Plotting predictions from an ordinal regression using the plot_predictions() function

Thank you for the information, that makes sense that, given it being an ordinal model, the graphs make those odd-looking stepped jumps from one integer level to the next. Your last point I definitely can speak to: as long as you are using the same data and those data are integers, you can compare a gaussian version of the model to an ordinal. It’s also worth noting that when the data are bounded counts, you can also compare gaussian to binomial models.

see here and here for an explanation by Aki Vehtari.