Tuesday, November 04, 2008

Obama is going to win

A month ago, I tentatively predicted an Obama win. I don’t believe in ‘jinxing’, so all I risk here is making myself look stupid. Obama is going to win. (I’m not the only one saying this: I’ve cunningly positioned myself inside a giraffe enclosure before sticking my neck out.)

First, and briefly, the national polls (all numbers via RealClearPolitics). Here are the averages of Obama’s leads in the Rasmussen, Gallup, Zogby and Hotline daily tracking polls for the last three weeks – he has consistently been ahead, and has stretched his lead in the last few days:


But it’s all about the Electoral College, of course. You need 270 votes to win. How are they both doing?

Obama has been ahead in all the states John Kerry won in 2004: Oregon, California, New York, Massachusetts, Connecticut, Illinois, Hawaii, Vermont, Rhode Island, Delaware, Maryland, DC, Maine, Pennsylvania, Michigan, Minnesota, New Hampshire, New Jersey, Washington and Wisconsin. All these leads have been comfortable, although in the last week Pennsylvania has narrowed – let’s come back to it later, so without it that’s 231 electoral votes for Obama.

Of the states Bush won in 2004, McCain has been more or less solidly ahead in Idaho, Alabama, Tennessee, Alaska, Kentucky, Texas, Mississippi, S Carolina, Arkansas, W Virginia, S Dakota, Kansas, Louisiana, Wyoming, Nebraska, Utah, Oklahoma and Georgia. In Indiana, Arizona, N Dakota and Montana, Obama has narrowed McCain’s earlier lead substantially. Obama had been leading narrowly in N Carolina but McCain has very recently edged back ahead. Missouri has been swinging back and forth for a month and is now neck-and-neck. Let’s give all of these to McCain (I’m being a touch generous to him here), for a total of 200 electoral votes in the bank.

Iowa has gone strongly behind Obama. New Mexico has gone fairly well for him, as has Nevada. That takes Obama to 248 electoral votes – 22 short of a win.

What’s left? Five states. Here they are, with electoral votes and averages of the latest polls:

  • Florida: 27 ev, Obama +1.8%
  • Ohio: 20 ev, Obama +3.2%
  • Pennsylvania: 21 ev, Obama +7.6%
  • Virginia: 13 ev, Obama +4.3%
  • Colorado: 9 ev, Obama +5.5%

Florida alone, or any two of the others, will do it for Obama. It looks good for him. He could well sweep them all.

So: what if the polls are wrong?

It would take a lot of polls, by a lot of different pollsters, being wrong for McCain to really be ahead. In 2004, the averages of the final polls were fairly accurate:

  • National: Bush +2.0 (result Bush +2.4)
  • Florida: Bush +0.6 (result Bush +5.0)
  • Ohio: Bush +2.1 (result Bush +2.1)
  • Pennsylvania: Kerry +0.9 (result Kerry +2.5)

Florida was notably off, but McCain needs the polls to be quite a bit wronger than last time. If they are, Obama still has a decent cushion: he can drop five points off his lead in every state and still win with Colorado and Pennsylvania. I truly don’t think that race is biasing the polls towards Obama, and I’m not sure what other new factor might.

Finally, there’s one other way the polls could be wrong. They might be understating Obama’s support. They mostly survey so-called ‘likely voters’, who are generally defined demographically based on previous turnout among various social groups. But if Obama’s candidacy really can get young people and African-Americans to the polls in better than the usual low numbers, then he’d be on for a landslide. And the signs are that turnout will be higher than usual.

Barring massive electoral fraud, Obama is going to win. I’m confident. But rather more than my reputation as a seer is on the line, of course. And my fingers are still crossed...

3pm Update: A few extra polls came out this morning. They put Obama ahead by 2% in Ohio, 9 and 10% in Pennsylvania, 1 and 3% in Florida, 4 and 7% in Virginia; N Carolina and Missouri are tied; and nationally, Obama is 8, 9 and 11% ahead.

4 comments:

Anonymous said...

Florida: Bush +0.6 (result Bush +5.0)

With that margin of error he really might be only ahead in two of the states he needs to win. Still, I think you're right although the story might be about how it's amazing that there's still any doubt. And it might not turn out to be a landslide, with all the stuff about mandates that entails.

And, of course, we all remember what The Simpsons taught us.

But I think talk of electoral fraud is silly. Would that be the same electoral fraud in the New Hampshire primary? If Obama loses and the second American Civil War breaks out it'll be your fault. Maybe you should turn around three times and spit on the ground, just in case.

Tom Freeman said...

Polling for primaries is much trickier than for general elections - the limited electorate and the likelihood of turning out among different sections of it is hard to gauge. And there is a much bigger polling consensus to upset this time.

I don't expect "massive electoral fraud"! Although, that said, the fact that elections are run by state party political appointees, who have a lot of leeway in what they can do, means there's a lot of scope for perfectly legal yet dodgy shenanigans (on both sides) all over the place. US electoral practice is not brilliant by developed-country standards.

Andrew R said...

As far as the Electoral College goes, fivethirtyeight is predicting 341 votes - that's based on crunching all the polls from all the states. If McCain's going to win, it'll be because several polls in several states are badly wrong.

The biggest oddity for me in US electoral practice is that voter registration is not automatic, nor handled by the state. It's a purely voluntary process, involving varying numbers of bureaucratic hoops. Hence the lawyers lining up challenge voter registration in marginal districts.

Matt M said...

I'm going to go out on a limb and say that I reckon he'll win it. :-)

Now let the crushing sense of disappointment, sliding gradually into bitterness, begin!