topic Re: Compare 2 measuring equipment in Discussions

Compare 2 measuring equipment

ThoYala — Mon, 24 Dec 2018 04:10:07 GMT

I have 2 measuring equipment. We would like to make sure that same sample from 2 measuring equipment give us same data or not.

I plan to have sample 100 pcs measuring by both equipment and use match pair to analysis whether data from both machine is statistic equal or not.
If it is not, I plan to use main machine as a master. Both machine have good correlation (Rsquar > 0.85), I will use the equation from scatter diagram to conver result from minor machine.

Please advise, my method is good enough or not. If not, what should be a good method.

Re: Compare 2 measuring equipment

ian_jmp — Mon, 24 Dec 2018 10:45:08 GMT

It's good that you can measure the same test piece on both pieces of equipment.

The 'Mastering JMP' series has lots of recorded webinars that you can review. For instance there is one on MSA by @Peter_Bartell. Also, don't forget 'Help > Books > Quality and Process Methods'.

Re: Compare 2 measuring equipment

Mark_Bailey — Mon, 24 Dec 2018 13:20:31 GMT

Just adding to Ian's suggestion...

This question can be aided by a hypothesis test. In this case, the (assumed) null hypothesis H0 is that the results from the two equipment are different and the (proposed) alternative hypothesis H1 is that they are the same. You must define a tolerable or practical difference because the two samples will never give an exact difference of zero. This test is actually based on two one-side t-tests. If the observed difference is simultaneously significant for (1) difference is greater than -tolerance and (2) difference is less than tolerance, you would decide that there is no practical difference.

If you entered the paired data on the same rows in adjacent columns then you can use a third column with a formula to compute the row-wise difference. (Unfortunately, the Matched Pairs platform does not offer a test for equivalence.) Then select Analyze > Distribution. Enter the difference data column in the Y role. Click OK and then click the red triangle next to the difference (not the top most red triangle) and select Test Equivalence.

Please see Help > Books > Basic Analysis then Distributions chapter and Test Equivalence topic for more information and examples.