PredictionsMock2012 Presidential Predictions - darthpi (D-PA) ResultsPolls
Note: The Google advertisement links below may advocate political positions that this site does not endorse.
Date of Prediction: 2012-11-05 Version:30

Prediction Map
darthpi MapPrediction Key

Confidence Map
darthpi MapConfidence Key

Prediction States Won
270 |
538 |

Confidence States Won
270 |
538 |

State Pick-ups

Gain Loss Hold Net Gain

Prediction Score (max Score = 112)

ScoreState WinsState PercentagesCD WinsCD Percentages


Alright everybody, here is my final prediction map. I see President Obama as very likely to win re-election, as I find it quite unlikely that the polls will be off by enough of a margin for Romney to win. While Romney has a marginally more viable path to a popular vote victory than an electoral college victory, I don't see either as particularly probable at this point.

My thoughts on a few key subjects

On the polls: I know a number of people feel the polls are likely to be off this year, including the Romney campaign (based on reports about the turnout model for their internal polls) and the poll unskewers crowd. While the poll unskewers are basing their argument around the notion that the polling companies want to see the President re-elected and are trying to depress Republican turnout (an absurd argument, in my view), there is a slight chance that the polls could be badly off this year, or any year, because they misjudged the composition of the electorate. This was likely a key reason behind the outcome in 1980, when Ronald Reagan was a favorite to win based on most of the late polling, but not by the kind of landslide margin that he ended up winning by - though an unusually large number of undecided voters making up their minds at the end probably played a large part that year as well. This year, however, I think the chance of the polling companies being badly off is less than typical, partly because there aren't that many remaining undecided voters or voters considering voting third party, and partly because of how the Wisconsin recall turned out. While Walker won that recall, the polls were pretty much spot on. If there was some sort of unknown factor in the composition of the electorate this year that the polling companies were missing, I would have expected it to show up in that election, and thus for Walker to have significantly beaten his polls (or for him to have come much closer to losing, if the unknown factor favored Democrats). While some sort of unknown factor could have developed in the last five months, I don't see any evidence that it has, and I can't really come up with a reason for why it should have.

If I am wrong: If I'm wrong about the polls, and Romney does pull off an upset victory, then the states he could pick up (in decreasing order of likelihood) are: Florida, Virginia, Iowa, Ohio (these are all very likely part of any Romney victory scenario), New Hampshire, Wisconsin, Nevada, Pennsylvania, and the 2nd Congressional District in Maine; I would note that Florida, Virginia, and Iowa could be for Romney even if the polls are only slightly wrong, hence why I have them listed as tossups. I find the probability of the polls being off so wildly that Romney wins states like Minnesota or Oregon so low that it isn't worth contemplating those scenarios.

If I am wrong AND happy: On the equally unlikely chance that President Obama beats his polls substantially, then the most likely states he gains from this map (in decreasing order of likelihood) are: Colorado, North Carolina, the 2nd Congressional District in Nebraska, Arizona, and (perhaps) Montana. As with Mitt Romney, any other states that I have President Obama losing are so unlikely to go to him that I don't really see the point in considering them.

Tossup states vs strong states: The states that I have labeled as tossups are the states that I would not consider it a substantial upset if the candidate I have losing ended up winning. The strong states are the states where I would consider such an outcome to be a substantial upset. In the future I might start using the lean category in my final prediction map as a way of representing states that could potentially flip if the polls are badly wrong, but I said in previous maps this cycle that I wouldn't use the lean category in my final Presidential map, so I'm going to hold to that commitment. For Iowa specifically, I decided to make a change at the last moment to swing status due to tightening in some of the polls. While ordinarily I might have dismissed those as just statistical noise, I think it is possible that there has been a move of some voters following the Des Moines Register endorsement of Romney. I still think President Obama remains a modest favorite there, but the likelihood of a Romney upset has increased from my previous map.

