Discussion questions#
Consider the situation where you have a Uniform prior on your model (i.e., an unknown prior). This effectively nullifies the relevance of the prior distribution in your calculations of the Bayes Factor, turning it into a ratio of likelihoods. Now compare this with the p-value. What is the key difference between a Bayes Factor and p-value in this situation and how does it impact the interpretation of your results?
Consider the relevance of Bayes Factors in the context of the ATOM approach discussed by Wasserstein et al. 2019 (discussed in the p-value lecture). Does the use of Bayes Factors nullify the need to calculate a p-value if you are adopting the ATOM approach? Why or why not?