Non-estimable parameters under RD

[Stephen]

I’m trying to estimate N of fish populations under robust design. I have 3 primary samples of 4 days each and around 15% of recapture.

When I perform most of the selected models, S and γ parameters are not estimable, but especially S. The only way to obtain S estimates is to perform models without emigration (all gamma parameters are set to zero).

I tried a lot of things like modifying design matrix, but I don’t have enough parameters to put a constrain on γ'' like in the example of chapter 16 (16.6.2)
Any comments or advices would be appreciate,
Thank you.