I am trying to model a conjoint experiment in Stan, which includes conditional pricing. Specifically, I am varying the price points shown conditioned on the presence of a level of a certain attribute. However, I am not sure how to consider this in my model.
When I include the price in Aggregate Logit packages such as mlogit, I get a collinearity error. Has anyone encountered this issue before, and if so, how did you address it in your model?
Any help or guidance would be greatly appreciated. Thank you!
Could you post your model and briefly describe the model and data?
I’m sure there are people who can help you with your issue here given the data structure and class of model you need, but (like me) they may neither be familiar with the system and the format of the observation, nor with the packages you mention (again, I for one am not even a user of “brms”, so I may miss the point or not know what the question is about just because I’m unfamiliar with the package).
I assume mlogit is shorthand for a multilogit model. Stan documentation includes an example of that that may be helpful, and I’d guess that’s also something you can find in stan-based packages like brms or others.
When implementing a multilogit model directly in Stan you may not get a colinearity error/warning, but instead get problems with identifiability, but the precise issue would require looking at the actual model.