Ocyan - I would respectfully submit that Dew is correct that such an issue does exist. Like much of Bayesian statistics it is very hard to define all the parameters - but that doesn't mean the general methodolgy is wrong. E.g. I think it is a safe bet that at least 50x more worthless drugs enter ph ii than worthwhile drug - and at that rate you might well expect that the majority of drugs with a stat sig ph ii are not really efficacious - and will fail the ph iii.
The real issue inre Dendreon is that they have a whole variety of positive trials or trials on the right side of HR=1.0. That significantly decreases the false positive rate. Is it subjective - yeah, somewhat. But even Dew agrees that Provenge probably has some efficacy - although he would undoubtedly assign a lower probabilty than us.
News 
Market Data 
Discover 
AI
