Questions Test3.Sm, Test2.CT

questions concerning analysis/theory using programs M-SURGE, E-SURGE and U-CARE

Questions Test3.Sm, Test2.CT

Postby TimvdS89 » Mon Oct 22, 2012 11:37 am

Hi there,

I'm currently busy with data on a dolphin species, which I have broken up into 6 ageclasses. Regarding my first ageclass, I have the following 9 encounter histories:

0 0 1 0
0 1 0 0
0 1 1 1
0 1 1 1
1 1 1 0
1 1 1 0
1 1 1 0
1 1 1 1
1 1 1 1

For Test3.Sr I get decent results, however, when running Test3.Sm, I'm presented with the following:
Image
I don't understand this - is there no df and chi-square value given due to the low number of encounter histories? And when reading these numbers, what power does my p-level = 1.000 have? The strange thing is, that with other ageclasses (and more encounter histories), my p-value is álways 0.80125. But when I increase the number of occasions (by adding years for example), the p-value for Test3.Sm differs more.

Regarding Test2.CT, I don't even get an output:
Image

I am at a loss here, do I not get an output due to too sparse data? Because I do get results for Test3.Sr.
Any help would be greatly appreciated!! As you can probably tell, I'm really new with all this stuff, and I'm eager to learn!

Kind regards!
TimvdS89
 
Posts: 20
Joined: Mon Aug 13, 2012 5:20 am

Re: Questions Test3.Sm, Test2.CT

Postby CHOQUET » Tue Oct 23, 2012 5:00 am

Hello,

As you want to test for the adequacy of the CJS model, you don't have to cut your data set.
As a consequence, you will obtain more power for testing the CJS model
on the full data set.

Rémi
CHOQUET
 
Posts: 211
Joined: Thu Nov 24, 2005 4:58 am
Location: CEFE, Montpellier, FRANCE.

Re: Questions Test3.Sm, Test2.CT

Postby TimvdS89 » Tue Oct 23, 2012 5:47 am

Hi Rémi,

Thanks for the response! By not cutting my data, do you mean I do not have to test seperately for the different ageclasses or that I need not divide my dataset into ageclasses? Because I tried running U-CARE on all groups simultaneously, and I get the same values. But I will try running U-CARE on my data without splitting it up into ageclasses.

Thanks!
TimvdS89
 
Posts: 20
Joined: Mon Aug 13, 2012 5:20 am

Re: Questions Test3.Sm, Test2.CT

Postby CHOQUET » Tue Oct 23, 2012 10:00 am

Yes, to group data except young individuals which have low survival (lower than adult).
CHOQUET
 
Posts: 211
Joined: Thu Nov 24, 2005 4:58 am
Location: CEFE, Montpellier, FRANCE.

Re: Questions Test3.Sm, Test2.CT

Postby TimvdS89 » Thu Oct 25, 2012 9:47 am

Hi Rémi,

Thanks again for the reply. I tried pooling the data that way, but no luck. I think the main problem was the low number of encounter histories or perhaps the low number of occasions (as said, increasing the number of occasions did result in the p-value of Test3.Sm differing more). Hence what I tried afterwards was to double the amount of occasions, and pooling calves with juveniles (group 1 + 2). I got some good results, the only significant results I found were:

Test2.CT group 3: p-value = 1.719*10-5 (c-hat = 5,93446)
Test2.CT group 4: p-value = 0.0053846 (c-hat = 3,673)
Test2.CT group 5: p-value = 0.0763 (c-hat = 2,74004)

So, please correct me if I state this incorrectly, but this means that for these groups (which are the "adult" groups of this population), there is a significant difference in when they were seen and whether or not they were seen before. So now my problem lies with Test2.CT, and not Test3.Sm (though I do get 1 significant p-value for group 5, but this is explained by 1 specific occasion). So now I need to try and understand why I get significant results for only these groups for this test. Any suggestions on where the 'problem' could be? Too low resighting rate?

Also, one rather stupid question I assume: I want to analyse this data using the POPAN model, but I believe I read somewhere that U-CARe is mainly used to test CJS model, and that there is currently no optimal GOF test for POPAN. However, POPAN is, afaik, a parameterization of CJS models. My question is therefore, can I, or can I not use U-CARE to test which general model is the best one to start of with?

Again, many thanks in advance!
TimvdS89
 
Posts: 20
Joined: Mon Aug 13, 2012 5:20 am

Re: Questions Test3.Sm, Test2.CT

Postby CHOQUET » Wed Oct 31, 2012 10:19 am

Test2.CT is in general significant (This is the case for groups 3 and 4) because of a trap-effect.
For the Dolphin, is there a possibility that the capture be influenced by the previous trap ?.

U-CARE only test for the the CJS model, if some test are significant then you have to modify
the model (it should be more general) to take into account the overdispersion.
In our case, you should used a model with trap-effects.

Rémi
CHOQUET
 
Posts: 211
Joined: Thu Nov 24, 2005 4:58 am
Location: CEFE, Montpellier, FRANCE.


Return to analysis help

Who is online

Users browsing this forum: No registered users and 18 guests