uncertain how to preceed after ambiguous GOF

[elli]

Hallo all,
I have a GOF problem.
First, I intend to estimate survival and lambda (Pradel-Lambda-Model), further I’m interested in prob. of entry and N (POPAN).
Now the GOF problem:
I used U-Care. The Global Test was n.s. (but indicated a trend for trap happiness). Test3.SR was significant for one group. So I used RELEASE (in MARK) to pin point the occasions, and fond transients effects in 3 occasions where I can explain them. The deviance of residuals has a slight trend, but has a tolerable distribution, thus it seems acceptable. C-hat (obtained manually: Sum Chi²/df) was 0.9409 (GIM Manual advises not to adjust).
Question: given the described setting, do I have to adjust for transience (TSM-Model?), or is it ok when I proceed as intended?
anybody a helpfull sugesstion?

Elli

[Eric Janney]

So, if I understand you correctly, you have 3 occasions for one group where test 3.SR indicates a lack of fit? How many occasions is your study and how many groups do you have? Are the data sufficient for those 3 occasions to conduct the GOF test? Based, on your c-hat estimate It doesn't appear that the overall heterogeneity in your dataset is much to be concerned with. It is not possible to run a TSM model for the temporal symmetry (i.e. Pradel) models. So, if you have some transience or a permanent trap response, but it is only for one group and only on a couple of occasions, there is not much you can do except be aware that the recruitment and lambda estimates for that group on those particular occasions are likely to be biased. Maybe someone with more expertise than myself can chime in with there thoughts?