Well obviously it doesn't "prove" anything, but I don't think it gives a "high level view" of what's happening either. You shouldn't take the mean of a highly

Well obviously it doesn't "prove" anything, but I don't think it gives a "high level view" of what's happening either. You shouldn't take the mean of a highly non-normal distribution. It's virtually meaningless (no pun intended). When the population you're looking at is completely arbitrary anyway (top 50), the results are further nullified. And taking the mean of a group of percentages is... not wrong, but dangerous IMO.