Known Mortality of Tagged Fish...

[abreton]

Left a confusing typo in there - replace suggestion (3) with: when a biologically plausible value is available to fix a parameter that is normally estimable given the model but, due to data sparsity or some other issue with the data, it cannot be estimated with your dataset.

[Fish_Boy]

Andre,

I wil try a different the link function and see what happens. You are correct about constructive (in the design sense) criticism. Like many other people I am always asked to produce Nt, whether or not the original study was designed to do so. I have retuned the sampling protocols over the last couple years to better fit the questions (i.e. the estimation of Nt).

I fixed pi based on the fact the other two reproductive populations being assessed had essentially identical pi values.

It would be easier if real data followed the modeled scenarios. In all scenarios a 'random' movement model is always much better than a "Markovian' movement model.

There are/will be a lot of issues in the future due to the longevity of lake sturgeon. Somdays I wish that I was working on a species with a shorter life span at least I could begin to address some age and cohort questions on a reasonable time scale. Some of these fish will be dying when I am retiring LOL!

At any rate, I will post some results of the different link functions regardin gthe parameters.

Thanks,

Patrick

[Fish_Boy]

OK...

Changing the link function does allow pi to be estimated, however, the CI's are either 0-1 or there is no SE and therefore CI's = parameter estimate.

Another issue is that the AIC ranks certain link function models low, but others share top spot e.g. sin low while logit is very low while loglog, Cloglog, log, and identity rank the same, but each have there own issues with the CI's and parameter.

I will keep at this... to be continued