportunity. The rule carried into the AIAG manual is that kappa of 0.75 or higher is good to excellent, and below 0.40 is poor. Use it alongside effectiveness, not instead of it.
Method and its limits
This calculator uses the AIAG MSA attribute agreement math, the same figures Minitab's attribute agreement analysis reports: within-appraiser, between-appraiser, each appraiser versus the reference, all appraisers versus the reference, plus miss and false-alarm rates. It reports a Cohen kappa per appraiser as a separate cross-check, which differs from the Fleiss kappa Minitab uses, so treat it as directional rather than an exact match. Its honest limits are the study you feed it. A 10-part study is a quick screen; a study you act on wants more parts, commonly around 50, weighted toward the accept and reject boundary where inspectors actually disagree, because that is where a weak inspection shows up. The tool reports Cohen's kappa on the pooled appraiser-versus-reference table rather than a Fleiss kappa across appraisers, so a value here can differ slightly from other software. References: AIAG Measurement Systems Analysis (MSA), 4th edition.
Common questions
What is an attribute Gage R&R (attribute agreement) study?
An attribute Gage R&R, or attribute agreement analysis, checks a pass/fail (go/no-go) inspection instead of a numeric measurement. Several appraisers judge the same set of parts more than once, each part has a known correct answer (the reference), and you look at how often the appraisers agree with themselves, agree with each other, and agree with the reference. It is the attribute half of measurement systems analysis (MSA), alongside the variable Gage R&R you run on numeric gages.
What is effectiveness and what is an acceptable value?
Effectiveness is the share of parts an appraiser classifies correctly across every trial, that is, every one of that appraiser's repeat calls on a part matches the reference. The common AIAG guidance reads it against a traffic light: 90% or higher is acceptable, 80% to 90% is marginal and may need improvement, and below 80% is unacceptable. Every appraiser should clear the bar, not just the average, so this tool drives the verdict off the weakest appraiser.
What is the difference between the miss rate and the false-alarm rate?
A miss is calling a truly bad part good: a reference-fail item judged pass. A false alarm is calling a truly good part bad: a reference-pass item judged fail. The miss rate is misses divided by all reference-fail opportunities, and the false-alarm rate is false alarms divided by all reference-pass opportunities. A miss is usually the costlier error because a nonconforming part escapes, so the AIAG guideline is tighter on misses (2% and 5% bands) than on false alarms (5% and 10% bands).
What is kappa and what value is good?
Kappa measures agreement corrected for the agreement you would expect by chance, on a scale where 1 is perfect and 0 is no better than guessing. This tool reports Cohen's kappa for each appraiser against the reference, pooled over every trial. The widely cited rule, given in the AIAG manual, is that a kappa of 0.75 or higher indicates good to excellent agreement, and below 0.40 is poor. It is a useful cross-check on effectiveness because it accounts for how many good and bad parts were in the study.
What do within-appraiser, between-appraiser, and versus-reference agreement mean?
Within-appraiser agreement asks whether one appraiser repeats their own call on a part across trials (consistency). Between-appraiser agreement asks whether all appraisers give the same call on a part (reproducibility across people). Agreement versus the reference asks whether the calls match the known correct answer (effectiveness, or accuracy). An appraiser can be perfectly consistent and still wrong, so you need all three, and effectiveness against the reference is the one that decides whether the inspection is trustworthy.
How many parts, appraisers, and trials should I use?
A common AIAG attribute study uses 3 appraisers, about 50 parts, and 3 trials each. Choose parts that span the decision: include clearly good parts, clearly bad parts, and a good number near the accept/reject boundary, since the borderline parts are where inspectors disagree. Small studies run faster but give a coarse, unstable kappa, so treat a 10-part result as a screening check and use more parts, weighted toward the limit, for a study you will act on.
This calculator applies the AIAG MSA attribute agreement method to the calls you enter and reports effectiveness, miss and false-alarm rates, within and between-appraiser agreement, and Cohen's kappa. It does not replace the AIAG Measurement Systems Analysis manual and is not your quality system. Treat the result as a working estimate, and validate it against your documented MSA procedure before acting on it.
Keep the study with the record
Axiospec keeps every calibration, reading, and certificate in one tamper-evident ledger built for ISO/IEC 17025. See it on real calibration data, no signup.