Estimable parameters within CJS with Co-Variables

questions concerning analysis/theory using program MARK

Estimable parameters within CJS with Co-Variables

Postby Halfyard » Mon Feb 14, 2011 2:42 pm

I’m new to MARK, and while the manual is well written, I am still a touch confused with regard to the estimable parameters of a CJS with individual covariates.

For clarity, I’ll ask the question using a ‘dummy ’ dataset, with 4 encounter occasions, 1 group and 1 covariate.
With a basic CJS (i.e. no CoVar’s), and a fully time dependent model, I understand that would have 5 estimable parameters (Phi 1, Phi 2, p2,p3 and the product of Phi3*p4). The design matrix would have 6 columns.


If I incorporated a simple covariate (i.e. mass), and ran the model Phi (t + mass) p (t), with different INTERCEPTS and a common SLOPE, how many estimable parameters would there be? In the DM, there would be a total of 9 columns, but does the non-estimability of the product of Phi3 * p4 affect BOTH the ‘dummy’ coding column and the CoVar coding column for Phi 3?



I hope this makes sense. Any help at all will be appreciated!
Halfyard
 
Posts: 7
Joined: Thu Feb 10, 2011 1:31 pm

Re: Estimable parameters within CJS with Co-Variables

Postby cooch » Mon Feb 14, 2011 3:16 pm

Halfyard wrote:I’m new to MARK, and while the manual is well written,


Thanks. We try. ;-)

I am still a touch confused with regard to the estimable parameters of a CJS with individual covariates.

For clarity, I’ll ask the question using a ‘dummy ’ dataset, with 4 encounter occasions, 1 group and 1 covariate.
With a basic CJS (i.e. no CoVar’s), and a fully time dependent model, I understand that would have 5 estimable parameters (Phi 1, Phi 2, p2,p3 and the product of Phi3*p4). The design matrix would have 6 columns.


If I incorporated a simple covariate (i.e. mass), and ran the model Phi (t + mass) p (t), with different INTERCEPTS and a common SLOPE, how many estimable parameters would there be? In the DM, there would be a total of 9 columns, but does the non-estimability of the product of Phi3 * p4 affect BOTH the ‘dummy’ coding column and the CoVar coding column for Phi 3?



I hope this makes sense. Any help at all will be appreciated!


Have a look at this thread:

viewtopic.php?f=34&t=1482&p=4194&hilit=individual+covariates+confounding#p4194
cooch
 
Posts: 1654
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Estimable parameters within CJS with Co-Variables

Postby Halfyard » Mon Feb 14, 2011 3:21 pm

Wow - lightening fast!

Much appreciated. I apologize for not being able to root out that previous post.
Halfyard
 
Posts: 7
Joined: Thu Feb 10, 2011 1:31 pm

Re: Estimable parameters within CJS with Co-Variables

Postby cooch » Mon Feb 14, 2011 3:24 pm

Halfyard wrote:Wow - lightening fast!

Much appreciated. I apologize for not being able to root out that previous post.


No worries -- as with all search engines, there is an element of luck' involved in finding what you are looking for. However, kudos on at least attempting a search.

This particular issue is perhaps sufficiently important as to warrant some 'discussion' of it in the book. Next time that chapter gets tweaked.
cooch
 
Posts: 1654
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 2 guests

cron