As long as I have been writing about it, and cataloguing about 3000 polls from it, it is pretty stunning to me to say the following five words: the 2012 elections are over.
The confetti has now been swept up. The votes (well, most of them anyway) have been tallied. All that is left are the inevitable post-mortems.
This post-mortem comes from somewhere between anger and amusement. That's because the dominant theme is how much of the conventional wisdom about this election was ultimately, and profoundly, disproven by the judgment of the voters on Tuesday night.
For most of the past month, when writers on the left (here and elsewhere) criticized much of the media CW on this election as errant, those critiques were roundly dismissed, and even met with a level of annoyance. From my perspective, the most glaring example was the high-intensity hissyfit that many establishment journalists had on Twitter when those on the left accused them of basing their reportage on the perpetuation of campaign narratives, rather on the available evidence.
What the election proved, though, is in spite of their hypersensitivity, we were right and they were wrong. The swing states were never truly in doubt. The key swing states in the South were not lost to Obama. There was no "Mitt-mentum" from about October 15th onward. This was not a center-right electorate, or even really a centrist electorate.
In the final analysis, with a doff of the cap to former NFL coach Dennis Green, the 2012 election was what we thought it was. The reason folks like Nate Silver (and, for that matter, Markos Moulitsas) wound up looking like oracles is because they constructed their narrative around the data, and not the other way around, like far too many people did.
What follows are three of the more persistent bits of conventional wisdom, and how the results on election night ultimately, and forcefully disproved them.
(Continue reading below the fold)
MYTH #1: Within two weeks of the election, the presidential election was tied.
This one comes down to a simple definition of terms, but that didn't stop a huge proportion of the public conversation on the election from being errant.
If you look at a simple average of the national polling, then...yes...from about October 8th onward, the trial heats between President Obama and Mitt Romney were about even. Indeed, in the national polls, neither candidate had an edge of more than 1.5 percentage points from that point onward.
Average of released national polls, 2012 presidential election, October 1-November 5:However, as everyone who even has minimal knowledge of presidential politics already knows, that's not how you elect a president. And the polling on a state-by-state level was quite consistent. Which is why Nate Silver never had Mitt Romney any better than a 40 percent prospect for victory, and why he had Barack Obama as a 2-to-1 favorite, or better, for virtually that entire time period where the national polls were even.
Mon 10/1: Obama +3.7
Tue 10/2: Obama +3.3
Wed 10/3: Obama +4.4
Thu 10/4: Obama +4.0
Fri 10/5: Obama +3.0
Mon 10/8: Obama +0.4
Tue 10/9: Romney +0.6
Wed 10/10: Romney +0.8
Thu 10/11: Romney +1.3
Fri 10/12: Romney +1.0
Mon 10/15: Romney +0.9
Tue 10/16: Romney +1.2
Wed 10/17: Romney +0.2
Thu 10/18: Romney +1.2
Fri 10/19: Obama +0.1
Mon 10/22: Romney +0.2
Tue 10/23: Romney +0.8
Wed 10/24: Romney +0.1
Thu 10/25: Romney +1.3
Fri 10/26: Romney +1.0
Mon 10/29: Romney +0.4
Tue 10/30: Romney +0.1
Wed 10/31: Obama +0.7
Thu 11/1: Obama +0.6
Fri 11/2: Obama +0.1
Sun 11/4: Obama +1.5
Mon 11/5: Obama +1.1
CURRENT MARGIN: Obama +2.7
And, with actual votes to look at now, we finally understand that national-swing state polling disparity. Of course, some of us knew it all along. As Markos noted three weeks ago:
Romney's entire advantage in [the Gallup tracking] poll comes from a massive lead in the South. Now sure, some of that may be Florida, but the state-level polling certainly doesn't show that. So Romney is driving up big margins in Texas, Alabama, Oklahoma, Mississippi and other such presidentially irrelevant states? Good for him! I'm sure that'll be cold comfort as he loses the states that actually matter in the Midwest and West.The election proved Markos to be correct on this score. If you exclude the Southern states (which, for our purposes, we will define as: Texas, Oklahoma, Arkansas, Louisiana, Mississippi, Tennessee, Kentucky, West Virginia, Virginia, North Carolina, South Carolina, Georgia, Alabama, and Florida), the popular vote was even more lopsided:
NATIONAL POPULAR VOTE (as of 10 AM 11/10/12):What this means, essentially, is that while the national polls were close for most of October, the swing states were considerably less so, at least in any calculus that could get Mitt Romney to 270 electoral votes. Sure, individual data points surfaced here and there to give the Republican campaign hope, but if you took the state-by-state polling as a mass, there was never really a point at which Mitt Romney got any closer than about 2-3 percentage points from cobbling together a coalition of states that would land him the vital 270 electoral votes. Furthermore, that modest Obama advantage in the electoral college was incredibly durable.
Obama : 61,820,108 (50.54%)
Romney : 58,563,101 (47.88%)
Others : 1,927,014 (1.58%)
NATIONAL POPULAR VOTE OUTSIDE THE SOUTH (as of 10 AM 11/10/12):
Obama : 42,769,338 (53.95%)
Romney : 35,100,308 (44.28%)
Others : 1,405,601 (1.77%)
Another reason the national polls were errant: to use the most oft-overused and trite phrase of the election, many of them were...ahem...skewed. Which provides us with an excellent segue, as it were.
