I am working on some dataset simulations using POPAN package in MARK. For example I simulated 1000 datasets with different models and then used the same models to estimate abundance from the simulated datasets.
I had no problem in simulating datasets with all POPAN models, except for model phi(t)p(t)pent(t). I tried many different combinations of parameters (at the end of this message there is an example) but I always obtained estimated abundances of 0 or very big numbers like 6013525 (while the true abundance value was 700).
I have been studying all the theory behind the Schwarz and Arnason (1996, 2006) models but I was not able to determine what is the problem. I suspect it is due to the fact that in the fully time dependent model some parameters are not identifiable but I am not able to go further.
My questions are: does anybody know why the dataset simulations of model phi(t)p(t)pent(t) do not work? Is there any way I could set up the parameters in a way that will make the simulations work?
EXAMPLE OF SIMULATION SETTINGS
Capture occasions = 6
Data type = POPAN
True model = phi(t)p(t)pent(t)
Link function = Identity
Beta values
phi1 = 0.4
phi2 = 0.6
phi3 = 0.5
phi4 = 0.4
phi5 = 0.5
p1 = 0.4
p2 = 0.5
p3 = 0.4
p4 = 0.3
p5 = 0.4
p6 = 0.4
pent 1-5 = 0.16 (I decided to setting this up as 1/6, since the pent parameters have to sum up to 1)
N = 700
Thank you in advance to every person that will take the time to read this message!
Nina