Not one of those things is directly correlated to the power. But those factors (and the many, many others that I did not mention) can add up to skew the results in one way. The next 1000 people from the same area that came might have had an extraordinarily large number of women capable of being Aes Sedai. Just look at the Two Rivers. The point is, non-random samples are biased. There doesn't have to be one mechanism of bias, when the sample is self-selected.