Most accurate national polls? Democracy Corps, RAND, ABC/WaPo and Gallup*

[Lief 🗽]

Obama's final national margin is going to end up somewhere between 3.5 and 4%. Democracy Corps final poll showed Obama +4 (49 to 45) and RAND's final poll showed Obama at +3.32%. ABC/WaPo was also pretty good, showing Obama +3 (50-47).

Ironically, Gallup's registered voter poll was also pretty good, showing Obama +3 (49-46). Their likely voter poll, showing Romney +1 was garbage.

[Antonio the Sixth]

The hilarious part is that their Romney+1 poll was actually Obama's best performance in a month or so.

[Eraserhead]

The hilarious part is that their Romney+1 poll was actually Obama's best performance in a month or so.

Yep. That was their "okay, let's try to save face" poll. They had Obama down 3-7% in their likely voter poll for weeks. Can you say clueless?

[Meeker]

RAND is partially Internet-based, I believe. Probably time to adjust the "it's an online poll, trash it" mentality that many have adopted.

[Tender Branson]

New York should push Obama's margin to about 51-47.

So, any pollster with a 3-5% margin was pretty good.

[Fuzzybigfoot]

New York should push Obama's margin to about 51-47.

So, any pollster with a 3-5% margin was pretty good.

How many votes ar left in NY again? 

[YL]

RAND is partially Internet-based, I believe. Probably time to adjust the "it's an online poll, trash it" mentality that many have adopted.

"Online poll" can mean a range of different things.  The likes of Yougov, and I believe RAND, have a sample recruited via the Internet and carefully weighted to match the population demographically; this seems to work reasonably well most of the time, though there are some interesting differences in the UK between online and phone polls.  (For example, online polls tend to show higher support for UKIP for some reason.)  At the other extreme, you frequently get "polls" on local newspaper websites and the like, which are entirely self-selecting samples and hence essentially meaningless.  (IIRC there was one in the last UK general election which showed the Lib Dems winning Witney - if only!)  I think there are also "online polls" which aren't quite that bad but still with major problems with their sampling methods.

[Tender Branson]

New York should push Obama's margin to about 51-47.

So, any pollster with a 3-5% margin was pretty good.

How many votes ar left in NY again? 

1 million at least I'd say. Probably even 1.2-1.4 million.

[Eraserhead]

New York should push Obama's margin to about 51-47.

So, any pollster with a 3-5% margin was pretty good.

How many votes ar left in NY again? 

1 million at least I'd say. Probably even 1.2-1.4 million.

Think NY will end up swinging to Obama (vs. 2008)?

[○∙◄☻¥tπ[╪AV┼cVê└]

New York should push Obama's margin to about 51-47.

So, any pollster with a 3-5% margin was pretty good.

How many votes ar left in NY again? 

1 million at least I'd say. Probably even 1.2-1.4 million.

Wow, is that more than the heavily absentee ballot west coast?

[BlueSwan]

It's a bit strange, that throughout the last months of the campaign Obama was polling better in Ohio than he was nationally. Infact, his strong polling in Ohio was one of the main reasons why so many remained fairly confident of an Obama win even when he was trailing in the national polls.

Yet, Obama ended up doing significantly better nationally than in Ohio...

Or rather: The Ohio state polls were pretty much correct, while the national polls were generally way off.