The Analytical Method for attribute msa on page 135 of MSA manual 3rd edition is the preferred method per AIAG. Other than the MSA manual, I have found very little information on this method. Are there many other people out there using this? Does anyone have a spreadsheet for the analytical method that they would be willing to share?
For years we have been using the 2 person x 20 part x2 times short version. This may have had some shortcomings but is was quick and straightforward. It was not unreasonable to expect personnel at various plants to be able to implement it. With the analytical method, I fear the complexity will overwhelm many sites that we have.

For years we have been using the 2 person x 20 part x2 times short version. This may have had some shortcomings but is was quick and straightforward. It was not unreasonable to expect personnel at various plants to be able to implement it. With the analytical method, I fear the complexity will overwhelm many sites that we have.IMO you may continue using this if it is still acceptable to your customers.

IMO you may continue using this if it is still acceptable to your customers.

I couldn't agree with you more!:bigwave:

The Analytical Method for attribute msa on page 135 of MSA manual 3rd edition is the preferred method per AIAG. Other than the MSA manual, I have found very little information on this method. Are there many other people out there using this? Does anyone have a spreadsheet for the analytical method that they would be willing to share?
For years we have been using the 2 person x 20 part x2 times short version. This may have had some shortcomings but is was quick and straightforward. It was not unreasonable to expect personnel at various plants to be able to implement it. With the analytical method, I fear the complexity will overwhelm many sites that we have.

As long as your customer approves the way you work you are safe.

But, if you deliver to e.g. Ford you have the following guidelines from their customer specific requirements :

Quote:
Variable gauge studies should utilize 10 parts, 3 operators and 3 trials. Attribute gauge studies should utilize 50 parts, 3 operators, 3 trials. Effective attribute gauge study samples include parts within specification and parts outside specification for each criterion being measured and within the expected range of manufacturing variability.
Unquote

Best regards,

Antoine

But, if you deliver to e.g. Ford you have the following guidelines from their customer specific requirements :

Quote:
Variable gauge studies should utilize 10 parts, 3 operators and 3 trials. Attribute gauge studies should utilize 50 parts, 3 operators, 3 trials. Effective attribute gauge study samples include parts within specification and parts outside specification for each criterion being measured and within the expected range of manufacturing variability.
Unquote


Does that mean Ford considers a "should" as a "shall"? :biglaugh: :bonk:

Unfortunately, my customer (GM) has not been too responsive to my request for approval of the short form. It seems that the people who make the decisions in supplier quality do not know enough about MSA to make an informed decision so I have been passed onto to some else and have not gotten any further. If anyone has Ford, Chrysler or GM as a customer, what attribute study have you been using? Thanks.

Quote:
Attribute gauge studies should utilize 50 parts, 3 operators, 3 trials. Effective

hi all,

This is the first time i'm joing in this forum... is there any method that using less than 50 parts..? i saw a presentation slide by Rolls-Royce.. they using a method that i'm not familiar with but acceptable to TS16949 standard.. the minimum recommendation is 20 parts.. i'm not sure whether this is the same method as cbeatty is using.. can someone comment on this..

TQ

Any takers on answering this question?

Quote:
Attribute gauge studies should utilize 50 parts, 3 operators, 3 trials. Effective

hi all,

This is the first time i'm joing in this forum... is there any method that using less than 50 parts..? i saw a presentation slide by Rolls-Royce.. they using a method that i'm not familiar with but acceptable to TS16949 standard.. the minimum recommendation is 20 parts.. i'm not sure whether this is the same method as cbeatty is using.. can someone comment on this..

TQWelcome to posting at the Cove, amirodemachio !:bigwave:

The Rolls Royce training document you are referring, says select up to 30 parts with minimum 20 recommended. They are calculating the effectiveness scores and not the kappa scores. Please read the attribute kappa crosstab method described in the MSA manual - in addition to kappa statistic, it does show effectiveness calculations as well as false alarm and miss rates.

Also See previous discussions on this topic::read:

Attribute Gage R&R - Number of Samples - OFI during our TS-16949 surveillance audit (http://elsmar.com/Forums/showthread.php?t=11981)

and

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

Hope this helps...If not, just ask and we would try to answer any specific query.

Unfortunately, my customer (GM) has not been too responsive to my request for approval of the short form. It seems that the people who make the decisions in supplier quality do not know enough about MSA to make an informed decision so I have been passed onto to some else and have not gotten any further. If anyone has Ford, Chrysler or GM as a customer, what attribute study have you been using? Thanks.

Cheatty - Welcome to the Cove :bigwave:

We are Tier 1, 2, and 3 to GM and we still use the 20 parts, 2 operators, 2 trials method with the acceptance being that all 4 measurements for each part agree. Haven't had a problem so far :cfingers: .

Hello, I am a new user. According to what I am reading, per MSA ver 3 the analytical method is the only method approved without customer consent. (page 125, first couple of paragraphs.

