r - Error in `contrasts<-`(`*tmp*`,value=contr.funs[1+isOF[nn]]): contrasts can be applied only to factors with 2 or more

admin管理员组

文章数量:1123412

I'm facing the following issue in R and I don't know how to solve it:

I want to run linear mixed model to model reaction time as function of group and task. When trying to run the model

lmm.reaction.time.LT.RT.PD.CG <- lmer(
  reaction_time ~ group * task +
    PANDA.total + FSS + education.years + 
    PANDA.mood + 
    (1 + group * task | subject_nr) +       
    (1 + group * task | sentence),
  data = df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy)

I get this error warning:

Error in contrasts<-(tmp, value = contr.funs[1 + isOF[nn]]) : contrasts can be applied only to factors with 2 or more levels

All my predictors have at least two levels:

contrasts(df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy$group)
contrasts(df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy$task)
    [,1]
CG    1
PD   -1  
    [,1]
LT    1
RT   -1

Simplifying the random structures did not work as following model led to same error warning:

lmm.reaction.time.LT.RT.PD.CG <- lmer(
  reaction_time ~ group * task +
    PANDA.total + FSS + education.years + 
    PANDA.mood + 
    (1 | subject_nr) +       
    (1 | sentence),
  data = df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy)


Error in `contrasts<-`(`*tmp*`, value = contr.funs[1 + isOF[nn]]) : 
  contrasts can be applied only to factors with 2 or more levels

I tried to run the model with an other predictor instead of task.sum and it worked, so I think the issue comes from this variable, which has two levels (RT: reaction time task, and LT: language task)

As the RT had fewer trials than the LT there is a strong imbalance between these two levels, which might cause this issues?

table(df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy$group, +    
df.master.correct.responses.no.rt.outliers.LT.RT.PD.CG.copy$task)
    
       LT   RT
  CG 4341  724
  PD 3955  791

My data frame looks something like this (I'm only showing the relevant columns and some example rows):

subject_nr group sentence condition reaction_time education.years PANDA.total PANDA.mood FSS.a task
CG01       CG    43lf     lf        3.433987      12              25          2          23    LT
CG01       CG    43sf     sf        7.078342      12              25          2          23    LT
CG01       CG    45sc     sc        6.137727      12              25          2          23    LT
CG01       CG    NA       left      7.198184      12              25          2          23    RT
PD12       PD    47sf     sf        6.658011      11              28          3          57    LT
PD12       PD    13lc     lc        5.986452      11              28          3          57    LT
PD12       PD    NA       right     6.920672      11              28          3          57    RT

Is there another reason that could cause this error?

Maybe the NAs in column "sentence" as "sentence" is used for random structure?

If the error occurs due to the imbalance of data points between these two tasks is there a way to structure my data frame in another way to be able to run the model?

本文标签：