[WoW] PVE-->PVP xfers now allowed!

[Vyndree]

7044 PvE guilds in the top 10% / 152859 total guilds = 0.046 = ~4.6%
8241 PvP guilds in the top 10% / [/font]152859 total guilds = 0.053 = ~5.3%

Because otherwise, the numbers and the percentages don't match up, you can't say that there are more PvE guilds in the top 10% because there simply aren't, there are 1197 more PvP guilds in the top 10% and more PvP guilds overall, the percentages mean little when talking about more and less.

But you're not taking into account the number of raid guilds per population. The number of PvP servers in existence is not the same as the number of PvE servers in existence.

Yes, there are more PvP servers in the top ten NUMERICALLY, but what if there are also more PvP SERVERS in existence? If you were to join a PvP server, would you have a statistical chance of getting into the top 10% guild just the same as you would if you rolled on a PvE server? What if the PvP server's population is higher than the average PvE server? This is all data that will skew the numbers if you take them at face value. Therefore, we have to look at the total value and just quantify the difference in DISTRIBUTION.


Let me put it in words so it's a bit easier to grasp.

There are 5 sunwell raid guilds on 1 pvp server -- and 200 karazhan raiders on 200 pve servers. Which is better?
There are 5 karazhan raid guilds on 1 pvp server -- and 200 sunwell raiders on 200 pve servers. Which is better?
There are 5 sunwell raid guilds on 1 low population pvp server -- and 200 karazhan raiders on 200 high population pve servers. Which is better?
There are 5 karazhan raid guilds on 1 low population pvp server -- and 200 sunwell raiders on 200 high population pve servers. Which is better?


You see how server number, progression (i.e. sunwell vs karazhan), and server population can skew a simple 200 > 5?


So what I did is said -- hey, let the population be population and number of servers be number of servers.

We DO have percentile data for progression. We have top 10% of the total number of guilds, we have top 10% of pve guilds, we have top 10% of pvp guilds. These numbers, in an ideal world, would be 10% = 10% = 10%.

We know x many total guilds raid. We also know y of those are pvp servers, and z of those are pve servers.
If I have a set of the top 10% of raiders -- both pve and pvp, that means ideally there should be 10% of the total pvp servers present in that number, and 10% of the total pve servers present in that number to make it a perfect "even" average.


Essentially, I'm taking a set of groups and seeing how the groups are DISTRIBUTED.

Therefore I am taking

# pvp guilds in top z% / # total pvp guilds = x
# pve guilds in top z% / # total pve guilds = y

And then comparing these two numbers to their ideal, which is just the z%.

So if we're looking at the top z = 10%, then these two equations (x and y) should both, ideally, be 10%.

I can completely ignore the total number of guilds, this way, because the total is irrelevant. I only used the total to formulate my BossKillers guild query -- "count all guilds in the top 10%". To get to 10% I had to find the total number of guilds they knew existed, and multiply it by 0.1 -- the total number of guilds is otherwise meaningless to me.