[R] new to R: question about nlme

Douglas Bates bates at stat.wisc.edu
Thu Aug 16 16:24:59 CEST 2001


Bill Simpson <wsi at gcal.ac.uk> writes:

> > I am a PhD student in sociology, studying in Essen, Germany. I am doing an
> > empirical research with data on three (hierarchical) levels: Individuals
> > within schools within regions. The data are very sparse with a lot of
> > schools containing only one individual. In addition the dependent variable
> > is dichotomous, so that a logistic regression would be appropriate. There
> > are independent variables on all three levels. After studying the related
> > literature it seems to me that a (non-linear) random-effects model might be
> > most appropriate, there being not enough cases to do a more detailed model
> > (that distinguishes within- and between-cluster regression) like a
> > random-coefficients model. I also found out that this might be done with the
> > NLME-package in R. But having no one in my environment who is working on
> > similar models or with the R software in general and having no experience at
> > all in programming, calculating this model with R is more than I can handle
> > on my own. Has anyone got an idea how I could speed up the process? Are
> > there any workshops? Other people working on related topics for cooperation
> > or help with the model specification in R language?
> 
> 1. It's your PhD supervisor's job to help you with the data analysis.
> Get him to help you. If he won't help, you need a new supervisor.
> 
> 2. Get the book and copy the example most similar to yours. You and your
> supervisor can learn this together.
> `Mixed-Effects Models in S and S-PLUS'
> J. C. Pinheiro and D. M. Bates
> Springer. ISBN 0-387-98957-9, 2000.

Mark,

As much as I appreciate Bill's recommendation that you buy our book, I
think you may be disappointed in trying to apply the methods described
there to your problem.  You say that the dependent variable is
dichotomous and a logistic regression would be appropriate.  The
combination of a logistic regression and random-effects (nested random
effects, in your case) is called a Generalized Linear Mixed Model or
GLMM.  The nlme package contains functions for fitting linear
mixed-effects models and nonlinear mixed-effects models but not
generalized linear mixed models.  There is another R package GLMMGibbs
that may be more suitable.

--
Douglas Bates                            bates at stat.wisc.edu
Statistics Department                    608/262-2598
University of Wisconsin - Madison        http://www.stat.wisc.edu/~bates/
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._



More information about the R-help mailing list