So I had originally said I would leave this be, that this wasn’t a rap battle, after writing my last detailed criticism of the recent Brad Schoenfeld study. Well clearly that’s not the case.
More on the Statistics
First let me point readers to a thorough analysis of the statistics used in Brad’s paper by Brian Bucher. Basically he takes them apart and shows that none of the THREE metrics supports their strongly worded conclusions.
None of them.
In this vein, here’s something interesting.
Brad and his group have NEVER used Bayesian statistics until this paper. I searched on my folder of his papers and the term Bayesian shows up 4 times. Three are papers that Menno Henselmans was on and it’s his email address. The fourth is one of James Krieger’s meta analyses. At best James has used them before.
Now I find this interesting because there is no way to know if Brad and James had planned to use this approach ahead of time. … Keep Reading