Who is Favourite to Win the Keys to the White House?

The conventional wisdom for the last several days is that the betting markets are pointing to the re-election of the President while the polls apparently have it too close to call, or have Mitt Romney as the narrow favourite.

So what’s the evidence?

Certainly, a glance at the betting markets reveals a pretty strong consensus that President Obama has a strong edge with the bookmakers, the spread betting operators and the betting exchange, Betfair.com, each of these currently converging at a probability of re-election of about 75%. Intrade.com has it a bit lower, but still make President Obama the very clear favourite.

So what about the opinion polls? Here the idea that Mr. Romney is ahead is in great part derived from an inspection of the headline national poll findings published on the RealClearPolitics.com (RCP) web-site. RCP publish a selection of the latest polls and take an average of this selection to come up with their overall figure. This average had Romney leading by up to a point or so over the last few days, and currently has it about tied.

But is this what the opinion polls are actually showing? The answer depends on whether you look at all the published polls, as broadly available on Pollster, or whether you exclude the polls you don’t favour, a methodology favoured by RCP. Not only do they not include a number of these polls in their overall average, they do not recognize them at all!

Taking average of all the national polls still has it close, but close in the President’s favour.

More important than the national picture, as Al Gore found to his cost, are the state findings. An analysis of all the state polls shows that President Obama is rather more comfortably ahead, including in the key swing states of Ohio, Virginia, Wisconsin, Colorado, Iowa, Nevada and New Hampshire.

Combining the national and state polls, and awarding them a similar weighting, reveals Mr. Obama to be polling close to 2.5% ahead of Mr. Romney – more in the state polls and less in the national polls.

Depending on the statistical analysis used to convert polling advantage into probability of winning, current polling evidence would seem to give the President somewhere above an 8 in 10 chance of winning (based on national and state polls) to rather more than a 90% chance (based on the state polls alone). FiveThirtyEight.com and the Princeton Election Consortium (http://election.princeton.edu/) represent good examples of this sort of analysis.

So the betting markets, contrary to conventional wisdom, are in fact more conservative in their evaluation of Mr. Obama’s chances than are the polls.

Why so?

There are a number of possibilities, but I will highlight three of them.

First is the well-established favourite-longshot bias which occurs in most betting markets, whereby the markets tend to under-estimate the true likelihood of the favourite winning, and vice-versa.

Secondly, the markets may be allowing for the uncertainties a model can’t allow for them, such as some late surprise. The point is that volatility is the friend of the underdog, and the enemy of the favourite.

Thirdly, the markets may be wary that one of the outlier polling organisations might just possibly be right, and the overwhelming polling consensus might just possibly be wrong. This is what I call the ‘Rasmussen effect’, after the polling company that seems to have leaned Republican by a few points in its national and state polling for some time now (http://www.huffingtonpost.com/alan-abramowitz/the-rasmussen-difference_b_2030330.html?utm_hp_ref=@pollster).This effect was particularly prominent in the 2010 elections (http://fivethirtyeight.blogs.nytimes.com/2010/11/04/rasmussen-polls-were-biased-and-inaccurate-quinnipiac-surveyusa-performed-strongly/), and this is the pollster that had Mr. Bush leading by nine per cent on election day, 2000, as evidenced by the National Council on Public Polling (http://www.ncpp.org/?q=node/20). That was about ten points out!

If Rasmussen is right, though, the race is very tight in the key states and Obama would be much less of a favourite than the polling consensus suggests.

So Rasmussen must be right and the other polls wrong, and Obama must lose Ohio, which Rasmussen is calling a pure toss-up, and he must not win elsewhere to compensate. All other routes are regarded as far more unlikely by the markets, though by no means impossible.

So what is the most likely outcome of this election? However you cut the data, the answer is the same. It’s a win for the President.

Polls don’t decide elections, however, and neither do the betting markets. People do, and that’s why anything can happen, and sometimes does!