Marginal Effects plot of ordinal data - unclear on x axis

Interfaces brms
1

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

  • Operating System: OSX 10.11.6
  • brms Version: 2.4.0

Hi forum,

I am using a cumulative() family to model membership for 3 discrete states (state) based on a single continuous variable (pheno).

When plotting the marginal effects I am confused by the x-axis. The probabilities do not match up with my continuous data. My “pheno” variable has a range:

> range(data$pheno)
[1] 0.4676923 0.9285714

yet the highest probability for inclusion in state 1 is from 0.0 - 0.2, which is out of the range of my variable.

13%20PM
Screen Shot 2018-08-31 at 12.23.13 PM.png1110×734 41.2 KB

I am not sure if this is an error in the plotting of the x-axis scale or a broad misinterpretation of the plot on my part. Does something need to be re-scaled? I know from looking at the data that there is a trend. Here is a plot of the raw data:

23%20PM
Screen Shot 2018-08-31 at 12.26.23 PM.png691×588 9.55 KB

Below is the model, and the data are attached.

Thanks for your time,
Jon

o_test <- brm(
  state ~ pheno, data = data, 
  family = cumulative(),
  iter = 5000)

ordinal_test.csv (3.1 KB)

2

Indeed that’s a weird behavior. I am not really sure why that’s happening, but I think it’s not a bug in brms although it looks like one. Actually, I deprecated these kinds of plots in the latest dev version of brms. I recommend installing brms from github via

if (!requireNamespace("devtools")) {
  install.packages("devtools")
}
devtools::install_github("paul-buerkner/brms")

and then go for marginal_effects(..., categorical = TRUE)