Adam - Without resorting to the rather rich lmer/lme4 discussion realm, you need to base anova() comparisons of lme models with different fixed effects on maximum liklihood estimates rather tham REML.
anova(update(l2,method="ML"),update(l2,~.-useful:nusience,method="ML")) should avoid the error and give a conservative estimate of the significance of your interaction. see also: http://tolstoy.newcastle.edu.au/R/e2/help/06/10/3565.html and related posts. A --------------------------------------------------------------------------------- Dr. Andrew Beckerman Department of Animal and Plant Sciences, University of Sheffield, Alfred Denny Building, Western Bank, Sheffield S10 2TN, UK ph +44 (0)114 222 0026; fx +44 (0)114 222 0002 http://www.beckslab.staff.shef.ac.uk/ ---------------------------------------------------------------------------------- On 10 Jan 2008, at 22:32, Adam D. I. Kramer wrote: > Hello, > > I've been running some HLMs using the lme function quite happily; it > does what I want and I'm pretty sure I understand it. > > The issue is that I'm currently trying to estimate a model with a > 14-level "nusiance" factor as an independent variable...which makes > the > output quite ugly. All I'm really interested in is the question of > whether > these factor as a whole (and its interactions with other factors) are > significant. > > The summary.aov function provides this sort of aggregation for lm > objects, but does not run on lme objects. I've also tried estimating > the > full model and restricted model, leaving out a main effect or > interaction > term and then using anova.lme to compare the models, but these > models appear > to be being fit differently. Say I have l2, and then > > l3 <- update(l2, .~.-useful:nusience) > anova.lme(l3,l2) > > ...to see whether the interaction term is significant, produces the > error, > "Fitted objects with different fixed effects. REML comparisons are not > meaningful." Upon examination using summary(l3), it seems that the > fixed > factors are indeed different. > > So, my question is this: How do I estimate omnibus main effects for > multi-level factors and multi-level factor interactions in lme models? > > Many thanks, > Adam D. I. Kramer > Ph.D. Student, Social and Personality Psychology > University of Oregon > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
