variance components in 'marked'

[dtempel]

I've run some CJS models in 'marked' for a 20-year study where we had a lot of annual covariates. Most of the covariates were spatial in nature (e.g., habitat covariates within an individual's territory in a given year). My question is whether it's possible to perform a variance components analysis in 'marked' (or an alternate software) to separate spatial process variance from sampling variance? Thanks!

[jlaake]

This forum is not for marked (at least yet). I've not discussed that with Evan and most folks outside of the R world will not know of its existence. The short answer to your question is no. Although you can write your own software to do that or adapt var.components or var.components.reml in RMark to the marked package. The marked package is in its very early stages of development. I put it out there as a venue for development of new models that are not currently in MARK and to explore alternative approaches to capture-recapture model fitting such as Hidden Markov Models and also to allow more general mixed-effects analysis. At least for the time being I highly recommend users stick with MARK unless you have a special circumstance or are a very capable R programmer.

regards --jeff

[jlaake]

PS for the time being, any questions about marked should be directed to me individually via email. When there is sufficient traffic Evan has agreed to create a sub-forum.

dtempel I tried to contact you via email but it was rejected.

regards --jeff