Gage R&R Results

Navigation Path: APQP > MSA Studies > [Select Study] > Gage R&R Results

The "Gage R&R Results" page presents a calculated study: whether the measurement system is fit for use, the numbers behind that judgment, the charts, the readings themselves, and the study's history. Any study that is Complete, Approved or Cancelled opens here from the studies list.

Page Header

The header names the study number and its type — "Variable Gage R&R" or "Attribute Gage R&R" — alongside the study's Status, Part, Feature, Tool, Spec and Plan, and the dates it was Created, Completed and Approved.

Page Actions

Approve
Approves a study in "Complete" status, setting it to "Approved". Shown only while the study is Complete.

Cancel Study
Cancels a study that is In Progress or Complete. A prompt asks for a reason, which is recorded against the Cancelled event on the Audit Trail. A cancelled study cannot be resumed.

Print
Opens a printable report of the study in a new tab.

Return to List
Returns to the MSA Studies list.

Verdicts

Every gated figure carries one of three verdicts, and the study's overall verdict is the worst of them.

• Acceptable: The figure meets the AIAG acceptance criterion.
• Marginal: The figure falls in the range AIAG treats as conditionally acceptable, depending on the application, the cost of the gage and the cost of repair.
• Unacceptable: The figure fails the AIAG acceptance criterion. The measurement system needs improvement before it is relied on.

Summary Tab — Variable Studies

The tab opens with the result strip: the gated figures, each with its value and verdict badge. Selecting any value opens an explanation showing the formula and the exact values substituted into it.

Gage R&R (% Study Var)
Measurement variation as a percentage of total study variation. The headline figure. AIAG gates it at 10% and 30% — under 10% Acceptable, 10% to 30% Marginal, over 30% Unacceptable.

% Tolerance (P/T)
Shown only when the feature has a tolerance. Measurement variation as a percentage of the tolerance, gated on the same 10% and 30% thresholds.

% Contribution
The variance-scale twin of %GR&R, gated at 1% and 9% — the 10% and 30% gates squared — so it always carries the same verdict as % Study Var.

Distinct Categories (ndc)
How many distinct groups of parts the measurement system can tell apart. Marked "Pass" at 5 or more, and "Low" below that, following the AIAG recommendation.

Beneath the strip, an acceptance gate bar plots %GR&R and P/T on the shared AIAG 0–10–30–100% scale. % Contribution and ndc are gated on their own scales and are not plotted on it.

Variance components

Variance attributed to each source, with each source's share of the total. Where a study reports both methods, a toggle switches the table between the method that produced the verdict — marked "used" — and Average & Range as a cross-check. Only one method's figures are ever on screen, so there is no ambiguity about which produced the verdict.

When a historical process sigma was entered at setup, a note under the heading says so: total variation is that sigma, and part-to-part is back-calculated from it.

Gage evaluation

The same sources expressed as standard deviations: StdDev (SD), Study Var (6 × SD), % Study Var, and % Tolerance when the feature has one. Study variation is 6 standard deviations, the spread covering 99.73% of readings. Standard deviations combine in quadrature, so the % Study Var column does not sum to 100% the way % Contribution does.

Probabilities of misclassification

How often this measurement system sorts a part into the wrong group, in two blocks: joint probabilities over a randomly selected part, and conditional probabilities within the group of truly bad, and truly good, parts. The four lines are the chance that a randomly selected part is bad but accepted, that a randomly selected part is good but rejected, that a part from a group of bad products is accepted, and that a part from a group of good products is rejected.

The true part value and the measurement error are taken as independent and normal, and each probability is that pair integrated over the spec region. A note under the heading says whether the part variation behind the figures came from the parts measured in this study or from the historical process sigma entered at setup — a ten-part sample and a real process can imply very different scrap and escape rates.

This section is hidden entirely on a study whose feature has no spec limits to classify against, and on attribute studies.

ANOVA table

Degrees of freedom, sum of squares, mean square, F-ratio and p for each variation source. On a two-way study a toggle switches between the model with the operator by part interaction term and the model without it, marking which one fed the reported variance components. The interaction is pooled into repeatability when it is not significant at α = 0.25.

