I predict 80% for Obama.
It is again worth nothing that Silver’s “projected” EV is not actually what he is predicting Obama will win (rather obviously, since it’s out to a decimal place, and you can’t get 0.7 of an electoral vote) but is an aggregate mean of all simulated elections he runs.
In fact, if the election were to go as Silver’s projections *for each state *are, it would be 303-235 Obama. This is the map he predicts, albeit with very low confidence in many of the states:
http://www.270towin.com/2012_election_predictions.php?mapid=bbnr
I’ll go 81.4%.
[freeper]Of course you think Obama can get a fractional electoral vote. He’s your Messiah![/freeper]
I think the forecast will be up over 90%, if I understand what the forecast is doing. Personally, I’m still predicting 303 EVs for Obama.
I predict the standard error of the guesses for the aggregate results will be 4.
9pm EST – Obama / Romney
Electoral Vote: 314.4 / 224.6
Chance of Winning: 91.4% / 8.6%
Popular Vote: 50.9% / 48.2%
Romney Swing State chances:
FL = Romney 56% v Obama 44%
NC = Romney 77% v Obama 23%
Nebraska CD2/Dist2 = Romney 88% v Obama 12%
Everything else to Obama with Ohio = Obama 87% v Romney 13%
Actually FL is now Obama 52.5% with an adjusted polling average 48.4 to 47.9. Huh.
They must have added in more data since 9pm. That’s why I put a time stamp on when I pulled those number, they’re moving really fast now that it is within hours of results.
Do you think he might be adjusting for the voting kerfufle as drawing out more people who are angry about the polling issues? To flip from Romney to Obama is one thing, but it has flipped by 8+ points.
I definitely think FL will be a test of Silver’s processes.
Right now (10:30pm EST), he has Obama at 315.3 EV; 92.2% chance; and 50.9% PV.
FL numbers are still what was posted above.
Correction: FL = 53.2% Obama Chance w/ 49.9% Obama vote to 49.7% Romney vote at 10:35pm
So, this must be an effect of the clock running out. Ohio is now marked safe Obama w/92% chance Obama wins it.
I’m hoping to see some post articulating why his model altered (is altering?) FL to that degree in these final hours. Perhaps there are new polls he has access to that are not up on the sights we see yet?
There’s a new PPP poll (11/4) that has Obama ahead and is weighted more than the online polls. Then there are early voting results that may be added into the mix (but don’t have a line item on his display). There has to be a factor for the clock running out. If you add a factor for the barrage of news stories about polls being in accurate, the open/close/open and the people willing to wait on queue for 6 hours… it seems like there is a high ‘enthusiasm’ factor going on down there.
It would be great of Silver posted an explanation so that his process could be better understood. At the rate his numbers are moving, it’s looking like his numbers area heading at putting a stick in the sand prior to the opening of the polls tomorrow.
He’ll probably make one more post tonight and explain the huge move.
The Nov 6th cast, must inherently approach the NowCast and the NowCast has always been more volatile. With more polls coming in as the election approaches, certainty in the outcome probably increases a lot. A state may still poll with only a 1% difference, but the margin of error may be less than 1% with increased polling. That will increase the outcome prediction a lot.
The prior PPP was O+1 too.
While I wait to read Silver’s logic I offer this: that last tied Gravis FL poll states this:
Half have already voted and of those who have voted Obama is, per this poll, ahead by 5%. The other half that brings it to the tie may or may not come out but of note pretty bad thunderstorms are expected pretty widely across the state tomorrow, especially in GOP leaning portions.
Obama’s lead with female voters really comes through … Mother Nature loves him.
A few possibilities on Florida:
-
538 adjusts the state numbers based on the national polls as well as doing the reverse. Obama is now up by a point or so up nationally which will help in the state calculations.
-
Linked to the above, 538 may have a momentum variable if one candidate moves significantly near the end as Obama has clearly done in the last day or two.
-
There is a “state fundamentals” variable which favors Romney in Florida. Perhaps this slides to zero at the end of the race on the assumption that it is now baked in the poll numbers.
Silver has made one more late update - as of 11:29 PM ET on Nov. 5, the Forecast and the Now-cast are now the same with probability of an Obama win at 92%.
@ 12:10a Nov 6.
Obama has slid from 92.2% to only 92% as posted above. It looks like they added two polls to FL. These two were not there earlier:
InsiderAdvantage 11/4 47.0 52.0 Romney +5.0
Ipsos (online) 11/4 46.0 46.0 Tie
Those have drug Obama in FL down to 53.1% (vs 53.2% earlier), but the vote percentage numbers haven’t changed and FL is still leans Obama.
Well that’s it. It’s over folks. At this rate Romney will be up to a 90% chance of victory by December 27th or so.
This is just the break the McCain campaign has been looking for!