Informational Gage R&R: Swapping operators

#1
Variable Gage R&R question:
I have 3 good operators where each operator has been at the company five or more years and have been making the same part and use calipers to measure a particular feature on the widget.

Each operator has been trained on how to measure the feature.

Question 1
One of the operators cannot pass repeatability and reproducibility (trained several times) - is it allowable to remove that operator from the Gage R&R and replace with another operator? Is there any set of circumstances that allow this? This seems to violate the very nature of Gage R&R's.​
Does the swapping of operators violate anything in the AIAG, ITAF, or ISO 9001?​
Question 2
Same situation but all 3 operators fail the Gage R&R - is it allowable to swap all three good operators for anther three operators?​
Any particular standards that this violates?

Responses from IATF 16949 auditors will be helpful.
 

Miner

Forum Moderator
Staff member
Admin
#2
Both the intent behind an R&R study and the type of ANOVA used to analyze the results assumes that all operators were randomly selected and are representative of all the operators that might use the gage. Therefore, if reproducibility is identified as a problem (as you have indicated), a corrective action, or improvement of the system should be taken and the study repeated to validate the effectiveness of the action. Simply swapping operators under either scenario is not sufficient unless the intent was that none of these operators are allowed to use the gage and make decisions regarding the acceptability of the product. Even then, you would need to ensure that the remaining operators were acceptable.

Doing so violates the intent of the R&R study as well as good practice.
 

Ninja

Looking for Reality
Trusted
#3
As Miner said, the intent is that the test method is repeatable and reproducible REGARDLESS of which operator allowed to perform it is doing so.

Cherry picking operators is like cherry picking the individual data points and keeping only the ones that help you pass, defeating the whole purpose of the GRR. You still end up with a test method that is unreliable to tell good from bad.

Don't swap operators, fix the test method.
 
#4
My $.02. If you're really using a caliper for measurement, you may want to change the device. Getting a good gage R&R with a caliper can be very difficult.
 

optomist1

A Sea of Statistics
Trusted
#5
Both the intent behind an R&R study and the type of ANOVA used to analyze the results assumes that all operators were randomly selected and are representative of all the operators that might use the gage. Therefore, if reproducibility is identified as a problem (as you have indicated), a corrective action, or improvement of the system should be taken and the study repeated to validate the effectiveness of the action. Simply swapping operators under either scenario is not sufficient unless the intent was that none of these operators are allowed to use the gage and make decisions regarding the acceptability of the product. Even then, you would need to ensure that the remaining operators were acceptable.

Doing so violates the intent of the R&R study as well as good practice.
Well said, and in some circles, doing so is akin to "stacking the deck" for purposes of hopefully passing GRR, I have witnessed and stopped the same actions, when suppliers prepare to pass a customer witnessed PDR or Production Demonstration Run...aka Run At Rate
 

Top Bottom