WILKES-BARRE – On the last day before the vote, with the last bit of new polling data, two King's College professors took one last stab Monday at predicting who will win the presidential election, and came up with the same results they had when they prognosticated last week: Too many states are still too close to call.
On Thursday, mathematics chair Daniel Ghezzi and dean of faculty Joe Evan explained their relatively simple yet highly accurate – so far – system of projecting which states each candidate will win, based on multiple recent polls of likely voters. At the time, they declared 11 states up for grabs. Of those, their system showed President Barack Obama taking Ohio and Wisconsin and Gov. Mitt Romney winning North Carolina.
At the end of Thursday's presentation, Ghezzi and Evan promised to crunch all the numbers Monday for one more crack at clarity.
The only state where, according to their system, things cleared up was Iowa, which they said should go to Obama. That was offset by Nevada staying in the tossup column, despite the promise of a late Thursday poll last week.
If the new predictions hold Obama would be assured of 235 total electoral votes and Romney 206, with 270 needed to win. States that were still too close to call using their system as of Monday were Colorado, Florida, Michigan, Nevada, New Hampshire, Pennsylvania and Virginia.
Of those seven, the duo's mega-poll math showed Obama ahead in every state except Florida, where Romney had slightly more than a 1 point edge. But in all seven cases, the margin of error is greater than the edge either candidate has, thus keeping them as tossups.
The math mavens rely on two methods. First, they combine polls and look for the overall winner to see if his victory exceeds the margin of error. If that doesn't provide a projected winner, they look at how many of the polls each candidate won, regardless of how close the win is. Enough wins and the candidate is declared Tuesday's likely victor.
Evan went out of a limb and made predictions beyond their well tested method. He said New Hampshire, Nevada, Michigan and Pennsylvania should go to Obama because he has won far more polls in those states than Romney, even though the victories didn't surpass the margins of errors. In the remaining three states, he said there wasn't any solid trend data, but still predicted Romney should win Florida while Obama takes Colroado and Virginia. The math behind those prognostications? A look at the national popular vote in 2008 and predictions for this one, with some statistical hedging built in.