model selection philosophy

[ant]

Dear folks,

this discussion is particularly interesting for me. So I'll put my situation in the table for my personal interest and to put another more or less concrete situation.

I've 13 primary occasions with 5 trap-night for small mammals, with 2 months interval. Well, the point is that I'm interested both in (re)capture probabilities and survival and migration parameters and N, so I'm trying to use the robust design. The problem is that I think that 'p's and 'c's don't behavior equally in the primary's closed occasion's (in one occasion they could be dots, in the other p(t)=c(t), or vary between sexes, I'd put juveniles in another group, but my N is not that great). So it would be exhaustive to put variations of p and c (worst: with different combinations between the trap occasions).

Then, in my narrow mind (when Mark is the subject), I thought of doing closed models for deal with p, c and N, for each primary occasion. Then I would do the robust designs (fixing p and c - first I was thinking in fixing them in the same manner, but now maybe fixing each occasion according with the results of the closed model) to deal with survival and gamma's.

Ok, I agree that the better way it's the one that Evan is advocating. But in my case I think I'll treat the two situations separately. So I ask:

A) the estimation of p and c would be more accurate in the robust because there is the influence of S and Gamma's?

B) the estimation of N would be better in the simple closed model or in the robust?

C) Concerning just S and gamma's: In theory, by fixing p and c the better model could be different if I didn't fix them?

D) If it's ok to do both the closed and the robust fixing the p and c in the robust, there'll be any difference in how I'll fix them (all fixed as constant, pe. or fixed according the result of the closed model)?

I hope I've made contributions in the discussion and that you help me to elucidate the better path for me,

Regards,
Antonio

PS: pardon me if there are too many errors in my writing, which is a bit rusty in the moment!