I would like to point out that this study is very difficult. Because, you have to factor in the skill level of all the players.
If, a good player plays people of lesser skills his ratio will jump, however, if he is in a room with the same level of players, he would have a kill ratio of 1:1. The ratio would be dramatically increased when facing newbies or the less skilled players, it could be as much as 10:1.
So when you are doing this study, you may want to consider that into the factor. This is game is too hard to put a simple statistics anaylsis together.
You need to have your 100 samples, divid them into 3 separate skill levels. Maybe, have 30 in each group. Then conduct your study.
Otherwise, your results are meaningless. With what i suggested, if you see a spike in the lower skilled group, you will know what's happening. However, the higher skilled people may jump from 300 when playing the same skilled to 700 by player lower skilled people. Just watch snakdaddy. Me, Tali, and snake. scores were around 200-300, then when we stopped and started playing in other rooms, our scores were inflated back to 500-600 points. It's difficult, because, there are just too many variables.
I've seen Pancho, go from 300 to 750. He wasn't banking, he was just playing good, maybe on a team or playing lesser skilled tanxters, but not banking.
Good luck with your study, However you need to rethink your sampling.
fast_yz400 WG