"iii) the bar is 1/2 what was seen in prior studies (where the effect was retroactively studied)"
That is simply not true. The "bar" is or course the P value.
The statement that the trial is 85% powered based on 1/2 the prior effect is missing a really important point. Namely, there has to be a std dev (or such) assumption in there. Given that nobody I know of has references to actuall data from the P1/2s (just summary), we have no clue on what this is the earlier trials.
I guess if clark want'd to waste some time he could he could make a much better comment on this.
BTW, what is a "registrational quality" trial?
Market Data 
Markets 
AI
