Saturday, November 3, 2012

And this is why you remember to hit "publish" when you're done.

Originally written just prior to the 2012 election.

I just googled "Romney landslide."  Given that ol' Nate Silver is predicting an 80% chance of Obama winning (83% as of this writing) I just want to get the results down, for the record:  (Wayne Allyn Root is a capitalist evangelist and serial entrepreneur.)  Gives a lot of hunches and perceived trends, but mostly focuses on, "hey, I've called races correctly before".

Mark Tooley predicts a Romney landslide, but only among the (increasingly irrelevant) white evangelical demographic.  Doesn't really count. :  Dick Morris, possibly the wrongest person in politics, predicts Romney landslide.  Nate Silver's response to the prediction?  "Sell Romney on InTrade!"

For the record, he's predicting (as of 10/30, eight days before the election), that Romney will take Indiana, North Carolina,  Florida, Virginia, and Colorado, perhaps also Ohio, New Hampshire, and Iowa.  He's calling Pennsylvania, Michigan, Minnesota, and Wisconsin the new battleground states.  He is also predicting 53 Republicans, 47 Democrats in the Senate.  I believe he'll be proven somewhere between "thoroughly" and "laughably" on the wrongness scale, but we'll see.  David W. Thornton has Romney winning 331 to 207.    Prediction seems to hinge entirely on "the Incumbent Rule", which says that undecided voters mostly break for the challenger.  There have been a few famous elections where the rule obviously held, but for the most part it seems to be a myth.

"David W. Thornton is a freelance writer and commercial pilot. He writes from the perspective of a conservative Christian and economic libertarian. He is a graduate of the University of Georgia and Emmanuel College."

The laughably awful Unskewed Polls is predicting the widest Electoral College margin of anyone I can find, 337-201 in favor of Romney.  Their prediction strategy -- actually, their entire reason for existing -- is to recalculate actual polls, based on the assumption that every single polling organization (including right-leaning Rassmussen and Fox News) are oversampling Democrats and undersampling Republicans.

Other Unskewed predictions to test against:  Romney wins Florida 52%-44%.  Romney wins Colorado, 52%-46%.  Romney wins Iowa, 49.4%-48.6% (too close to round).  Romney wins Nevada, 57%-42%.  Romney wins New Hampshire, 48%-47%,  North Carolina 57%-42%, Ohio 50-49%, Pennsylvania 50-47%, VA 52-46%.  Note, these predictions not only include the "unskewed" poll numbers, but also assume a 3/1 advantage for Romney among undecided voters.  It's hard to imagine this being even close to right.

For a similar exercise, visit: 

No comments: