6 ms·

4,820 is a large enough sample size to be representative of the population within a few percentage points.

• dominotw 5y ago
  > is a large enough sample size What logic are you using to say that( for my future reference). I am having hard time wrapping my head around how .001% of something can be representative.
  • jfengel 5y ago
    If you assume that your sampling is random, it's actually not too hard. If you were able to pick randomly from all over the planet and ask who was male and who was female, you'd probably get a number of around 50-50 with a sample size of just 10. Maybe it's 60-40 or 40-60, but it's unlikely to be all of one or all of the other. That would happen about 1 time in a thousand. If you asked a thousand people there's basically zero chance you'd get all of one or the other. Your low number might be 490 or even 480, but it would take a bizarre coincidence if it were even as low as 450. The big "if" is assuming that you can sample randomly. That's the hard part. There are a ton of ways they go about trying to ensure that they're doing a random sample, but yeah, you do have to take it with a grain of salt. But setting that aside, the fact that they sample only a tiny percentage isn't really a problem. The exponential nature of coincidences makes significant departures from reality unlikely. One coincidence is rare; hundreds is basically impossible. Most of the math behind statistics assumes that there are infinite populations of people. It requires special corrections if you're dealing with only small groups like hundreds. Once you get into the millions, it might as well be infinite. And as long as your sampling is random, it still works out to give reasonably close answers.