WASHINGTON -- Once upon a time, polls were conducted to learn what people think. Now, political scientists wonder whether polls are telling people what to think.
Has a simple tool of commerce intended to help match products with markets evolved into a force of its own in this society obsessed with focus groups and numbers? When a candidate such as Bob Dole trails so far behind in public opinion surveys that his cause is declared by the press to be all but lost two months before an election, does that become a self-fulfilling prophecy?
No one knows for certain. But news reports based on daily waves of ever more finely-tuned polls clearly have an impact on elections.
"I think what's happened is polling has gotten so good that people have come to rely on it," says John M. Barry, assistant director of the Roper Center on polling at the University of Connecticut. "Not only in politics, but it's a tool that's used to make huge decisions in the corporate world every day."
Polling is a relatively recent phenomenon that has become pervasive. Less than a century ago Americans made it through whole days without being asked: For? Against? Undecided? Don't know?
Elmo Roper, a New York jeweler who in the 1920s was stuck with merchandise that wouldn't sell, became a pioneer in the field when he decided to ask potential costumers what kind of jewelry they liked rather than presuming to decide for them.
Newspapers have been taking political surveys since the 19th century. But until the 1930s those were mostly polls of readers and were conducted haphazardly and required respondents to return ballots or questionnaires. An early theory was that accuracy of the results was a function of the size of sample: the more responses you received, the more likely they were to be on the mark.
In 1936, The Sun mailed a poll ballot to every registered voter in the state, asking their preference in the presidential contest between Democrat Franklin D. Roosevelt and Republican Alfred M. "Alf" Landon. Fewer than a third of the voters returned their ballots, but the results came within 1 percentage point of predicting the Maryland vote of 63 percent for Roosevelt to 37 percent for Landon.
But the straw poll method was discredited in the same election, when the Literary Digest predicted a Landon victory based on nearly 2.4 million ballots returned from a list of voters drawn from car registrations and telephone listings. People with cars and telephones during the Depression were better off than the voting population as a whole, and excluded were many Roosevelt supporters at the other end of the economic scale. The survey had predicted a close race in Maryland.
George H. Gallup, coming from the market research tradition of Elmo Roper, became the father of modern-day newspaper polling in 1935 when he began writing a newspaper column based on his polling. He developed the system of weighing poll samples according to factors such as income, age, race, gender, education and urban or rural residence that match the proportions of those categories in the electorate.
Gallup found that a well-selected group of respondents would yield accurate results even if the number of people surveyed was relatively small. Fifteen hundred people or fewer can signal the -- preferences of the entire nation.
Politicians have become so dependent on the use of polls to determine public attitudes that presidents since the days of Jimmy Carter have included pollsters among their top advisers. Critics claim Bill Clinton's presidency is entirely poll-driven.
"We don't use polls to make policy decisions, but we do use them to determine how to explain those decisions to the public," says George Stephanopoulos, a top Clinton adviser.
Polls that reveal how voters feel about certain issues -- or even react to certain words or pictures -- can be enormously valuable to political leaders.
But for a candidate now in the position of Bob Dole, the burden of being tagged a long-shot in a daily avalanche of polls can make a last-minute surge to victory more difficult.
"None of this is fair to Dole," says Kathleen Hall Jamieson, dean of the University of Pennsylvania's Annenberg School for Communication. She cites a study of 20 news reports on Dole in the first week of September, all of which said he was "struggling."
Focusing on the poll standings is "easy journalism," Jamieson says -- journalism that detracts from what Dole is saying on issues that could boost his position, and conveys a message to voters that it's a waste of time to listen to him.
Then there's the case of Ross Perot, the third-party candidate who appears to have been excluded this week from participating in the presidential debates, although he received 19 percent of the vote in 1992 and is drawing public campaign money. Perot's poll numbers are so low this year that the Commission on Presidential Debates has determined he has "no realistic" chance of being elected and recommended that he not be admitted to the debates.
Whatever chance he had is probably smaller now.
Pollsters say there is no evidence that predicting the likely outcome of an election contest -- or more precisely providing a snapshot of public opinion on that contest at the time the poll was conducted -- has ever influenced the result.
One such opportunity might have come in the 1992 presidential contest when a Gallup Poll published a week before the election reported that Republican President George Bush had cut by half the 10 percentage point lead held by Democratic challenger Bill Clinton. But none of the other surveys at the time showed the same trend, and the Gallup numbers on Bush dropped a few days later.
"If publication of a poll could affect public opinion, Bush might have been able to get some momentum out of the Gallup Poll," says John Nienstedt, president of Competitive Edge Research, a polling firm based in San Diego. "But that didn't happen."
A negative poll hurts more than positive poll helps. A poor showing dries up campaign contributions and thus the ability to advertise on television.
"Polls have a huge impact on money," says Barry of the Roper Center. "If your numbers drop low enough, all of a sudden the checks stop coming."
Even so, blaming the polls is like blaming the messenger, says Andrew C. Kohut of the Pew Research Center for the People and the Press.
"He [Dole] is bucking against the definition of Bob Dole as a loser," Kohut says of the Republican presidential nominee. "But Bob Dole's problem is not with the polls," but with weaknesses in his campaign reflected by the polls, Kohut said. Dole is not complaining about the heavy media focus on polls because he tTC figures his turn will come.
"We've always thought that this was going to be a late-breaking race," says senior Dole aide John Buckley. "So, when the race tightens and Senator Dole's numbers begin to climb we would expect those polls to get the same sort of attention as when he was 20 points behind."
Pub Date: 9/19/96