Performing a GOF test with sparse data

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

Performing a GOF test with sparse data

Postby JBLecomte » Wed Sep 26, 2018 10:20 am

Hi all,

I am quite a beginner in analyzing capture-mark-recapture data with E-Surge and U-care.
I build a multi-state CR model in E-Surge to study movement of a fish population between five areas. One of the particularities of the dataset is that a fish history only consists of one or two observations: at tagging and at recapture. In fact, recapture corresponds to dead fish caught by the fisheries, and fish cannot be observed alive between tagging and recapture.

A fish history is like:
0002000005 or 0000000100 or 05000300000 for a detection in areas 1 to 5.

There is a lot of fish that are not recaptured, making the dataset quite sparse.

I wanted to perform a GOF test as it is usually done in the literature. I used U-care and its GOF for multistate model, and it returns statistics equal to 0 or NaN.

So I decided to simplify my GOF test by using a single state model instead. When I do so, I have results for the and 2.ct tests, but I am not able to compute the two tests 3.SR and 3.SM: U-care only returns zeroes.

I also try to remove the temporal effect by starting all fish histories with the tagging event without success.

I don’t think I have transient individuals in my dataset due to the species ecology and sampling design of the study, but I would like to test for an effect of newly marked versus previously marked individuals. Do you have any suggestions, which could help me in performing a GOF test ?

Many thanks,

Posts: 2
Joined: Tue Sep 25, 2018 7:08 am

Re: Performing a GOF test with sparse data

Postby CHOQUET » Thu Sep 27, 2018 5:45 am

Hello Jean-Baptiste,
your data is a mixture of capture and recovery (found dead) so there is some particular models
and GOF that can handle these data (coded 1 for tagging, 2 for recovery).
First you should say yes when U-CARE ask you if this such a mixture.
For the model, there is some huge litterature and easy to find about how to handle such a mixture.
Posts: 210
Joined: Thu Nov 24, 2005 4:58 am
Location: CEFE, Montpellier, FRANCE.

Re: Performing a GOF test with sparse data

Postby JBLecomte » Fri Sep 28, 2018 11:29 am

Thanks for your answer, i will check that out.
Posts: 2
Joined: Tue Sep 25, 2018 7:08 am

Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest