[S] split-plot or multilevel models with lme

Douglas Bates (bates@stat.wisc.edu)
23 Apr 1998 12:46:02 -0500


Given that there have been two messages to s-news in the last 24 hours
that have asked about a split-plot analysis and about multilevel
modelling in S-PLUS I will make a pre-announcement of a beta test of
version 3.0 of the nlme library.

This test version of nlme 3.0 will have only the linear mixed-effects
modelling code in it but we hope that will be of interest and of use
by itself. Among other things it handles multilevel models. It will
handle essentially any model that SAS PROC MIXED will. (OK, it
doesn't handle "panel data" with crossed random effects but other than
that ...). The computational methods are new and rather neat. They
are described in
http://franz.stat.wisc.edu/pub/NLME/CompMulti.ps
or
http://franz.stat.wisc.edu/pub/NLME/CompMulti.pdf

Jose Pinheiro and I are both hideously busy with other things right
now but we will try to get a proper announcement out "soon". The
brave (or the foolish, depending on your point of view) can take a
look around
http://franz.stat.wisc.edu/pub/NLME
ftp://ftp.stat.wisc.edu/src/NLME/Beta/

If you decide to take a look at it we recommend you subscribe to
nlme-announce@stat.wisc.edu and also to nlme-help@stat.wisc.edu. Send
a message with the word "subscribe" in the body to
nlme-announce-request@stat.wisc.edu and to
nlme-help-request@stat.wisc.edu. The first list should be very low
traffic as it will just provide announcements of new beta versions
(can you say "bug fixes"?). The second is the recommended list for
beta-testers to ask questions about why things don't work the way they
expect them to.
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news