DATS V3 variability in Rub and Buzz tests

I bought a Dayton Audio DATS V3 system a month ago. Mostly so I wouldn't have to build the correct jig for impedance and TS parameter testing in REW...

But I also got it for the Rub and Buzz test capabilities. Their theory on it seems sound - shifts in impedance and resonating frequencies with different input levels indicate voice coil drag. Seems simple enough.

The issue is that if I test one driver and it passes with say 10% Zmax variance, if I immediately test it again I'll get a completely different set of numbers. The next several tests back to back will show Zmax variance - 30%, 40%, then 15%, 50% - all over the place. Same with the F(s) numbers. If I test one driver up to ten times, I'll even get one test that fails.

Makes me question the validity of the test. I would expect the same driver tested back to back to produce similar results.

Anyone else have this issue or concern?
 
It's free space - sunny side up. The back is not vented - these have a screen dome for venting.

I would think if you don't move the speaker and test it repeatedly the position/mounting shouldn't matter with respect to the variability I'm seeing.