Hi Sundar,
Yes, I missed that. Thank you. However, here is my real problem:
This model works:
fit5<-lme(log(f0mean)~cat*gender*strength, random=~1|subid/emotion,
weights=varIdent(form=~gender),
correlation=corCompSymm(form=~1|subid/emotion))
While copying and pasting the model with the addition of a (-1)
produces:
Fit6<-lme(log(f0mean)~cat*gender*strength-1, random=~1|subid/emotion,
weights=varIdent(form=~gender),
correlation=corCompSymm(form=~1|subid/emotion))
Problem in .C("mixed_EM",: Singularity in backsolve, while calling
subroutine mixed_EM
Use traceback() to see the call stack
Oddly, if I do a strange combination of fixed effects like
cat+strength+cat:gender - 1 it is okay with that, but as soon as I close
a model with all possible effects minus the intercept I get the error.
This is true even if I only include cat*gender as the fixed effects.
Perhaps it is a non-issue. I wanted to exclude the intercept so that
the contrasts would be performed on the "cell means". In this case, are
only the numbers easier to interpret or is the meaning of the contrast
different when the intercept is removed. If the meaning is essentially
the same I can live with not being sure what the contrast value
represents.
Thanks,
Michael
> -----Original Message-----
> From: s-news-owner@lists.biostat.wustl.edu [mailto:s-news-
> owner@lists.biostat.wustl.edu] On Behalf Of Sundar Dorai-Raj
> Sent: Monday, January 20, 2003 10:35 AM
> To: Michael Schmidt
> Cc: 'S-news'
> Subject: Re: [S] error removing intercept in lme
>
> Hi Michael,
>
> Michael Schmidt wrote:
> > Using the example on page 18 of "Mixed-Effects Models in S and
S-Plus",
> > attempting to remove the intercept from the model produces the
> > subsequent error.
> >
> >
> >>attach(ergoStool)
> >>fm3Stool<-lme(effort ~ Type - 1, data=ergoStool, random=~1|subject)
> >
> > Problem in lme.formula(effort ~ Type - 1, data = er..: Length of
subject
> > (variable 3) is 24 != length of others (36)
> >
> > Have I failed to do something??
> >
>
>
> You have `subject' misspelled. Try `Subject' instead. S-PLUS is case
> sensitive.
>
> Regards,
> Sundar
>
>
> --------------------------------------------------------------------
> This message was distributed by s-news@lists.biostat.wustl.edu. To
> unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with
> the BODY of the message: unsubscribe s-news
|