i didnt think so:
"is roughly as demanding as an a priori requirement that the entire set of 100 measurements show p<0.05."
it all depends on the variance of the populations.
for a small subset, the variance will tend to be higher, making a p .0125 just about impossible
but with 100 measures, pooled a .05 is easier to obtain. most of the time populations of small number tend to "smooth themselves out" if you collect more data.
if indeed the .0125 measure MUST be met, then the studies would have to be designed with sufficient people in each that would allow making this probability, which you calculate from asking:
what do you expect your response to be?
how well can you measure it?
what variance will be int eh data?
than a stat head simply calculates the patient population size
in your first message, you said .0125 for each OR .05 for all.
which is really meant or both?
Market Data 
Markets 
AI
