I have established a global model, estimated median c-hat, run candidate models, and evaluated the top-ranked model for my full population (adults and juveniles) in the CJS framework. As a next step, I am working on Pradel and POPAN models for only a subset of my data (just the adults). At this point, it is okay to:

1) Proceed with using my top-ranked Phi and p structure and original c-hat while constructing/comparing JS models for the "adults only" data,

Or should I...

2) Run a new CJS "global" model, estimate a new c-hat, and establish a new, best model structure for Phi and p for my "adults only" data subset?

Thank you!

-B