www.phidot.org

Just a simple question.
Is there a way to run random effects models from RMark?
If not yet - is it possible in general?
Thanks!
Eldar

Eldar wrote:Just a simple question.
Is there a way to run random effects models from RMark?
If not yet - is it possible in general?
Thanks!
Eldar

1\ ' is there a way to run random effects models in MARK?' should be the question. RMark doesn't do anything on its own, besdies providing an (arguably) more powerful\extensible 'front-end'.

2\ appendix D.

Thanks, Evan.
I've read appendix D already, and was not able to figure it out from there.
Let me formulate my questions more precise.
1. Which part of mark is doing estimations for random effect models - mark.exe or mark GUI?
2. If mark.exe then is it possible to create mixed effects input file in RMark?
Eldar

The function var.components in RMark implements the approach in Appendix D and the function var.components.reml uses a restricted maximum likelihood approach. See the examples for these. They use the fitted parameters and their v-c matrix from the MARK model and estimate a mixed-effects model for those estimates. The functions do not output all of the stats that MARK does but with some minor changes that could happen.

--jeff

Eldar wrote:Thanks, Evan.
I've read appendix D already, and was not able to figure it out from there.
Let me formulate my questions more precise.
1. Which part of mark is doing estimations for random effect models - mark.exe or mark GUI?

As you might expect, the GUI doesn't do calculations. All numerical calculations are handled by mark.ex

2. If mark.exe then is it possible to create mixed effects input file in RMark?
Eldar

If you mean, can RMark interface with mark.exe, and handle the VC and random effects models as described in Appendi D, then 'yes', I believe. If you mean full blown hierarchical random effects, then no -- you can't do this in MARK (whether or not you use the classical GUI, or RMark). For that, you're looking at BUGS or JAGS or equivalent.

Great!
Thanks a lot!