    
jrcuesta (Unregistered Guest)
Unregistered guest
| | Posted on Thursday, March 09, 2006 - 3:53 pm: | |
Hi,
Sometimes the reference methods used to develop a calibration are no accurate enougth (they are not the official ones,....), and that brings to have a lot of different samples with the same reference value. This is even more evident if the range of the calibration is low (1 or 2 %). If the accuracy of the reference method used for the calibration is not good enough, with not too many samples we can cover the range of the calibration. As far as more samples are added is more evident how the horizontal lines with the same Lab value appear crossing the slope line. The SEP obtained from validations can be close to the Lab error, but I would like to know how more can be done when managing this kind of situations(take out redundancy, live all the samples,..... ,).
I would apreciate any comment about it.
Best Regards
J. Ram�n Cuesta |