Help with the right analysis tool

#1
Good afternoon,

I am trying to compare results from 2 pieces of test equipment to see if their results are statistically different. Problem is the experiment looks like this example;

Test kit 1 Test Kit 2
Sample 1 12.62 11.97
Sample 2 13.33 14.01
Sample 3 10.55 10.87
Sample 4 15.52 15.01
etc

All samples are 100% wool but each sample is from a different breed, type etc. So we are not comparing the means and variances of two true sets of data i.e. 20 replicates of Sample 1 on both bits of kit, so a simple t test will not help.

I am trying to understand if both bits of kit are significantly different or if the variance between results is normal so the kits could be viewed as being 'in calibration'. I suppose it is a bit like a laboratory correlation trial.

Any advice much appreciated.

Thanks,

David
 

Miner

TS Contributor
#2
There are several approaches that you could take. One is the Bland Altman plot. Another is the Youden plot. You can also perform a 1-sample test for equivalency on the differences. I find the Bland Altman or Youden plot to provide more useful information.