Debate Drops Obama Chance of Winning to 99.3%
Five Thirty Nine blog – by Nate Silverado
After a ho-hum performance in his first Presidential debate against Mitt Romney, the 539 model now puts President Obama’s likelihood of winning at 99.3%. This precipitous 0.6% drop from last month was caused by declines in purple states Texas, where Obama’s chance of winning slipped from 85% to 75%, and Utah, where the race is now scored a statistical dead heat.
My prediction model, which has never been wrong, was adjusted for this week’s Gallup poll that showed Mitt Romney up six points at 51%. Some claim no one ever lost when over 50% in the Gallup so late in an election cycle. I disagree, the Gallup poll is invalid.
Although the precise components of my proprietary algorithm shall remain a trade secret, I share highlights to quiet those who claim my high probability of Obama winning is inconsistent with popular polls.
My “Dewey Beats Truman” adjustment fixes the problem of 1948 where national polls incorrectly predicted a win by Dewey, the Republican. I shift all polls +5% in Democrat favor to account for this “Give ‘Em Hell” effect.
Second, using the complicated Black-Sholes formula, I adjust the weighting of all known polls. Extra weighting is given to polls touted by anyone with the last names Axelrod or Plouffe. Polls containing the words “Fox” or “Rasmussen” are underweighted by 90%.
In order to “look like America,” the 539 poll is adjusted by age, race, party ID and gender. Because of the “gender gap”, I reduce the weighting of men in all polls by 20% with an equal offset to increase women. Since I don’t like old people, I downsize the proportion of voters over 65 by half. You never know how many will die before Election Day.
My “vote fraud” adjustment doubles the proportion of any poll sample taken in Chicago. Dead people vote but they don’t answer pollster’s calls.
My algorithm applies certain national economic and social trends to the model. I score unemployment but remove all layoffs announced on Fridays, when most are, and Mondays. The year 2009 is removed as an outlier. The year 2010 is also removed. As are 2011 and 2012. My prediction of 400 million new jobs created by President Obama next year is imputed by scoring all independent voters in the Obama camp.
Finally, my racial adjustment is made. In order to get the desired Obama landslide I want, I drop all whites from the poll samples. I trust you now understand why recent polls from Gallup and others cannot be taken at face value. Obama is a lock. Especially after Joe Biden’s laughs in the VP debate, for which, I added +2% to every poll.
Picture from Wikipedia Commons.