Saturday Afternoon

A note of reminder to those of you watching the polls closely: When pollsters brag about how accurate they were in previous elections — or when someone says, "This pollster was on the money last time so he's probably right this time" — they're comparing the final predictions by the pollster to what happened on Election Day. They're not evaluating in any way how correct that pollster was in what he said two months before the election, two weeks before or usually even two days before. They can't. No one will ever know how precise those predictions were. At most, you can say how much they matched what other pollsters said at that point.

I can say right this minute that all the pollsters are wrong — that Roseanne Barr is going to win 344 electoral votes — and no one can prove I'm wrong. Maybe if the election were held today, she would. And then November 5, my polling "detects" a massive switch to Obama or Romney — one of them winning by two points — and I'm back in the running. I might even wind up being the most accurate pollster of all.

Collectively, when properly chosen and averaged, the major pollsters usually chart a stable trendline to Election Day…though surprises are always possible, especially when you forget that there is a margin of error in all these polls. If you're two points ahead of me in a poll with a two point margin of error and I win, that should not be a shocker. It isn't even much evidence that the poll was wrong.

One other thing about polls. A friend of mine who used to be part of the Washington press corps once told me you should always ignore anything having to do with a campaign's "internal polling," meaning the polls the candidates themselves do and sometimes cite or leak. For one thing, he said, they're usually lies. A pundit or reporter who claims to have seen internal polls may very well be lying or the source which told him of the poll may well have been lying to him. At best, according to my friend, the campaign did eight internal polls, six were bad news and two were good, then they circulate only the two good ones, omitting any real data on how the polls were conducted.

At this moment, despite some claims of "Mittmentum," it looks to me like Obama has a narrow but significant lead in enough states to make it to 270 electoral votes, the popular vote looks closer…and Romney ain't going up. Something, perhaps relating to that big hurricane that's about to hammer the East Coast, will happen that might be a game-changer. I doubt the game will change much but it could.

Whoops! This just in: My polling shows that Roseanne Barr is now poised for a 50 state sweep, winning all the electoral votes this time plus several that were cast for McCain-Palin four years ago. Stunning. This may not hold up until Election Day but you can't prove it ain't true at this moment.