Speaker: Richard McElreath.
Applied Bayesian data analysis benefits responsible action inside the model, outside the model, and in numerical implementation. I'll provide computational examples. Inside the model, the design of priors benefits from the simulation of prior predictions. Outside the model, deriving valid causal inference from a Bayesian model always requires more than the assumptions necessary for the model itself. Finally, efficient implementation of ordinary multilevel models often requires very extraordinary parameterizations.