The curious incident of the presidential campaign in the night-time

by Doctor Science

The 2012 Presidential campaign can be explained with a single image:

Created by me at 270 To Win.

Huh? you say: that looks almost exactly like the final result:

From The New York Times' election coverage.

Yes. And yet, that first map is based on predictions made by David Rothschild and Chris Wilson at Yahoo's The Signal blog, in February.

With fewer than nine months to go before Election Day, The Signal predicts that Barack Obama will win the presidential contest with 303 electoral votes to the Republican nominee's 235.

This was not a fluke. On May 31, Wiesman and inuyesta at Some Disagree prepared their predictions:

In other words, they agreed -- correctly -- on 49 out of 50 states (and also on the District of Columbia), and they agreed that Florida was a true tossup.

Another data point: on July 26 , Nate Silver's electoral vote simulations showed:

That big peak is for 332 electoral votes, the total we actually arrived at. This is one reason for all the jokes about "Is Nate Silver magic?" -- because he apparently predicted the outcome of the election 4 months in advance.

But it wasn't just him, and it wasn't only since mid-summer. I've look at The Signal's data in more detail (Google spreadsheet link). When I take their state-by-state February predictions for Obama's share of the vote, multiply them by the number of actual votes cast in each state in November, and add them all up I get a predicted vote share for Obama of 50.18%. His actual share? 50.37%. So 9 months ahead of time they predicted the popular vote to within a quarter of a percent.

Now *that* is getting to be creepy. It's as though we had an entire Presidential campaign and nothing happened.

All those stories you're seeing about how "Obama won the election because he did X, Romney lost because he didn't do Y" -- they are, statistically speaking, fairy stories. There is no evidence so far that *anything* made a difference, not even the choice of Romney as the Republican nominee. The whole bloody hellishly expensive campaign did nothing.

Here's another crucially informative graph, this from Votamatic:

Those lines are what we technically call "flat". The uncertainties decreased as election day neared, but the trend line was unchanged. Statistically speaking, nothing happened.

Seth Masket noticed, too, that I spent $6 billion and all I got was this lousy economic retrospection election, but he says

So was this all a waste of everyone's time and money?

No, not really. A large part of the reason the results are so close to the line is that the campaigns were pretty well matched. Spending, if you include independent expenditures, was remarkably similar across party lines. The campaigns largely identified the same states and the same demographic groups as persuadable and they threw everything at them. And they were both quite effective in turning out their bases.

And if the amount of money that went into this irks you, keep in mind that this is what it takes to inform and mobilize a large electorate.

This can only be true, I think, if all the efforts to inform and mobilize were "baked in" to the February model. In particular, it means that, if the Obama campaign's GOTV organization was excellent and the Romney campaign's was inadequate -- as all seem to agree -- this was only to be expected. Good GOTV by the Democrats and poor GOTV by Republicans did not have a significant effect on the national-level results, they seem to be built into the data used to build the model.

I suspect that neither Romney's winning the Republican nomination nor the fact that the economy slowly improved changed the game because both events were undramatic and likely, and so were also built into the model.

I find it really emotionally difficult to let go of the idea that things that happened during the last 9 months *mattered* for the election, but the math is staring me in the face. I'd say we can *certainly* rule out the following factors:

• Hurricane Sandy
• Benghazi
• Romney's choice of Ryan for VP
• Aken and the other rapologists
• the debates
• the conventions
• the ads
• the Supreme Court decision on Obamacare
• any other development that pundits have called "game-changing".

We can tell that there were no game-changing developments, because the game never changed. Which means that something like 99.9% of the narratives going around about the campaign are disproven bunk. One of the few observers who got it right was Wiesman, who wondered in July if this was the least volatile election ever. I think he was completely right when he said then that People are pretty much settled on who they will be supporting in November.

Most people -- and pretty much all journalists -- never noticed that nothing was happening during the campaign, because the sound and fury had been cranked up to record levels. We continued to talk about it as though it was a horse race, even though that's clearly the wrong metaphor.

So what's a good one? Was the campaign like a race with this kind of horse:

Coin-operated kiddie riding horse. My brother and I loved these things. Photo by Furry Eyebrows, no indication where it was taken.

Noise, excitement, bouncing and jolting -- but no actual movement. In this metaphor, almost all the money and human effort of the campaign were essentially burned up, wasted uselessly.

Or was it this kind of race:

Google found this illustration of the Red Queen running with Alice in the Jerusalem Post, but I can't find out where they got it from. I guess it's a Chinese edition of Through the Looking Glass, but I don't know when or where.

"Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!" That is, was the situation a high-energy stalemate: each campaign put out an enormous effort, which was precisely counterbalanced by the other campaign. Net movement: zero -- but if either campaign had slackened off, the other would have steamrollered to victory.

Comments

[JustMe]

So pollsters employ their inaccurate models early in order to......????????????????????????

This is basically projection on Brett's part. Rasmussen, for example, seems to have a "house effect" that leans to the right through most of the election season and then tightens up in line with most other polls the week before the actual election. Brett simply assumes that all other polls do the same thing.

[Brett Bellmore]

Mostly to save money, I suppose. It's expensive to run a proper poll. And if you're applying filters to try to increase the accuracy, you have to increase the number of people you contact, to compensate.

