On this page, Will and I have an ongoing, gentleman’s disagreement about the proper way to measure the election. I think the state-by-state polling is a better way of looking at what the outcome will be, and Will points to the national polls. Here, we are pleasant and civilized about such things. On Twitter, not so much. I can’t read Twitter anymore.
After all, this disagreement was at the root of what looks in retrospect like a little bout of “media bias rage” from the rightosphere that took place yesterday and the day before, aimed at statistician Nate Silver and his invaluable blog, fivethirtyeight. Apparently math and poll results are somehow biased in favor of Democrats.
Now, if I’m wrong, then the race is too close to call because the national polls are all very close and most are within their margins of error. But if I’m right, then Nate Silver’s pithy justification for giving President Obama such high odds for re-election is dead on:
As far as I can see, there isn’t much more to think about than these four words. If you look at the states where it just plain isn’t competitive anymore, you get Obama with a foundation of 253 electoral votes† to Romney’s 206.‡
The Electoral College math is simple from there. Governor Romney must win both Florida and Ohio in order to be the next President. But he is trailing in every five out of six of those swing states. He’s more likely to get Florida than Obama, but of the others, really, only Colorado is competitive. If Romney manages to win every one of these six states but Ohio, he still loses in the electoral college, by two votes.
What I think will actually happen is that Romney will win Florida and Colorado, and lose Iowa, Virginia, New Hampshire, and fatally, Ohio. That result is 294-244 in favor of the President’s re-election. I can’t say I’m terribly exercised about this. While there is lots — lots — to criticize about President Obama and I’m averse to the ‘six year curse’ by which all two-term Presidents get bogged down in their sixth continuous year of administration, I can’t see President Romney doing any better on the things that I both care about and that the President can influence — other than Supreme Court nominees. Here, I’d expect Obama to nominate Justices more liberal than himself, or Romney to nominate Justices more conservative than himself. And the fact of the matter is that except for the new Second Amendment cases I expect to see percolating up over the next ten years or so, the kinds of Justices Obama would nominate would be more expansive in their interpretation of individual rights than the kinds of Justices Romney would nominate.
By the way, in what I’ve called the most interesting Congressional race of the year, Arizona’s Ninth District, there seem to be hints that Democrat Kyrsten Sinema is narrowly leading Republican Vernon Parker, and more than a million and a half dollars were spent on that race last week alone, with that same amount of spending currently underway in the last days of the campaign. And that money is paying for some very questionable commercials. The anti-Sinema ads look more hamfisted and stink of tonedeaf desparation more than the anti-Parker ad, in my opinion. But I can’t find good polling data anywhere, unless I pay for it, and come on, I’m not that interested in something that in four days, we’ll have a definitive result for anyway.
And with that, I may well find it within my willpower not to blog about politics any more until the election. So there.
___________________________
† California, Connecticut, D.C., Hawaii, Illinois, Maine, Maryland, Massachusetts Michigan, Minnesota, New Jersey, New Mexico, New York, Oregon, Pennsylvania, Rhode Island, Vermont, Washington, Wisconsin.
‡ Alabama, Alaska, Arizona, Arkansas, Georgia, Idaho, Indiana, Kansas, Kentucky, Louisiana, Montana, Mississippi, Missouri, Nebraska, North Carolina, North Dakota, Oklahoma, South Dakota, South Carolina, Texas, Tennessee, Utah, West Virginia, Wyoming.
I’m going to cheat and wait until Monday. My Spring prediction had Obama at 284, last year’s prediction at 293 (I can’t remember what states). There’s a good chance Monday’s map will actually have him at over 300 (I’m leaning towards Virginia-for-Obama). But the relative closeness of both national and state numbers, and the sheer unknowns of who shows up on a particular day to cast their ballot, lead to greater uncertainty on my part than Silver’s numbers presently suggest.
I’m wondering if the Bradley Effect will reappear after a brief hiatus.
Obama does play basketball, but not at that level.
Thanks for a great post. I don’t know how there is any sort of a debate over national vs. state polls– since electoral votes are at the state level, the best prediction would have to come from well-sampled state polls.
“I don’t know how there is any sort of a debate over national vs. state polls”
Since the debate only seems to ever occur immediately before presidential elections (and since the party that is behind is invariably the one making the argument), I suspect that the reason is that one side doesn’t always like what those polls say.
We’ve learned that the rightosphere has the same grasp of statistics that they have of climate science and evolutionary biology.
It’s not just them. Lots of people don’t get statistics. It’s just generally not that important in their day to day life, so it doesn’t matter or they just accept it or get angry or in denial over tiny, personal incidents.
I’ve been a little surprised at how many college educated folks (and our pundit class is, by and large, college educated and quite thoroughly at that) don’t get the difference between “stastitical outcomes” and “predictions”.
I’ve found the whole thing really fascinating, though. Watching pollsters move (over the last decade or three) from straight up sampling into the weeds of averaging and Bayesian analysis. It’s some really exciting stuff.
I think the funny thing about the anti-Nate stuff is — he’s a stats geek. He’s not in it to shape elections or write narratives. He wants to see if his model works. He’s in it for the geekery of it. (Even if, yes indeed, he also votes).
Would you say 80 percent of people don’t get statistics?
/Ducks
No.
It’s higher than that.
lol. 🙂 One of my favorite t-shirts reads “There are 10 types of people in the world. Those that understand binary, and those that don’t”.
