Author |
Message |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 173 Registered: 10-2002
| Posted on Thursday, October 28, 2010 - 10:07 am: | |
Hi Leire, I had another question. How are you preparing the samples for measurement by NIR? It may be desirable to measure the samples with no preparation, so that the measurements are maximally sensitive to the coating of the pellets. Grinding the pellets to a uniform mixture might bury the signals you wish to measure. Best wishes, Dave |
David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 172 Registered: 10-2002
| Posted on Thursday, October 28, 2010 - 10:01 am: | |
Hi Leire, Welcome to the Discussion Group. I think you have a very challenging application here, and fortunately you have an excellent resource in Cordoba to help you, Ana Garido-Varo. Are you studying with her? It is good that you have selected to do qualitative identifications rather than quantitative determinations of the rabbit secretions. You have no choice but to use the reference values provided by your panel of judges. You can only hope that they are well-trained. Perhaps you can request that they be challenged with several samples at least 3 times, so that you can obtain a measure of the accuracy of the reference method. Another thought I had is, perhaps the judges can provide a 3-level estimate of coating, as in 0=none, 1=evidence visible, 2= evidence strongly visible. Then you could test the ability to discriminate 1 from 0 and if that doesn't work, perhaps 2 from 0 would work. You will have to accept that NIR is not the answer for all discriminatory tests. However, Ana has been involved in some tough ones, so it is worth trying. Best regards, Dave |
Leire Ruiz (leire)
New member Username: leire
Post Number: 1 Registered: 10-2010
| Posted on Monday, October 18, 2010 - 2:56 am: | |
I have started a feasibility study of NIRS as a tool for detection of some glandular secretions and pheromones covering rabbit pellets in the wild, but these secretions are too much and too complex for chemometrics, I can't choose a hallmark to analyze, so I use inter-judge agreement as reliability measure of qualitative data, and bibliographic background, as references in discriminant analysis between pellets coated with glandular secretion and pellets without it. But I have been told that these are very poor reference values indeed, and I don't know how could I improve them. Help me, please. Thank you. |
|