The problem, diomedes, is that when the numbers are as small as the numbers we are talking about relative to other numbers in the reference frame, the amount of data that needs to be taken to get a representative sample of the smaller number is much greater than the amount of data it takes to get a representative sample of the larger number.
Think of it in terms of medical statistics. By taking a sample of ~100 people (chosen randomly, of course), you could probably get a decentish percentage of people who have broken their leg.
That same 100 person sample, on the other hand, is *not* going to be anywhere near accurate for people with facial ectodermal dysplasia. (I don't know what this is, but i got it from a google search for 'rare diseases'.
So while you *think* that 100 is relevant when its a 30-40% difference, the added difficulty getting a more representative sample for a rarer 'condition' means that generally, when dealing with those small numbers, you are *going* to have a large standard error. It's just not worth it to get the percent error down, in most cases.
Please, PLEASE do NOT feed the trolls.
fksumot tag: "Sheik had to become a man to be useful. Or less useful. Might depend if you're bi."
--Predictions--
1) WiiFit will outsell the pokemans.
Current Status: 2009.01.10 70k till PKMN Yellow (Passed: Emerald, Crystal, FR/LG)
