[R] multilevel basic lme question

Dieter Menne dieter.menne at menne-biomed.de
Sun Jun 15 20:54:38 CEST 2008


eugen pircalabelu <eugen_pircalabelu <at> yahoo.com> writes:

> I want to use the lme package for a multilevel analysis on the 
following example:
> > math<-c(2, 3,2, 5, 6 ,7 , 7)
> > sex<-c(1, 2, 1, 2, 2, 2, 1)
> > school_A<-c(1,1,1,2,2,2,2)
> > school_B<-c(10,10,10,20,20,20,20)
> > mydata<-data.frame(math, sex, school_A, school_B)
> > mydata
> 
> School_A and school_B are two different school characteristics, math is 
>> an individual score for each
> subject and sex is also an individual characteristic. 
> Now i want to specify in R the following equations:
> 
> math=a +b *sex +error1
> a= aa+ bb*school_A+ cc*school_B+ error2
> b= aaa+ bbb*school_A+ ccc*school_B+ error3
> 
> but i don't know how. I searched nabble but could not find anything.
> 

Looks like you mixed up aov and lme. There is no "error" term in lme, 
its counterpart is the "random=" term, and you probably read some of the 
aov literature.

I recommend having a look book by Pinheiro/Bates on mixed models, or at 
least checking the examples in the nlme directory.

Dieter



More information about the R-help mailing list