Error in contrasts(this_fixed_data) : contrasts apply only to factors

[Michael Bishop]

Thanks so much for your continued help.   The subject lists the error I'm getting when I attempt ezPredict as shown below.

continuous variables: avrank.zgs  and gpa.n.z

factors:  race5f and sschlcde2 (technically its "labelled")

lme.avrank.zgs <- lmer(avrank.zgs ~ 1 + (1|sschlcde2) + race5f + gpa.n.z

                    , data=avr, na.action="na.exclude" )

> p.lme.avrank.zgs = ezPredict(

+   fit = lme.avrank.zgs

+   , numeric_res = 10

+ )

Error in contrasts(this_fixed_data) : contrasts apply only to factors

[Michael Bishop]

I'm still getting:

Error in contrasts(this_fixed_data) : contrasts apply only to factors

when I try to use ezPredict on my data.  I have tried using a few different continuous outcome variables, and a number of different fixed and random predictors and always get the same error... I don't get the error when I use a gamm4 model, and set "fit = gam.avrank.zgs$gam"  nor do I have a problem when I use the ANT data and an lmer model.

Many sincere thanks if you can give me any help.

Mike

[Joerg Stephan]

On Sunday, 9 December 2012 21:13:11 UTC+1, Michael Bishop wrote:

I'm still getting:

Error in contrasts(this_fixed_data) : contrasts apply only to factors

when I try to use ezPredict on my data.  I have tried using a few different continuous outcome variables, and a number of different fixed and random predictors and always get the same error... I don't get the error when I use a gamm4 model, and set "fit = gam.avrank.zgs$gam"  nor do I have a problem when I use the ANT data and an lmer model.

Many sincere thanks if you can give me any help.

Mike

Hi,
I got the same message ("Error in contrasts(this_fixed_data) : contrasts apply only to factors"). However, I realized it was because I centered and scaled the continuous variable within the model. The solution was to scale the variable within the data frame (e.g. data$x.sc<-scale(x)) and than use x.sc in the model instead of x.
Best
jörg