I mean, you ARE aware that the polls several months out from an election aren't conducted in the same way those several days out are, right? They don't pretend otherwise. There are live body polls without callbacks. Registered voter polls. "Likely voter" polls, with different models for who is likely to vote. Escalating efforts to get a reply, so your response rate won't be absurdly low.

There's nothing necessarily nefarious about their doing inaccurate polling in August. But you should be aware that's what they're doing.

[Brett Bellmore]

I assume they do the same thing, because the other polls DID show Obama way ahead earlier, and then tightened up. You can assume that the race closed, or that the polls got more accurate. Considering they went in steps from warm body to registered voter to likely voter, I think the latter interpretation reasonable.

[Duff Clarity]

That was, in case you didn't notice, in response to earlier polls showing not a close election, (Such as we actually experienced.) but an Obama blowout

No, it wasn't. It was in response to wj saying about the pick of Ryan and the potential for that pick to turn off voters due to Ryan's budget plan, "But in a close race, do you really want to kiss off anybody you didn't have to?"

Then you made the first comment I quoted. After that Phil was all, "LOL Rasmussen" and wj said, "Nobody who's paying attention thinks Romney is a lock to win. Even people who are convinced he will win admit that it will be a near thing."

Didn't I say that the polls would tighten as we got closer to the election

Here's a poll tracking average for the election. On August 12 when you made the comments, it had the average poll at 45.9% Obama against 44.8% Romney. On the last entry it had before the election, it had 48.2% Obama vs 46.7% Romney. That is not tightening up.

the polsters (sic) started employing their accurate models to avoid being embarassed (sic) on election day?

Now you may be on to something. Let's see what Rasmussen was up to the week before the election.

Wow, they had Romney by 4% or so a week out (no not in August, a week out), then their polls got closer to other pollsters results by election day.

LOL Rasmussen.

Everybody in that thread was talking about the election being close. They weren't talking about an Obama blowout. Somebody is in an echo chamber around here. I hope they stay there along with all the Republicans. If you can't face facts, you can't win elections.

[Duff Clarity]

Oops, lost the italics.

That was, in case you didn't notice, in response to earlier polls showing not a close election, (Such as we actually experienced.) but an Obama blowout

Didn't I say that the polls would tighten as we got closer to the election

and

the polsters (sic) started employing their accurate models to avoid being embarassed (sic) on election day?

were Brett's comments.

[Phil]

BTW, since "ZOMG zero Romney votes in some precincts = TEH FRAUD" is fast becoming an article of faith on the part of conservatives, I went back and checked the Cuyahoga County results for the 2008 election. There were 19 precincts reporting 0 votes for McCain, and one precinct reporting 0 votes for Obama. It was apparently considered unremarkable at the time.

[Julian]

"It was apparently considered unremarkable at the time."

Is it just me, or has hypocrisy in public debate become more tolerated at precisely the same time as exposing it has become easier and easier?

I'm relatively young so maybe this only seems new.

[Duff Clarity]

It was tolerated during Nixonland. It has certainly become easier to expose, but I don't think the level of tolerance for it has changed.

It's always OK when your guys do it.

[ral]

Phil, watch your spelling! It's "zOMG" not "ZOMG".

[hairshirthedonist]

Since this thread has died down, I'm going to go way off topic. I received and e-mail from my sister that somehow aliased my e-mail address to my handle here. I didn't notice it until I replied to the e-mail this morning. I don't use a real e-mail address when posting here. Any techies around here have a possible explanation for that?

To my knowledge, I've never associated any login for any service with my e-mail address using my ObWi handle as a username. I did have a username associated with my e-mail at one time for posting comments on sites run by my e-mail provider, but it wasn't the handle I use here.

Creepy, no?

[wj]

HSH, security of anything posted on the Internet is somewhere between weak and non-existant. Weak is usually enough, because things like e-mail addresses are not much use to hackers. Low resale value ==> low rate of hacking.

But if you ever put anything out anywhere, it is available somewhere forever. And as software for crunching "big data" gets better, finding connections between things gets easier. Even if the connection was only ever mentioned once, and was indirect (i.e. several steps).

The sad part of the creepy is that there are an awful lot of teenagers posting stuff which will come back to haunt their future job searches. And they either are blissfully unaware, or simply don't believe what they have been told.

Privacy -- rapidly becoming, if not an outdated concept, at least a near impossibility.

[Blackhawk]

"Any techies around here have a possible explanation for that?"

It's that yottabyte processing power in the NSA facility. Your sister should be careful using official means for personal ends.

Oh yeah, that's right, according to that Turbulence these things are all so mightely encypted that what you're telling us is impossible.

[Turbulence]

Awww, I have my very own thanksgiving troll who drags my name into random discussions so he can lie about me! Blackhawk, I know I humiliated you and demonstrated that you just make shit up because you're pig-ignorant and delusional, but you don't have to obsess over me...really, you don't.

[Julian]

Is there a reason Blackhawk isn't banned?

[liberal japonicus]

Because it's not worth the bother?...

[Phil]

Pretty soon he'll start trying to imply that you are Teh Gay, Turb, and that's when you know you've really got a friend forever.