GOF and robust design - recommendations in 2020?

questions concerning analysis/theory using program MARK

GOF and robust design - recommendations in 2020?

Postby Tana » Tue Jun 16, 2020 1:42 pm

Hello everyone,

I’d like to do a goodness of fit test on a robust design model (Full Likelihood Heterogeneity pi, p, and c). Mark (version 9.0) does not allow to run a bootstrap GOF or the median c-hat on that model. When searching the forum here for GOF and RD, I found the recommendation to collapse the secondary occasions, treat it like a CJS model and run a GOF test on that in Mark (see here: http://www.phidot.org/forum/viewtopic.php?f=1&t=189&hilit=goodness+of+fit+robust+design).

Since that is from 2010, I would like to know if this is still recommended in 2020, or if there is another method either in Mark or RMark in the meantime?
The markbook (Ch. 15) mentions only a Pearson Chi² test in program ORDSURVIV, which I think refers to an MSORD model in an open robust design approach.

Any hint would be appreciated. Thank you very much for an answer.
Tana
 
Posts: 2
Joined: Mon Jun 15, 2020 3:23 pm

Re: GOF and robust design - recommendations in 2020?

Postby ehileman » Thu Jun 25, 2020 5:34 am

Hi Tana,

I don't believe a formal GOF test is available for robust design. In addition to the strategy you suggested, you could also evaluate candidate model sensitivity to changes in c-hat. See section 5.10.1 in the mark book for details. Hope this helps! -Eric
ehileman
 
Posts: 39
Joined: Sat Nov 26, 2011 6:40 pm
Location: USGS, Fort Collins Science Center, Yigo, Guam

Re: GOF and robust design - recommendations in 2020?

Postby Tana » Sun Jun 28, 2020 6:17 am

Hi Eric,

this is something I haven't thought about, but it seems very reasonable to me.
Thank you very much for your answer.

Tana
Tana
 
Posts: 2
Joined: Mon Jun 15, 2020 3:23 pm

Re: GOF and robust design - recommendations in 2020?

Postby ehileman » Tue Jun 30, 2020 5:00 am

You are very welcome! I use that strategy fairly often as it is informative, especially when a formal GOF test isn't available for a particular model type I happen to be using.

Cheers,

Eric
ehileman
 
Posts: 39
Joined: Sat Nov 26, 2011 6:40 pm
Location: USGS, Fort Collins Science Center, Yigo, Guam

Re: GOF and robust design - recommendations in 2020?

Postby Bill Kendall » Thu Jul 02, 2020 8:02 pm

That is a good suggestion. An alternative I've used is to apply a multistate model structure (if you are allowing for temporary emigration, otherwise CJS) to the entire data set, setting S=1 and psi=0 where you have closure (within primary period), and then run the median c-hat option. This is equivalent to the close robust design model, except for estimating abundance. If you were including trap response in p, then under the MS structure you could use the priorcapl() function in the design matrix to identify prior capture effects on p. However, if you have found that the heterogeneity (mixture) model for p is important, this MS alternative won't work for that.
Bill Kendall
 
Posts: 83
Joined: Wed Jun 04, 2003 8:58 am


Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest

cron