2016 Postmortem
Related: About this forumPooling the Ohio Polls
The last 10 Ohio/Obama-Romney polls show a range of results, from an Obama-Romney tie, to a +5 Obama advantage. The average (unweighted) result reveals a 2.6 percentage advantage for Obama.
If we pool these polls, we have a sample of 8,738 likely voters. A simple one-sample t-test of the poll results suggests a 99.4 (one-tailed) probability of an Obama victory in Ohio. The test is based on the unweighted average. If I were to weight the results based on individual sample sizes, the results wouldn't change meaningfully.
I did this after getting annoyed at a CNN reporter for suggesting that the race was a dead heat.
It doesn't appear to be that way.
aaaaaa5a
(4,667 posts)...with the polls, Obama is predicted to get 49.1 (+/-1.0) percent, and Romney 46.5 (+/-1.0) percent.
LeftInTX
(25,556 posts)He said that their internal polls show Romney ahead by +2
He said that the other polls were wrong because they were using the "2008 model"
I'm like????
alcibiades_mystery
(36,437 posts)We're getting out clocks cleaned? We might win Minnesota?
They're running out of spin.
When you're disputing EVERY SINGLE poll out there with your "internal polls," you're losing and spinning. Or your internal polls are being run by a con artist.
frazee
(61 posts)it's doubtful that they are all wrong. And with Rasmussen of all firms showing a tie, yeah I don't think so.