[Tom]

I messed up your example in an attempt to explain why it misses the point (unsuccessfully as it turns out).

I understand exactly what you're saying, but its not the point I am making. Its correct that you don't draw an inference about why a team lost after 1 game. You try as hard as you can to isolate the variable (or variables) to determine what effect they have on the outcome. Think about something as complicated as the health impacts of smoking. Some people smoke their whole life and nothing happens, whilst others appear to develop all sorts of problems. We cannot tell exactly what effect smoking will have on any particular person, because a range of genetic and lifestyle factors also significantly affect the result. But, after looking at millions of smokers, we can confidently state that, generally speaking, smoking is bad for you.

This is why, in looking at Starcraft, you should look at the results of thousands or millions of games, played on a range of maps, at a range of skill levels. If players using one race consistently underperform across these variables, it is fair to draw an inference that race is having some effect on the outcome (even though it would be difficult to use race to predict the outcome in any single case).

I am not going to waste any more time trying to explain it.

*************************

For the same reason that the GMs in this thread refuse to debate balance/gameplay with silver league players, I am no longer going to debate statistics/scientific method with anyone who does not hold (at least) an undergraduate degree in mathematics, engineering or science. Sorry to sound like a douche, but its just not worth the aggravation.

Whilst I am happy to debate genuine methodological flaws, I am getting tired of justifying basic principles of science/statistics over and over. Please read the OP!