Error in r.squaredGLMM()

Question

I am constructing GLMMs (using glmer() of "lme4" R package) and sometimes I get an error when estimating R2 values (using r.squaredGLMM() from "MuMIn" package).

The model I am trying to fit is simmilar to this one:

library(lme4)
lmA <- glmer(x~y+(1|w)+(1|w/k), data = data1, family = binomial(link="logit"))

Then, to estime R2, I use:

library(MuMIn)
r.squaredGLMM(lmA)

And I get this:

 The result is correct only if all data used by the model has not changed since model was fitted. Error in .rsqGLMM(fam = family(x),
 varFx = var(fxpred), varRe = varRe,  :    'names' attribute [2] must be the same length as the vector [0]

Do you have any idea why this error appears? For instance, If I use only a single random factor (in this case, (1|w)) this error does not appear.

Here is my dataset:

data1 <- 
structure(list(w = structure(c(2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 
1L, 2L, 1L), .Label = c("CA", "CB"), class = "factor"), k = structure(c(4L, 
4L, 3L, 3L, 3L, 4L, 1L, 3L, 2L, 3L, 2L), .Label = c("CAF01-CAM01", 
"CAM01", "CBF01-CBM01", "CBM01"), class = "factor"), x = c(1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 0L, 0L), y = c(-0.034973549, 
0.671720643, 4.557044729, 5.347170897, 2.634240583, -0.555740207, 
4.118277809, 2.599825716, 0.95853864, 4.327804344, 0.057331718
)), .Names = c("w", "k", "x", "y"), class = "data.frame", row.names = c(NA, 
-11L))

Any thoughts?

Answer 1

This was a bug that has been fixed in version >= 1.15.8 (soon on CRAN, currently on R-Forge).