I am trying to get predicted probabilities across a range of observed predictor variables in a MClogit model with 2 continuous covariates.
I’ve been at it for 3 days straight and can’t get it to work correctly. The ONLY way I can get it to operate is if I leave out the newdata argument.

The program insists that the newdata have exactly the same number of rows as the dataset used to build the model. Doesn’t make any sense. I have tried changing every object associated with variables and newdata lists it to: lists, data.frame, data.frame.table, et al..
If I let it have its way without newdata, the predicted values don’t make any sense whatsoever. I’ve tried every example I can find, but for some reason my model is “special” and won’t work….

I don’t understand why it’s so difficult to take the given coefficient and error term from the model and plot the predictions across the range of observed values.