POLLS: opinion-taker or opinion-maker?

When this statement is read to the renowned pollster George Gallup, he throws his head back and laughs out loud for a full minute. "That's an interesting comment coming from him," he says. "As close as he is to the President. Calling people in Washington gutless. I think that's just nonsense. I think we have done just the opposite. We have taken political decisions out of the smoke-filled rooms of yesteryear. We've opened up the process. People now have a chance to have their views known."

Dr. Gallup -- founder of the Gallup poll and one of the great-granddaddies of the profession -- belts out his views with all the ardor of a crusader battling the enemies of democracy. "I'm an evangelist," he readily admits. "I like nothing more than to be out on the stump."

When he's not out on the stump or vacationing at his home in Switzerland, Mr. Gallup does his "evangelizing" in the two-story, ivy-covered Gallup Building surrounded by little shops on a quaint, winding street in Princeton, N.J. Hardly the setting where you'd expect to encounter this ham-handed bear of a man who grabs you firmly by the hand, putting one large mitt on your shoulder, and sort of pulls you with a friendly squeeze and a deliberate momentum into his dark, paneled office.

His lined face and large, energetic features suggest the dynamism behind a New Jersey trucking company rather than the brains and brilliance behind a premier national polling organization. His mind is aggressive and youthful. As he talks he gestures vehemently, occasionally hitting the desk smartly with an open palm to emphasize a point.

Lately, there is a lot to slap the desk about. There is growing uneasiness -- a feeling that polls lead statesmen and journalists alike to misread the public mind, that they may in fact corrupt the political process itself.

Dr. Gallup couldn't disagree more.

"The greatest advance this country could make," he says, sitting squarely behind his massive desk and staring out under bushy eyebrows, "is to adopt the initiative and referendum form of government. This would lead us to a pure democracy. I mean, as great a thinker as Walter Lippmann considered Congress a perfect representation of American opinion. Well, it's not. And I am convinced that, at any point in time, the collective judgment of the people [as measured by polls] is likely to be better than of any smaller group. History has proven this.

"We've always had polls," he argues forcefully. "Newspapers have been conducting polls since 1824, 150 years ago! They were doing street corner polls , they were doing all kinds of polls. The only problem was, they weren't doing them properly. They weren't following scientific procedures. The only difference between today and 100 years ago is that the polls are more accurate."

Dr. Gallup is well qualified to offer this little history lesson. He helped establish modern polling techniques in 1932. At the time, he had taught journalism at Drake University and Northwestern University (and would soon join the faculty of Columbia university). He still considers himself to be in the journalism field, which he says he entered when the local school board at his high school in Jefferson, Iowa, did not include a basketball gym in its new building. "I wrote a scathing attack on the school board. My father looked at it and said, 'You ought to be in journalism.'"

Dr. Gallup went into polling after he had been teaching and doing newspaper survey work for almost 10 years. "It was strictly experimental. The first election we tried out these new sampling procedures was the 1934 congressional contest. The first presidential election was 1936." These early polls were met with the skepticism accorded anything new, a fact that brought out the advocate in Dr. Gallup, and eventually led him to write a frequently updated, book-length defense of the polls.

He has been defending polls against all comers, ever since.

Back in 1970, Gallup's British subsidiary, along with every other major pollster in Britain, completely misread the prospects for that nation's election , predicting that Harold Wilson would win the prime ministership handily. (He lost by almost a million votes.)

Some more recent domestic surveys have been extraordinarily inaccurate. In Pennsylvania in 1978, for instance, Gallup predicted that Rep. Richard Thornburgh would lose the race for governor by four percentage points when in fact he won by six percentage points. But perhaps the most outstanding example emerged from this year's New York Democratic primary, when Louis Harris and Associates did a survey for the New York Daily News that was off by 38 percentage points, an incredible amount by any pollster's standards

Mr. Harris attributed his error to a miscalculation of the voter turnout. Some academics and polling professionals blamed his sampling and interpretation techniques.

