I see on Stocktwits that Mayo gives the ABCLEAR3 30mg subgroup at 55-56 percent by measuring at 48 weeks rather than at the outset, which I calculated as ~47 percent.
I think you guys get the point.
Anavex is not Texas sharpshooting -- since they are not changing any trial endpoint numbers -- but I have to wonder, given the obvious fact that the response distribution graph must be bimodal and that these data will entirely populate one mode, whether the post hoc nature of the ABCLEAR3 analysis is all that statistically unsound. Wouldn't the possibility of this extreme special case as random be much more unlikely than a traditionally (i.e., Gaussian) random post hoc finding? Where are our stat guys?