Clive Crook at the Financial Times has a good look at why the polls in July are virtually meaningless for elections in November. I find his argument compelling.

One cannot help but be struck by the current disconnection in U.S. presidential politics between, on one hand, the excitement and enthusiasm that attend Barack Obama’s candidacy and, on the other, the tightness of the race according to recent polls. The first suggests a sweeping victory for the Democrat in November, the second a close election and the distinct possibility of a win for John McCain.

Most would agree that Senator Obama has so far waged a polished and efficient campaign. He stumbles occasionally. His endless iterations on troop withdrawals from Iraq, for instance, have given Senator McCain a valuable opening. But he recovers quickly from his missteps and despite them projects an air of competence and assurance that belies his lack of experience.

Mr McCain, in contrast, has all the experience one could wish but little, these days, of the composure and gravitas that it is supposed to confer. When he trips up, the error seems to stick and he looks ridiculous. Mr Obama looks presidential. Mr McCain, sad to say, does not.

So, what gives? If this is the picture — and it is as far as I can tell — why are the polls so close?

The simple answer may get me ejected from the guild of political commentators, who have a lot of space to fill between now and November – but I report it nonetheless. It is that these early head-to-head polls and the vast enterprise of political analysis, nit-picking and minute speculation they support, are, to a first order of approximation, worthless. In short, you resolve the paradox by ignoring them.

Right, but then what?

Alan Abramowitz, a politics scholar at Emory University, has shown that summer head-to-head polls convey almost no information about the forthcoming election. (Subsequent head-to-head polls are not much better.) Instead, he has a simple “electoral barometer” that weighs together the approval rating of the incumbent president, the economy’s economic growth rate and whether the president’s party has controlled the White House for two terms (the “time for a change” factor). This laughably simple metric has correctly forecast the winner of the popular vote in 14 out of 15 postwar presidential elections.

All of this speaks for Obama. With his obvious intelligence, support from the troops in his trip to Afghanistan and disaffection with Bush, Obama is a virtual shoe-in for the election.

Read Crook’s column to get the full story. It’s a good read, although it’s a paid site.

Source
One simple way to predict a victor, FT