Linear mixed model - matrix is rank deficient

Question

I'm trying to apply a linear mixed model to my dataset and I keep getting a strange message:

"fixed-effect model matrix is rank deficient so dropping 1 column / coefficient"

Right now my model code is:

 m0 <- lmer(y ~ var1 + var2 + var3 + (1|var4))

where y is a numeric response variable and variables 1-4 are factors.

What could this possibly mean?

It means that `var1`, `var2`, and `var3` (the fixed effects) are multicollinear, i.e. there is some linear combination of them that's constant. — Ben Bolker, Apr 01 '14 at 14:19
http://stats.stackexchange.com/questions/35071/what-is-rank-deficiency-and-how-to-deal-with-it — Ben Bolker, Apr 01 '14 at 14:20

score 0 · Answer 1 · edited Aug 12 '16 at 15:11

0

I had the same issue with my initial run so I checked the variables I had included and one of them (a factor variable) only had 1 level. So I removed it and it worked. Also make sure you don't have NA's in your variables.

To check for NA's:

sapply(data[, c(variable_names), with=FALSE], function(x) sum(is.na(x)))

To drop NA's:

data_no_NAs <- na.omit(data[, c(variable_names), with = FALSE])

edited Aug 12 '16 at 15:11

Billal Begueradj

20,717
43
112
130

answered Aug 12 '16 at 14:39

Preston Zifko

1
2

Linear mixed model - matrix is rank deficient

1 Answers1