Why should we trust this man?

Frank Luntz is king of the pollster pundits, but don't ask him where his numbers come from.

Published May 26, 2000 6:00PM (EDT)

Whatever it is about the pollster Frank Luntz -- the baby face, the unctuous charm, the ever-changing hairdo -- no one seems to be able to get enough of him. Even since last week, when his sole candidate, Rudy Giuliani, made a tortured exit from the campaign trail, Luntz still makes the media rounds, from CNBC's "Rivera Live" to the New York Times.

He's possibly the best example of what we could call the pollster pundit: someone who both purports to scientifically poll the opinions of the public, and then also interpret that data to support his own -- in Luntz's case, conservative -- point of view. This is what allows Luntz to face a room full of journalists and, in all seriousness, proclaim George W. Bush's jittery, time-delayed appearance on David Letterman -- the one which prompted boos from the audience -- a total success.

It's what allows Luntz to proclaim that Giuliani would've been no more hurt from his admission of marital difficulties than his admission that he has cancer. "He beat crime, he beat drugs, he beat unemployment, he beat welfare, he beat trash in the streets, he beat the squeegee guy," Luntz said. "He's like a mayor machine."

What's more, it's what allows Luntz to do this without citing a single polling result, a single number, and yet still be taken quite seriously as a pollster.

In a city full of cynical, number-hungry journalists who pride themselves for taking no one at their word, this would seem to fall under the category of a neat trick. Then again, Luntz is a special case. Since about 1994, he has been a rising star in politics (Time listed him that year among "50 with the
requisite ambition, vision and community spirit to help guide us in the new millennium"), part of a new class of media personality, the celebrity pollster. Along with fellow Republican darling Kellyanne Fitzpatrick, he's regularly called upon to explain candidate strategies and voter reaction.

But while the media can be disdainful of Fitzpatrick (through the occasional referring to her as a "pundette"), Luntz gets the heavy-hitter treatment, frequently getting called in by the networks to offer color commentary on politics even when he has no poll to cite.

The day Luntz spoke of W's Letterman appearance, he wasn't just beguiling an uncombative TV audience. He was addressing the Sperling Breakfast, the long-held meeting of powerful journalists organized by Godfrey Sperling of the Christian Science Monitor. (Full disclosure: I'm a Monitor political columnist.) These breakfasts can certainly be stuffy, but there's no denying that an invitation to speak at one is tantamount to a stamp of approval by Washington's establishment press, with previous invitees including Al Gore and Ken Starr.

The morning of the breakfast, Luntz showed up ready to talk about politics armed with nothing more than a tape of himself conducting a focus group on MSNBC. He spun fast and furiously, and then told everyone in the room the reason he was giving them his information (the less-than-shocking revelation that Bush needs a good chunk of McCain's supporters to win the presidency) is that he wanted the press to pass along the news to Bush. He would have told the Bush campaign himself, he said, but his commitment to his job as a media pollster -- with occasional gigs for outlets like MSNBC and U.S. News & World Report -- forbids it. The crowd chuckled, nodding knowingly and kept writing.

Of course it's hard to know how seriously to take Luntz. His "polling" and "analysis" always seems half-serious and half performance art. Whatever he really means, however, the media generally seems to take him at his word. Much of the polling industry, however, has been more circumspect.

In 1997, Luntz was formally reprimanded by the American Association for Public Opinion Research for his work polling on the GOP's 1994 "Contract with America" campaign document.

Luntz told the media that everything in the contract had the support of at least 60 percent of the general public. Considering the elementary phrasing of that document (stop violent criminals, protect our kids, strong national defense), it seems almost laughably uncontroversial. But one of AAPOR's 1,400 members wasn't so amused, and filed a complaint requesting to see Luntz's research and a verification of the figure.

Luntz's response? He couldn't reveal the information because of client confidentiality.

"None of those people have ever worked for a private client," Luntz says now. When told that some members of AAPOR do actually work in the private sector, he replied: "Then they should understand about confidentiality."

In fact, Luntz says, the AAPOR slap had a surprising effect. "Look, I shouldn't say this," he says, "but I made money off that incident. People basically said, 'If you're willing to go through that to honor your commitment, I want to work with you.'"

But what about AAPOR's claim that when you make results public, you owe it to people to release all the data. "I don't agree," he says. "Say you poll on an environmental issue, and on eight of the 10 questions the numbers are in your favor. Why release the other two? It's like being a lawyer ... This is my case, and these are the strong arguments and these are the weak ones. You go with your strongest case."

There are a few problems with this analogy. First, pollsters aren't lawyers; they are (in theory) researchers and are treated by journalists as such. Second, in a trial there are prosecutors and defense lawyers and everyone is working off the same page. There is an established pool of evidence that either side can argue over. What Luntz proposes is a trial in which a lawyer makes his case with no opposition and no opportunity for a jury to consider the source.

