DO POLLS HELP DEMOCRACY?

LEADERS must know what their people are thinking. If France's Charles de Gaulle or Columbia's Grayson Kirk had followed that simple rule, they might have saved themselves a lot of grief. Therein lies the chief justification for opinion polls. Yet there is also something vaguely troubling about the polls, those incessant readings of the U.S. voter's psyche.

The classic demagogue swayed the crowd through oratory. Polls sometimes suggest a kind of demagogy in reverse: the crowd seems to sway the politicians through the polls. One expects those who seek high office to speak out with courage and conviction, to teach the people, to lead. Instead, the candidates seem increasingly guided by pollsters--semivisible oracles who claim to know what millions of U.S. voters think and feel. How many Americans have ever talked to a pollster, or even seen one? Yet, pollsters have acquired huge and growing power.

In varying degree, polls account for the biggest surprises of this surprising election year. George Romney dropped out of the presidential race because his private polls showed him losing badly to Richard Nixon. Robert Kennedy dropped in only after his polls convinced him that he could beat Lyndon Johnson in the California primary. In renouncing a second term, Johnson was influenced by a Gallup poll showing that only 26% of the people approved his handling of the Viet Nam war.

During the months ahead, the polls will be more important than the primaries. While Kennedy must surge ahead in both to capture the Democratic nomination, Hubert Humphrey's best hope of braking a Kennedy bandwagon is to continue outpolling all other Democrats. In re-entering the Republican race, Rockefeller has become the first presidential candidate to base his campaign almost wholly on polls. He can win the nomination only if national surveys convince G.O.P. regulars that Nixon will lose in November. But those surveys now put both Republican contenders ahead of all Democrats--and if that reading continues, the regulars may well stick with Dick.

All this raises obvious questions about the role and relevance of the pollsters. Do their samplings really reflect public opinion or create it? Are they scientists or self-fulfilling prophets? Do they enhance democracy or menace it?

America Is a Bowl of Beans

The present phenomenon of polling arises from the politicians' ancient need to know what the public thinks. Finding out was easy in the days when strong party machines relayed the information. Today, the weakening of political allegiances has disrupted communications. Polls have partly bridged the gap.

The pollsters rose to fame and influence on the basis of two celebrated debacles. During the 1936 presidential campaign, the old Literary Digest ran a mail poll and was wrong, while three more scientific pollsters were right. Those three--George H. Gallup, Elmo Roper and Archibald Crossley--conducted interviews among a predetermined mix of ethnic, income and age groups that seemed representative of the U.S. population. The other turning point was in 1948, when the pollsters again used this "quota system" of sampling--but were wrong. The U.S. had become so complex that picking just the right population mix was too difficult.

So the pollsters switched to the more accurate "random sample," which relies on the theory of probability and owes its development to Galileo, Pascal, some expert gamblers and the U.S. Census Bureau. Probability theory says that if a jar contains 1,000,000 beans--half black and half white--and somebody scoops up 100 of them, he will almost always draw half black and half white, within a 3% margin of error. Gallup views the nation as a big bowl of beans. On a strictly random basis, he picks 300 sections of the U.S. and selects five voters in each section. Then he sends his interviewers--who are mostly middle-aged women working part time--to poll those five voters. For the entire nation, and some large states, the random sample is 1,500 voters. Pollsters have found that they cannot significantly increase accuracy unless they go very far above 1,500, and that is usually uneconomic. In the national elections since 1948, Gallup has erred only 2.7%. Since 1954, his election polls have been off only 1.5%.

How to Read a Poll

Despite their relative accuracy, polls fluctuate so sharply that since last August, every major candidate has at one time or another emerged as the people's presidential choice. The reason is clear: during pre-election months, the volatile voters' sentiments seesaw along with news events and the candidates' pronouncements--which is what campaigns are all about. Party loyalties are also weaker in May than in November. One consequence is that extremist, third-party candidates often run stronger in polls and primaries than in national elections--as may well be the case in 1968 with George Wallace, who has been garnering 11% to 19% of the poll vote.

The fact is that a poll, like a photograph, is accurate ' only at the moment it is taken. Moreover, there is usually a lag of a week or two between the time that interviewers ask their questions and the time that the results are analyzed and published. In the interim, public attitudes may be radically changed by all kinds of events. Although the latest Gallup poll shows Robert Kennedy's popularity declining, for example, this reading may be outdated because it was taken before Kennedy's primary victories in Indiana and Nebraska, which in turn may be outdated by this week's Oregon primary.

Equally tricky is the effect of the pollster's questions. A vaguely worded or blatantly biased question can alter the results by 10% to 40%. A sound question gives the respondent a real choice: "Do you favor or oppose President Johnson's bombing pause?"

Unfortunately, the public's answers can be as misleading as the questions. Ashamed to admit racial or religious prejudice, people who answer polls sometimes resort to artful lying. Though 80% tell interviewers that they will vote, only 65% do so. To prevent bias, interviewers ask trip-up questions ("When did you vote last?" "Where are you registered?"), and toss out roughly one-fifth of the respondents on the ground that they are unlikely to vote.

What most baffles pollsters is the great block of "undecided" voters who swing most elections. In 1952, Gallup's last pre-election poll turned up 13% undecided. On the basis of past voting patterns, he "allocated" the undecideds more than 2 to 1 for the Democrats, which put the Eisenhower-Stevenson election into fifty-fifty country. Had Gallup instead discarded the undecideds and prorated the rest of the vote, his poll would have shown Ike over Adlai by 54% to 46%--Eisenhower won the actual election with 55% of the vote.

