What's In a Poll?

What's In a Poll?

By Susan Estrich - August 21, 2008

Can John McCain really be up by five?

That's the headline on Drudge Report, based on a new Reuters/Zogby report, and whether it's "true" or not, it will almost certainly help McCain with his fundraising and hopefully rattle anyone inside the Barack Obama tent who was busy picking out the drapes for their office in the West Wing.

Of course, by the end of next week, even if Obama has made a mistake in giving half his convention to my friends Bill and Hill, and going for the roll call vote (instead of just taking the debt) and even if his football stadium speech does play right into the hands of the McCain "He's just a rock star" message, the senator from Illinois should still be ahead. And then, by the end of the next week, barring any speeches about Holy Wars by the likes of Pat Buchanan, McCain could be up again.

There's an old saying in show business: No one knows anything. Movie executives, at least the honest ones, will tell you that if they'd given green lights to half (actually way less than half) of the pictures they said no to, and said no to all the ones they said yes to, they'd probably be in exactly the same position they are. If anyone tells you that they know, at this point, how the election will turn out, then you can be sure that they really don't know anything at all.

That's not to say that elections are inherently unpredictable. Sometimes you know. Believe me, I knew in 1984, as I traipsed around the country with Fritz Mondale, that at the end I'd be going back to my house in Cambridge and my job as a law professor. And I knew, earlier than I'd like to admit, that I wasn't on my way to a corner office in the White House in 1988, even when we were still up in the polls in the summer. There was a reason I gave out the Ambassadorship to the Court of St. James to anyone who asked: I knew no one would ever hold me to it. Sometimes you can look at the numbers, and doom stares you in the face. Always believe your own polls, especially when they're bad, my friend and pollster Tubby Harrison used to tell me. By the summer of 1988, the country had turned from believing we were on the wrong track to thinking we were on the right track. They thought my candidate, Governor Dukakis, was more conservative than he actually was -- that's what beating Jesse Jackson every Tuesday will do for you.

By the fall, it was clear: right or wrong, they didn't like him. Of course, you never say that to the candidate; you tell him "they don't know you," not that "they don't like you." You say it's a communications problem, which is why being the communications director is the hot seat in a losing campaign.

But this isn't one of those elections. The polls don't tell you what you need to know, at least not yet. In most of the key states, neither candidate has reached the 50 percent mark, the point where all you have to do is hold on to the people who already like you, as opposed to persuading or dissuading the ones who don't. To be sure, the underlying indicators favor a Democrat: unpopular war, uncertain economy, ridiculous gas prices, home foreclosures, not to mention a Republican president who is down to the immediate family when it comes to support.

The generic Democrat beats the generic Republican by as much as 15 points in generic match-ups for Congressional seats. But generic candidates don't run: real ones do, and Barack Obama is running behind the generic Democrat and John McCain is running ahead of the generic Republican, at least if you believe the current polls. I'm happy to bet the mortgage that Democrats will do very well in both the Senate and House races this fall, but I don't think I'll find any takers. But I'm not making any bets, at least not yet, as to who will be moving in to 1600 Pennsylvania.

There are plenty of reasons not to trust the polls, but they cut both ways. On the one hand, my Democratic friends will tell you, with good reason, that the polls are a better indicator of who would have won the last election than who will win the next one; that they simply don't tell you, because the people taking them can't know, what the electorate will look like come November. Will African-Americans turn out in the kind of numbers that make any sample based on past percentage obsolete? Maybe. Will all those young people and non-voters that seem to flock to Obama rallies actually pull it together to register and vote this time around? Maybe. Maybe the polls underestimate Obama's true strength.

Or maybe they overestimate it. What worries me, as a Democrat, is the pattern I got very used to every Tuesday during primary season. Fool me once, shame on you; fool me twice, shame on me. The night before the New Hampshire primary, I was as certain as you can be in politics that Obama would win and Hillary would lose. So were many of my friends in Hillary-land, who were looking at the same polls I was and telling me that the "line" was that anything within single digits (meaning a loss by a margin of single digits) should be considered a victory. I thought Obama was on a wave. He wasn't. The polls were, quite simply, wrong. They overstated his strength. After that, I stopped believing exit polls in big states. "Too close to call," Drudge Report claimed every Tuesday, complete with the little flashing siren at the top. My friends in Hillary-land would e-mail, panicked. Not me. I knew.

California was not too close to call. Neither was Massachusetts, or New Jersey or Pennsylvania or Ohio. I could go on. The point is not that the polls were "wrong," but that people weren't honest with the pollsters. Was it because they didn't want to admit that they weren't voting for the black candidate? Maybe. In California, we call it the "Bradley effect," after the former Los Angeles Mayor who led in every poll for governor (including the exit polls), but lost on the secret ballots. Whatever it was -- or is -- there's certainly reason for Democrats to worry that polls that show this race being too close to call mean that McCain is actually ahead, that he would win if the election were held today.

Of course, it's not being held today. That's the problem with these horse race polls. And it's not being held at the end of next week, at the conclusion of the Democratic Convention, or at the end of the following week, at the end of the Republican Convention. It's still two debates, and countless ads and events, away. And the people who will most likely decide it are watching the Olympics right now.

Susan Estrich

Author Archive

Follow Real Clear Politics

Latest On Twitter