In that exercise you assume that 25% of the Bavi that was to be administered to 1mg arm patients ended up in the control arm, right ?
Would that still comply with the fact that the MOS compared to historical MOS implies that only a small amount of Bavi made it into the control arm.
Unless I misunderstand the 25% (based on your 0.25 and 0.75 arms that represent 1mg combined) if what you suggest would be true then the 0.25mg arm will show that such low amounts of Bavi have nearly NO effect but that 0.75mg does have quite some decent results.
That would be an alternate way of proving that Thorpe's 1mg optimal dosage theory would be quite close. Question is, will the FDA accept such shifting around.
Interesting approach!!!! though I wonder how often patients had their dose. If it is a 1 shot deal then you are absolutely right.... Unless the mistake was being continously being made