During a fundraising breakfast at Manhattan's swanky Metropolitan Club yesterday morning, Barack Obama deliveredan unusually dour message to some of his top contributors. "For those of you who are feeling giddy or cocky or think this is all set, I just have two words for you: New Hampshire," he said. "I've been in these positions before when we were favored and the press starts getting carried away and we end up getting spanked." Obama was referring to his nearly three-point loss to Hillary Clinton in January's Granite State primary, which came after the final pre-primary polls showed him ahead by an average of 8.3 percent. The senator felt so strongly about the analogy, in fact, that he repeatedit this afternoon on a conference call with his staffers. "Run scared," he said. "Remember New Hampshire." It's worth nothing that some Republicans have told me the same thing. Ignore Obama's seven-point advantage in the national polls, they say. He could get always hit with another New Hampshire.
These warnings--which imply that current polling averages could be off by ten or more points--serve obvious political purposes. For Republicans, they help keep hope alive amid a daily flood of surveys that, collectively, show Obama ahead by an estimated 190 votes in the Electoral College. For Democrats, they help ensure that no one--staffers, volunteers, potential voters--takes anything for granted down the homestretch. Both parties are, in effect, encouraging supporters to show up on Election Day no matter how bad (or good) the polls may look.
my question: Is it reasonable to believe--based solely on what happened
in New Hampshire--that Obama could wake up on Nov. 4 with a seven- or
eight-percent lead and wind up losing to McCain by a couple of points?
The answer: I don't think so. The first problem, of course, is that comparing primary polling to general-election polling is a total apples-and-oranges maneuver; you're talking about two different electorates (Democratic primary voters vs. Democrats, Republicans and everyone else). Setting that consideration aside, however, there's simply no reason to use New Hampshire--as opposed to the other Democratic primaries--to predict what could or couldn't happen on Election Night. Given that we talking about a national election, a more rational approach would involve comparing Obama's actual primary performance to pre-primary polling trends across the entire spectrum of primary states. Luckily, the indispensable Nate Silver of FiveThirtyEight compiled these statisticsback in August. What he found is that New Hampshire was the exception, not the rule. Over the course of the entire primary season, Obama actually outperformed the polls by an average of 3.3 percentage points. In the West, he did 1.1 percent better than pollsters predicted (on average). In the Midwest, he surpassed the surveys by 3.1 percent. And in the South, he exceeded expectations by a whopping 7.2 percent. The only region where Obama underperformed, in fact, was the Northeast; there, he finished about two points below the pre-primary trendline. That said, Obama is currently winning every northeastern state--including New Hampshire--by more than 10 points; he'd probably be willing to trade a few points in New England for a few points in Virginia, Colorado, Ohio or Florida.
That's not to say that Obama couldn't lose his lead on Election Day. It's just that the Democratic primary results--again, including New Hampshire--give us no reason to believe that he will. In fact, they suggest (if anything) that the senator could outperform the polls by a point or two on Nov. 4. That said, things could always change. According to Blumenthal, the most compelling explanation for what happened in New Hampshire isn't that Granite Staters lied to pollsters about race--i.e., the Bradley Effect--or that pollsters didn't screen for likely voters as well as they should have. It's that there was simply a last-minute shift toward Clinton. Now, that's much less likely to occur in the general election. Back in January, 80 percent of New Hampshire's Democratic primary electorate approved of all three major candidates--Obama, Clinton and John Edwards--which meant that switching sides wasn't much of a leap. The general electorate, on the other hand, is sharply divided (and largely decided) on Obama and McCain. Still, if the polls narrow in the final weeks of the presidential race--as they're wont to do--a defining moment could move the needle in McCain's direction. (Remember Hillary's tears?) Ultimately, says Blumenthal, reminding voters of New Hampshire is an "entirely prudent thing for Obama to do. There's enough uncertainty about this race that it would be foolish for anyone to say it's done."
Even if New Hampshire itself is sort of irrelevant.
UPDATE, Oct. 18:In the comments section, reader "Philosopher" floats the theory that
Obama only overperformed in caucus states. This isn't true. In fact,
only Iowa and Nevada were polled heavily enough to be included in
Silver's analysis; otherwise, the numbers above are based entirely on
Obama's performance in primaries.