nordlead on 17 December 2009
@jammy
statistical sampling mostly relies on a truly random sample. The difference between having 1% and 20% sample size doesn't matter if 1% is truely a random sample (and large enough, as obviously 1 person out of 100 isn't a large enough sample to deduce the trend of all 100 people). Every poll place does this, from Nintendo's research to presedential gallop polls. Of course, the toughest part is finding a truly random sample. Also, traditionally with extrapolations you'll find information on how accurate they feel their numbers are (withing 5% typically). But given the graph above, even if it is slightly off, is enough evidence to disprove that 70% of the DS population are boys 12 & under.
