Chicago (IL) - Election day isn't until November 4, 2008, and yet already politicians are using math to determine that McCain is ahead by 28 electoral votes. One forecast model shows Democrat Senator Barrack Obama weighing in at 255.2 (47%), while Republican Senator John McCain pulls in a hefty 282.8 (53%) or about 27 more votes. A recent Gallup poll shows the popular trend reversed - Obama receiving 48% of the vote with McCain receiving 44%. If those votes could translate directly into the electoral college, it would be Obama 210 and McCain 192 (the remaining votes would go to 3rd party candidates). So, do we believe the model or the poll?
The mathematical model research team is being led by professor Sheldon Jacobson of the University of Illinois-Urbana. It involves others from his university as well as the University of Southern Illinois-Edwardsville. They've put together a website which shows the forecasting trend. Save computations are also shown, placing the race a little closer at Obama 233, McCain 247, only 14 votes apart.
The model factors in "lessons learned" from the 2000 and 2004 election, namely that the winner cannot be predicted based on on the popular vote. In 2000, for example, Democrat Al Gore received a greater percentage of the nationwide popular vote, yet because of the way electoral votes are cast by each state, he lost the election. A similar situation happened in 2004 when Kerry generally polled slightly ahead of Bush, yet lost the general election.
Still, researchers believe it is possible to get a reasonably accurate pictures just from math alone, even if that picture differs notably from the popular polls currently being taken.
Specific details of the mathematical model will be presented at an annual meeting of the Institute for Operations Research and the Management Sciences (INFORMS). The meeting will be held less than three weeks before the election, October 13-15 in Washington DC at both the Marriot Waldman Park Hotel and the Omni Shoreham Hotel. More than 4000 analysts and experts in analytics are expected to be on hand.
According to the report, "Jacobson's model employs Bayesian estimators (which help scientists make decisions when conditions are uncertain) to determine the probability that a candidate will win each state. He obtains state polling data from Rasmussen Reports, the Quinnipiac University Poll, and SurveyUSA." State-by-state probabilities are then used in a dynamic programming algorithm to determine a probability distribution for the number of Electoral College votes that each candidate will win in the 2008 presidential election.
"We take into account 'safe' states— states that each candidate is basically guaranteed to win," says Jacobson. "In 2004, once you took into account Bush's 'safe' states, he had a much narrower gap to close to get to 270 electoral votes than Kerry."
Regardless of past performance with even "safe state" models, there are several factors which may change voter's minds this time, such as fuel prices and the continued Iraq War. Whether or not the model can hold up to the real-world attitudes of the U.S. voters is a story we'll all find out in about six weeks.
The mathematical model research team is being led by professor Sheldon Jacobson of the University of Illinois-Urbana. It involves others from his university as well as the University of Southern Illinois-Edwardsville. They've put together a website which shows the forecasting trend. Save computations are also shown, placing the race a little closer at Obama 233, McCain 247, only 14 votes apart.
The model factors in "lessons learned" from the 2000 and 2004 election, namely that the winner cannot be predicted based on on the popular vote. In 2000, for example, Democrat Al Gore received a greater percentage of the nationwide popular vote, yet because of the way electoral votes are cast by each state, he lost the election. A similar situation happened in 2004 when Kerry generally polled slightly ahead of Bush, yet lost the general election.
Still, researchers believe it is possible to get a reasonably accurate pictures just from math alone, even if that picture differs notably from the popular polls currently being taken.
Specific details of the mathematical model will be presented at an annual meeting of the Institute for Operations Research and the Management Sciences (INFORMS). The meeting will be held less than three weeks before the election, October 13-15 in Washington DC at both the Marriot Waldman Park Hotel and the Omni Shoreham Hotel. More than 4000 analysts and experts in analytics are expected to be on hand.
According to the report, "Jacobson's model employs Bayesian estimators (which help scientists make decisions when conditions are uncertain) to determine the probability that a candidate will win each state. He obtains state polling data from Rasmussen Reports, the Quinnipiac University Poll, and SurveyUSA." State-by-state probabilities are then used in a dynamic programming algorithm to determine a probability distribution for the number of Electoral College votes that each candidate will win in the 2008 presidential election.
"We take into account 'safe' states— states that each candidate is basically guaranteed to win," says Jacobson. "In 2004, once you took into account Bush's 'safe' states, he had a much narrower gap to close to get to 270 electoral votes than Kerry."
Regardless of past performance with even "safe state" models, there are several factors which may change voter's minds this time, such as fuel prices and the continued Iraq War. Whether or not the model can hold up to the real-world attitudes of the U.S. voters is a story we'll all find out in about six weeks.