Formula parameters

questions concerning analysis/theory using program MARK

Formula parameters

Postby j.harv3y » Wed Mar 06, 2019 10:22 am

Hi,

Apologies for such a simple question, however I have read and reread the mark book and am struggling to understand a basic concept and was wondering if someone could explain please?

Essentially are there any guidelines for which formula parameters should be included in a model?

For example I am running a multi-state open robust design and am interested in looking at the survival ect of the population over time. This population isn't static, so when I am inputting Phi parameters, do I need to include Phi dot (~-1) as I know biologically this isnt true? All examples in the book seem to include this, and when I include in my formulas my it has the lowest AIC value (however biologically this model wouldn't make sense and doesn't answer the question!)

Thank you so much in advance,

Jess
j.harv3y
 
Posts: 45
Joined: Mon Oct 08, 2018 4:45 am

Re: Formula parameters

Postby cooch » Wed Mar 06, 2019 12:38 pm

j.harv3y wrote:For example I am running a multi-state open robust design and am interested in looking at the survival ect of the population over time. This population isn't static, so when I am inputting Phi parameters, do I need to include Phi dot (~-1) as I know biologically this isnt true? All examples in the book seem to include this, and when I include in my formulas my it has the lowest AIC value (however biologically this model wouldn't make sense and doesn't answer the question!)
s


First, MARK doesn't use formulas in the manner you've indicated -- RMark does. So, if the question at some stage becomes formula-specifc, you should post in the RMark sub-forum.

Second, a 'dot' model for a non-detection parameter (like survival, or movement) may not be realistic biologically, but it is often included in the candidate model set either (i) as a check on the ability of the data set to support more complex models, or (ii) occasionally, as a 'structural constraint' imposed by the user to make 'other parmaeter' estimable. (translation: my data are lousy in some respect, but I really want time-dependent estimates for parameter A, so I'll see if I can get there from here by assuming/pretending that parameter B is fixed -- i.e., 'dot').

If a 'dot' model is your 'winning model', by some criterion, then it is probably a very strong indication of the former, i.e., that your data aren't sufficient to model anything more complex than just that, a 'dot' model.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University


Return to analysis help

Who is online

Users browsing this forum: No registered users and 7 guests