A thank you to Obama's Grandmother:

As you may have heard, Obama's grandmother passed away this morning. I just wanted to take a second to thank her for raising the next great president of these United States.

So, thank you Madelyn Dunham. Rest in peace knowing that you have left us in good hands.

My final predictions:

The more things changed this week, the more they stayed the same. Though a tightening has occurred on a national level and in some state races, the hardening of support for either candidate has put Obama over the top where he needs it most: the swing states.

Taking into account the wide range of information available, I predict Obama will win a minimum of 291 electoral votes. That means he will cross the 270 electoral college vote threshold and win the election. Obama will win all of the states that John Kerry won, including New Hampshire and Pennsylvania, plus:

Nevada, where he has been over the 50 percent mark for several days;
New Mexico, where has had a comfortable lead for quite some time;
Colorado, where the race is close, but again, Obama is over the 50 percent mark and;
Virginia, where he is over 50 percent over time.

That gives him a minimum of 291. The rest remains largely unknown. If one believes the polls that show Obama up by wide margins because of a supposed increase Democratic turnout, than Obama could pick up Missouri, Ohio, Florida, Indiana and North Carolina. That would give him 375 ECV's. I tend to think that is optimistic.

My original projection was Obama 313 - McCain 225, with McCain keeping Ohio, Missouri, Nevada, North Carolina, Georgia and Indiana. But, I think it is clear that Obama has a good chance of picking up Nevada, giving him 318. I'm sticking with my prediction that Obama wins Florida, and adding Ohio to his column - giving him 338.

Indiana, North Carolina and Missouri appear to be pure toss-ups, but I tend to think that McCain has a bit of momentum.

That makes my final tally Obama 338 McCain 200

Nate Silver reads my blog?

I could just be imagining things, but it would seem that Nate Silver over at fivethirtyeight.com has been viewing my blog. He wrote an article today talking about the "myth of the lag." My post from yesterday was also on the 'lag.'

Silver was referencing a slightly different aspect of the problem, but he accidentally reinforces my point:

"One of the most bizarre but oft-repeated myths about polling is the notion that state polls 'lag' behind national polls, and particularly lag behind national trackers. This is largely a relic of a bygone era in which newspapers might commission a poll in their state, and then sit on it for several days until their Sunday editions or some other ebb in the news cycle. Thus, you might be seeing state polls that were in the field a week or so ago, whereas the national trackers were more up-to-date.This largely is not true today, however. Rasmussen and SurveyUSA, perhaps the two most prolific public pollsters, generally release their data no later than 24 hours after it has left the field; likewise with other pollsters like InsiderAdvantage and Public Policy Polling. A
couple of other pollsters like Quinnipiac and Mason-Dixon will occasionally sit on a poll for 24-48 hours, but generally not more than that. Every now and then, you'll have some small college or some fledgling marketing firm release a poll that is a couple of weeks old, but this is unusual, and it's easy to notate the exceptions. Most of the big, business-savvy pollsters recognize the importance of timeliness in this era of 24/7 news cycles.Conversely, some of the national tracking polls are actually not all that fresh. IBD/TIPP has a 5-day polling window. Battleground also has a 5-day polling window, and they don't poll weekends, meaning that they're usually including some data that is a full week old.Certainly, there is some mystery as to why the state polls and the national trackers seem to have diverged somewhat of late, with John McCain picking up perhaps 2 points in the tracking polls versus last week's averages, whereas the state polls haven't really budged one iota. The 'lag' effect, however, is not a valid explanation.

So even though there is probably not a difference between state and national polling, there is a significant lag between state polling and and all of the wonderful but wonky election prediction models - like Nate's. Most of the sites, like fivethirtyeight.com, electoral-vote.com and pollyvote.com, are at least two, sometimes three days behind. And Nate also forgot to mention that while national polls are out daily, state polls are considerably less frequent. Sometimes there are days between polling samples in the less contentious states - meaning that all of those beautiful maps and algorithms are operating on less-than-perfect data, which furthermore, is essentially averaged with previous polls to limit wild swings caused by outliers.

It doesn't take a statictician to figure out why the national polls are tightening, while the prediction models all show a blowout.

Add to that, the whole issue of voter turnout models and differences between national voter turnout models and those used by individual pollsters in individual states, and one is left with a fairly low confidence level in the standard views about the upcoming election.

Again, I stand by my previous prediction of an Obama victory. But, the race may be closer than some are willing to admit. For Obama's sake, I hope I am wrong.