Making Sense of the Polls

By LAURENCE I. BARRETT WASHINGTON

PEROT LEADS IN NEW POLL" has become so frequent a bulletin that the Texas billionaire's image has changed from interesting maverick to serious presidential contender. If Ross Perot does endure as a major force into autumn, one large reason will be the opinion surveys of spring, despite their notorious fragility during this period. Says pollster Peter Hart: "More than any other person I can think of in American politics, Perot has been aided and abetted by the polls."

Headlines trumpeting Perot's apparent popularity offset what is normally a huge liability for a little-known independent -- skepticism that he has any chance to win. In Perot's case, poll results feed on themselves. High ratings help beget higher ratings even while he remains an elusive figure who declines to state his views.

Yet the numbers that seem firm can be illusory, as a survey by the Times Mirror Center for The People & The Press showed last week. To explore the quirkiness of the public's mood, the center matched Operation Desert Storm hero Norman Schwarzkopf against George Bush and Bill Clinton. The retired Army general placed second, with 29%, vs. 35% for Bush and 27% for Clinton. Andrew Kohut, who ran the poll, thinks that result "underscores the difficulty of judging how much of Perot's standing is really support for Perot rather than a yearning for a nonpolitical alternative." In another experiment, Kohut found that Perot fell from first place to second in a three-way test when the questionnaire omitted a preliminary item comparing only Bush and Clinton. When confronted solely with choosing among the trio, some voters apparently move from Perot to "undecided." When opinion is in as mercurial a phase as it is now, small changes in polling methods affect results.

Such nuances are familiar to pollsters and political reporters but meaningless to the public. Also opaque are differences between types of surveys. Perot got a large boost earlier this month when, in the final round of primaries, the networks included his name in exit polls -- interviews with those who have just cast ballots. Such samplings usually provide reliable demographic data and allow speedy projection of the winners. But those who come out for primary elections are not representative of the larger electorate.

It is also risky, in terms of eliciting firm opinion, to mix questions about what people have just done in voting booths and what they would do in a different election. And Perot, who had not run in the primaries, had been spared the criticism and intense scrutiny inflicted on the active candidates. His strong showing in the exit polls so dominated news coverage that he won a publicity victory in contests he had not entered.

Even in more serene elections early polls often prove ephemeral, because voters' preferences are, in pollsterspeak, "lightly held." In 1988 Michael Dukakis' 17-point lead over George Bush disappeared in a twinkle. This year the public's extraordinarily sour mood makes horse-race numbers still more suspect. "In this atmosphere," says Everett Carll Ladd, director of the Roper Center for Public Opinion Research, "polls often become a source of misinformation rather than insight into what's happening."