Page 1 of 1

Re-running c-hat test for subset of data?

PostPosted: Tue Nov 14, 2017 11:30 am
by bjohn28
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

Re: Re-running c-hat test for subset of data?

PostPosted: Sun Feb 18, 2018 5:53 pm
by Bryan Hamilton
I'm assuming you've also ran goodness of fit tests in CAPTURE?

I would recommend #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?


Whether or not to re-establish best model structure for Phi and p depends on your goals. Are you testing hypotheses or trying to get the "best" parameter estimates?