Error: Can only combine group-level terms of the same grouping factors

Interfaces brms
1

Please also provide the following information in addition to your question:

  • Operating System: Mac OS High Sierra v10.13.6
  • brms Version: 2.10.0

Hello again!

PROBLEM
While running my imputation multilevel model I have been receiving the error:

“Error: Can only combine group-level terms of the same grouping factors”.

MODEL

I am trying to model hormone values as the result of a series of demographic/social parameters in 2 primate species.

A simplified version of the model is:

brm(log_hormone ~ (1|q|Harem) * Species1Dummy + Species + Sex + Age)

Basically here is the issue: One of the species I am researching forms harems while the other does not. I wanted to include a random effect for harem membership since it is possible that some female hormone values may be the result of which harem they belong to; however, I was unsure how to do this considering that the other species does not even form this social grouping. It was suggested to me to include a dummy variable so that the harem effect would only be calculated for one of the species and ignored for the other.

However, I get the above error when I try to run this model.

Does anyone have a suggestion of how to circumvent this problem? I would prefer to do as much as possible in a single model rather than making multiple models, especially if that means I would have to reduce my sample size by half to focus on just the harem-forming species.

Thanks for the help!

2

Can you provide a minimaly fully reproducible example for me to check the problem myself?

3

Hey! Sure!

I just generated this example dataset and when I run it, I get the same error as before:

log_hormone   <- log(sample(1:100,100, replace = T))
Species       <- c(rep("A",50), rep("B",50))
Sex           <- sample(c("M","F"),100, replace = T )
Age           <- log(sample(5:25,100, replace = T))
Species1Dummy <- c(rep(1,50), rep(0,50))
Harem         <- c(sample(c("Prince","Hendrix","Elvis"), 50, replace = T), rep("No_Harem",50))
  
  d <- cbind(log_hormone,Species,Sex,Age,Species1Dummy,Harem)
  
  model <- brm(log_hormone ~ (1|q|Harem) * Species1Dummy + Species + Sex + Age, data = d)

Error: Can only combine group-level terms of the same grouping factors.

Thanks for helping me with this!

4

I see it now. You cannot multiple group level terms with other predictors. This is not supported by the syntax.

5

Ah! Thanks for letting me know. Are there any plans to potentially alter the syntax to allow this sort of computation in the future?