[R-sig-ME] specification of corAR1()

array chip arrayprofile at yahoo.com
Wed Aug 10 19:07:59 CEST 2011


Hi all, I am struggling to learn to use corAR1() correlation structure. Let me use the Machines data set from nlme package as an example:

I added a time variable to create a new dataset:

> Machines2<-cbind(Machines, time=c(1,2,3))

This one runs without any problem:


> lme(score~Machine,data=Machines2,random=list(Worker=pdSymm(~0+Machine)),correlation=corAR1())


But all of the following generated the same error message:
> lme(score~Machine,data=Machines2,random=list(Worker=pdSymm(~0+Machine)),correlation=corAR1(~time))
> lme(score~Machine,data=Machines2,random=list(Worker=pdSymm(~0+Machine)),correlation=corAR1(~time|Worker))
> lme(score~Machine,data=Machines2,random=list(Worker=pdSymm(~0+Machine)),correlation=corAR1(~time|Worker/Machine))

"Error in abs(value) : Non-numeric argument to mathematical function"

Did I misunderstand something here?

Thanks very much!

John




More information about the R-sig-mixed-models mailing list