A nested study uses a nested model with parts nested within each appraiser, and has no interaction term. A study analyzed by pooled standard deviation has no ANOVA table.

Summary Tab — Attribute Studies

Attribute studies are gated per appraiser, and the study takes the worst appraiser's verdict. The figures in the strip are that appraiser's, so each one matches the verdict beside it. Every appraiser's figures appear in the table below the strip.

Kappa (κ)
Agreement corrected for chance. Measured against the recorded reference when the study has one; otherwise it is the between-appraiser Fleiss kappa.

Effectiveness (against reference)
Shown only when a known reference was recorded. The share of parts an appraiser matched against the reference on every trial.

Miss Rate (customer risk)
Shown only when a known reference was recorded. How often a bad part was accepted.

False-Alarm Rate (producer risk)
Shown only when a known reference was recorded. How often a good part was rejected.

A second, quieter row carries the diagnostic figures, which are reported without a verdict:

Within-Appraiser (repeatability)
How consistently each appraiser repeated their own judgment across trials.

Between-Appraiser (reproducibility)
How consistently the appraisers agreed with one another.

All Appraisers vs Reference
Shown only when a known reference was recorded. The share of parts on which every appraiser agreed with the reference on every trial. It is reported without a verdict because the AIAG acceptance criteria are written against an individual appraiser, and this figure is stricter than any of them.

Per-appraiser agreement

One row per appraiser, with Within, and — where a reference was recorded — Effectiveness, κ, Miss, False alarm and that appraiser's verdict. Each appraiser's verdict is their worst gate.

Outcomes against the reference

Shown only when a known reference was recorded. The counts behind the rates above: Good accepted, Bad rejected, Bad accepted (miss), Good rejected (false alarm), and total Assessments. Counts are assessment-level, so each part contributes one row per trial.

Charts Tab

Variable studies show the standard Gage R&R chart set: Components of Variation, Measurement by Part, Response by Part (Operator), R Chart by Operator, Measurement by Operator, Xbar Chart by Operator, and Part × Operator Interaction.

Attribute studies show Within Appraisers and Appraiser vs Standard. Each chart carries its own explanation on the help icon beside its title.

Measurements Tab

The readings as recorded, one table per appraiser, with trials down and parts across. Variable studies close each table with a "Part avg" row; attribute studies with a known reference close it with a "Reference" row instead. Cell shading follows the same tolerance colouring used on the Record Measurements grid elsewhere in Quality Management.

On a nested study a note above the tables records that each appraiser's parts are distinct.

Audit Trail Tab

Every event recorded against the study, with Timestamp, User, Action and Detail. The event types are Created, Entered, Calculated, Reviewed, Approved, Plan Changed, Cancelled, Completed, Linked and Unlinked. Where a reason was captured — on a cancellation, for instance — it appears in quotes in the Detail column. Use Export to download the history as an Excel file.

Access

Supplier Users:

• Full access to every tab on their own company's studies
• Can approve a study when its status is "Complete", with the Calibrator, Setup Administrator or Super Administrator role
• Can cancel a study when its status is "In Progress" or "Complete", with the same roles

Customer Users:

• Can open a supplier's study read-only when they reach it through a linked APQP/PPAP checklist item or a Bill of Characteristics
• Can view the Summary, Charts and Measurements tabs
• Cannot approve, cancel or print the study, and do not see the Audit Trail tab or the return link to the supplier's studies list

If the role you need is unavailable, contact your setup administrator. See this article to find a setup administrator.

Common Issues & Quick Solutions

"No results have been calculated for this study."
The study has not finished collection. Open it from the studies list to continue recording measurements; results are calculated once every planned measurement has been saved.

"This study was cancelled before a result was calculated."
The study was cancelled during collection, so there is nothing to report. The reason recorded at cancellation appears on the Audit Trail tab.

"This page cannot be accessed due to the following issue:"
Your user does not have access to this study — most often because it belongs to another company and is not linked to an APQP/PPAP record or Bill of Characteristics you can see. The message names the specific issue.