When I was introduced to Bayesian statistics, I heard the phrase: "for Bayesians, parameters are random". I took this at face value for years even though I didn't believe it. Instead, I thought of parameters as fixed, but our uncertainty about their value assigned a probability distribution. I was happy to find this arXiv article to be published as a discussion paper in The American Statistician that states "Actually, for Bayesians as much as any other statistician, parameters are fixed but unknown. It is the knowledge about these unknowns that Bayesians model as random." On a different note, I still don't understand the doomsday argument discussed in the paper.

blog comments powered by Disqus


15 June 2012