But John McLaughlin, chief political correspondent for the Daily News, felt the problem went much deeper. In an angry column he wrote that "the polls have become an issue unto themselves, almost as controversial as the candidates and in some ways more decisive. The polls helped generate money and they turned it off. They persuaded political reporters to reject their own findings and instincts. they put Sen. Howard Baker in a hole he could never climb out of. . . . Politicians, reporters, and the candidates have been bewildered by the polls this year and all have made misjudgments by relying too heavily on them."

He is not alone in his assessment. Other political analysts, such as Dr. Everett Ladd at the Roper Center for Public Opinion Research and Dr. Albert Golin of the Newspaper Advertising Bureau, have been wondering out loud about the effect polls have on the opinion they are supposed to measure. Books have been written on the subject, including the graphically titled "Lies, Damn Lies, and Statistics," by Michael Wheeler (professor of law at the New England School of Law), a book Dr. Ladd calls "a solid and reliable summary."

Writing in Esquire magazine in 1977, Tomothy Crouse (author of "The Boys on the Bus") concluded that this book's findings "confirm what some of us have always suspected: that the polls are often inaccurate and misleading; that the pollsters' alibis are bunk; and that the political polling business is not only unnecessary to the continued welfare of the Republic, it is a potential menace."

To George Gallup, these are fighting words. Bobbing and weaving behind his large cluttered desk, he energetically taps his foot, twists a pencil nervously in his hand, and shoves stacks of paper around in front of him as he sails into his critics.

"He is dead wrong," he retorts. "I could produce stacks of objective data to prove that this is all absolute pure speculation. All one man's opinion. There is not one scientific shred of evidence to back these charges up.

"I would say that the title ['Lies, Damn Lies,'] should apply to the book itself. The charges in it were absolutely, totally false. He has a long lot of nonsense in it about the 1948 election [which pollsters and reporters all miscalled]. It's just preposterous. Usually, these things are said by people who are, in one sense, in competition. In the early days, political writers hated our guts. Because we seemed to be taking away from them their prerogative: predicting the political races."

Not all questions raised about polls come from Dr. Gallup's competition, however. Professors at various universities, and some respected figures in the research field, have been studying the effects of opinion research on the political process; and some are convinced that polls have an unsettling effect on election results.

There's no question as to whether polls affect politics," Dr. Gollin of the Newspaper Advertising Bureau observes. He maintains that polls are a determining factor in "the insider's game," the one played among campaign workers, donors, and other professionals to see where to throw their support.

"Polling defines an inescapable reality to these people. [Hubert] Humphrey did not have enough money to counter Richard Nixon's early television spending in 1968 as a result of his standing in the polls. And the outcome of the election was markedly affected by his inability to get money early enough to commit for television time."

Other analysts point out that a strong showing in the polls is essential to a candidate in party primaries. And they are troubled by the fact that these highly determinative primary polls are often unreliable, by the pollsters' own admission, because it is so difficult to predict voter turnouts.

"One could have made exactly the same charge against some political writer in 1908 who said that [William Jennings] Bryan was going to win, . . . that he was interfering with the process," Dr. Gallup argues.

"In the entire history of polling there never has been one single scientific study ever made which shows that polls influence election results," he maintains with a well-timed slap of the desk. "Believe me, we have gone through every election since 1936. And you can spin all kinds of theories. But you can't find a single study in the history of polling -- and the academics have not been too much sold on polling -- that has been able to find any."

Maybe not. But there is a growing body of informed opinion that polls are more than the passive mirror of the people's thoughts described by Dr. Gallup, that they are in fact an integral, shaping force in the course of political campaigns.

Reportedly, his showing in the polls influenced Sen. Edward Kennedy to enter the 1980 Democratic primaries. "At the time," says Dr. Ladd at the Roper Center , "the polls showed Kennedy a 2-to-1 favorite over Jimmy Carter for the nomination. They said the nomination was his for the taking."

