Statistical Analysis: This is NOT a close election: Romney’s chances worse than McCain, Kerry
Tonight will be the first presidential election debate of 2012 and major news outlets are claiming that the campaign is a close race. The Wall Street Journal yesterday called Obama’s lead “slim.” Just how close is this election?
Unfortunately for Mitt Romney and Paul Ryan, the answer is clear: statistically, this is NOT a close election. While Romney may only be behind by a few points in national popular vote polls, we do not elect a president by the popular vote. Because of the way the electoral map is drawn, Romney has less than a 10% chance of winning at this point.
Absolute Closeness: Electoral Map
Obama has 201 electoral votes in “safe” blue states while Romney has 181. Romney must therefore win the majority of the votes in the dozen toss up states to make up the 20 electoral vote deficit and win the election. But there are only 156 electoral votes up for grabs so making up 20 is statistically very difficult, even for a candidate who is polling ahead. To make matters worse, Romney is polling behind in 11 of the 12 states that are still in play. All of this adds up to odds of about 7% of Romney being elected.
Relative Closeness: Now vs. Past Election
What about past elections? Have candidates come from this far behind to win? The short answer is no, at least not recently. It is less than 5 weeks until Election Day. In 2008, McCain’s odds of being Obama at this point in the campaign were 28% and in 2004, Kerry’s odds at this point were 15%. Neither man was able to overcome those odds, but there’s always a chance (a 7% chance) that Romney will be able to pull it off. Still, it is worth noting that this is the least “close” election since at least 1996.
There has been much talk recently about “oversampling” of Democrats in election polls. Polling bias is a real issue and no polling-based prediction would be complete without considering it. The NerdWallet model used to calculate presidential election odds projects polling error based on past polls’ ability to predict election outcomes. If there had been directional bias in the poling for previous elections, the model would assume the same bias would exist in current polls and correct for it. Therefore, the only polling bias concern is whether or not the bias has changed relative to previous elections. If polls are more likely to oversample Democrats this election than in previous elections, then the model could be underestimating Romney’s chances. At this time, however, there is no evidence that polling bias has increased. Further, conservative Fox News has released polling that is in line with more progressive media sources. Nevertheless, we will definitely keep an eye on this issue and adjust as necessary.
This election may be a lot of things, but “close” is not one of them. From a purely statistical point of view, Romney needs to increase the volatility of his candidacy to improve his odds of winning. Let’s see if tonight’s debate can provide the shake-up the Republican campaign so desperately needs!
UPDATE (October 22): This IS a closer election than 3 weeks ago
After strong debate performances Mitt Romney’s odds have greatly improved over the past 3 weeks. He remains an underdog, but has momentum on his side. This is now a more close election than the 2004 and 2008 elections, but is not as close as the 2000 election.
2 weeks before the election
2000: Gore’s odds of beating Bush: roughly 50% (data insufficient for greater precision)
2004: Kerry’s odds of beating Bush: 24.1% (up 9% over 3 weeks)
2008: McCain’s odds of beating Obama: 16.0% (down 12% over 3 weeks)
2012: Romney’s odds of beating Obama: 29.5% (up 23% over 3 weeks)