"These are not complicated questions," says Diane Colasanto, who was president of the AAPOR when it reprimanded Luntz. "It is simply wanting to know: 'How many people did you question? What were the questions?' He did finally give us some information, but it wasn't enough. It didn't really explain what the figures were based on. All we could tell was it seemed like there might have been some survey done.

"We understand the need for confidentiality, but once a pollster makes results public, the information needs to be public. People need to be able to evaluate whether it was sound research."

Warren Mitofsky, the current standards chair at AAPOR, says the complaint against Luntz was a rarity. There are only one or two complaints filed against pollsters each year and they hardly ever go as far as Luntz's.

Of course, critics of the AAPOR complaint note that the group is not really the domain of political pollsters. The group's membership mostly includes academic pollsters and commercial research firms. But even in the world of political pollsters, Luntz is a special case, says David W. Moore, author of the book "The Super Pollsters."

"And it's not just a question of being a political pollster," Moore says. "Both Peter Hart and Bob Teeter are political pollsters and both are very solid. When you hear them talk, they genuinely try to analyze the data." (Hart, a Democratic pollster, and Teeter, a Republican, work together to poll for NBC and The Wall Street Journal.)

Luntz's work and that of other celebrity pollsters, such as Fitzpatrick, is nothing more than "propaganda" masked as research, Moore says.

"What bothers me is they are given so much prominence. One of the reasons media organizations started doing their own polling was to make sure they wouldn't get biased data." Now, the media pay these people to poll, he says. "The whole trend is really a backward step."

Luntz has risen to media stardom on the strength of his work with focus groups, typically gatherings of a dozen or so people, carefully screened to be representative of a larger population. A moderator leads the group in a discussion, and carefully chooses questions to elicit the participants' deep feelings about candidates or issues.

And Luntz is an able moderator. Watching him work a room is like watching a good politician: He's bright, funny, amiable and connects with his subjects. He speaks in simple, direct sentences, and asks questions like "Is Bush a smart guy?" or "Does he have what it takes to be president?" He's a first-rate empathizer, all grins and furrowed brows.

This has undoubtedly helped the media pine for him, but it's also fallen in love with those wonderful little gizmos he often gives his groups that allow voters to instantly make their opinion known during a speech or debate -- a little dial that they can turn one way for approval and the other for disapproval.

As an event unfolds, he sits back and watches a constantly shifting fever-line that shows which statements people favor or dislike. If you have ever wondered about the fickleness of the American public, just watch one of Luntz's dial tapes as I did during the 1996 presidential debate in Hartford, Conn. It's an experience in terror. One line may elicit a negative response, while a different line meaning basically the same thing is all positive.

Of course, political focus groups aren't new. Their use dates back to World War II. They allow the moderator to dig into specific questions more deeply than "Do you support abortion rights?" or "Do you use Clorox?" and check for emotional responses.

They are not, however, substitutes for polling data. They create no hard numbers, and since the groups are usually small, it's hard to extrapolate any definitive results. "You cannot generalize from the results of a focus group, period," says Mitofsky. "You can get results you can explore in a real survey, but that's all. A lot of people do research on the cheap and that's a good way to get in trouble."

But Luntz argues that regular polling often misses what's really going on with voters. "Human behavior studies have consistently proven that people will reveal their innermost thoughts only to those [with whom] they believe they share a common bond," he wrote in a 1994 polling-report article called "Voices of Victory." "If conventional wisdom and telephone polls were accurate, Ross Perot should have barely scraped into double digits in 1992."

Luntz's reasoning is thought-provoking, and sounds kind of sensible -- except that the polls leading up to the 1992 election were actually pretty on target. Just days before the voters went to the polls, ABC had Perot with 18 percent. A Harris Poll had Perot at 17 percent. He wound up with 19 percent.

Some days after his appearance at the Sperling Breakfast, I ran into another journalist who was at the table that morning. Didn't it bother him, I wondered, that Luntz offered no numbers to back up his assertions and that he said he was using the press to pass a message to Bush, regardless of how silly that message was?

"Luntz is different," he told me. "He's not a regular pollster. He's more interpretive."

But the whole point of polling is to collect data. It is not a perfect science but, done properly and explained well, polls yield hard numbers that can be discussed.

What Luntz does is something else: scientific man-on-the-street interviews. And regardless of how scientific the samples are, how serious can one take the results when there are no numbers to point to and the moderator happily admits he's doing what he can to aid a specific party?

"We call people pollsters when they poll," Moore says. "Why don't we all call Luntz a focus-groupster?" Probably because that wouldn't sound nearly as good on TV.

By Dante Chinni

MORE FROM Dante Chinni

Related Topics ------------------------------------------