Multivariate Cumulative Probit with mixed ordinal and continuous data

Modeling
3

Thank you so much! This is wonderful - I have spent the last few days going over the GHK algorithm, Albert and Chib 1993, and even stumbled upon @CerulloE’s paper. As to the code above I think it all makes sense to me. Few questions:

  1. Could mus be written as vector [K] mus [N]. Thus, it would equate to my mean function g(x) = x^r? In turn, as you have it written Ys would be my ordinal observed responses variables (0,1,2,3,etc.) as vector[K]Ys[N]? Is this correct?

  2. Would Threshholds be defined in the parameter block as and N_thresh in the data block? If so, I’d have to read in N_thres into the data block and define vector [N_thresh] Thresholds [N,K]? Something like this?

  3. I could then out a prior on Thresholds like Betancourt’s induced dirichlet? If I wanted to return the latent values of z would I move those to the generated quantities block?

I suppose my big question is where/how do I declare n_thresh and thresholds. I’ve been looking at some of @CerulloE code as well and I’m trying to figure out how to modify. Let’s say I have categoris labelled 0,1,2 so 3 ordered categories. How would this translate?

Thank you!! Also, I’ll start to play around with this code and then see what I can do about the mixed setting.

1 Like