Mixed effects

[kmk]

I am trying to model adult survival from ragged telemetry data using the Nest Survival models in MARK. A total of 58 birds were tracked across two field seasons in my study (2010 and 2011). However, 7 of these birds were tracked in both years. I understand that there is a way to account for this in a mixed effects model in SAS, but does anybody know of a way to code a random effect of 'bird id' in MARK?

[abreton]

There are two random effects options implemented in MARK, the first (variance components) will not accommodate a random effect of bird, the second (MCMC) may but I'm not familiar enough with this option to be sure. See appendix D in the MARK book, Variance components and random effects models in MARK, for details.

andre

[bacollier]

Andre is correct on his response regarding MARK unless something has change recently that we missed.

I think that one could use the 'logistic exposure' (e.g., Shaffer) link function which is hard coded in R's family() and you could probably used one of the various GLMM approaches in R to run the analysis similar to the GLM approach presented by Terry:

http://www.npwrc.usgs.gov/resource/bird ... reFamily.R

But, without tinkering around with the code I don't know exactly how it would work off-hand, but I seem to remember a few R threads about 2004-2005 on this topic you could search for.

However, as a matter of practicality, I don't think that having 7 birds between years will be enough for you to look at a RE modeling approach.

Bret