Kennedy's decision to enter the race "may have been the most significant result of polls this year," says Dr. Ladd, who is also professor of political science at the University of Connecticut.

According to Dr. Gollin, the polls showing Kennedy with such a commanding lead never really examined the important issues relating to the senator's candidacy. They simply posed a horse-race question pairing Kennedy and Carter in a hypothetical runoff. "There was nothing linked with the question," complains Dr. Gollin, "not Chappaquiddick, not the national issues on voters' minds, not the emergence of a Republican candidate with the stature of Ronald Reagan."

A number of analysts feel that such polls are basically too unreliable a measure of public sentiment by which to evaluate candidates.

A Changing Times magazine article, which Mr. Gallup himself credits with a high degree of accuracy, pointed out: "Over the years the use of computers, the refinement of survey techniques, and the emphasis on 'scientific' samples have helped make public opinion polling more reliable and objective -- but not totally reliable and objective. That's because, despite the technology, polling remains essentially an art. Though some survey samples are constructed using the laws of probability, not all of them are. And nearly every other aspect of polling, from composing the questions to conducting the survey to evaluating the results, is based on human judgment. Where human judgment is involved, bias, whether intentional or not, is almost inevitable."

Other analysts flatly charge that national pollsters are influenced by their own political philosophies and their ties to political figures: "Pollsters, like political journalists, sometimes become so close to powerful politicians they forfeit their claim to objectivity," Timothy Crouse writes.

Does the judgment of the man who designs the survey ever enter into the polls?

"Of course it does," Dr. Gallup says. "It can enter in at every stage. But so does it in every scientific operation. Even in the physical laboratory judgment enters at some point. Even in reading a thermometer, there's a factor of judgment and human error. So this is not unique to polling.

"But there's one good, solid way the public can determine how good a poll is, and that is by its results. It's one of the few areas of social science, almost the only one, where you know how good the procedures are down to the last decimal point."

Has the track record really been that good?

He leaps up and shuffles through a stack of books until he comes up with a brochure that shows the Gallup polling record going back to the 1936 Roosevelt-Landon race. "There's the record, right from Day 1," he exclaims triumphantly.

Then, sitting down, he grumbles, "The one thing we've never been able to get reporters or journalists to do is look at the final results. If you can find one single article in the history of polling that recites the error to polls by the difference [between poll predictions and actual results], I'll give you a thousand dollars. This guy who arote the book 'Lies, Damn Lies, and Statistics' didn't have guts enough to put in the record. He didn't have guts enough to look at the final figures, which anyone can do."

The final figures are remarkably close to the actual election result in each presidential race (except for the notorious Dewey-Truman in 1948, when every pollster and columnist elected Dewey over Truman); and they get closer as the years progress. But calling a vote the day before the election is not the most precarious aspect of polling, nor the one that is judged to have the most impact on electoral campaigns.

It is the string of polls during the campaign -- jumping around from week to week and causing candidates to rethink their strategies -- that most critics question. These polls are impossible to corroborate, as are the polls sampling public opinion on important national issues.

How do we know if pollsters are right or wrong when they disagree about public opinion during a campaign or during national debate over the SALT II and the Panama Canal treaties?

"Unfortunately, you simply have to place your trust in us or the other guy. The other thing you can do, a simple thing, is to look at the question that was asked, look at the sample to see just what the polling universe [the community surveyed] was, and the timing, and the sequence of questions.

"On television, however, all you see are the figures popping up. You don't get the question, you don't get any of the detail which you need as background to interpret the poll. Obviously it's a matter of time [available to explain these issues in a TV news segment], and they don't want to do it, but it's a violation, when they don't do it, of the rules."

These rules -- the ones supposed to govern fair and ethical polling -- are important to George Gallup. He lives by them. And, according to his rules, well-run polls are an instrument of democracy, the vox populim by which politicians and statesmen alike can determine what voters think is the greatest good for the greatest number. The issue may never be settled. After all, you can't prove it with a poll. And without statistics to back up the argument, who would believe you?