Design matrix and link function

questions concerning analysis/theory using program MARK

Design matrix and link function

Postby wijw » Wed Feb 10, 2021 2:05 pm

This question is about CJS models and link functions.

I have been reading chapters 4 and 6 about link functions and design matrices and it is my understanding that with the identity design matrix, you usually use the identity link function or log link function. I also read that in this particular situation the parameters will not be constrained to [0,1].

My question is, is there anything stopping me from using a sin or logit link function with an identity design matrix? (I want the parameters to be constrained to [0,1]). Is it because the parameters may not be esteemable close to either 0 or 1 boundaries?

I ran the same CJS model using sin, logit, and identity link functions and got the same results except for the last phi and p parameters at the end.
wijw
 
Posts: 23
Joined: Mon Oct 26, 2020 2:27 pm

Re: Design matrix and link function

Postby jlaake » Wed Feb 10, 2021 2:12 pm

Not sure where you read that or maybe you misread. The default for probability parameters with an identity design matrix is the sin link. If you fit phi(t)p(t) the last parameters are confounded and not separately estimable.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Re: Design matrix and link function

Postby wijw » Wed Feb 10, 2021 4:44 pm

You are correct. I misread a paragraph in the book. Please feel free to take down this post. :oops:
wijw
 
Posts: 23
Joined: Mon Oct 26, 2020 2:27 pm

Re: Design matrix and link function

Postby egc » Wed Feb 10, 2021 4:52 pm

Nah -- all posts (even the embrassing ones) are left online (otherwise Cooch and Laake would have at least 50% of their posts deleted ;-)
egc
Site Admin
 
Posts: 201
Joined: Thu May 15, 2003 3:25 pm


Return to analysis help

Who is online

Users browsing this forum: Google [Bot] and 15 guests

cron