identifying age class

questions concerning analysis/theory using program MARK

identifying age class

Postby bjimenez » Tue May 07, 2019 6:11 pm

Dear forum members:

I am working with a database in which all individuals were marked as adults. CJS GOF analysis indicates that the best models include 2 - 3 age groups. Every individual in the .INP file has its own code and I would like to learn if there is a way in which MARK can produce an output about which individuals belong to each group? I thank you very much in advance for your consideration.
bjimenez
 
Posts: 7
Joined: Thu Apr 07, 2016 11:09 pm

Re: identifying age class

Postby cooch » Tue May 07, 2019 7:48 pm

bjimenez wrote:Dear forum members:

I am working with a database in which all individuals were marked as adults. CJS GOF analysis indicates that the best models include 2 - 3 age groups.


Exactly how did you do this?

Every individual in the .INP file has its own code and I would like to learn if there is a way in which MARK can produce an output about which individuals belong to each group? I thank you very much in advance for your consideration.


Simply put, no. Taken at face value, the fact that a data set consisting of 'marked as adults only' shows a better 'fit' by some measure if the model has multiple age classes is fairly diagnostic of heterogeneity amongst individuals. You have a couple of options -- finite mixture models, or continuous mu/sigma random effect models (introduced in Chapter 14 in the context of closed population abundance models, but the approaches can be applied to CJS models as well). Some people like to try to interpret the pi parameter from mixture models as indicating something about the proportion of indiviuals in a given group, but this is dangerous, and works only in a few specific types of situations. But, to your specific question - no. There is no simple way to assign a given individual to a particular 'group' (age, heterogeneity group, etc.).
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: identifying age class

Postby bjimenez » Wed Jun 05, 2019 11:32 am

Thank you very much for your response and suggestions, Dr. Cooch. I will work on that.

When I mentioned that every individual in the .INP file "has its own code", I refer to the /* code */ that we assigned for identification of that individual and that I included at the beginning of each row in the .INP file, between /* */ symbols. I was thinking of a way to detect/identify "heterogeneous" individuals in a similar way to the residuals plot.

Again, thank you very much
bjimenez
 
Posts: 7
Joined: Thu Apr 07, 2016 11:09 pm

Re: identifying age class

Postby cooch » Wed Jun 05, 2019 11:46 am

bjimenez wrote: I was thinking of a way to detect/identify "heterogeneous" individuals in a similar way to the residuals plot.


You might be able to explore residuals 'manually' to get some sense as to which individuals are 'deviant' in some respect (its a fairly standard part of the MARK workshop to identify a putative 'pair' of individuals -- a pair likely violating the independence of fate assumptions -- by looking at the residuals). But, to do it categorically, with the intent being to retrospectively assign individuals to a particular class might be tough.

Having said that, there *are* patterns in the residuals that you tend to see if you fit a model with 'incorrect structure' to the data. Working out this is some sort of formal way would be a project in and of itself.
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 9 guests