OK, I'm not a statistician and I presume you are....but that still doesn't sound right. That would mean a sample of 1000 would have a margin of error of 1/10,000th of one percent (1/1,000,000)? I've seen polls with sample sizes larger than that, showing margins of error between 1 and 2 percent.
But I don't pretend to be an expert. I'd be interested in seeing where I'm wrong.
Hank
What he’s saying is take the square root of 1000, 31.6227766, and then take the reciprocal.
1/31.6227766 to get 0.316227766 or 3.2%.
No, the margin for n=1000 is 3.16.
Not a statistician, but I use statistics a lot. If you had a group made up of subgroups and you are combining them, margins can be different. Also, there are different ways to get that number, but the one I showed is basic.
It does seem bogus to poll 1000 people and declare that it represents 300,000,000 people to 3%, but that’s the math.