My question is.....what are the acceptance limits?

Sorry, I have to ask experts here some repeated questions, It seems to me a little bit difficult to me. And my questions are:
1. how do you get 93% for 95%UCL,71% for 95% LCL? What's the formula?
2.how do you get Miss Rate, False Alarm Rate for A,B,C respectively?

Any response would be greatly appreciated.:o

Hello, I am a new user. According to what I am reading, per MSA ver 3 the analytical method is the only method approved without customer consent. (page 125, first couple of paragraphs.

My question is.....what are the acceptance limits?

Welcome to the cove.

Acceptance criteria are explained in the manual.

- Look at page 129 for kappa limits
( > 0.75 is considered good and <0.40 is considered bad )

- Look at page 132 for limits for Effectiveness, False rate and Missed parts.

Hope this helps,

If not, please come again.

Best regards,

Antoine

Sorry, I have to ask experts here some repeated questions, It seems to me a little bit difficult to me. And my questions are:
1. how do you get 93% for 95%UCL,71% for 95% LCL? What's the formula?
2.how do you get Miss Rate, False Alarm Rate for A,B,C respectively?

Any response would be greatly appreciated.:o

Please check in attached file.

Best regards,

Antoine

There seems to be some confusion between the Analytic Method (AIAG pg 135) and the Crosstab Method (AIAG pg 128). They are not the same. The Crosstab Method requires customer consent. The Analytic Method does not require customer consent. The Crosstab method calculates kappa. The Analytic method calculates bias using Gage Performance Curve. I have been unsuccessful in finding a spreadsheet for the Analytic Method so I started my own. If anyone has something better, please post it. Thanks.

Hi Miner

I think you can use less number of pieces. But, remember nothing is free.
The bigger your sample is. Better the information you got form your measurement system.

If for some reason you can not get as many pieces as is recommended, you may have no choice to take a lower number with the understanding that the conclusions made out of this data are less reliable.
Let's suppose that You need to make a study to test your 3 inspectors one per shift, because someone told you there are some differences and disagreement between them.
Let's suppose the following scenarios:

You select 20 pieces for the MSA study, and after being measured in your lab the results are as follows:

1st
18 pieces are acceptable at nominal
2 pieces are rejected beyond limits
Now you run a attribute study using a "go" "no go" gage, running three trials/

2nd
You select 50 pieces with following characteristics after been analyzed by you lab:

35 pieces are at nominal size
4 pieces are acceptable but at border line within limits
6 pieces are rejected and border line out of limits and
5 pieces are rejected beyond any doubt.
And you run the study same conditions as above

Now guess what study will reflect better the possible differences between inspectors?

Hi everyone,

I´m new in the forum.

I have a question, I need to do by request of my costumer an Attribute Gage R&R. I have a lot of part numbers (different products) running in the same assembly line. I´m wondering if I can mix parts in the Attribute Gage R&R study. For instance, I have 3 power cable assemblies with different lengths (let's say 50", 75" and 90") but with the same connector in all of them. My operation procedure indicates me to do an inspection for the cable location in the connector (go-no go inspection).

Can I mix the different lengths in the study or I have to do a separte Attribute Gage R&R study for each product?

Please, help me with this matter

Hi, Ruido
Welcome to the Cove.
Let me see if I understand it correctly. You like to run an Attribute MSA study using the analytical method.
If you are measuring length, you must have a “GO” “NO GO” gage for each different size.
Is this is the case them you need to run a study for each gage. Remember you are qualifying your measurement system that includes your gage and the inspectors as well.

I hope this helps

Vicman

Ray Black

MSA 3 states that Hypothesis Test Analysis and Signal Detection Theory should only be used with the consent of the customer so use the attribute method shown on page 127 but I have to warn you about something. With the information shown in MSA, you won't be able to do it.

I e-mailed AIAG when this verson came out and they did send me the information needed to complete the study. I really don't know how anyone could figure it out with their original information. The math is easy but how did one get the counts, etc. I, eventually, developed a seminar.

Anyway, less than 0.4 is considered bad and above 0.75 kappa is considered excellent. By deductive reasoning, your company can determine where the "good" should be and also "OK" level.

Here is a suggestion for everyone struggling with this gauge study. Do the study ONCE for families of gauges. Got an attribute gauge for positional tolerances. Do a study on ONE gauge and use that for all positional attribute gauges.

One does NOT have to have 50 samples unless it is a customer specific such as Ford. Some companies use down to 20 parts but make sure that there are some BAD ones in the sample.

Just some info from an guy who has been doing this stuff for quite a while.

Dear David,

I have an excel sheet I use to conduct attribute Gauge RnR. I believe it was created based on MSA edition 2. I have interest in knowing whether you can share with me your presentation, also whether you can review my excel sheet I have. I need to email it to an email address.

Thanks,

Omar Shubailat
shubailat(at)trostel.com