Do We Ask Too Much of Polls?

By ELLIS COSE

In these skeptical times, polls may be the one remaining authority that the press customarily accepts without question. The subject may be the Panama invasion (the public supported it), the arrest of Mayor Marion Barry (Washingtonians thought he should resign), or Jane Pauley's treatment by NBC (PEOPLE readers who answered a call-in survey found it unfair), but editors rarely meet a poll they don't like. Polls have even been published reporting the number of California drivers with paraphernalia hanging from their rearview mirrors (8%), and Iowans with ornaments on their lawns (24%).

The seemingly insatiable appetite for polls arises because they satisfy so many editorial needs. In times of uncertainty they offer apparent objectivity and precision. On otherwise slow news days they track the excitement of public opinion on the march. ("Watch out! Here comes 'Big Mo.' ") They promise a window into private thoughts without the inconvenience of intimacy. And, in a world of ten-second sound bites and shrinking news stories, poll-derived graphics can be wonderfully concise.

Surveys have become a staple of stories examining presidential popularity (George Bush, so far, is doing better than Ronald Reagan), foreign policy (Americans are upbeat on Mikhail Gorbachev but remain down on communism) and race (blacks are less optimistic than whites but believe more strongly in education). Editors have even employed polls to study journalism itself. In the mid-1980s, with newspaper readership declining relative to population growth, researchers diagnosed widespread public skepticism about journalists' methods and motives. Confounded by inconsistencies in those surveys, Times Mirror, the publisher of the Los Angeles Times, Newsday and several other papers, hired the Gallup organization to get to the truth. Gallup reassuringly reported in 1985 that no credibility crisis existed.

More recently, however, the news from Gallup has not been so encouraging. In a report published last November, the 16 news organizations rated in the survey had collectively lost 9 percentage points from their believability. By at least one standard, journalism was not doing so badly: among individuals and institutions rated, only Pope John Paul II was found to be more believable than the media. When the poll was narrowed to specific news organizations and journalists (including the Wall Street Journal, Cable News Network, Peter Jennings, Dan Rather and Ted Koppel), several actually outscored the Pope -- and left President Bush far behind. Still, the overall decline in media credibility was enough to prompt somber newspaper reports. PUBLIC CONFIDENCE IN PRESS DIPS SHARPLY, SURVEY FINDS, headlined the Los Angeles Times.

Such hand wringing may be treating the numbers with more respect than they deserve. After all, polls are no more accurate on press credibility than they are on any other subject. When pre-election polls in New York and Virginia went awry last fall, the almost unanimous press query was, How could they have been so wrong? That question has plagued journalism since at least 1936, when the Literary Digest predicted that Alf Landon would become President of the U.S. A more appropriate question might be, Why do we so expect them to be right?

As polling methods have advanced, the press has gradually elevated pollsters to the status of prophets. And journalists sometimes forget that their prophecies come not from the heavens but from a branch of mathematics called probability theory, whose most obvious application is to gambling. The concepts are commonly introduced in statistic classes with reference to coin tosses and dice. It is hardly an exact science. Roughly one time out of 20 the typical pollster's finding will fall outside the stated margin of error. And even that assumes a flawless sample that will be exactly representative of the whole population and a 100% response rate -- conditions that are never met in the real world. Accurate polling also supposes that the questions are unambiguous, the interviewers perfectly interchangeable, and that the answers are freighted with the same meaning the analyst believes they have. These conditions too are virtually impossible to satisfy.

The flaws inherent in polling methods have caused some news organizations to become more cautious. During the latest New York City mayoral race, and for the first time in memory, the New York Times did not poll prior to Election Day. Adam Clymer, who at the time was the paper's polling specialist, explained that there was "simply no decent track record." No one could predict with confidence exactly which and how many of New York's registered voters would actually go to the polls. Why, then, did the Times report on polls carried by others? "These polls were part of the equation," says Clymer. "People were talking about them." In short, though the Times had little confidence in the surveys, polls have assumed such a central roll in elections that ignoring them was not a real option.

The fad has even caught on in the Third World, with much of the news of recent elections in South America and India dominated by reports about polls. That is more than a little disturbing. For, as the best pollsters recognize, the deepest questions of life -- or politics, or journalism -- can be probed only in the most primitive manner with the blunt instrument of a poll. Thus readers entering upon stories peppered with numbers and percentage signs should arm themselves with a mental note: POLL AHEAD -- PROCEED AT YOUR OWN RISK!