Conditional likelihood VS full likelihood

[Pascal P.]

Hi everyone,

I'm working on density estimation of a black bear population with hair snag genotyping and I wondering which type of likelihood is better to use  (CL= TRUE or CL=FALSE)? I'm not sure about when to use one method rather than the other.

Thank you!

Pascal

[Eric Howe]

Hi Pascal,

The two methods give identical results in most cases. With the full likelihood, density is a parameter in the model, but with the conditional likelihood, density must be estimated as a derived parameter after fitting the model. It's faster to fit models with CL = TRUE. However...

If you want to model spatial variation in density, you need the full likelihood.

If you want to model continuous individual covariates, you need the conditional likelihood, but categorical individual covariates can be usually be handled as groups or sessions in full likelihood models.

Note that you can't compare AICc between models fitted using different likelihoods, so you'll want to make this decision early.

All the best,

Eric

--
You received this message because you are subscribed to the Google Groups "secr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to secrgroup+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.