Andrew’s joke
In what is apparently becoming a tradition, Andrew Gelman posted once again on this First of April a well-written text that eloquently express the opposite of his opinions. Last year, he had listed a series of arguments against Bayesian statistics that ended up in a discussion paper published by Bayesian Analysis; this year, the point is against (read, in favour of) using graphs rather than bland tables of numbers. As an editor, I cannot agree more with the (true) position of Andrew, as most tables are close to unreadable and would benefit both the reader and the writer much more when turned into (good) graphs. Tables can be stored in supplementary material sites as proofs that experiments were run the way the paper says they were, but as such, in a printed format, they are close to impossible to use.
Leave a Reply