Polling and the US Presidential election: who will win?

According to the Pew Research Centre's weekend poll President Barrack Obama holds a 48 per cent to 45 per cent led over Republican Mitt Romney. Taking into account undecided voters, Pew determines Obama likely to secure 50 per cent and Romney to secure 47 per cent of the national vote. A week earlier, Pew had both candidates on a tie at 47 per cent. A 30^th October ABC News and Washington Post poll had both candidates at 49 per cent each. On 28 October Gallup recorded Romney ahead 51 per cent to Obama's 46 per cent while on 26 October Reuters gave Obama a one point lead over Romney, 47 to 46 per cent. Polls conducted on the 4^th of November by CNN declared a tie at 47 per cent each. While NBC, ABC News and the Washington Post called a one point win for Obama. If there is anything to be found in polls it is that the political environment in America is changing almost hourly and at least daily.

Polling is considered to have a variance of anywhere from two to four percentage points, depending on the methodology and sample size. Responding to the view that polling in this election is much more volatile that previous elections, Huffington Post developed model-based poll averaging. It is a combination of the most current national trends in voting intentions and in-house polling. Even with modelling, The Huffington Post is reporting the national popular vote remains "too close to call". Polling of the popular vote needs to be considered in how the U.S. President is elected.

Presidents are elected via an electoral college system. Each state is allocated a number of votes based on population. For example, California with almost 38 million people has 55 electoral college votes. Montana with a population of one million people has three electoral college votes. The only exceptions are Nebraska and Maine that divides their electoral college votes based on the winners of congressional districts and the statewide popular Presidential vote. Apart from this exception, the candidate who wins the popular vote in that state secures all the electoral college votes. To be elected President, a candidate must secure 270 electoral college votes. The broader problem for both Obama and Romney is that neither is assured of securing 270 electoral college votes. A repeat of the 2000 presidential poll would be unthinkable for Democrat strategists. In that poll George W. Bush won the electoral college vote, including the votes from Florida (then governed by his brother Jeb Bush), even though he lost the popular vote by more than half a million votes.

Much of the uncertainty in the polls is reflecting historical voting trends in American battleground states. BBC reports that Republican stronghold states are likely to return 191 electoral college votes to Romney from mainly south and mid-west states while Democrat stronghold states, such as California, are likely to return 186 electoral college votes. Obama's vote in California, for example, is solid at 51 per cent to Romney's 35 per cent. There are at least 13 swing states, with their accompanying electoral college votes, up for grabs in Tuesday's poll. Iowa, Ohio and Florida are as always in the mix. Also highly contested are the states of Colorado, Michigan, Minnesota, Nevada, New Hampshire, North Carolina, Pennsylvania, Virginia and Wisconsin. Of these states, the ones to watch are those with the double-digital electoral college votes, notably Pennsylvania (20 votes), Ohio (18 votes), Virginia (13 votes), and Wisconsin (10 votes).

Politico.com reports 50 per cent of Iowa and Ohio voters supporting Obama and 48 per cent supporting Romney but local polling from the Miami Herald suggest Florida is already a Romney win at 51 per cent compared to Obama at 45 per cent. National polling by NBC gives Florida to Obama (49 per cent to Romney's 47 per cent) while Reuters polling is calling a draw at 47 per cent a piece.

Obama will need Florida's 27 electoral college votes if he is to win another term in the White House. In 2008, Florida supported Obama 50.9 per cent to Republican candidate John McCain's 48.4 per cent. In voting terms this represents the votes of a little over 200, 000 Florida voters. Florida of course was the issue for Democrat John Kerry in 2004 were the official vote was recorded at 47.1 per cent to Kerry against the 52.1 per cent recorded for then President George W. Bush. Obama's win in Florida was largely attributed to high voter turnout: the highest since the 2000 election between Bush and Democrat Al Gore where the Florida popular vote and its electoral college votes were undecided for 36 days. As with the 2004 poll there were claims of faulty voting machines and electoral fraud.

Huffington Post's model-based polling of Florida indicates the fluid nature of voting intentions. Beginning in January 2012, it had Obama at 45.3 per cent compared to Romney 44.5 per cent. In any poll, the variance of two to four per cent would make the result statistically void. Results for the 1^st of March revealed a clear Obama lead (48.1 per cent to Romney's 42.7 per cent). By the start of May and the middle of August, similar polling to January showed both candidates almost tied. By the third week of September, Obama showed a 3.2 per cent lead over Romney but by the first week of October the gap narrowed to a 0.3 per cent lead for Obama. On 3^rd November, Florida voters had returned to the impasse of January with 48.5 per cent in support of Obama and 47.7 per cent in favour of Romney.

One key factor in this, as with most U.S. presidential elections, is voter turnout. Low voter turnout historically bodes badly for Democrat candidates and this election is likely to see voter turnout lower than in both 2004 and 2008. Of the expected 90 million eligible voters who wont vote in this election, USA Today reports the majority are likely Obama-Democrat voters. Only about 55 per cent of all eligible voters participate in American elections. Curtis Gans from the Centre for the Study of the American Electorate at Washington's American University cites the reasons for low voter turnout as a lack of trust in political leaders, a lack of positive feelings towards political institutions, a lack of civic education, the fragmenting effect of communications technology, and the cynicism of political coverage.

The need to get voters out early as a political strategy has its upside in locking in partisan votes, leaving candidates to concentrate on swing or undecided voters, which is exactly what Obama has been doing in recent days. This view probably drove Obama to himself vote early, the first U.S president to do so. Returning to Florida, the vote early strategy has polarised the state. Florida's Democratic Party has filed a federal lawsuit over long delays at polling stations and polling stations closing early. In Miami's Dade County, polling stations were supposed to be open on Sunday from 1pm to 5pm, however the polling station was closed at 2pm with officials claiming they were "swamped by the unexpected high [voter] turnout". It has been reported that 4.5 million Florida voters have cast early ballots, likely to favour Democrat candidates over Republicans.

The second factor that will likely impact on Tuesday's presidential poll is Hurricane Sandy. Hurricane Sandy is considered by pollsters as having a positive impact on Obama's chances with 46 per cent of Romney voters and 63 per cent of swing voters approving of Obama's handling of the crisis. It will directly effect voters in New York, New Jersey and Connecticut who are still facing power outages meaning polling stations are not operating and will disrupt postal deliveries of absentee ballots. All three have Obama leading in the polls: 59 per cent in New York, 53 per cent in New Jersey, and 54 per cent in Connecticut. Had these voters voted early, the vote shouldn't negatively impact Obama. Even if these voters have not voted, and assuming electoral officials can open polling stations in time, Obama will gain an advantage over Romney and likely across America for his presidential handling of the crisis.

The most far-flung view of the 2012 presidential election is that it may be a tie. With polling so close, it is electorally possible that both Obama and Romney could each end up with 269 electoral college votes. This is possible if Obama wins in Ohio, New Hampshire and Wisconsin and Romney wins all the other swing states. Vote re-counts will be the first order of business followed by a series of judicial actions. Should the electoral college vote remain unchanged, the 435 elected members of the House of Representatives are sworn in January 2013, and under the 12^th amendment, the new Congress choose the new President. Republicans currently control the House of Representatives, this is unlikely to change hence Romney would likely win a House vote. This is a very improbable scenario but one of interest to psephologists.