To deal with the double problem of indecision and hidden prejudice, Gallup has voters mark secret ballots and deposit them in sealed boxes in the last few polls just before election day. The ballot obliges them to make a choice. Equally helpful is the new "intensity question." Using a scale ranging from plus 5 to minus 5, pollsters ask voters to indicate how strongly they feel about candidates and issues. Plus 5 indicates a firm attachment to a candidate; plus 1 suggests that the voter might well swing to the other side. Even in very close contests, pollsters can usually spot the winner by measuring not only the size but also the intensity of his support.

All the same, no one should assume that all polls are created equal. They vary widely in reliability. For that reason, poll readers should ask five basic questions: Who sponsored the poll? What pollster conducted it? When and where was it conducted? How big was the sample? Was the question objective or suggestive?

For one thing, any poll sponsored by a candidate is quite possibly slanted. Polls conducted by newspapers may also be unreliable because they do not test a true random sample but measure a floating population on street corners and in bars, tending to overlook housewives, elderly people and stay-at-homes. The best polls are those conducted by established, well-known polling organizations that regularly publish results. Even these may be suspect if the sample was less than 1,000, the question is unstated, and the poll was taken more than two weeks before publication.

Uses & Abuses

While published polls get all the attention (giving the pollsters entree to far more lucrative market research), private pollsters work for candidates much as augurs labored for Caesar. No longer sure of his personal instincts, the modern politician has upgraded the pollster from the rank of technician to that of campaign tactician. Says a leading pollster, Joseph A. Napolitan: "Polls never won an election, but you can win an election with what you do with your polls."

The key question that pols put to their pollsters is not "Who's going to win?" but "What's bugging the voters?" Polls help candidates to identify their own negatives, and then change those characteristics that voters find unattractive. Taking advice from their pollsters, California Democratic Chief Jesse Unruh peeled off 90 lbs. to reshape his corpulent boss image, and Pennsylvania Democrat Milton Shapp discarded his maroon socks (but lost the 1966 gubernatorial race anyway). Candidates also use private polls to find out where the large and decisive mass of swing voters is located, and then concentrate their campaigning in those areas. Most important, polls tell what issues the voters really care about and how deeply they care.

Though every candidate swears that he makes up his own mind on the issues, the fact is that polls have a marked influence on political pronouncements. Dick Nixon's pollster, Joe Bachelder, reported that Oregon primary voters were worried about education, public works and inflation--and Nixon quickly stressed those issues. Pollster John F. Kraft warned Robert Kennedy before the Indiana primary that he had the Negro vote sewed up but faced trouble from blue-collar whites; Kennedy shifted his campaign emphasis from help for the poor to law and order.

The candidate also can--and does--leak his private polls to the press, particularly when they show him ahead. A favorable poll more than pays for its cost (about $10,000 for a medium-sized state) by swelling the campaign war chest. Says Nixon: "When the polls go good for me, the cash register really rings." On the other hand, a candidate does not want to appear to be a shoo-in, lest his campaign workers slow down and his voters stay home on Election Day. The art of selective leaking is to make it appear that the candidate's strength is steadily rising but always slightly below the expected margin of victory. In the New Hampshire primary, for example, Nixon said that he would be lucky to get 50% of the Republican vote, although his private polls showed him winning with at least 70%, and that made his victory--with 79%--seem all the more smashing.

Of course, politicians can blatantly abuse polls. One case was last November's leaked report that a poll of a "bellwether" county in New Hampshire had put Johnson far ahead of all possible competitors. Not stated, however, was the fact that the county was historically Democratic and had given 68% of its vote to Johnson in 1964. Occasionally, pollsters also abuse polls. Some shifty operators claim to interview 1,000 people but only reach 400. A few conduct surveys for one party, then sell them under the table to the other side.

Toward the Instant Referendum

Some critics have called for state control of polls, or outlawing them altogether, but that would probably amount to unconstitutional censorship of what has become a lively branch of journalism. Polls are here to stay, and pollsters have an obligation to make them even more honest and accurate. Gallup, Roper, Crossley, Mervin Field, Joe Belden and others have begun a drive for self-regulation, calling on their colleagues to disclose exactly what question was put to how many people, as well as when and where.

Possibly in a generation, polls may lead to instant national referendums. Every voter would have a small electronic box with "yes" and "no" buttons. The President could ask for public opinion on any issue--Should the nation invest $50 billion to send men to Mars?--and the presumably informed electorate would flash back an immediate response. Technically, this is feasible right now. Automated democracy might dilute the power of a lot of Congressmen--no loss to democracy in some cases.

Do polls build band wagons? The evidence thus far suggests that they may do just the reverse--as in 1948, when Harry Truman urged the public to "prove the polls wrong." If polls really sway voters, argues Gallup, Dewey would have won. But polls do present other problems. They give an edge to rich candidates, who can afford more and deeper polls than less affluent candidates. Old-line party chieftains worry that the polls have robbed them of some of their previous powers to dictate nominations--though few people would complain about that.

The greatest danger remains that polls tempt candidates to be popular rather than right. Yet in a democracy, there is always a conflict between responsiveness and responsibility. And quite often the public is far more alert to the need for new policies than are self-justifying politicians, who may be loath to alter stand-pat positions. So for all the flaws and abuses of present-day polls, they do stimulate the dialogue between the people and their elected officials.

In trying to measure what the public wants, the pollsters argue that a wise politician can trust his own instincts about 75% of the time, but a sound poll will give him a correct reading 90% of the time. That makes a 10% margin of error--and there is no guarantee that, in fact, the margin is not apt to be much larger. Thus the powerful polls can help democracy triumph, or at least muddle through--as long as the politicians and the public remember the margin of error and refuse to be hypnotized by the augurs.

This file is automatically generated by a robot program, so reader's discretion is required.