How Good Are Polls? We Refuse To Answer

Ross Perot's quirky run for the White House is turning into a full-fledged gallop. That's if you believe the polls: The Texas billionaire leads George Bush 36% to 31% in a recent survey taken by Times Mirror Co. Bill Clinton is in third place with 27%.

But wait. A recent New York Times/CBS News poll had Bush leading Perot 32% to 30%, while an ABC News survey says Perot is ahead of Bush by 36% to 30%. Maybe, as ABC News polling director Jeff Alderman puts it, "no one really has any idea who's ahead."

KEEPING MUM. Why all the confusion? Voter opinions are more fickle than usual, says Andrew Kohut, director of surveys for Times Mirror. In such a climate, pollsters can get different results merely by shuffling their questions. And it's not just political polls: Marketers also now are doubting their research.

Campaign '92 has undercut many cherished assumptions, one of which is the reliability of public opinion research. The polls have been so muddy and contradictory that some handicappers may feel more comfortable using the Daily Racing Form than the Gallup Poll. Nobody has expected polls to be infallible since they predicted Dewey would defeat Truman. But several trends are conspiring to place voter surveys under a fresh cloud of suspicion.

For starters, the validity of polls is being undermined by the rising number of voters who refuse to participate in them. "The refusal rate is increasing with every election because people have been inundated with polls," says Frank I. Luntz, research director for Perot. Pollsters won't cite a figure, but many agree that refusal rates are only slightly less than for consumer-product surveys. Walker Research Inc., an Indianapolis firm, estimates that 36% of those asked to respond declined to answer consumer-product phone surveys in 1990--a 12-percentage-point increase over 1986.

BEAN COUNT. Worse, those who duck surveys tend to be "market makers"--industry jargon for the younger, more affluent consumers who are most likely to try new products. Since companies often decide whether to launch products based on their reactions, missing them adds risk to an already chancy process.

One who passed on political polls is Janice Wayne, an actress in Rye, N. H. Wayne says that she received 10 calls from pollsters in the three weeks before the New Hampshire primary, and "it got to the point where I felt rebellious."

Even when voters do cooperate, pollsters are taking their answers with a grain of salt. Some argue that fierce antipolitical sentiment is prompting many voters to say they support Perot when they'll probably vote for Bush or Clinton in November. "People are tired of cornflakes," says Charles Black, a senior adviser to the Bush-Quayle campaign. "Naturally, they jump up and down and say they want Perot."

Some pollsters remain undaunted. Manuel Corella, who runs the Texas Frijole Poll out of his Mexican restaurant in Galveston, Tex., says he has called every Presidential race correctly since 1976. Corella asks each patron to toss a pinto bean into a wine carafe bearing the name of his or her preferred candidate. At last count, Perot had 60% of the beans, with Bush at 25% and Clinton at 14%. But the bean poll doesn't allow for undecided voters. "Some of the customers want to throw it into an undecided carafe," Corella explains, "but we ask them to make a choice." If only every pollster could be so pushy.

Before it's here, it's on the Bloomberg Terminal.