E-minus 2 days and counting....
Poll Composite: Bush 47.8%, Kerry 46.2% (Bush +1.6 – no change)
The great thing about poll analysis is that there are so many prisms through which you can interpret the data. Also the rotten thing.
The naked partisan in a Bushie's position can launch chapter and verse on polling methodology and skewed samples and the like. This, of course, can be at least somewhat negated by averaging polls to net out those differences and biases between surveys, which was the genesis of my polling composite.
The sadder but wiser analyst, no matter how pro-Bush s/he is, has to take other factors, like current events and an avalanche of last-minute Big Media "gotchas," into account. That's what I've striven to do over the past few days.
So, what does today's composite mean? Well, it suggests to me that whatever negative impact the OBL video had on Dubya was short-lived, and since 60 Minutes didn't drop any fresh daisy-cutters tonight, Big Media just might be out of ammunition. It also might mean that the "tracking polls on weekends are as credible as Ted Kennedy addressing an Alcoholics Anonymous convention" argument is a significantly bigger factor in all of this.
Let's take the Fox News/Opinion Dynamics tracking poll in particular. Today's number shows Bush and Kerry tied at 46%. That's a five point move toward Kerry in the space of three or four days. Not a good sign, right?
Ah, but look at those delightful internals. Specifically this one noticed by the KerrySpot:
"I am told the poll shows Kerry leading with men 47-45, Bush leading by 1 with women. That strikes me as a bit odd. Were many men out watching college football during this period?"
Indeed. I will personally shag footballs for Kerry on the White House south lawn before I'll buy that Mr. French is leading among male voters. Bush by ten (to use a nice, round figure) is a lot closer to reality.
I did some back-of-a-bookmark arithmetic this morning between Sunday School and worship services to plug in that adjustment figuring the daily sample size of 400 respondents. The result? This purported 46-46 tie becomes a 49-43 Bush lead, which approximates the 50-45 lead that Fox posted for the President in their normal survey released last Tuesday. And plugged into my composite, that moves GDub's lead back to three points, well outside the vote fraud discount zone (see below).
Ergo, by this theory, tomorrow the tracking polls should show some movement back towards the President as Friday's numbers pass off the board and are replaced by Monday's.
At the very least, barring a massive al Qaeda strike tomorrow, there shouldn't be any more outside events - genuine or nakedly contrived - to influence the race.
Electoral College: Bush 223, Kerry 185 (tossups 130)
-New Jersey moves from tossup to "leaning Kerry"
-Pennsylvania moves from "leaning Kerry" to tossup
Popular Vote: Bush 50.2%, Kerry 48.5% (Bush +1.7 - down 0.1%)
Electoral College: Bush 301, Kerry 237
Minnesota and Wisconsin switch places again.
All you can do is chuckle at this point. Zogby shifted Iowa four points and New Mexico ***NINE*** points toward Kerry overnight. He's still over the Kerry cliff on Wisconsin (seven points), but is neatly balanced by Gallup which has Bush up there by eight. But by the same token, Gallup has Kerry ahead by eight in Minnesota, which is the one and only reason Minnesota flipped again in my polling average. In this they echo the notorious Minneapolis Star-Tribune, which now has an assortment of assholes out of which to defecate thanks to the gents at Powerline.
Florida: Bush +0.8
Hawaii: Bush +0.9
Iowa: Bush +0.4 (+1.3 w/o Zogby)
Michigan: Kerry +3.3
Minnesota: Kerry +0.8 (Bush +1.0 w/o Gallup)
Nevada: Bush +4.0
New Hampshire: Kerry +2.0
New Mexico: Bush +2.0
Ohio: Bush +1.0
Pennsylvania: Kerry +1.0 (+3.5 w/o Gallup, which has Bush up four)
Wisconsin: Bush +0.2
Today's bottom line: It's all but impossible to figure Democrat vote fraud into election forecasting because there's no way to tell how much of it will succeed and how big an impact it will have on the outcome from state to state. Accepting that arbitrariness as the foundational premise, I've used a twofold rule of thumb: if Bush's national lead falls below two points, but his projected final number is above fifty percent, any state in which his lead is under a point is assumed to be stolen by Kerry. If the President's projected final number falls below fifty percent, any state in which his lead is under two points is assumed to be stolen by Kerry.
As of tonight, Bush leads in my composite by 1.7% with a projected 50.2% of the popular vote. Accordingly, Florida, Hawaii, Iowa, and Wisconsin would, by these criteria, be larcenously flipped over to Kerry, giving him the Electoral College 285-253.
I've heard experts who know a lot more about this stuff than I do suggest the same two percent fraud buffer, so I'm assuming it has some validity.
I've gone on at some length about this tonight to save myself keystrokes tomorrow, when I'll be making my final election predictions, both nationally and state by state ANNNNNND the Senate races as well.
Suffice it to say, tomorrow the President has to get back above a two-point lead nationally, or he - and America - are going to get robbed.