Pre-defined Models Results vs Example Results- Sage Grouse

questions concerning analysis/theory using program MARK

Pre-defined Models Results vs Example Results- Sage Grouse

Postby db » Thu Feb 18, 2010 9:38 pm

I was running through the sage grouse example for a class I am teaching but when I compared the results in the SGROUSE1.pdf that shows the results table for that example, the AIC scores do not match up with the AIC scores from running the pre-defined models for the same models. In addition, the parameter count does not match up for the S(g*t)r(g*t) model. I wondered whether anyone might have encountered this problem? Is this a glitch with running pre-defined models or are the example results wrong or what? I like to choose the pre-defined models option as a short cut to getting started in the modeling process and then build more complex models from there.
db
 
Posts: 4
Joined: Sat Jun 07, 2008 8:03 pm
Location: University of Tennessee

Re: Pre-defined Models Results vs Example Results- Sage Grouse

Postby cooch » Thu Feb 18, 2010 9:57 pm

db wrote:I was running through the sage grouse example for a class I am teaching but when I compared the results in the SGROUSE1.pdf that shows the results table for that example, the AIC scores do not match up with the AIC scores from running the pre-defined models for the same models. In addition, the parameter count does not match up for the S(g*t)r(g*t) model. I wondered whether anyone might have encountered this problem? Is this a glitch with running pre-defined models or are the example results wrong or what? I like to choose the pre-defined models option as a short cut to getting started in the modeling process and then build more complex models from there.


Short of looking the the file (I'm away at the moment).

1. for pre-defined models, you have a choice as to whether or not the identify DM is used. This in turn determines the default link function used. Which, in turn, often determines the number of parameters which MARK reports.

2. given (1), compare model deviances. If the deviances are the same, then you're fine. If the deviances are the same, but the AIC values differ, then the reason is differences in parameter count. This basic point is mentioned several times in the book

3. don't use pre-defined models in the first place. Especially if you're teaching people how to use MARK. The only reason they're in there is as a convenience for people who know what they're doing. But, in general, avoid them.
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: Bing [Bot] and 2 guests