We are using data from the Global Drug Survey to analyse the impact of different alcohol information messages across countries. Using a bayesian multivariate mixed modelling approach we are estimating posterior probabilities using partial pooling of random effects. One challenge is how to convey / report the complex results.