[R-sig-ME] FW: linear mixed model for non-normal negative andcontinous dataa

Titus von der Malsburg malsburg at posteo.de
Mon Apr 28 14:07:28 CEST 2014

On 2014-04-28 Mon 12:31, Caroline Lustenberger <caroline.lustenberger at hotmail.com> wrote:
> mod_Knmn<-lmer(Knmn~Group*Timepoint+(1|VPnr),data=data)
> When performing a qq-plot my residuals are clearly deviant from the
> norm (long-tailed). Due to negative values I cannot perform classical
> transformation methods (e.g. log transformation). How could I proccede
> with this data. Is there a possibility to use a generalized linear
> model?

Just multiply with -1 (turning the measurements into strictly positive
values) and then do a log transform or box-cox transform.


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