[R] Gibbs sampling in GLMMs: Beta testers required

Jonathan Myles mylesj at icrf.icnet.uk
Fri Dec 15 16:55:22 CET 2000

Sort of a warning before I start: This post may be considered to
describe a rather amateurish approach to distributing software
which may annoy  some people, but I sincerely hope it doesn't.

I've been working for some years with David Clayton on a project which
started life as
an S package but has now turned into an R library. It is (now)
called  GLMMGibbs and estimates the parameters of the Generalised
Linear Mixed Model by Gibbs Sampling (using the Bayesian

The basic idea is that one declares a particular factor to be
a random effect with one statement and then performs the Gibbs sampling
with another statement with syntax somewhat like the glm() statement.

I no longer work on the project as part of my job but I wouldn't like
the work to go to waste, so I would like to offer it now for beta-testing
and make a collection of bugs/problems/comments which I could work on, 
ideally over the Christmas period, (although I've probably now left
it a bit late for that) or at weekends.  I
should emphasise that the software is
rather in a beta state which is why I propose to email it to people
who ask me for it rather than put in on CRAN yet. (My new job doesn't seem
to have any webspace where I could put it either)  At present we are currently
recommending that people do


before using the package because of the large amount of C involved,
although GLMMGibbs hasn't actually crashed R for months (-;). We already
know that very small numbers of trials in binomial models seem to cause
numerical errors (which don't crash R)---this is something I'd like to look
into over Christmas, but I'd appreciate knowing about other problems.

At the moment GLMMGibbs has only been produced on UNIX/LINUX machines.

If you would be interested in helping to beta-test GLMMGibbs, please 
email me at

mylesj at icrf.icnet.uk

Jonathan Myles
Imperial Cancer Research Fund

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