5:00 pm Thursday, April 17, 2014
Back Porch Seminar : And the Bayesians and the frequentists shall lie down together... by Keith Winstein (MIT) in Schedler and Lawn's back porch
See email for the location. As usual, homemade quiche and beverages will be served! In 2007, an academic cardiologist downloaded 42 medical studies from the Web site of drug giant GlaxoSmithKline, combined them in a meta-analysis, and found that the world's best-selling diabetes drug caused heart attacks. GSK lost about $12 billion in sales and market value. But a different way to analyze the same data -- a "Bayesian" way -- finds that the drug actually reduces heart attacks. Or does it? We often hear of this conflict, between Bayesian and "frequentist" statistics. But much of the conflict is misguided. Viewed formally, on the same axes, the two schools of statistics turn out to share a tight symmetry. Criticisms of each can be transformed into a corresponding criticism of the other. I'll also show results from a new numerical algorithm that calculates the performance of "exact" hypothesis tests that hadn't previously been tractable to characterize. This algorithm has found unexpected behavior in the best such methods to date (calculated by tools like StatXact), and suggests a path forward that borrows from digital filter design.
Submitted by