Gage R&R (Repeatability and Reproducibility) attribute analysis is a powerful methodology used in quality control to assess the accuracy and reliability of a measurement process when qualitative characteristics are involved.
1. Carry out an R&R study
To perform a measurement method analysis with attribute data.
- Click on the Gage R&R menu
- In the horizontal bar: click on the Gage R&R attribute submenu.
- Click on "add a Gage R&R".
- Complete the information on the method.
⇒ Range name: Designation of the qualitative characteristic measured. This can encompass various responses with two modalities. example: aspects (Conforms/Nonconforms), criteria (Acceptable/Not Acceptable), and so on.
⇒ Products: Enter the quantity of products (compliant or non-compliant).
⇒ Operators: number of operators.
⇒ Repetitions: number of measurement repetitions per operator per part.
⇒Measurement process: The name of the inspection method equipment or inspection machine.
- Pressing validate opens the R&R attribute pledge page below. This page is composed of two zones
⇒ Zone 1All the information entered in the previous step. Such as number of parts, number of operators, number of repeats.
⇒ Zone 2 In this zone, complete the "Expert" column and the results obtained by operator:
- left-click for "CF" ⇒ a "CF" result will be in green.
- left-click+Ctrl for "NCF" ⇒ an "NCF" result will be highlighted in red
2. Using the results
In zone 2, you can access 3 tabs. "measurements", "analysis" and "graphics".
2.1 Analysis
In the analysis tab, several results are presented in different tables:
⇒ Table of Cohen's Kappa indicatorsThis table is a symmetrical matrix (with no diagonal) used to evaluate the level of consistency of judgments between operators, or between operators and the expert.
⇒ Matching by operator This is assessed by taking into account the total number of measurements carried out by each operator (the number of parts multiplied by the number of repetitions). In other words, we analyze the proportion of times the operator has obtained results identical to the reference, without necessarily checking whether the operator was repeatable or reproducible in relation to the expert.
⇒ Efficiency Repeatability: is assessed by examining the operators' overall agreement with the expert, focusing solely on the number of parts. In other words, we analyze the proportion of times the operator has demonstrated both repeatability and reproducibility in relation to the reference, in relation to the total number of parts examined.
⇒ R&R Three indicators are calculated in this table.
Global concordance
Global concordance, also known as overall precisionmeasures the percentage of correct decisions made by all operators. In other words, it assesses the extent to which all operators agree with each other, and whether they arrive at a consistent conclusion in relation to reality or the given references.
- If each operator is able to classify the samples identically on each attempt, and their assessments agree with the correct reference, overall agreement will be high.
- This measure provides an overview of the measurement system's ability to deliver accurate and consistent decisions.
R&R between operators
The R&R between operators or reproducibility measures the extent to which operators differ from each other when measuring the same attribute, for the same sample. The idea is to see whether operator A evaluates in the same way as operator B for a given sample.
- If operators disagree significantly with each other, this suggests that the measurement system has a reproducibility problem.
- A good R&R score between operators indicates that all operators perform measurements in a similar and consistent way.
R&R vs. the reference
The R&R vs. reference evaluates the concordance between operators' decisions and a predefined reference or standard. This reference is often considered to be a "true" or "ideal" measure. Here, the aim is to measure whether operators' assessments correspond to this reference.
- If operators give answers that deviate from the reference, this reveals a bias or a lack of accuracy in the measuring system.
- A good R&R vs. reference score means that operators are able to evaluate attributes in line with expected standards.
2.2 Graphics
In the Chart tab, you'll find pie charts of concordance analyses by operator:
The video below shows how to perform an attribute R&R pledge analysis.