Why the pollsters have been wrong

Byline: | Category: 2012 | Posted at: Monday, 15 October 2012

Republican presidential nominee Mitt Romney is just recently sneaking ahead of the incumbent.  This is a surprise to many, but it shouldn’t be.  That’s because Barack Obama has always been behind.

Many poll watchers have posited that the 2012 election won’t match the historically high turnout of usually low-turnout voters like minorities and youth that occurred four years ago.  Those critics are right that 2012 won’t resemble 2008–but for the wrong reason.  There is another cause, oft forgotten, of why 2008 was an unusual year, and thus unsuitable for mathematical extrapolation:  It was the first election since 1952 when neither an incumbent President nor a sitting Vice President was on the ballot.

While poll skeptics have pointed at the 2008-like turnout models as a source of modeling error, the 2012 polls include an even greater fundamental flaw:  they rely heavily on a 2008 methodology that assumes that there are only two candidates in the race.  The relevant data point that aggregators like RealClearPolitics and Nate Silver track is the difference between Obama’s support and Romney’s support.  That is the wrong metric for most presidential elections.  This year’s election returns to the historical norm where there are three candidates in the runup to the presidential election:  the incumbent, the challenger, and not-the-incumbent.  And in these types of elections the single best predictor of re-election is the incumbent’s level of support.  The difference between it and 50% is what poll watchers must watch. 

When you look at 2012 this way you will see how weak Barack Obama has been all along.  The RCP average has stayed remarkably consistent with the incumbent scoring between 46 and 48 percent support for most of two years.  Even while through most of 2012 Mitt Romney has trailed Barack Obama’s level of support, the incumbent has rarely and only briefly exceeded this narrow range.  The undecideds are not entirely undecided and are actually quite stable in their opinion.  They have decided that they do not like the President.  They just haven’t yet decided that they support the challenger.  Of the major poll watchers, I think only Jay Cost has been mindful of this gap.

When you factor in the fact that most organizations have polled registered voters until only recently switching to a likely voter model, you understand just how difficult the President’s path to reelection has long been.  This is not 2004, when President George W. Bush, rarely dropped below 50%, and could afford to demonize the challenger and focus on turning out the base.  Instead, 2012 bears greater resemblance to 1980.  Barack Obama just doesn’t have enough base to turn out to defeat the combination of Romney supporters and not-the-incumbent voters. 

The fundamentals of this election have always been against the incumbent Obama.  Romney’s debate win was not the cause of his recent rise in the polls.  It was simply a catalyst that precipitated what was always there:  Barack Obama does not have, and long hasn’t had, the support of the majority of the nation’s voters.  Absent a significant shift, Obama is going to lose this election, and by a bigger amount than any have expected.  48% is his upper limit and he is not going to exceed it on November 6th.

MORE:

Ed Morrissey brings us an anecdote that succinctly demonstrates the point of this post: 

“I was not undecided between Obama and Romney.  I was undecided between Romney and not voting.”

Undecideds in this election, aren’t undecided about Obama.  52% or more of them are not going to vote for him.  His only hope of reelection is to keep many of them home on election day.

Share this post:

Comments are closed.