Can I be confident of seemingly reasonable phi estimates when recapture p estimates are obviously poor?
I have a small mark-recapture dataset (4 years, 12 sites) with anywhere from 10 to 80 individuals captured (flying squirrels, chipmunks) at a site. When I run open capture models for a site, even the simplest phi(.)p(.) model, I get reasonable phi estimates (CV ~20%) but the recapture p estimates are wild (some 0, some 1, some with very high SE) even for sites with >50 indivdiuals captured. This occured even when GOF tests indicate the fully parameterized model fits. Can I use the phi values with confidence?