GOF with unobservable state in U-Care

[pascaline lg]

Hello,
I am using U care to test GOF of multistate models with three states: alive, unobservable and dead. To run the test I need to fix the probability of capture of the unobservable state to 0, how can I do this in U-care?
Thank you so much for your help.
Pascaline

[cooch]

Hello,
I am using U care to test GOF of multistate models with three states: alive, unobservable and dead. To run the test I need to fix the probability of capture of the unobservable state to 0, how can I do this in U-care?
Thank you so much for your help.
Pascaline

I don't believe you can in U-CARE (Remi will advise otherwise). You can, however, run a median -chat in MARK, using simulated annealing as the optimization method.

[CHOQUET]

Hello,

Evan gived you the good reply, there is currently no GOF for models with unobservable site. However in your case, you can run the single site test and compute an overdispersion coefficient from it.
This overdispersion coefficient will be worse than it should be.

Rémi

[pascaline lg]

Thank you both for your help!
I run the median chat with MARK with several data set structures to compare them, but it took me more than one week to have estimates for one model (I have 35 occasions, and according to the data set 2 to 3 states and 2 to 4 groups). Is there any way to reduce the simulation time?
thanks,
Pascaline