confident

[cos]

Ever since mid-March, when it became clear that Barack Obama would be the Democratic nominee, I've been cautiously confident that he'd be elected. Today, I think it's time to change that to "confident".

It's now less than a week before election day, which means asking people "who would you vote for if the election were held today?" in a poll is relevant to predicting how people will actually vote next week. Currently, there are 286 electoral votes in states where Obama has led by more than 5% in multiple recent polls and has been over 51% in the pollster.com average for at least a week. (It takes 270 to win)

Obama can afford to lose up to 16 EV worth of states from the "over 51%, greater than 5% margin" set and still win. And that set of states does not include Ohio or Florida. The only state in the set that's large enough that isn't really safe is PA, where Obama leads by only ~9%-11% in recent polling averages; other than PA, he'd have to lose at least two states from that set to go under 270. And of course winning Ohio or Florida would make up for it, and he's been leading in polls in both of those. But Obama wins without either Florida or Ohio as long as he holds at least all but one of the states where polls show him stable at above 51% and more than 5% above McCain (as long as that one isn't PA).

Also, of course, this confidence presumes:
- No completely unexpectable major event happens in the next few days (chances of that: very low)
- Obama's campaign continues at the pace it's set so far, including fieldwork (I'm confident of that)

If neither of those changes, I think it very likely that he will win.

P.S. Of the "above 51% and margin over 5%" states, CO is the only one where the stable margin is below 8%. If you only count states where Obama has been above 51% and has more than an 8% lead in stable poll averages recently, that's 277 EV.

Comments

[lil-brown-bat.livejournal.com]

The only possible flaw I can see in your numbers, and I stress it's only a possible flaw, is the discrepancy between "all voters" and "likely voters". If your data is based off the former rather than the latter, there's a chance of a very different outcome. However, even if that's the case, there's good reason to believe that the gap between "all voters" and "likely voters" isn't as meaningful in this election as it usually is (particularly given that "likely voters" is largely based on past voting history, and that the Obama campaign shows every sign of having successfully mobilized a group of voters (many but not all young) who haven't voted in the past).

I found out yesterday that I need to be in Boston on Tuesday, so I quick called up my town clerk and voted absentee yesterday -- none of this "I live in Massachusetts, my vote doesn't count" bullshit for me. I also called my brother in Colorado and told him I was counting on him to do his part to turn Colorado blue. He thinks it's going to happen.

I'm also concerned about suppression of democratic voting. I have some hopes that the awareness is higher among the voting population, and that both election officials and political operatives know that the level of scrutiny is higher. I also have hopes that the average voter standing in line at the polls is going to be determined enough to stand there all day if need be. I know I would.

[cos]

Different pollsters use different models, and some are "likely voter" while others are "registered voter". They also use different demographic and party ID corrections. pollster.com averages them in what I believe is a statistically sound manner. If the pollster.com running average stays stable above 51%, then I think it's fairly reliable even if some pollster's models aren't as good as others.

Vote suppression is happening, but it will be lower this year than in 2000 and 2004. In particular, Florida has a governor who genuinely wants people to vote this time around (and he's a Republican!), and Ohio has not only a governor who wants that, but a secretary of state who's actively working to make that happen. Both states have Republican party vote suppression efforts again, of course, but this time state government is trying to limit the damage rather than cooperate. Also, both states kicked out their diebold touchscreen machines.

CA an NH have some of the best secretaries of state on these issues, but of course Obama was gonna win CA anyway, and in NH this is no change from 2004, and vote suppression was not a big problem there then either. But OH and FL are greatly improved since last time.

Colorado, unfortunately, has a yucky Ken Blackwell style secretary of state who's been trying to come up with excuses to deny voter registrations. The Democratic Party in Florida has been very good about beating this stuff back, but if vote supporession plays a major role in swinging a swing state this year, I think Colorado is the most likely suspect.