Message-ID: <4C9F445B.8030502@mail.iastate.edu>
Date: 2010-09-26T13:02:19Z
From: Dixon, Philip M [STAT]
Subject: ci's for fixed effects in lme
In-Reply-To: <mailman.1.1285495201.10757.r-sig-ecology@r-project.org>
Manuel,
The problem is that tempo is declared as both a fixed effect and a
random effect in modelo2. If tempo were a factor, that would be a
disaster because the same effect (differences among groups) would be
modeled two ways simultaneously.
In your model, tempo (fixed effect) defines a regression while tempo
(random effect) defines deviations of tempo group means from that
regression. That should not be a problem (I say should only because I
haven't checked that this model is correctly fit by lme).
The solution is to create a copy of tempo and use one variable to define
the fixed effect and the second to define the random effect:
ipa$tempog <- ipa$tempo
modelo2 = lme(ipa ~ tempo, random = ~1 | tempog/localidad, data=ipa)
Best wishes,
Philip Dixon