Team Blog

Election forecasting

President Obama appears to hold a lock on the Electoral College ahead of Tuesday’s election. EPA/Jim Lo Scalzo

Yes, I am going to tell who is going to win and by how much. But first, in honour of both the devastating weather and crushing polling uncertainty that America has been suffering, I am going to announce that my forecast for the remaining days of the campaign is a 100% chance of “Purple Haze”.

I don’t mean the Hendrix song or any type of cannabis associated with it, but that:

  1. Hurricane Sandy has clouded the picture both literally and politically (although not as much you might think);

  2. The aggregate of public polling paints a very hazy picture of a tight race – but that is only when you do not take into account that progressive and conservative sponsored polls are painting entirely different pictures and one of them has to be wrong and;

  3. That the entire election is really going to come down to just two “purple” swing states.

The tragic loss of life and extensive disruption caused by Sandy are regrettable, but should not significantly disrupt the election. There were only two states directly impacted by the hurricane that are truly contested in this race: Virginia and North Carolina. By one week out from the election, their early voting stations were already back up to nearly completely operational.

It seems highly unlikely that Sandy will have any effect where the election really matters, which is in the Electoral College. Remember, Americans do not have a national vote for president, we have 51 simultaneous votes in the states and the District of Columbia for electors. New York, where many residents will presumably still be without electricity next Tuesday, is a lock for President Obama.

At this point the only way that the storm could hurt Obama is if it is very close election and he wins the electoral vote by states but more Americans in total vote for Romney. For example, George W. Bush received one more vote than necessary for a bare majority in the Electoral College in 2000, even though half a million more Americans voted for Al Gore than for him. In a tight race, if hundreds of thousands of reliable Democratic votes do not materialise then it is possible that Obama could win but with a lesser degree of legitimacy in the eyes of critics and less of a governing mandate. Although, this certainly did not stop Bush from pursuing his agenda.

Instead, polls show that Obama has benefitted politically from Sandy. There is a well-documented “rally around the flag effect” in times of crisis, during which presidential approval ratings increase by several points on average. That Obama has received positive marks for handling the emergency competently – even from majorities of Republicans – and that it contrasts with his predecessor’s handling of Hurricane Katrina in New Orleans could actually boost the president’s standing on election day and in ongoing early voting. Romney, by contrast, has been inevitably somewhat sidelined.

Even before Sandy, the polls had been moving in Obama’s favour, both nationally and in key battleground states (the fewer than 10 out of 50 that have been considered genuinely competitive). Obama has apparently rebounded from his nadir after the first presidential debate and was ahead once again nationally and in “swing” states like Colorado that have been won by both parties at different points over the last few elections.

But the evidence has been extremely inconsistent, with a few polls still showing Romney pulling further ahead. The picture is murky for some marquee senate races as well, with different pollsters in the same races showing either the Democrat or the Republican ahead by several points. Progressive-sponsored polls tend to show Democratic candidates ahead more often than not, while conservative-sponsored polls have more Republicans winning. Part of the explanation is cynical: Campaigns will release their own, privately-funded research showing good results to reassure donors about keeping their wallet open or to influence the media narratives.

But the public, “neutral” polls are also showing wildly diverging results right now. And the reason in both cases is that pollsters make very different assumptions about the composition of the electorate that will actually turn out to vote (or, really, will less reliable Democratic youth and minority voters bother to vote or not). Simply screening “likely voters” against all “registered voters” makes a difference of a few points. Obama has led consistently among registered voters throughout the past few months, but different pollsters make different projections about expected actual voters. Some of the major public polling firms, such as Rasmussen, tend to produce much more Republican-friendly results on a consistent basis because of their likely voter screens.

Of course, the release of polled results influences media coverage of the race and campaign donations, which in turn further influence the poll results. In September, when Obama sported comfortable leads nationally and in swing states, some conservatives claimed a media conspiracy to sink Romney, in effect charging the media and polling firms with deliberately creating a self-fulfilling prophecy of an Obama victory, and some began a cottage industry to “unskew” the public polls by imposing their own likely voter screens to recalibrate them to what they “should” show.

As a result of this unprecedented epistemological storm, it is difficult to say who is really ahead nationally, in swing states, and in key senate races. Really, it comes down to the preponderance of the evidence and which pollsters’ assumptions you trust. We are now in the age of post-modern political polling, far less predictable than hurricane tracking.

That said, the aggregate evidence points to an almost certain Obama electoral victory on Tuesday. Romney is just not winning enough “purple” (neither reliably Republican Red nor Democratic Blue) swing states to attain a majority of the Electoral College. Obama has led relatively comfortably in crucial Ohio even through his doldrums, and he appears to have retaken the lead in Virginia, Colorado, and perhaps Florida. As before the debates, the only battleground state that now appears to favour Romney is North Carolina, the closest contest in 2008. Only if you assume that Romney wins every close contest but Ohio – Nevada, Colorado, Iowa, New Hampshire, Virginia, North Carolina, and Florida – do you get a Romney victory.

And if you give New Hampshire back to Obama then you get the unprecedented 269-269 tie that would throw the election to the Republican-controlled House of Representatives and thus to Romney. You can play with the numbers yourself at www.270towin.com, and the reality is that Obama has always been ahead in Nevada and Iowa too, so without those states Romney would have to win both Ohio and Colorado.

In the end, however, there are really only two states that are probably relevant to determining whether or not 2012 will be a suspenseful election: Virginia and Florida. If Obama wins either of these two east coast states then there will not be any need to wait around for returns farther west. If Romney loses either of these two tight contests, it will not be a night that sees him pick up other states that have been more solidly behind Obama. Obama should get somewhere in the 290-300 electoral vote range, or around 330 if he wins Florida.

The Democrats will make gains in the House of Representatives, but the conventional wisdom espoused by all but Nancy Pelosi is that they will not come close to winning the chamber back in this deeply divided electorate. In the Senate, where Democrats had been expected to lose their 53-47 majority, the likeliest outcome at this point is a swing of no more than one seat either way.

In other words, when the purple haze lifts, the political landscape in Washington, D.C. will probably look exactly as it has for the last two years. And then it will be time for the 2016 prognostications to begin.