[R-sig-ME] Contrasts with nlme
bates at stat.wisc.edu
Wed Nov 11 18:52:51 CET 2009
I think you will need to check with the author of the contrast package
regarding this error. It is entirely possible that the methods in the
package do not accept lme models.
On Tue, Nov 10, 2009 at 2:11 PM, Jacques Ropers <jropers at gmail.com> wrote:
> Dear All,
> I'm modelling some longitudinal data (1 continuous
> response variable measured at 6 times (covariate time taken as factor), 1
> continuous baseline, one "treatment" variable also taken as a factor)
> collected in the same subjects using the following model:
> model.lme<- lme(response ~ V0+ time + tt + tt:time, random =
> ~1|subject,correlation = corSymm(form = ~ 1 | subject), na.action=na.omit,
> Then I would lke to estimate the effect of "treatment" at time = 6 using the
> package "contrast"
> contrast(model.lme ,
> a = list(tt="1",time="6"),
> b = list(tt="0",time="6")
> But I get the following error message
> "Erreur dans gendata.default(fit = list(modelStruct = list(reStruct = list(:
> not enough factors"
> I'm obviously doing something wrong.
> Thanks for your help.
> R-sig-mixed-models at r-project.org mailing list
More information about the R-sig-mixed-models