Okay, Zogby has Obama ahead by two points (as did Diageo, which has been in the tank for Obama), and Hotline has it at one point. I've written recently about what we should do (fight like h--l) "when we take the lead again." I said I thought that would happen sometime around the beginning of the third week of October but perhaps as late as early November. Obama all through the primaries "over-polled" and "under-performed in the voting booth."

In the PA Primary (April 22), Wolf Blitzer and others said the race was "tightening" and we even heard their favorite cliche ("too close to call"). I said that Hillary Clinton would win by 10 percentage points. On April 23, it was at 10 points, although it fell to 9.2% by the end of the day. I do know PA voters (ornery!).

I also predicted Clinton could win OH, TX and IN, in all of which Obama "over-polled." Sen. Clinton won all three.

In 2004, Gallup came under heavy criticism from the far-Left for having "too many Republicans" in its survey template. Like terrified rabbits, Gallup changed its legendary template ("the gold standard"). As a result, it got both the national and the state races wrong.

It said in early November, 2004 that Bush would win PA by two points and Kerry would win Ohio by two points. I told the editor of the Pittsburgh daily that maybe Gallup had gotten the states confused. Kerry won PA by three points, and Bush won Ohio by two points. Gallup also predicted that Kerry could win FL, which he lost by hundreds of thousands of votes.

There has been a lot of talk about "the Bradley Effect," referring to the fact that voters tell pollsters that they will vote for a Black man -- and then don't. It does exist in this campaign, and there are a (very) few people trying to figure out its effect on this year's election. Why do people tell pollsters "fibs?" Because they think (correctly) that's what the pollsters want to hear.

One great caller in Beaver County, PA (Audrey) is so sweet and friendly that when she asks voters if they support John and Sarah, EVERYBODY SAYS YES!

Gallup's "internal numbers" have had McCain ahead as recently as last week (45-39), but it's buying the heavy "new" (Black mostly) vote theory, so it put McCain behind. Say what?? In the PA primary, Obama lost approximately 60 counties out of 67. He won some suburban Philly counties by relatively narrow margins. He won Philly (big), Pittsburgh (city), Harrisburg, and downtown Lancaster. Otherwise, he didn't win squat. His support in PA is very soft.

Beware the polls (especially CNN), my friends. I love those polls of course, but I don't really believe them on a day-to-day basis.