[R-sig-ME] Imputation methods mixed model analysis

Karl Ove Hufthammer k@r| @end|ng |rom hu|t|@@org
Thu Apr 16 20:33:50 CEST 2020

Breugelmans, S. (Sara) skreiv 16.04.2020 15:33:
> So then I thought it would be better to use some kind of imputation strategy. […] I was wondering if there is a build-in function in lmer() to do this. Or is it better to manually impute the data before analysing.

If you only have missing data in your response variable (LHS of the 
formula), you probably don’t need to impute your data. If you (also) 
have missing data in your explanatory variables, imputation is probably 
a good idea.

> I already found a function called mice(). Does anyone of you have experience with mice() and would you recommend using it?

Yes, the ‘mice’ package is great. But doing proper multiple imputation 
for mixed models can be surpringly tricky. I recommend reading the 
‘Flexible Imputation of Missing Data’ book, written by the author of the 
‘mice’ package. It’s available online, for free:


Chapter 7 deals with imputation for multilevel data.

Karl Ove Hufthammer

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