[R-sig-ME] 'fixed' covariates in nlmer
gentz at ensat.fr
Fri Jan 21 09:51:36 CET 2011
I ended with a similar conclusion than yours. Yet, from a scientific
point of view, the population effects that I would like to add are fixed
It is really unfortunate that this functionality is not implemented into
the nlmeR (lme4) function , precluding the use of the package for my
work. I asked D. Bates but didn't get any answer.
I'm still "surviving" using the nlme package, with all the problems
related to numerical estimation and convergence. It is not perfect and
sometimes frustrating, yet it allows results to be analyzed.
I thus need to play with two packages : lme4 for 'simple' linear mixed
models and nlme for non-linear mixed models.
> Dear Laurent,
> Did you ever get a reply about your message below?
> I am also investigating the same problem.
> My impression is that nlmeR (lme4) is currently intended
> just for the individual fixed effect estimates and that the population
> values you describe would need
> to be added in the random effects.
> So I think that if for some reason you need these in fixed effects
> rather than random then we would need to use nlme.
> Is that also your impression and what you
> ended up doing?
> Many thanks
Prof. L. GENTZBITTEL Phone: +33 (0)5 34 32 38 96
INP - Agro Toulouse Fax : +33 (0)5 34 32 38 89
EcoLab, UMR 5245 CNRS-UPS-INP
18, Chemin de Borde Rouge - Auzeville Tolosane
31326 CASTANET TOLOSAN - FRANCE
E-mail: gentz at ensat.fr
More information about the R-sig-mixed-models