Dear colleagues, I am doing a little bit of work using the BMA package to examine predictive models for types of pain - the outcome variable is a binary (Yes/No) for the type of pain experienced by a patient. Our observation is that BMA makes a lot of sense in this application, as the data suggests that there are several well-fitting possible logistic models, each with posterior probabilities close to 0.1, as well as a straggle of somewhat less likely models. I am unsure how best to produce predictions from the BMA output - i.e. the posterior means of model coefficients. There isn't a predict.BMA or similar, that I can see. What's the recommended way to produce predictions (i.e. fitted values, or estimates from new data) from these BMA models? Best wishes, Anthony Staines -- Anthony Staines, Professor of Health Systems Research, School of Nursing, Dublin City University, Dublin 9,Ireland. Tel:- +353 1 700 7807. Mobile:- +353 86 606 9713