Thursday, November 8, 2012

2012 Presidential election postmortem

So, in my August 2nd prediction, I called out an Obama win with 328 electoral votes and 51.5% of the popular vote.  And that was before I started following Nate Silver.

At the time, though I didn't say it, I thought that Evangelicals and other white Christians would move their votes to Gary Johnson or some other third-party vote, in an effort to register their displeasure with the possibility of a Mormon in the White House.  As a result, I thought that Romney would only get 46.5% of the popular national vote.  After that first debate however, there was a real momentum swing as notable Evangelical leaders openly supported Mitt Romney (even if they didn't address the relationship of Mormons to Evangelicals).

When the near-final tally came in (assuming Florida, which has Obama up, is counted), I was not surprised that my August prediction held up, with the exception of New Hampshire.  (If you go back to early August, there were few polls taken in NH, and while the newer polls showed Obama with a lead, hypothetical polls from the previous year pointed to solid red streak for Romney, even though he hadn't yet entered the primary season.  In hindsight, I probably should have just picked Obama for NH and stick with what the aggregate polling had shown, rather than buck it.)

That I was able to guess 49 of 50 states correctly, on August 2nd, should highlight the fact that poll aggregation and applying even a simple algorithm to the aggregate data, yielded an exceptional story about this year's election: stability.  Because the polling data was telling me that the polls were uncannily stable, I decided to take a modest risk and run with an early prediction, so that I could later tell you this story about how all of those political pundits on TV were a bunch of fools, Fox News, especially.

My prediction on August 2nd.
Final outcome (pending Florida)
If you look at the state-level data of some of the battleground states, the more sophisticated your algorithm, the clearer the picture, and that picture was of incredible stability.  Obama seemed to hold a -- relatively -- tight margin that was steady throughout the year!  Question the methods, question the bias, question whatever you want, while in the middle of the election season, but the fact of the matter is, when you aggregate more data, the margin of error shrinks.  (There is no harm in including Rasmussen polling data, if you aggregate it with all others; if you weight Rasmussen accordingly, then your aggregation should produce solid numbers, right?)

The problem has always been, that the mainstream media likes to cherry pick which poll it wants to highlight, in order to tell whatever story it wants to, and excludes the context of the other polling data.







Now, there is no doubting that the state of the economy, regardless of where we had come from and where we were going, was miserable. It is just absolutely remarkable that, even as the state of the economy was terrible, Obama was able to hold onto a steady lead. There are at least two or three different reasons why this occurred:

  • Mitt Romney could not avoid being Mitt Romney.  For every good week or good event, there was at least two bad weeks and bad events.  His foreign policy trip turned into a foreign policy nightmare.  Even after that bump from the first presidential debate, the gaffes and stumbles did not stop.  Once he was successfully labeled as an out of touch rich guy, he did nothing to counter it, and in fact only reinforced it with his comments and actions (think 14% income tax rate).
  • The Obama camp really knew what was going on because they had really smart people who built very good algorithms, and were tracking data in ways that it has never been tracked before; meanwhile the Romney camp, despite its attempts to keep up, just didn't have the powerful machine needed to compete in real time.   It gave the Obama camp the ability to hit back with the right tone, in the right states, with the right message, all year.
  • It was always about more than just the current state of the economy.
    • It was about women's rights.  Mitt Romney paid lip service to the Lilly Ledbetter Act, insisting that a resurgent economy was the only meaningful concern for women.  Throw in the odd rape comments from other male Republican politicians, and you could see that many women were completely offended.
    • It was about gay rights.  This was not a decade ago and the rise of the Defense of Marriage push.  2012 marked widespread support for gay marriage, just as President Obama became the first sitting president to express support for the cause.  It seemed incredulous that a sitting president could express this view, without being punished at the polls, and yet if anything, he helped push Americans into accepting this as a civil rights issue.
    • It was about minorities.  In no small part because of the changing demographics, many older and working-age white Americans engaged in a counter-revolution, known as the Tea Party movement.  Their charge was to take America back to the whiter version of itself; they spoke openly of disdain and with disrespect for the authenticity of Obama's birth in America.  When Mitt told folks in Michigan that everyone knew where he was born, that was a coded reference to the legitimacy of Obama's heritage.
    • It was about integrity and the big American moderate pie.  Many people just simply could not trust Mitt Romney.  A lot of people thought Mitt was moderate, but in the last several years it seemed that he had embraced the far right.  Some people decided to cross their fingers and hope that he would not do what he said he'd do; most people refused to accept that risk.  Once they heard him embrace the far right, they stopped trusting him and his message.
So what made conservative pundits think that Mitt was going to win, or worse, why did some of them think Mitt would win in a landslide?  All of them gave one basic premise for their predictions: the polls were wrong, because there was a building momentum for Mitt Romney in the independent vote.

That was a terrible error, and one solidly entrenched in a party that has a disdain for science and math.


A quick look at both RCP and HuffPo will tell you slightly different stories, but neither showed a momentum for Romney in the closing days; the difference between the two, is that HuffPo took into account more polling data and used stronger smoothing to give a longer-trend view.  While RCP suggested that Mitt Romney took the lead, HuffPo showed that this was not really the case.  It was probably RCP's trendline, that Republicans were focused on.

Now, I also knew that these same Republicans would blame Hurricane Sandy and its political implications with Governor Chris Christie's praise of President Obama and FEMA.  While I'm sure some people were heartened to see proof that the President, in fact, could put away the partisan politics and put all Americans first, you can see that, as a matter of fact, even RCP's aggregate poll gap had peaked the previous week, specifically October 27th.

Because RCP is an aggregate of the previous week's polls, this means that by the 27th, there were already polls pointing towards Obama.  If you look at Gallup's daily tracking of likely voters, Obama bottomed out at 45% on October 21st.

So one side relied on the expectations of a momentum, bad polling data and suspected poll bias, while the other side continued to use polling data, algorithms and analysis.

And which side won?

No comments: