This is not currently supported. If you want to poke around and make some suggestions on lmer support, I'm open to it.

I think this should be working now:

library("lme4")
m <- lmer(Reaction~Days+(1|Subject), data=sleepstudy)
(mar <- margins(m))
## Average marginal effects
##   Days
##  10.47
## Warning message:
## In marginal_effects.merMod(model = model, data = data, type = type,  :
##   factor variables are not handled as factor for models of class 'merMod'
summary(mar)
##  factor     AME SE  z  p lower upper
##    Days 10.4673 NA NA NA    NA    NA

I've gotten the following error when using margins on a random intercept model (1 | location) with v 0.3.5

Error in model[["weights"]] : this S4 class is not subsettable

@jeffmax Ah, sorry. Apparently the margins.merMod() method was not exported, so this was hitting the default (which won't work). Can you reinstall from Github and try again?

@leeper, I am still getting the same error (confirmed help says 0.3.6). Thanks for looking into this so fast.

@jeffmax what's the class of the object you're passing to margins()?

> class(m_int)
[1] "glmerMod"
attr(,"package")
[1] "lme4

@jeffmax Can you try to make a reproducible example? The package works with examples from ? glmer so I need more information to track down what's happening here.

Hi @leeper, I am having the exact same issue. I re-downloaded margins from Github and continue to have the follow error as well even when I use the sleep study dataset.

install_github("leeper/margins")
library(margins)
m <- lmer(Reaction~Days+(1|Subject), data=sleepstudy)
(mar <- margins(m))
## Error in model[["weights"]] : this S4 class is not subsettable`

Thank you so much for all your work on this package!

Hi @leeper, this seems to be an awesome package. Much appreciate your work on this!
Unfortunately, I have the same issue like @jeffmax and @tamaravdd. Is there any update on the what might cause the error message?

yeh @leeper it is really great to see this package because it can get us average marginal effects. Please share if you have any updates. I am getting similar Error message
"defined for objects of type "S4" only for subclasses of environment"

With the latest Github versions of margins and prediction, I do not get any errors using the first example models from lmer and glmer.

I do get (for model fm1 from lmer examples):

Warning messages:
1: In marginal_effects.merMod(model = model, data = data, variables = variables,  :
  factor variables are not handled as factor for models of class 'merMod'
2: In predict.merMod(model, newdata = out, type = type, ...) :
  unused arguments ignored

and (for model gm1 from glmer examples):

Warning messages:
1: In marginal_effects.merMod(model = model, data = data, variables = variables,  :
  factor variables are not handled as factor for models of class 'merMod'
2: In predict.merMod(model, newdata = out, type = type, ...) :
  unused arguments ignored
3: In predict.merMod(model, newdata = out, type = type, ...) :
  unused arguments ignored
4: In predict.merMod(model, newdata = out, type = type, ...) :
  unused arguments ignored
5: In predict.merMod(model, newdata = out, type = type, ...) :
  unused arguments ignored

But at first glance these warnings are more or less spurious in terms of their bearing on the accuracy of the output.

The latest version on Github seems to have regressed from when @jacob-long commented. MWE:

library(margins)
library(lme4)
library(lattice)

gm1 <- glmer(cbind(incidence, size - incidence) ~ period +
                 (1 | herd),
             data = cbpp,
             family = binomial)

m <- margins(gm1, data = cbpp)
traceback()
sessionInfo()

and the output

> traceback()
5: terms(model$model)
4: find_terms_in_model.default(model, variables = variables)
3: find_terms_in_model(model, variables = variables)
2: margins.default(gm1, data = cbpp)
1: margins(gm1, data = cbpp)
> sessionInfo()
R version 3.4.2 (2017-09-28)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 17.10

Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.7.1
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.7.1

locale:
 [1] LC_CTYPE=en_GB.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_GB.UTF-8        LC_COLLATE=en_GB.UTF-8    
 [5] LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=en_GB.UTF-8   
 [7] LC_PAPER=en_GB.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] lattice_0.20-35 lme4_1.1-14     Matrix_1.2-11   margins_0.3.22 

loaded via a namespace (and not attached):
 [1] minqa_1.2.4         MASS_7.3-47         compiler_3.4.2     
 [4] tools_3.4.2         Rcpp_0.12.13        splines_3.4.2      
 [7] nlme_3.1-131        grid_3.4.2          data.table_1.10.4-3
[10] nloptr_1.0.4        prediction_0.3.2   
`

Thanks all for your bug reports. I'm about to push an update that will include much more extensive unit testing for mixed effects models so that we shouldn't encounter these regression issues again. I believe all of these examples will now run successfully. Let me know if not.