MYTH #2: The 2012 electorate was not the 2008 electorate.
Well, maybe there is a kernel of truth in that statement. As it happened, the 2012 electorate was actually incrementally better for Democrats than the 2008 one.
While the gap between Democrats and Republicans in the party ID of the exit polling was slightly narrower than in 2008 (D+6 instead of D+7), other indicators were even better for the Democrats. The 2012 electorate was less white (72% in 2012, 74% in 2008) and more liberal (25% in 2012, 22% in 2008) than the electorate that first sent Barack Obama to the White House.
The entire Unskewing phenomenon (which I took to task back in September) was predicated on the ultimately repudiated notion that the electorate had trended center-right between 2008-2012.
The problem with this article of faith among many conservatives, and a disturbing number of "impartial" journalists, was that it was based on two very faulty premises.
The first faulty premise was that the 2010 exit polls, which showed a dead-even D/R gap and a chasm between self-identified liberals and conservatives (20 percent liberal, 42 percent conservative), represented some kind of sea change in America's political trajectory. Seeing 2010 as a watershed election makes sense on some levels, but not on this one, for what should have been an obvious reason.
There were 40-45 million fewer voters in 2010 than there were in 2008.
It is universally assumed (and the data tends to back it up) that more habitual voters tend to be white and conservative. Ergo, if you know that you are going to get an additional crush of voters to the polls in a presidential election year, they are probably going to be a more Democratic-leaning constituency. Thus, basing any 2012 turnout assumptions on what happened in 2010 was unbelievably silly. But, as now seems evident, many conservatives (and some others, to be fair) saw 2010 as somewhat of a realigning election. This election, of course, disproved that pretty effectively.
The second faulty premise, meanwhile, was even more absurd.
The GOP poll deniers, led by Mr. Unskewed himself, seemed to presume the existence of two polling trends that, taken together, seemed entirely implausible. For one thing, hey scoffed at the notion that there could still be a sizable gap between self-identified Democrats and Republicans. At the same time, however, they also crowed that Mitt Romney had a lead (and, at times, a significant one) in the polls among Independent voters.
The conservative inability to put two-and-two together here was stunning. This is the quintessential example of having your cake, and wanting to eat it, too. These folks were steadfast in their insistence that Barack Obama's sheer awfulness had led to a 15-25 point swing with Independents (Obama carried Indies by 8 points in 2008), and at the same time led to millions of Americans either moving into the Republican Party or moving out of the Democratic Party.
What seemed infinitely more plausible, as many bright minds like David Karpf had noted, was that many conservatives, feeling their teabagging oats, had simply re-identified as Independents.
The exit polling seems to bear out Karpf's thesis. In the 2012 exit polls, Independents did indeed move markedly. An Obama edge of eight points in 2008 became a Romney edge of five points in 2012. This thirteen-point shift was far greater than the movement nationally (which should settle in, eventually, at a four-point shift between the two elections), which would seem to confirm the presence of the teabagger effect in the Independent electorate.
It also is confirmed, of course, by the fact that the D/R spread in 2012 was basically unchanged from 2008.
MYTH #3: Romney-mentum was robbed by Hurricane Sandy
This is one of those myths which makes sense in the realm of political science, but not as much in the actual data.
The "rally effect", the surge in support for government officials when they confront a crisis and deal with it, is a staple of American political science. And there is more than enough evidence to affirm that it usually does exist.
But the errancy here is not that Obama didn't get a lift from his handling of Hurricane Sandy. Indeed, that is indisputable: his approval rating in the exit polling was 54%, which is higher than it has been in virtually any polling in the past two years.
The mistake is in presuming that Romney was in a position to win before Sandy, and Sandy stalled his "momentum."
First things first: that graph of the national polling data above makes one thing evident. Whatever momentum Mitt Romney had already plateaued on or about the second "town hall" presidential debate. What's more: the very small national lead that Romney did briefly enjoy was already dissipating well before Sandy made landfall. Romney's peak lead (of 1.3 percent) came in the immediate wake of the Boca Raton presidential debate. From that point forward, his "lead" began to erode. Indeed, by the day that Sandy made landfall (October 29), Romney's "edge" was down to 0.4 points.
Here is a graphic representation of the same thing:
That is not to say that Sandy had no impact. Indeed, there was a palpable rally effect in several states. New Jersey performed, in the final analysis, about 5-7 points more Democratic at the presidential level than pre-election polling had indicated. Smaller, but very real, differences between the polls and the final results could also be seen in Maryland, New York, Connecticut, and Rhode Island. That may have also applied to Delaware, for all we know, but we don't actually know because no one ever bothered to poll Delaware.
But those half dozen states would not have too profound an impact on the national numbers, meaning that the idea that Sandy somehow rescued the Obama presidency is, in the final analysis, a media (and GOP operative) driven fallacy.
There are certainly more myths to be skewered. For example, at this point, any analyst who calls America a center-right nation probably should not be allowed to analyze American politics for a while (consider it the punditry equivalent of the penalty box). The three I've outlined here were among the most persistent, however, and the easiest to skewer. It is a given that partisan operatives will spin every election. One can only hope that journalists and analysts that should know better will have long memories of how badly the 2012 pre-election analysis was botched, by the time 2014 and 2016 roll around.