Polls apart

The polls out in the last two days prove one thing: some of the pollsters do not have a clue what is happening. Gallup has the lead by Obama narrowing to 2%. Zogby and Rasmussen also show a narrowing lead, albeit larger. Battleground has been steady for the last three days at 3%. Then you have Hotline at an 8% lead and Pew research at a 15% Obama lead. Clearly some of them are very wrong.

Of course, there are the usual excuses for the variations. Pew somehow has 10% fewer Republicans and uses registered voters rather than likely voters. Still, the results are too far apart for there to be an explanation other than some error in the numbers.

Right now, I have to believe that the race is tightening. Obviously, that would be my preference, but when you add yesterday's TIPP margin of 2.8% (which hopefully will go down today), there are too many polls with numbers in the 2-5% range to be ignored. In other words, I think that the big lead polls are outliers or biased/flawed samples. Right now the RCP national average is Obama over McCain by 6.8 %. I think that the reality is about half that amount. Indeed, the average of Gallup, Rassmussen, Tipp and Battleground is 3%. There are also an average of 5.2% undecided in those polls. My estimate is that the undecided will break for McCain. After all, these are folks who have not yet been able to bring themselves to vote for Obama, and since Obama has been the clear focus of the election, my guess is that those who have not yet chosen will go the other way towards McCain. If McCain wins the undecided by 2 to 1, he will cut about 1.7% off of the Obama margin, bringing the Obama edge down to 1.3%

Of course, none of these numbers mean anything in particular except for this: The election is still far from over. A late move up by McCain in which only 2% of the voters switch sides could put McCain over the top.