as some of you may know, brms is an R package based on Stan, which allows to fit Bayesian generalized (non-)linear multilevel models using extended lme4-like formula syntax. The beta version of brms 2.0 is now available on Github (https://github.com/paul-buerkner/brms).
Its main new feature are generalized multivariate models, which bring all the flexibility of brms’ univariate models to a whole new level. Each response variable may have its own set of predictors, response distribution (i.e. model “family”), and autocorrelation structure. Group-level (“random”) effects can be modeled as correlated across response variables if desired and residual correlations can be included in multivariate normal and student-t models.
Although I have many unit tests in place and have checked the new functionality quite extensively, brms 2.0 as currently available on github is still in the “beta” phase. I would be grateful for anyone willing to give it a try and help me iron out problems that I assume are still in the updated code base. That way, we can make sure that the CRAN release version of brms 2.0 is as stable and of as high quality as possible.
Many thanks and all the best,