Interpreting The National Polls

Chris Cillizza writes that the "surprising closeness" between Barack Obama and John McCain in recent poll has "emboldened" some Republicans who believe that John McCain ought to be trailing by double-digits.

Should Republicans be so emboldened?

First, don't fall for the availability heuristic and obsess over just the latest poll, which comes courtesy of Newsweek. The poll of polls and internal campaign polling for both campaign gives Obama a lead of anywhere between four and nine points using a tight likely voter model.

It's almost an axiom of modern political polling that, depending on the question and the sample, surveyees extremely susceptible to daily, even hourly, shifts in elite opinion. Voters are paying attention to the big picture, but they're not paying attention to the details just yet. Methodological differences account for random fluctuations, as do random fluctuations. As both candidates have had rough weeks, it's entirely predictable that these polls would regress to a mean.

Much more useful this year would be polls of the 18 or so competitive states. When last polled, Obama was beating or tying McCain in places like Indiana, Georgia and North Carolina. If, over the next few weeks, we see that he continues to perform above Democratic presidential levels there, then we can pretty much disregard the national surveys until the fall. The polls don't account for the force multiplier effect that Obama's campaign will almost certainly bring to bear with its millions of volunteers and thousands of paid staffers. Whether that effect is 1.01, 1.05 or even 1.3 -- we don't know yet. But even the McCain campaign acknowledges its existence.

Also: I think McCain is leading among independents because many Republican leaners now identify as independents. Self-described voter identification is useful, but when figuring out where independents are coming from, it'd be more useful to have the independents broken down by ideology. Second, the Dukakis analogy doesn't hold; by the same logic, Obama could be up 25 points by election day if he's even now's up by only four.

The election pits a likeable man who belongs to a party no one likes against a likeable man about whom a lot of Americans don't know what to think. Opinion is going see-saw. The McCain campaign deserves technical points for lowering expectations, and the Obama campaign deserves technical points for almost preternatural self-confidence about their strategy.