Value of K1, K2, K3 in GR&R - I want to measure Repeatability and Reproducibility

[Tabrej]

Hey There,
What should be the value of K1, K2 and K3 if I want to measure Repeatability and Reproducibility for more than 3 trials, more than 3 Appraisers and more than 10 parts?
Thanks in advance.

[Stijloor]

Can someone help with this?

Thank you very much.

Jan.

[AdamP]

Hi - thanks for an interesting question!

I dug around and found that most references do not show how to go beyond 2 trials, 2 raters and 10 parts - but I think these links provide what you need to make your calculations:

First the K constants are determined according to which part of the variation you are assessing and use d2 and d*2 values. These can be found in Duncan's book, Quality Control and Industrial Statistics. Here's a link to the 4th ed. via Amazon:

http://www.amazon.com/Quality-contro.../dp/0256015589

You can possibly find the d2 tables within the current version of the MSA manual (don't have one, so guessing).

I found this excerpt from a SAS manual describing the actual formulas for each K constant:

http://www.okstate.edu/sas/v8/saspdf/qc/chapa.pdf

Scroll down the pdf to pages 21-23 for the K1, K2 and K3 formulas. I think it's encouraging for what you want because it's clear that SAS allows for more than 2 trials, 2 raters and 10 parts.

Hope it helps.

Cheers,

Adam

[Sturmkind]

Yes, the d*2 Table is in 4th edition page 203 of the AIAG MSA.

Another alternative is to download Minitab trial version good for 30 days.

[Bev D]

The K factors are just cookbooked shortcuts to standard formulas for estimating the standard deviation given a range. I don't own any of the AIAG manuals so I don't have the exact formulas in front of me. Miner probably does though.

How about a different approach: why more than 3 trials? I know AIAG asks for 3 repeated readings for 10 parts, but really 3 repeats provides no practical improvement in the estimate of the repeatability standard deviation over 2 repeats. And moving to 4 or 5 repeats doesn't help much either - you are better off increasing the number of parts and using only 2 repeats...

[edubry]

hi you can help me about who calculate de K1 k2 y k3 in the studies r yr

[Bev D]

may ask for a bit more clarity?
are you asking:

• who (which person) calculates the k factors during a study?
• who originally dervied them? (history of the factors)
• the formulas for how the factors are calculated
• or how to do the math involved in the formulas

[edubry]

how to calculate K1 as the example of MSA 4 ed as not meeting the precedimento explicit,
not really use the table as d2.

perform an identical spreadsheet, MSA Wing 4 ed p 118, but I have not heard the truth as it has been determined, the values ​​K1, K2, K 3 are the same, the attached example, spreadsheets, the study is a vernier, a sheet I'm honest the other was that I made myself, thank you,