Correct Tool for Gage R&R - Assay Dilutions

C

codename47

In a recent validation the robustness testing for a new method failed which indicated either the method or the equipment was not capable producing results consistently. The study was done using two instruments that were of the same brand but different model number. So I was told to figure out why the test failed. I was told to execute a Gage R&R which admittedly I'm not that familar with so I've been reading up on the matter but I'm still not for sure it is the right tool for the job. Basically I want to compare the two pieces of equipment to ensure they are both giving results that are the same or within a slight margin of error. Originally, I started by using Minitab and generating a random ordered test with 2 operators, 2 Parts (instruments), and 8 duplicates. I was going to have one operator perform the assay dilutions and then each operator use that solution to test on each piece of equipment. My boss suggested running a high and low standard instead of just a single sample to get more distinct categories.

After reading more there were comments on the net about using an attribute Gage R&R but then I started thinking maybe a different hypothesis test might be better suited. Any ideas how to compare these instruments and get the best information possible on the ability of these two instruments to measure a known value?
 

Miner

Forum Moderator
Leader
Admin
I recommend that you start by reading my blog on the Comparison of Two Gauges. You can also search for posts by Bev D on this subject.

I agree with your boss. You do want a distinct spread. In fact, I recommend at least three distinct levels. Stick with continuous data. The information content of continuous data is much greater than attribute. I do not recommend an attribute study.

You mentioned the new method was inconsistent. Consider performing a Stability study.
 
C

codename47

Thanks for the quick reply. The method is being questioned but after speaking with R&D they are recommending everyone upgrade their UV/Vis instruments b/c they believe the current instruments are not capable of being robust enough to run the assay. In my long past laboratory days an assay is developed and run on a piece of equipment and not the other way around. Either way it should be an interesting study. I'll read up on your blog and try running three different test values, high, low, and maybe the actual assay which should split the difference.

Thanks again.
 
Top Bottom