Directly testing your expectations or testing the null hypothesis? A comparison between classical null hypothesis testing and Bayesian model selection