Results of last-minute Gailup and Roper polls released yesterday, indicate that Stevenson could pick up enough of the large "shaky and undecided vote" to carry the election, according to Samuel A. Stouffer, director of the Laboratory of Social Relations.
Stouffer was a member of the Committee on Pre-Election Polls and Forecasts of the Social Science Research Council which published a 400-page analysis of the failure of the polls in 1948.
Gallup yesterday gave Eisenhower 47 percent, Stevenson 40 percent, with 13 percent undecided. Roper sees one out of ten potential voters as still undecided.
The figures, Stouffer added, show that as of the end of last week, Ike's "aggressive" campaigning had not succeeded in stopping the recent trend to Stevenson. All it may have done, he said, is to prevent the trend from going more sharply.
Stouffer cautioned, however, that "these last minute trends put the polishers in a tougher spot then ever. If the trend toward Stevenson had demonstrably been stopped or reversed, Ike's lead would have made the odds in his favor. Now, as far as one can tall from the polls, the odds are about even."
As it stands now, he said, either candidate could win by a "whopping majority" of electoral votes even if the popular vote is very close.
Ignorance of this fact led the polls to their most serious mistake in the forecasting of the 1948 election, Stouffer declared. "They misled themselves and the public into thinking that a close national election can be accurately forecast," he explained.
He pointed out that the pollsters are being considerably more careful this time. They are making no definite predictions of the winner, because they realize that with the popular vote as close as it is, very small margins in the popular votes in large states could result in a landslide for either candidate.
He went on to say that the polls could only be considered wrong if there is a real landslide in popular vote for either candidate. They should not be considered wrong if a close popular vote results in a landslide electoral vote.
"Whatever happens," he said, "the polls are going to be accused of hedging. But it would be misleading for them to go beyond their data, just as it would be misleading for the weather bureau which is following a hurricane off the Florida coast to predict three days in advance with certainty that it will hit the sea-board."
The polls have a very good record, on the whole, Stouffer said. "The public damned them in 1943, just as many people damned the weather bureau for failure to predict the New England hurricane and the New York blizzard."
Stouffer said that besides caution in prediction, the polls have remedied their 1948 mistakes in a number of other ways.
They are using "varsity improved sampling methods and are taking pains to dig deeply into the lower economic strata of the population," he said. They are interviewing in the evening rather than the daytime so as to catch working men at home.