[R-sig-ME] Fwd: Assigning random effects

[Emmanuel Curis]

Hello,

On Thu, Jan 10, 2013 at 11:55:18AM -0500, Nathan Eustis Rutenbeck
wrote:

« I still am left with the question as to whether to alter the random-effects
« structure for each gas-exchange response variable of interest (Anet, WUE,
« etc.), or whether to develop a random-effects structure that is sensible
« based on experimental design and stick to it? In other words, because of
« experimental design, is there potentially an argument for maintaining the
« more complex random effects structure of fm4 above through all my models,
« even though it does not necessarily provide the best fits in a given model?
« Does it even matter if I am mostly concerned with hypothesis testing of
« fixed-effects?

It may be a very (too much?) pragmatic point of view, but why not test
your fixed effects with both models and see if conclusions are the
same (or, for estimates, close enough for your usage)? If so, and if
really only fixed effects are of interest, may be it does not really
matter which one is used --- but I agree that on an understanding
point-of-view, it may be unsatisfactory.

NB: I'm also somehow a beginner in this field, so do not hesitate to
correct this if there are some traps or misconceptions in this idea...

-- 
                                Emmanuel CURIS
                                emmanuel.curis at univ-paris5.fr

Page WWW: http://emmanuel.curis.online.fr/index.html