Bootstrapping GOF

[Zach Peery]

Hi,

I am trying to determine if my data fits the global model (phi(t),p(t),lambda(t)) using Pradel's survival and lambda model. The deviance for this model is about 30 times as large as the mean deviance of 500 simulated datasets, meaning that c-hat is huge and the data does not fit. However, when I estimated the GOF using the same bootstrapping technique for the global model within a simple live-recaptures framework (phi(t),p(t)), c-hat is 1.0 indicating that the data fits fine.

Can anyone explain this discrepancy?

Thanks a bunch!