[S] LME question - answer

Laszlo Tothfalusi (l.tothfalusi@utoronto.ca)
Mon, 16 Feb 1998 17:11:44 -0500


Dear Splus users,

Some days ago I asked the following question:

>Is there any way to specify a heterogenous covariance structure of R
using LME ?
>For example, if I would like to analyze growth curves of boys and
girls, the following
>SAS code would do the job:

>Proc mixed data=test;
>class person gender;
>model y = gender age gender*age/s;
>repeated/ type=cs subject = person group = gender;

>Now I get different residual cluster error according to the gender.
>How can I do this type of analysis in Splus ?

I am very grateful for Jose' Pinheiro from Bell Laboratories giving
me the answer which is

fit <- lme(fixed = y ~ gender * age, random = ~age, cluster = ~person,
+ data = test, var.function = "strata", var.covariate = ~gender)

It should be noted , that the strata option is only mentioned at
lme.formula in the help files
but a bit more information can be found at nlme.formula. ( I am using
Splus 4 for Windows).

Laszlo Tothfalusi
Department of Pharmacology
University of Toronto

-----------------------------------------------------------------------
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