We received on OFI during our TS-16949 surveillance audit about Attribute gage R&R sample size. The software we currently use to perform our R&R only allows 20 parts to be used for this study.
Our auditor has told us that it an MSA 3rd Edition requirement that we do three appraisers - three tests - fifty parts.
I have looked for this requirement in MSA but the best I can find is the "example" shown on page 127 and detailed in the scenario beginning on page 125.

Is this 3X3X50 an actual requirement? If so, where is it detailed?

I didn't find any requirement for a 3X3X50 in the 3rd MSA manual .

On page 75 it is stated that " Although the number of appraisers, trials and parts may be varied ......"

Look also at note 59 on page 129 ( necessity of large number of parts )

No firm requirement found.

:rolleyes:
Indeed no "MUST" for 50 samples requirement. That's why you have this OFI not a NC.

We received on OFI during our TS-16949 surveillance audit about Attribute gage R&R sample size. The software we currently use to perform our R&R only allows 20 parts to be used for this study.
Our auditor has told us that it an MSA 3rd Edition requirement that we do three appraisers - three tests - fifty parts.
I have looked for this requirement in MSA but the best I can find is the "example" shown on page 127 and detailed in the scenario beginning on page 125.

Is this 3X3X50 an actual requirement? If so, where is it detailed?

In order for this to be an actual OFI, the auditor would have to demonstrate that the suggested method is more efficacious than the present method. My guess is that the auditor can't do this, so whether or not you implement the "opportunity" is your call.

IMO, there no 'requirement' to use 50 parts as such. Samples must be from all 3 zones :
1. Clearly bad samples (Below LSL & Above USL): Probability of acceptance=0
2. Clearly good samples / within limits: Probability of acceptance=1
AND most importantly,
3. From the Grey Zone.

This document (http://www.allison.com/support/supplier_qa/Attribute_MSA_Training.pdf) from Rolls-Royce Supplier QA recommends min. 20 and upto 30 parts.

Also see the thread: TS 16949 - AIAG's MSA Manual - Thread Inspection Gage - Attribute Gage R&R (http://elsmar.com/Forums/showthread.php?t=8119)

We have just had our preassessment and were told by the auditor that the sample size should be 50. This is the interperation they have been given. If we can agree a smaller number with our customer then that would be acceptable.

We have just had our preassessment and were told by the auditor that the sample size should be 50. This is the interperation they have been given. If we can agree a smaller number with our customer then that would be acceptable.

It sounds like you are getting caught by the OEM's sample size mandate versus the actual statistically valid sample size. This happened before when the automotive OEMs forced the sample size for a capability study from 30 to 50 to 100.

If you use KAPPA analysis for your attribute study, a statistically valid sample size is 20 to 50. However, if the OEMs mandate 50 , you have to measure 50.

For help on KAPPA analysis, see:
http://www.statsdirect.com/help/chi_square_tests/kappa.htm

and

http://elsmar.com/Forums/showthread.php?t=13188&page=1&pp=10&highlight=KAPPA

According to what I understand in the Ford Specific requirements 4.35 it states that "Attribute gauge studies should utilize 50 parts, 3 operators, 3 trials." This is the form on page 127 of the MSA 3rd edtition. If you are a supplier to Ford would you not have to meet this requirement?
:confused:

If we were a supplier to Ford...

Is not necessary use 30 samples, 3 assessors adn 3 trials, but it is when we have to recieve more accurate rezults.

Maintain the following minimums:
- 30 samples (S)
- 2 assessors (A) - minimum to be able to assess reproducibility
- 2 trials (T) - minimum to be able to assess repeatability
- S x A x T > 90

My english is not good but I want to help :D