Probably safe Democrat (179 EV) from 172 EV back on 20 Sep:

• Likely Democrat (37 EV):  CT, ME, NJ, WA
  
• Solid Democrat (142 EV):  CA, DC, DE, HI, IL, MA, MD, NY, RI, VT

• Lean Democrat (22 EV):  MN, NM, OR
  
• Toss Up (156 EV):  CO, FL, IA, MI, MO, NC, NH, NV, OH, PA, VA, WI
  
• Lean Republican (31 EV):  AZ, IN, SC

• Likely Republican (98 EV):  GA, KY, KS, MT, ND, NE, SD, TN, TX, WV
  
• Solid Republican (52 EV):  AL, AK, AR, ID, LA, MS, OK, UT, WY

Of those eleven states that have shifted from their aggregated position three weeks ago, only three (Connecticut, Missouri, and West Virginia) have moved into a position more favorable to the incumbent; the other eight are now more favorable to the Romney-Ryan ticket.  There also seems to be a building consensus that the "tipping point state" will be Ohio . . . big surprise.

Of course, opinion polling can be a bit tricky, and those of us familiar with the science of Statistics are well aware of the truth behind the adage that figures don't lie, but liars can certainly figure.  We now know that the sampling procedures being used by nearly all of the mainstream polls are oversampling the known democrat "likely voters" by between 6% and 10%.  Additionally, according to Dick Morris, almost all pollsters are using some variant of the 2008 turnout models in weighting their samples, apparently without bothering to account for how different the 2008 GOTV numbers were from 2004 or 2010.  So not only are democrats being deliberately overrepresented in these polls, but their projected turnout is also apparently being actively overestimated.

Apparently, Chris Matthews really wants his tingle back.

There's also an article from the pen of Jeffrey Lord, published in The American Spectator about two weeks ago, in which he provides us with a history lesson of just how blatantly the media elite will manipulate polling data to politically support the leading liberal in the race in question.  In the 1980 campaign, the New York Times ran a series of nine articles over the course of 18 days in October on "crucial states" (which we'd call battleground states today), showing how Jimmy Carter was in a close-but-winnable race with Ronald Reagan.  This was followed by an article on the only debate of the campaign, rating the respective performances as close.  The end result, of course, was that Carter was absolutely thumped by Reagan in the Electoral College, including carrying all nine of the "crucial states" which, with the exception of New York, weren't even close.

Yet I'll be willing to bet a steak dinner at my favorite restaurant against the doughnut hole of your choice that by the time Monday morning rolls around, mainstream polls and the papers that report them will somehow have President Barry climbing back to his September numbers, having gained a badly needed bounce from "Laughing Joe's" bar brawl style of debating.  And they will screw with those numbers any way they have to in order to get the results they want.

The problem with such manipulation is that it does occasionally come back to bite the manipulator in the arse.  And if the update to the University of Colorado election forecast model (published the morning after the first presidential debate) is as predictive as it has been historically, then all of these mainstream media pundits are going to be scrambling for the Excedrin while they try to explain how Mittens pulled off a win that he clearly should not have.

So far as I've been able to tell, only the Sydney Morning Herald, the Colorado Daily, the Augusta Chronicle, the Aspen Business Journal, Restoring Liberty, The Blaze, Peach Pundit, the Colorado Watchdog, and Newsmax have found this update to be worth reporting, though I did notice a link over on the Drudge Report, so maybe it'll eventually get some mainstream coverage.

Unlike contemporary POTUS election prediction models, the Berry-Bickers model starts with the presumption that all 51 voting jurisdictions must be predicted independently of each other, thus mimicking the Electoral College model set up by the Founding Fathers.  The CU professors also use six or seven economic indicators (such as the unemployment rate, per capita income, and so forth) as the primary means of determining how a state's POTUS PV will fall (and thus how its electoral votes will be apportioned).  The things considered by the opinion polls, and in fact even the opinion polls themselves, are strictly a secondary consideration.

By the way, did you know that a democrat POTUS has zero incumbency advantage if the unemployment rate is over 5.6%, and that no POTUS has been re-elected with an unemployment rate over 7.4%?  Seriously.

I referenced the updated projected outcome map in the commentary to my article three weeks ago, but now having received the revised state-by-state projections from the CU professors, I have what I need to nuance that outcome map a bit.  The state-by-state forecasts of the popular vote percentage, when sorted according to the incumbent's projected popular vote share, produce a rather interesting table picture.  Evidently, of the nine "barely there" states, seven of them are expected to be in the Obama-Biden column, and all save Washington State and Wisconsin will be within two points (with Nevada being in recount territory).  Also, apparently 266 of Romney's needed 270 electoral votes will come from states that he carries by a greater-than-ten-points margin (and the tipping point state will be Iowa).  The average ∆ PV% per state also shifts in Romney's favor by 2.73%, though the nationwide popular vote projection is still Romney - 52.86% / Obama - 47.14% (Δ PV = 5.72%).

It's like Laura Ingraham said in the video that I used back above the fold: there's no good reason that the Republican Party shouldn't walk away with this election 24 days from now, at least at the federal level.

>