alt Hacker NewsAssuming your samples are not biased, 1000 subjects are generally far more than are necessary to demonstrate an effect. People who complain about sample size are generally not that well-educated in statistics.
Assuming your samples are not biased, 1000 subjects are generally far more than are necessary to demonstrate an effect. People who complain about sample size are generally not that well-educated in statistics.
A well-educated person in statistics would also mention that it requires a certain class of distribution. This is one of Nassim Taleb's favourite subject (imagine computing the average net worth of a random group of people and suddenly Bill Gates is among them)