I have data from a discrete choice experiment and would like to use Stan to fit a generalized multinomial logit model, as described in Fiebig et al. (2010)
https://pubsonline.informs.org/doi/pdf/10.1287/mksc.1090.0508
This model encodes both coefficient heterogeneity (as in a mixed multinomial logit) and scale heterogeneity.
Does anyone have experience specifying this sort of model in Stan? It would be great to see an example GMNL model / see what an efficient specification would be before I get started on creating my own.
Thanks for any tips.