Minitab limit definition from R & R



Hi everyone,
I have recently started to use minitab and since 3-4 weeks I have a problem.
We have a new machine in factory and products.
The problem is we don't know how to qualify machine as we don't know the limits related to our products. We would like to detemine product limits experimentally.

The machine has 3 palettes which carry products.

My boss demanded me to run R & R first, and I have run it 3 palettes, 10 samples, 3 times and I have the values.

Now, client demand me to find limit values of the products from obtained R & R results.
I can simply give them the mean value and 6*standart deviaiton as limits but my boss warned me that will be only valid for the samples that we have tested and there is a risque that we will not be correct in future.

At this point I need your help dear professionnals.

How can I determine limits from my R & R results in minitab and determine which samples are our of limit. As currently the machine has no limits, it can say all samples are good so I don't know which samples are good in this test conditions.

I need a setup of the limits first from R & R values and also determine if some of the samples in R & R study is good / bad according to those experimentally determined limits.

For information, I am totally beginner in this minitab subject and statistical analysis. So, please help me and with specific explications so that I can understand.

Thank you very much.


Statistical Steven

Super Moderator
Gage R&R studies are not statements about the product, but the measurement system. You cannot or should not set product specifications based on the data from a R&R study.


A Sea of Statistics
Super Moderator
Amen, others in industry have attempted the same.....good call, its is more accurate (no pun intended) to use the Measurement System Analysis, in lieu of Gage R&R - Optomist1


Forum Moderator
The previous posters are correct. You need to perform a study that includes the process stream variation (i.e., pallet to pallet) and the longer term variation (e.g., setup to setup, batch to batch, etc.). Very few R&R studies would include all these sources of variation and the sample size of 10 is way too small. You need at least 100.

Since you do not have specifications yet, you can't perform a true capability study, but Minitab does have a "Tolerance Interval" analysis that will return tolerance intervals for a specified percentage of the population. If you specify 100%, it will return intervals that should yield a Cp/Cpk = 1. You will want to increase the tolerance interval to ensure you have better Cp/Cpk results.
Last edited:


Dear all,
Thank you very much for your input.
We have decided to conduct more studies. If the things willl change / improve, I wil llet you know. thank you for your help.
Top Bottom