Gage R&R: Selecting Number of Trials and Appraisers and Parts

G

Geets

Hi,
I am doing a research on paint finish. I thought of using R & R concept in my study. The data I have is both Ordinal ( Rank, Non - Normal ) and measurable data. My Questions are :

1) Can we use R & R for Data which has Ordinal data ?
2) Can the PArts be more than 10 , i.e., 15 , 18 or should we stick to the conventional 10 parts.
3) how is number of trials determined ?

I would appreciate if you could give me a lead in this.


Thanks,
Geets:eek:
 

Miner

Forum Moderator
Leader
Admin
Hi,
I am doing a research on paint finish. I thought of using R & R concept in my study. The data I have is both Ordinal ( Rank, Non - Normal ) and measurable data. My Questions are :

1) Can we use R & R for Data which has Ordinal data ?
2) Can the PArts be more than 10 , i.e., 15 , 18 or should we stick to the conventional 10 parts.
3) how is number of trials determined ?
1) Yes, you can do an MSA on ordinal data. This will not be the standard variable MSA, but rather an attribute MSA. I believe that the Kappa analysis is appropriate for ordinal data. However, my expertise is in variable MSA, not attribute.
2) Yes, you may use more than 10 parts. Most Excel-based MSA files are limited to 10, but software such as Minitab can handle more parts.
3) Trials(replicates) are typically three. There is a tradeoff between replicates, operators and parts. That is, more of one allows fewer of another. You must have a mimimum of 2 replicates and 2 operators. The more of each the greater the confidence in the results.
 
A

anurodh

Hi there,
I have one question on this.

I have around 16 Electronic Testers which are used to measure Electrical signals from a product.
Now i have 3 samples for a particular parameter under study & i have data of these 3 samples on each tester each measured 3 times.

Now these testers are independent from Appraiser effect. i.e. no matter how many appraisers i use, it won't affect my measurement (Being electronic Tester)

now can i use the attached sheet which is a regular MSA Sheet to perform R&R analysis?
If yes, then how will address Appraiser Variation?
Please help
Anurodh
 

Attachments

  • r%26r(1).xls
    29 KB · Views: 239
J

justncredible

3 samples is to small use 10 samples, and I think you had posted in another thread. miner had responded and he is the GR&R expert, maybe some others can explain familys.

As for AV on that sheet you have zero, so it would not effect the outcome.
 
A

anurodh

Thanks for the prompt response.

I'll use 10 samples

THis means that in my case Reproducibility cannot be calculated since appraiser variation is not available?

There are 16 test equipments & i can find only Repeatability on each tester?

But what about variation from one test equipment to another test equipment?
 
J

justncredible

You are doing a study on each gage.

The gage to gage variation is not part of a GR&R, but if you are seeing variation you might need to have the units calibrated back to original specs.
 
A

anurodh

thanks,
Can you provide me any standard sheet which i can use for this study.

Or should i use conventional MSa Sheet as attached

If i should use attached sheet, then i'll have to leave sections for Appraiser B & Appraiser C as blank?

In case i do so, i'll not be able to calculate X bar Difference & hence Appraiser Variation.

Will that be OK.

Also can you provide me standard D2* table

Please guide

THanks
Anurodh
 

Attachments

  • gage_r%2Br(1).xls
    72 KB · Views: 234
J

justncredible

Can you post the data for us?

Enter it in a simple excel sheet, a good format for us is.

op: part #: results:
a: 1: 1.1
a: 1: 1.1
a: 1: 1.1
b: 1: 1.1
b: 1: 1.1
b: 1: 1.1
c: 1: 1.1
c: 1: 1.1
c: 1: 1.1
a: 2: 1.2


Makes it very easy to just paste it in the minitab, so col 1 would be the operator, col 2 is the part number, and col 3 are the results.
 
A

anurodh

U See that what my problem is:

I dont have appraisers

I mean there are just 16 different Test equipments

One Person is performing measurement on all these 16 testers.

Even if appraisers are 2 nothing is going to change as appraiser has nothing to do except put the part in the test equipment & the equipment will measure it.

It seems that i can calculate only repeatability.

Reproducibility will not be applicable in this case.

Please correct me if i'm wrong

Thanks
Anurodh
 
J

justncredible

Your AV or appraiser variation will be zero 0. You do not care about reproducibility, since AV = reproducibility.

You will study the EV or equipment variation, EV is the repeatibility.

you can use the same form I think, but I would increase the sample size.

Sorry I am so busy.......
 
Top Bottom