[quote name='supershammy']"One amazing fact is that whether the survey researcher is studying the characteristics of people in a city or a whole country, about 1200 randomly-selected participants will enable the researcher to be 95 percent confident of describing the entire population with an error margin of 3 percent or less." - pg. 10, Myers, Social Psychology
It should also be worth noting that any poll that takes place on the Wii is not of a randomly-selected group of participants, and is therefore not accurate in reflecting the opinions of any region in which it takes place - only that of Wii owners.[/QUOTE]
I was going to say...The size of the sample is significantly less important than the MAKEUP of the sample. Thus, your method of choosing participants is paramount. So limiting polls to Wii owners, while interesting, will produce pretty skewed results...Not as skewed as polling, say, 360 owners, but skewed nonetheless.
It reminds me of the old "Dewey Defeats Truman!!!" story. That headline was predicated largely on a phone poll of registered voters. Only problem was, back then, only those in a very specific affluent demographic owned a phone. So the results they got were not representative of the actual electorate. Exit polls are a significant improvement, but they're still fallible (as the last two presidential elections proved).