I’ve also seen it as “Those that get this t-shirt, and those that don’t”.
I’m actually quite, quite horrible at math. I lack..something…in my brain, some ability to handle pure abstracts. If you hold my hand, I can follow you through complex math (my ability to work solo craps out somewhere around differential equations. I can differentiate and integrate with the best of them, although my algebra is sloppy), but only a handful of areas can I work alone — mostly stuff that’s big in computer science, and it’s not math skill that does it — it’s experience.
It’s funny — you’d think being crappy at abstract math would make me a bad coder, but…strangely not. I can hold state machines in my head, make code dance for me, but its because it’s not abstract to me.
So I find Nate fascinating not because he’s predicting an Obama win, but because of what he did for baseball and how he’s extending it to political polling and how he’s tackling the numbers.
I could wish he wrote more papers, and had less behind a paywall.
There’s a slight problem with that pseudo-geekism about 10 types of people who understand binary. If they did understand binary, the jokester would only need one bit.
Put it this way, when programmers teach their children to count, the wee ones start with Zero. Then One.
2 in binary is still 10.
Well, yeah. But the data is more amenable to a boolean typing than integer. There aren’t three types insofar as you either understand binary or you don’t.
I’m shocked at the number of preschool teachers or other early childhood educators who don’t teach zero. I make sure that my kids know what that means.
Eh, most of of math is setup — these days — along with how the average child’s brain develops. There’s no point teaching some concepts before the brain generally has the wiring to handle it.
Zero is, I think, a fairly difficult concept for kids to grasp until they can grasp abstracts. There’s a reason we did math for thousands of years without it. 🙂
You don’t really bring in zero until you bring in the number line.
It’s weirdly different and unique this time too. The usual slag on polling (from anyone who is behind in the polls) is that ‘I (we) don’t worry about polls’ or the more simply ignorant “I don’t see how 500/1000 people can determine the opinion of 300 million people”. This time though, with the gay-baiting of Silver, it’s gotten all Kultur War-y.
Michael Barone is predicting 315 electoral votes for Romney.
I will bet anyone a beer that Romney does not get 315 (or more) electoral votes. I’m giving two for one.
I’d need it to be two cases to one can before I’d even consider taking that wager. Barone claims to be going with electoral “fundamentals,” which to me looks to boil down to, “but here’s how it’s always seemed to me to happen before, so obviously that ‘data’ stuff must be all wrong.”
Hmm. Bet he’s going with “no President has ever been elected with unemployment above X”.
I think that’s a pretty bad way of looking at it — it doesn’t match people, for one, and it’s a hard static line on something that’s almost certainly dynamic.
Dynamically, you’d want to look at trends. Which WAY is unemployment heading — up or down? Which way is the economy going? Up or down?
Just from my gut, I’d say people don’t vote on what the unemployment number is — they’d vote (insofar as economic conditions dictate their vote) based on whether they thought things were getting better or worse (and probably with a lag — stuff getting better the last week doesn’t make up for months of it getting worse).
But 315? Jesus. Romney wins 280 or so, sure, okay i can see that. That’s not..ludicrous. But 315? Where’s the other 35 votes supposed to come from? he’d have to run the battleground tables, wouldn’t he? Including PA?
Barone includes Pennsylvania, Ohio, and Wisconsin. Don’t think so. I too could see The Mittster eking out a narrow EC victory, but not the landslide some on the right are predicting. Delusion has clearly set in.
Child’s play. Unskewed Polls has Romney winning Oregon.
That was so much fun to say, I’m going to say it again:
Unskewed Polls has Romney winning Oregon.
http://www.unskewedpolls.com/unskewed_projection_2012%20president_02.cfm
Even The Conservative RCP has Obama winning 290 to something.
If Romney wins, I don’t think that it can be without some degree of people lying to pollsters.
359 electoral votes for Romney?! Surely that’s a parody site?
“Surely that’s a parody site?”
No, which is what makes it a personal favorite of mine. But I think the Red Oregon is far better than the 359.
For those that don’t know much about Oregon, allow me to share a brief anecdote from lat week:
One of the things my wife and I do every election is to sit down with the voters guide that the State of Oregon provides to each registered voter along with their ballots. I’m pretty sure that most states have something similar, but on the off chance that they don’t a voter’s guide is a magazine made on cheap, ink-smearing news stock where the campaigns for every person on the ballot send in a one-page pitch for their candidate. These pages can include resumes, bios, endorsements, or arguments on why you should give that last rose to only them.
Last week when we sat down with a laptop (we used the online voters guide this), we discovered that most of the Republican candidates for state office didn’t even bother to send in anything for the state voters guide.
Yes, Oregon is so predictably and reflexively Blue that even our state’s own GOP finds it too depressing to submit material for the voters guide.
2-1? Cahalan, you cheap bastard, Ladbroke’s is offering 16-1.
http://sports.ladbrokes.com/en-gb/politics/us-presidential-election/2012-us-presidential-race-e212304268-m220922684
I closed my little AI forecaster when the Nov RBOB contract came in. But the East Coast gasoline situation has it all in a cocked hat. My model proved perfectly useless: this disaster drove the RBOB / pump price spread into the stratosphere. Before Sandy, it was giving Obama a slight edge, but of no statistical significance.