counting # of parameters with "even flow" movement in a RD

[amandarg]

I'm new to using Program MARK and I'm running a closed robust design model on my data. I've been using no movement, even flow and random movement as the possibilities of temporary emigration for my study species as Markovian really doesn't seem to apply to them. I've noticed that the parameter counts that come out for random movement and even flow tend to be smaller than they should. I think I can figure out the correct parameter count and adjust for random movement but I'm very confused with the even flow model. I set it up using the identity matrix and tend to get some very high standard errors in survival and gamma' and gamma'' as well as estimates of gamma'' practically at zero. I can attach some of my estimates of real parameters if that would be helpful. I'm just not sure where my beta terms are exactly and how to count them in this situation so I can adjust the parameter count accordingly. Thanks so much.

[Eurycea]

It sounds like a similar problem that I've had with my data set, where I would get nonsensical results with SE of 0 or 1. When this happens, MARK doesn't count the number of parameters correctly and so you must do it by hand. There is some guidance in the MARK book on counting the number of parameters, however you may have bigger problems because of this potential "partial convergence failure," where you didn't get proper estimates for some of your parameters. Your data may be too sparse to generate these estimates. Alternatively you can try using the Simulated Annealing option when you run the model, to help it from being stuck on local maxima. Hopefully this helps until someone else with more experience chimes in.

Nathan

[amandarg]

Nathan,
thanks for all your ideas. I definitely know I need to hand count the parameters with the even flow model but I am still unsure exactly how to do that as I cannot find any readings on it. I've only managed to find information on random movement and markovian models. I am running pretty basic models in an effort to avoid convergence issues but I will looking into the simulated annealing option and try it out.

thanks,
Amanda

[Eurycea]

What is the "even flow" model?

[cooch]

What is the "even flow" model?

Even Flow:

1/ annoying song by Pearl Jam that got way too much airplay at one point (IMO)

2/ a RD model that is fully described in Chapter 15 of 'the MARK book' - starting in section 15.3.4, and various places thereafter.