In a previous post I applied a Bayesian ordered-probit model to movie ratings and showed how the results differ from treating the data as if they were metric. The metric model used frequentist t tests (because that's what most applied researchers would...