estimate same value for parameter over 2 time periods

posts related to the RMark library, which may not be of general interest to users of 'classic' MARK

estimate same value for parameter over 2 time periods

Postby sharrod » Thu Aug 24, 2017 5:14 pm

Hi everyone,

I've just switched from MARK to RMark to conduct a forward selection analysis of how weather variables affect avian survival. I have run a few simple models and estimated the parameters in MARK, but since I have 34 covariates, it makes more sense to do the actual forward selection in RMark.

I ran a few basic models (Phi(.)p(.)) in RMark and compared the estimates to those obtained in MARK; so far, so good. However, the best model in MARK was Phi(t)P(2p.), meaning that the sampling period is divided into 3 sections. From 2004-2010, data were collected by previous students and a different PI. In 2011, no data were collected. Starting in 2012, the protocol changed slightly and data were collected by my PI. So I need years 2004-2010 to have the same capture probability, 2011 to have an estimate of 0, and years 2012-2015 to have the same estimate as well. In total, there will be 3 capture probabilities.

I can fix the capture probability of 2011 to 0, but am struggling to figure out how to tell RMark to calculate capture probability estimates over the 2 separate periods. I initially tried creating a dummy variable called Period, in which years 2004-2010 were classified as "first", 2011 "mid", and 2012-2015 "last", and tried the following code:
Code: Select all
p.three=list(formula=~time,fixed=list(p.Period="mid",value=0))


However, I found someone with a similar issue http://www.phidot.org/forum/viewtopic.php?f=21&t=1014, and Jeff pointed out that
The fixed notation only allows age, time, cohort or index specifications. It will not accept other specifications that you add to the design data...
Jeff also suggested using the index approach described in Appendix C to fix these variables; however, since I don't need to fix capture probability of the 2 time periods so much as specify "2004-2010 will have the same capture probability, and 2012-2015 will also have the same capture probability." It was pretty easy to specify these capture probabilities in MARK, and I hope I can do the same in RMark.

If anyone has any ideas, or knows of somewhere in the Appendix that describes how to do this, I'd appreciate it. Thank you for your help!
sharrod
 
Posts: 10
Joined: Thu Aug 24, 2017 12:05 pm

Return to RMark

Who is online

Users browsing this forum: No registered users and 13 guests