R help - Jan 2004 - R-analogue to the estimate and lsmeans statements in SAS

Dear all,
In connection with a stats course where we want to use R, I am trying to find
out if there are analogues to the ESTIMATE and LSMEANS ("population"
means) statements in SAS (in proc glm, proc genmod and proc mixed). The students
are agronomists, vets and that sort of people, and most of them have some
familiarity with SAS.

    The design package seems to go some of the way, but the model objects are
not "pure" R (as lm or glm) and for that reason I am a little
reluctant to embark on that train. The gregmisc package also provides some
features, but I can not figure out a simple of producing e.g. lsmeans.

- Are there any packages/features I have overlooked?
- Has anyone written a small note on how to obtain the same things from in SAS
and in R

Thanks in advance
Søren Højsgaard


(Just to show that I am prepared for harsh comments on the lsmeans: A lot of
people would probably argue that its a good thing that lsmeans is not there is
because lsmeans usually is not the "right thing" to compute. I agree
to some extent - but sometimes it is a sensible quantity to calculate, and I do
not want to deprive people the right to calculate lsmeans - I'd much rather
spend time on 1) telling people what it is and 2) discussing with them when its
a relevant quantity to calculate...)
	[[alternative HTML version deleted]]