Dear Allstat members, I have 13 possible predictor variables and I'm considering which method to use for selecting the best poisson regression model from those available. Within R I am looking at the following two procedures: 1. The step() function (stats package) which uses a stepwise algorithm and AIC 2. The bic.glm() function (BMA package) which uses Bayesian Model Averaging and BIC Are these both reasonable methods for model selection or is one more appropriate than the other? I hope someone can help. Thanks in advance. Best wishes, Des