Early voting: One factor that is a bit problematic for me to account for in certain states is early voting. Because the polls have moved by a fair bit in some of the swing states since President Obama's low point about a week after the Denver debate, it is possible that early voting could diminish his chances relative to the current polling. Of particular concern for me is Colorado, which has one of the highest early voting rates in the country (if not the highest), and which experienced one of the more dramatic shifts in Romney's favor. While ordinarily I would probably have President Obama ahead there based on the current polling (though certainly at a low confidence level), I think the early vote is going to hurt him there by enough that Romney wins the state. Florida could also hypothetically be problematic for President Obama because of early voting, though the anecdotal reports (which are always of admittedly questionable reliability) I'm hearing suggest that he is actually doing fairly well in the early vote there, and is also doing a decent job of turning out infrequent voters. It is because of this early voting factor that Florida and Colorado are my least confident states.

Momentum: In many of the swing states, as well as in the national polling, it appears that President Obama does have some momentum that could be key in a close race. In Florida and Virginia in particular, the momentum over the past two or so weeks seems fairly clear and seems to perhaps be more than the national move, hence why I'm marginally more confident about Virginia than I might otherwise be, and why I'm willing to predict a come-from-behind win for President Obama in Florida. Colorado, on the other hand, has seen an essentially flat trend over the same time period: after rebounding from his lows, President Obama has not had sustained momentum there, something that I think does not bode particularly well for his chances. I would suspect that momentum in the states where we are seeing it suggests that the formerly undecided vote is breaking in President Obama's favor, rather than any significant move on the part of Romney voters. Whether this is primarily because they think he is going to win and are going for him in a bandwagon effect, or if they are reacting positively to his response to Hurricane Sandy, or they have decided for any number of other reasons I don't know, though I think all are having at least some effect.


LATE UPDATE: I've decided to switch Colorado back to President Obama. Early voting there started later than I thought it had, and he has had a modest if inconsistent lead since it did. While his lack of clear momentum in the state troubles me, I can no longer say he is the underdog there.


National Popular Vote

Barack Obama: 50.6%
Mitt Romney: 48.3%
3rd Parties: 1.1%


As with 2008, one final word:


Prediction History
Prediction Graph

Comments History - show

Version History

Member Comments

User's Predictions

Prediction Score States Percent Total Accuracy Ver #D Rank#Pred
P 2020 President /56 /56 /112 % pie 454
P 2018 Senate 32/35 23/35 55/70 78.6% pie 26 1 67T483
P 2018 Governor 32/36 23/36 55/72 76.4% pie 26 3 122T372
P 2016 President 51/56 33/56 84/112 75.0% pie 37 1 87T678
P 2014 Senate 34/36 26/36 60/72 83.3% pie 11 4 21T382
P 2014 Governor 31/36 17/36 48/72 66.7% pie 6 3 73T300
P 2012 President 56/56 45/56 101/112 90.2% pie 30 1 77T760
P 2012 Senate 32/33 22/33 54/66 81.8% pie 7 1 40T343
P 2012 Rep Primary 45/52 19/52 64/104 61.5% pie 42 - 25T231
P 2010 Senate 35/37 30/37 65/74 87.8% pie 26 1 5456
P 2010 Governor 34/37 27/37 61/74 82.4% pie 6 1 29T312
P 2008 President 52/56 44/56 96/112 85.7% pie 19 1 74T1,505
P 2008 Dem Primary 37/52 19/52 56/104 53.8% pie 15 - 58T271
Aggregate Predictions 471/522 328/522 799/1044 76.5% pie

Alabama Alaska Alaska Alaska Alaska Alaska Alaska Alaska Arizona Arkansas California California California California Colorado Connecticut Connecticut Delaware District of Columbia Florida Georgia Hawaii Hawaii Hawaii Hawaii Hawaii Hawaii Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Maryland Maryland Massachusetts Michigan Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Virginia Washington Washington Washington West Virginia Wisconsin Wisconsin Wyoming

Back to 2012 Presidential Prediction Home - Predictions Home

Terms of Use - DCMA Policy - Privacy Policy and Cookies

© Dave Leip's Atlas of U.S. Elections, LLC 2019 All Rights Reserved