it cant be. an aggregate p value, when i was in the business, used the combined populations of the 4 subsets. so if there were 25 measures each i needed a .05 p for 100 measures this is much different than .0125 for each 25. how can we reconcile this?