Replies to post #1353 on Dendreon Corp fka DNDNQ

[iwfal]

I mean they are faced with reviewing and explaining to the non-stats guys (excluding D'Agostino, who I think may be one) on the panel whether a different result has it's merits.

It's hard to believe that after all those years that they do not know it as a fundamental. Perplexing.

a) D'Agostino is a statistician (although I don't believe that any AC members play a role in adjudication outside of the AC itself)

b) You may not see references to the improvement in Statistical Reviews/AC's but you will find reference to the added assumptions implicit in a Cox Regression. The p value improvement is the flip side of these added assumptions and their validity. In a way the actual median p value improvement is unimportant once you've validated the added Cox Regression model assumptions.