CFAR has done a nice job of evaluating the accuracy of the various pollsters and poll aggregators in the 2012 election, and the results are quite revealing. Some of the biggest takeaways:
1. Despite frequent claims from the Right that the polls were biased in Obama's favor, they were in fact biased in Romney's favor;
2. Rasmussen and Gallup, probably the most cited and most prolific pollsters, were among the least accurate and most biased pollsters in 2012. In fact, they were THE least accurate polling firms, with NPR placing a close third. All three had a strong pro-Romney bias;
3. Yougov and PPP were the two most accurate pollsters, with the former winning with a very small margin of error. Both had a very slight pro-Romney bias;
4. Most of the poll aggregators were very accurate. Nate Silver gets all the press, but others were close or better depending on which metrics you look at.
Was Nate Silver the Most Accurate 2012 Election Pundit? | Center for Applied Rationality
I disagree with the assessment that they were biased in Romney's favor...
We saw two major swings of momentum in the final month of the election... One after the first debate... Then a second after Sandy...
As of election night most polls had it just about right... But a week to two before that, all of their polls showed a separate story...
So, if we have learned that the polling aggregate sites are accurate, we have also learned that there was the momentum in Romney's favor, then the momentum swung back in cooincidence with Sandy (and the Christie praise as well as the Bloomberg/Powell endorsements, and Trump opening up his yapper)...
Nate Silver has had all the press, so much that people are approaching Chris Matthews praise of Obama status in thier admiration... However, as you said, across the board the polling aggregate sites had it... a close popular vote, which had swung in Obama's favor following Sandy, and an electoral vote which had a few close states, but was in Obama's favor...
We've still got provisional ballots to count, fraudulent votes to eliminate, and military ballots which won't ever get counted... so when we go awarding accuracy by minute percentages of pull fluctuation... it seems best to wait until that resolves itself and the final result tallies are reported by the states...
Especially, when the polling results represent Registered Voters, Likely Voters... as in existing votes...
Obama won huge with the youth vote... as in people who would'nt have been counted in polls...
Obama also won big with the same day voting registration states (which seems highly corrupt IMO, the process lends itself to fraud)... So same day registration states seeing a swing in votes which turned the result in a certain direction also would've been something out of the scope of the polls...
We all know PPP, DailyKos/SEIU/PPP, etc. have a huge Democratic lean... so them being able to accurately predict a Democratic leaning result is better... but... I don't think that's what occured... What i think, is there was a .5-1pt swing in momentum towards Obama in between the last released polls and election night... which is why the last released polls showed a 50-47 Obama advantage... and that .5-1pt swing in momentum in Obama's direction made the more accurate polls off by 1-2pts...
So as you can see, they have the "accuracy" line on those 3 polls... but that "accuracy" line is also from polls conducted a couple days before the election... not on election day... as the swing in momentum was underway... If you don't believe there was momentum... You're not paying attention...
Following the polls by date, you can see a huge shift in momentum following Sandy... the Halloween storm... that landed on the 28th... in those polls theres a swing from a tie pre-Sandy, to big Obama swing, to a lessening of that swing...
However, there's also the other swing, of bandwagon fans who want to support a winner... at that point all the polls point to an Obama win, so everybody rushes out to participate in it... and that widens the lead...
I would bet the real status of the election was right in between the massive group of other polls... that we DO have a liberal lean on the polls... but late momentum towards Obama from "climate change" after Sandy, and with key praise from Christie, endorsements from Powell, and Bloomberg, etc. have made the result jump the way the liberal leaning polls had inaccurately overstated the election to be...
I would even say, in defense of Gallup, who are already taking the brunt of criticism for this... They had a poll that was 5pts Romney favor from 2 weeks before the election, then down to 1pt Romney favor from 11/1-11/4 which shows they even picked up on the momentum swing in Obama's favor...
RealClearPolitics - Election 2012 - General Election: Romney vs. Obama
If you follow along there... you can see the fluctuation in the polls... and the general swings in trend... looking at both the individual polls and the composite average...
Romney was up in polls when he;
- won big in early primaries
- when he sealed up the nomination
- after the convention
- after the debates, especially the first 1
Obama was up,
- out of the gate
- as the Republican primaries got ugly
- when he pandered to gays and illegals
- after the convention
- after the Sandy response
So to suggest all polls were static, and that all polls predicted this race the way they had it at the final poll they released, when they were all over the map as far as release date and individual poll results is ridiculous...
This also points to how media can influence elections... At the debates, the only time when people saw the candidates face to face without media involvement, most people thought Romney was the better leader, and polls reflected that... However, when you look at the times when the advertisements and the national news media controlled the topics discussed... it's hurricane sandy proves that there is climate change (when it doesn't), its that Obama is leading in polls and likely to win, etc.
Then... the next day after the election, it's "Whoa watch out for the fiscal cliff!"... that suddenly, after not getting any mention during the election cycle, becomes the big priority that the country has to face...
Do you think the result would've been the same if they reported about this fiscal cliff the way they are right now prior to the election?
