stshroye wrote:Thank you! That's the kind of explanation I was looking for. I now realize that p from the CJS model with phi constrained to 1 does not even equal c from closed model Mt, let alone Mb or Mtb.
If you use Huggins closed capture model, which, like CJS also uses conditional likelihood, and assuming you have adequate data to model the extra parameter in the Huggins models, the 'p' in the CJS models phi (fixed to 1), p(.) and phi (fixed to 1) p(t) should be equivalent to the 'c' in Huggins models Mb and Mbt.