If the polling organizer did its job correctly and removed bias from their sampling, a beautiful law in statistics has been proven, stating that small-ish samples are representative of the whole population if it follows e.g. a normal distribution. It’s called “law of large numbers”.
Is 500 really big enough to get a proper spread though? Iirc my statistics course (which i don’t all that well tbf) you need a pretty significant sample still, would think a few thousand at least
497 more than 1% of Greenland’s voting population. For larger populations the rule of thumb is generally 1000 people for a good poll, 500 for a decent one.
If the polling organizer did its job correctly and removed bias from their sampling, a beautiful law in statistics has been proven, stating that small-ish samples are representative of the whole population if it follows e.g. a normal distribution. It’s called “law of large numbers”.
Is 500 really big enough to get a proper spread though? Iirc my statistics course (which i don’t all that well tbf) you need a pretty significant sample still, would think a few thousand at least
497 more than 1% of Greenland’s voting population. For larger populations the rule of thumb is generally 1000 people for a good poll, 500 for a decent one.
Fair enough, thanks.
2000 is the standard max. 500 is probably good enough for most things.