In these last days before the election, Democratic presidential nominee Walter F. Mondale has hoisted a new banner: "The polls are dead wrong!"
In one sense, he is dead right.
Throughout this political year, the variation in poll results has been so pronounced as to call into question the scientific precision of the popular art of opinion-taking.
At the beginning of the year, for example, the Gallup organization reported that Mondale and President Reagan were tied for the public's favor. Simultaneously, the CBS-New York Times poll found Reagan with an enormous lead of 16 percentage points. One of those polls was dead wrong. They sampled the same universe of potential voters at the same time.
A month later, a poll by USA Today gave Reagan a 22-point lead. ABC and The Washington Post, polling the same universe at the same time, came to a more modest conclusion: Reagan's lead was only 7 points. One of those polls was dead wrong.
The Chicago Sun-Times reported last week that Reagan led Mondale by 16 points. The Chicago Tribune reported at the same time that Reagan's lead was 6 points. One was wrong.
An interesting contradiction occurred in July following the Democratic National Convention. The Gallup organization, in a poll for its client, Newsweek magazine, reported that Mondale led Reagan by 2 points. Immediately thereafter, the Gallup organization, in a poll for its newspaper clients, reported that Reagan led Mondale by 12 points.
A trend chart published Monday in USA Today suggested that after all the time, money and energy expended on the 1984 presidential campaign, barely a mind has been changed, barely a vote affected.
The chart depicted its poll results between Feb. 16 and Oct. 27 as three approximately straight lines. Throughout those nine months, Reagan had about 60 percent of the vote, Mondale about 35 percent and 5 percent were undecided. Neither historical events, the primaries, the conventions nor the debates had a significant effect on those results.
The trend lines in other polls -- Newsweek/Gallup, for example -- look more like a roller coaster, with big ups and downs as the candidates' fortunes seemed to change.
One of those charts is wrong. One depicts a landslide from start to finish; the other depicts a volatile, competitive political contest.
Several explanations are offered for this inconsistency. Some poll takers, insisting on gentlemanly anonymity, accuse their brethren of "poor methodology." That covers a range of alleged sins.
In some cases, it is said, poll samples turn up too many Republicans or too many Democrats. In other cases, it is argued that questions are put in the wrong order.
For example, if the first question is: "Are you going to vote for Reagan or Mondale?" one will get a different result than if the question is asked later in the interview after questions about issues and policies.
Then there are the "margins of error" at work in all polling exercises. If a poll finds one candidate leading another by 55 to 45 percent and if the "margin of error" is plus or minus 5 percent, the real result could be 60-40 or 50-50, a very flexible range.
And there is always discussion within the polling community over the reliability of telephone polls versus personal interviews. Some have contended that telephone polls have a slight Republican bias because the poorest people, presumed to be Demcrats, will be underrepresented. Others argue that house-to-house interviews undercount the poor because interviewers refuse to go into the poorest neighborhoods.
Peter D. Hart, Mondale's principal pollster, is uncommunicative about his most recent findings. He is not predicting that Mondale, like Harry S Truman, will humiliate the poll takers in 1984. But he points out that on the eve of the New Hampshire primary this year, Mondale was a certain winner over Sen. Gary Hart (D-Colo.), according to the polls, and lost badly. Later in the year, polls gave the Illinois primary to Hart, but Mondale won.
"Sure," he said, "some of the polls have been wrong. When they produce the conflicting numbers we've seen this year, somebody was wrong."
As the election nears, poll takers tend to be nervous. Their reputations are at stake. They remember 1980 when Reagan's large victory caught many of them off base.
Barry Sussman, polling director for The Washington Post, recalls the dilemma he faced three days before the 1980 election. One of his national samples showed Reagan ahead, 43 to 39 percent, and his other showed President Jimmy Carter with an identical lead. Reagan won, 50.7 to 41 percent.
The common explanation for this large discrepancy was a massive voter shift over the campaign's final weekend. That is why several pollsters this year will be sampling as late as the night before the election.
It is now six days until Election Day. A consensus is forming in the polling community that Reagan will have a large victory: 24 points says Time magazine, 23 points says USA Today, 17 points says Newsweek/Gallup.
As Mondale insists that the polls are wrong, he evokes the image and experience of Truman, who, according to polls, was a certain loser in 1948. But there are differences between 1948 and 1984.
One is that polling was a cottage industry in 1948 and is today a substantial industry, providing politicians and citizens with a wide spectrum of options. Another significant difference is that pollsters in 1948 decided early that Thomas E. Dewey would win and stopped polling in the final weeks.
The last Elmo Roper poll, completed in September 1948, gave Dewey a 15-point advantage. George Gallup's "final" poll, published Nov. 1, was completed in mid-October. The results, Gallup predicted, would be 49.5 percent for Dewey and 44.5 percent for Truman. He had the figures right, but Truman won the 49.5 percent.
It is not impossible, in pollster Hart's view, that something of the sort can happen in 1984. He cites the voters' volatility in the primary season as a warning against excessive certitude. But he makes no predictions. He is still polling.