I cannot believe that this poll by the University of Iowa made news. I cannot even believe that the University of Iowa would let it be published under their name.
According to it, the biggest winner after the announcement of Elizabeth Edward’s cancer was Hillary Clinton, and the biggest loser was Barack Obama.
If you read all the way to the bottom of the press release, you find out the sample size of the poll for likely Democratic caucus-goers (and God only knows how they determined that status). That size? 128. They report a margin of error of 6% on that sample, though by my calculations it should be more like 8.7%. Even under the generous assumption that this sample was divided exactly in half for pre- and post- announcement polling, that leaves the before and after numbers with a margin of about 12%.
You know what that means? Nothing at all can be concluded from this poll. Nothing. You’d think a University would be able to figure that out, but maybe they don’t teach statistics 101 to Political Science Ph.Ds.
4 Comments
Not quite right
I agree it was a worthlessly small poll. But the total sample was 334 and the Democratic likely caucus goers was 206.
boringdad Sat 31 Mar 9:05 PM
You're Right
I misread it. So here’s my new question: how do they determine that one third of their sample of registered voters are likely caucusgoers? Last I heard, the caucuses have less than half of that for turnout on a good year.
simon-stevenson Sun 1 Apr 10:24 PM
Not quite right
I agree it was a worthlessly small poll. But the total sample was 334 and the Democratic likely caucus goers was 206 according to your link.
boringdad Sat 31 Mar 9:20 PM
Oooops.
boringdad Sat 31 Mar 9:20 PM