"[E]rroneous. But Unfortunately, It Permeates the Medical Research Community"
What is it? It's the view that "when a p-value is </= 0.05 then there is sufficient evidence that the drug works." In what I do it's also the claim that low p-values prove so-called general causation in a toxic tort case. The quotes come from a wonderfully readable paper just published in the Journal of the National Cancer Institute. It's: "Demystify Statistical Significance - Time to Move on From the P Value to Bayesian Analysis".
I particularly liked this: "Statistics in medicine has passed through its infancy and childhood. As it moves into its adolescence, the growing pains of reconciling frequentist and Bayesian views continue... Although the frequentist paradigm has been widely applied and is deeply rooted in medical research, it is time to move on and look for a better solution."
The frequentist paradigm spawned most of the toxic tort litigation of the last few decades and it also provides an easy way to gin up grant money for researchers and funds for "consumer research" so don't expect it to yield to reason without a long and bitter fight. I predict the Mt Sinai and Collegium Ramazzini folks will man the ramparts of the frequentists' final redoubt.
That having been said, perhaps the legal empiricists, learning from medicine's experience with touching hot stoves, can avoid the pain and skip straight to Bayesian analysis. Yeah. Right.
Anyway, expect the fight over whether a drug should or shouldn't be approved, or should or shouldn't have been approved, to move on to Bayesian grounds.
