BOB GARFIELD:

In political coverage, even infinitesimal changes in a candidate’s poll numbers make for headline news.

FEMALE CORRESPONDENT:

Newt Gingrich is emerging as a solid frontrunner in the Hawkeye State. A brand new Washington Post ABC poll has him 15 points…

MALE CORRESPONDENT:

Republican frontrunner Newt Gingrich had surged through November, but our new poll out tonight shows that December has not been so kind.

MALE CORRESPONDENT:

And now we’re seeing the rise of Ron Paul. The GOP presidential hopeful is surging in the latest polls, this as the frontrunner…

BOB GARFIELD:

Of course, as the data crunchers say, “Garbage in, garbage out.” According to Slate’s Sasha Issenberg, some of the assumptions made in media polls of self- described likely voters may skew the results. In a piece this week, Issenberg describes a study which found that more than half of the people who tell pollsters they won't vote in an upcoming election actually do vote. And since those people don't get counted in media polls, those polls aren't always as informative as you'd like. Issenberg says the problem lies in the faith media polls place in voters to be honest.

SASHA ISSENBERG:

Anybody can pick up on the other end of the line, somebody who’s not registered to vote, somebody who’s not 18 years old, somebody who’s not a citizen or somebody who is not likely to actually cast a ballot.

So what pollsters do is they have a battery of questions that they call a screen that’s designed to filter out those who are probably going to vote from those who won’t vote at all. And from that they're able to show you numbers that are supposed to represent what the electorate will look like on Election Day.

We've always known that people lie to claim that they’re more likely to vote or after elections, lie that suggest that they voted when they didn't. The real surprise is people who say the opposite, who tell a lie that theoretically makes them look bad, that they won't vote but end up doing it.

BOB GARFIELD:

It's kind of like telling a pollster, oh yeah, I use porn, or – I totally leave the water on when I brush my teeth. It seems like volunteering a socially unacceptable answer that isn't even true. Why would you lie against yourself?

SASHA ISSENBERG:

One idea is that people know that saying that they won't vote will get the pollster off the other end of the phone. But another theory is that people use it as a way of sort of articulating their embitterment about  political institutions or the media. It's a way of saying, you know, I don't want to be part of this.

BOB GARFIELD:

And the campaigns have a very different methodology than the one employed by the media pollsters. How do they get a better notion of who is actually likely to vote?

SASHA ISSENBERG:

They’re no longer reliant on people to be honest about what they will do. They’re relying on these massive databases that they've built of voters and their political histories to segment out people whose past behavior, notably whether they've regularly voted in the past, makes them look like somebody who’s likely to vote in an upcoming election.

BOB GARFIELD:

In Iowa, you write that the difference between the data generated by this kind of filtering and the data generated by the media pollsters is particularly significant. Why is that?

SASHA ISSENBERG:

It all comes down to the fact that in Iowa the caucuses are run by the parties and not by state officials, so the most valuable piece of information is a list of people who caucused in the past. And these are maintained by the Democratic or Republican parties. You sell them to candidates for often six figures. None of the news organizations actually have this.

So campaigns are building their samples out of a reliable list of people who look like voters, while media organizations are calling people randomly, and what I assume we're seeing is that there's far more volatility reflected in the media polls than the ones that the campaigns are saying internally.

BOB GARFIELD:

Does that mean that when we see the ebb and flow of Romney fortunes or Gingrich’s or Paul’s, at the moment that it's meaningless?

SASHA ISSENBERG:

You know, media pollsters putting out calls today and somebody who, let’s say, saw Jon Huntsman on The Tonight Show last night and is momentarily excited by Jon Huntsman or just got a piece of mail from Newt Gingrich might tell a pollster that they're likely to vote in two weeks. That person may not even be a Republican, may not be registered to vote and may not be that likely to turn out on a Tuesday night in January. And a public pollster has no way of differentiating that person from a likely voter other than what  they say on the other end of the line.

And so, that probably means that somebody like Ron Paul, who provoked a lot of enthusiasm from people who were untraditional caucus goers, looks a lot stronger in the polls that we’re seeing from newspapers and television networks than he does in the polls that his opponents are commissioning  privately.

BOB GARFIELD:

The last presidential election showed a flaw in the methodology of the private pollsters who did not really account for growth in the electorate.

SASHA ISSENBERG:

Well, what we saw in Obama’s victory in Iowa and around the country during the primaries and the general election was that he was bringing new people into the process, people who hadn't been previously registered.

And so, that's one place where public polls that call randomly and just ask for a sort of self-reported indicator of enthusiasm will pick up significant changes in the composition of the electorate faster than calling off of voter files.

BOB GARFIELD:

So the private pollsters have a significantly more accurate methodology, except when they don't.

SASHA ISSENBERG:

That's about right.

BOB GARFIELD:

All right, Sasha [LAUGHS]. Thank you very much.

SASHA ISSENBERG:

Thanks.

BOB GARFIELD:

Sasha Issenberg is a columnist for Slate and author of the ebook, Rick Perry and His Eggheads:  A Sneak Preview from the Victory Lab.