Well, I was mostly posting that tongue in cheek, because a lot of people say that sample sizes are too small without actually checking. I've got my old statistics text book buried in storage, so I went to an online tool.
First, what are we sampling? We'll assume US Citizens over the age of 18. Potential voters, in other words. Sure, we don't have any way to know that only US Citizens over 18 voted in the poll, but that's really the least of the problems with that kind of poll. (IOW, I agree it's a frivolous poll. Even more frivolous than one on a network news site.)
Checking the census website, I find that the estimated US population for 2013 is 316,128,839, and that 23.3% are under 18. That gives an estimate of 242,470,819 US citizens over 18.
So, I'll plug that into a sample size calculator. This is just the first one I found online. I'm not vouching for it, but the math isn't really that hard, so I'll assume that if whoever made it can do basic programming, they can do basic statistics.
For a confidence level of 95 with a confidence interval of 5, I get a valid sample size of . . .
384!
Much lower than you'd think. Heck, much lower than I thought, and I knew it was bound to be a low number.
For a confidence level of 99 and a confidence interval of 5, the number is 666. (Yikes!)
For a confidence level of 99 and a confidence interval of 1, the number is 16,640. That's a pretty stiff level for this sort of thing, though. You won't find too many studies or opinion polls attempting this level of confidence, though even some online polls will easily net over 16k responses.
Of course, size is only one measure of whether a sample is valid.
So, sample size can be surprising.
The usefulness of these polls is another matter, but it never hurts to express your opinion on a subject in a non-confrontational format, either.
First, what are we sampling? We'll assume US Citizens over the age of 18. Potential voters, in other words. Sure, we don't have any way to know that only US Citizens over 18 voted in the poll, but that's really the least of the problems with that kind of poll. (IOW, I agree it's a frivolous poll. Even more frivolous than one on a network news site.)
Checking the census website, I find that the estimated US population for 2013 is 316,128,839, and that 23.3% are under 18. That gives an estimate of 242,470,819 US citizens over 18.
So, I'll plug that into a sample size calculator. This is just the first one I found online. I'm not vouching for it, but the math isn't really that hard, so I'll assume that if whoever made it can do basic programming, they can do basic statistics.
For a confidence level of 95 with a confidence interval of 5, I get a valid sample size of . . .
384!
Much lower than you'd think. Heck, much lower than I thought, and I knew it was bound to be a low number.
For a confidence level of 99 and a confidence interval of 5, the number is 666. (Yikes!)
For a confidence level of 99 and a confidence interval of 1, the number is 16,640. That's a pretty stiff level for this sort of thing, though. You won't find too many studies or opinion polls attempting this level of confidence, though even some online polls will easily net over 16k responses.
Of course, size is only one measure of whether a sample is valid.
So, sample size can be surprising.
The usefulness of these polls is another matter, but it never hurts to express your opinion on a subject in a non-confrontational format, either.
