Can I use random effect analysis in hBayesDM?
봉수현
Hello to all those who develop and use hBayesDM,
I am planning to perform behavioral modeling using hBayesDM.
My hypothesis is that there may be a likelihood of choosing different models at the model level in a normal population and a clinical group, and I am testing this hypothesis by adding a new model to hBayesDM.
For example, I aim to prove that the clinical group will undergo bayesian model selection for bart_par4, while the normal group will undergo selection for bart_ewmv.
During the course of the research, I started wondering if there might be a way to treat the model as a random effect, given the difficulty of perfectly separating the clinical group from the normal group.
In situations where we can't split into Group A and Group B, would it be possible to test with hBayesDM a RFX-BMS (https://mbb-team.github.io/VBA-toolbox/wiki/BMS-for-group-studies/) analysis, which suggests that models are chosen at different frequencies depending on the group?
To do this, it seems necessary to perform model selection at the individual level instead of hierarchical Bayesian analysis, and look at the frequency at the group level. I am wondering if it is possible to perform such an analysis based on MCMC sampling using Stan.
