Puncturing the Polls
From• RICHARD H. ROVERE
NEW YORK
WE are at the start of a lively and perhaps useful debate on .public opinion polls. It was touched off by the publication, on August 17, of a Gallup survey which found 50 per cent. of the voters favouring Vice-President Nixon, 44 per cent. favouring Senator Kennedy, and 6 Per cent. undecided. It was explained that the sampling had been made in the immediate after- math of the Republican Convention and might be presumed to reflect fresh and favourable impressions of Mr. Nixon's acceptance speech. It was not, however, the lead enjoyed by Mr. Nixon that aroused suspicions and complaints, but the percentage of the 'undecided,' which seemed astonishingly small even to the members of Mr. Nixon's staff who commented on the poll. In a speech on the Senate floor, Albert Gore, a Democrat of Tennessee, recalled that a poll Published on election eve in 1952 had reported that 13 per cent. of those questioned had not made up their minds between General Eisen- hower and Mr. Stevenson. Senator Gore also said that he knew of another poll, conducted by an agency fully as reputable as Dr. Gallup's, Which showed 24 per cent. unready to commit themselves either way as of a few days after the 1960 Republican Convention.
An explanation of sorts was made by the columnist Joseph Alsop, who checked with some of Dr- Gallup's statisticians and found • that in the August 17 poll persons who did not profiss to have made a final choice but who inclined at the moment to One or another of the candidates were assigned to the side toward which they leaned. Mr. Alsop, himself a pulse-feeler of almost professional. standing, also entered objec- tions on other scores: the methOd of disconnt- Ing the opinions of non-voters seemed faulty; the Democra,tic South had not been given its proper Weight; the sampling—under 1,300 by his calcu- lations—was so small that a shift of thirty-nine votes toward Senator Kennedy would have pro- duced a tie.
Dr. Gallup has responded to Setaiiar Gore and Mr. Alsop with a defence of his method that rests primarily upon the correspondence, over the Years. of his figures with the actual vote in national elections. He claims that since 1948— the Year that he and almost everyone else pre- dicted a smashing defeat for Harry S. Tiuman- Gallup 'poll findings have deviated from election results . . . by an - average of 1.7 percentage Points.' To non-statisticians—and to at least a few of the statisticians who have joined the debate—this seems moderately impressive, and one imagines that it is in any case the best that statistical ingenuity can do. There are, after all, no percentage points for Dr. Gallup in being deliberately wrong. He and his competitors have nothing to sell but their approximation to accur- acy, and in point of fact the basic case against the polls is strengthened not so much by evidences of unreliability as by evidences of reliability. It is because the polls have been far more often right than wrong that their influence on the course of events is to be feared. 'The poll itself,' as Samuel Grafton wrote recently in the 'New York Times, '[has become] a political event, interfering with the thing observed.' The com- pletely trustworthy seer would be a terribly dangerous man to have around, and, in recogni- tion of this,, one of the country's leading measurers of public opinion, Elmo Roper, de- clined to conduct any political polls in 1959. 'I have always felt,' he wrote, 'that the publishing of the fact that some really good man was far down on the list in the public's esteem might have nipped some very promising candidacies in the bud, and the one thing I don't want this measuring rod to do is to influence the thing it is intended to measure.'
Dr. Gallup has pooh-poohed this by citing .Harry Truman's defeat of Thomas E. Dewey in 1948. 'If people had wished only to be with the winner,' he wrote last week, 'then Dewey would have won by the greatest majority in history.'
This hardly meets the objections raised by Mr.
Roper or by Senator Gore, who wants a Con- gressional investigation of the whole subject of polling. The man with a secret ballot to cast may feel no obligation to ride the wave of the future, simply because it has been identified for him by Dr. Gallup, but the man who is thinking of contributing his talents or his money to a particular cause or candidate would almost cer- tainly be put off if he was advised of the cer- tainty of defeat. And there are other dangers. According to Robert Day, Vice-President of Columbia Broadcasting System News, a privately conducted poll taken in Idaho in 1956 showed that an ageing Republican Senator known to be a heavy drinker could be defeated by a young Democrat whose virtuous ways were reflected in a virtuous face, particularly if the young Demo- crat 'ran with high moral indignation against his Republican Senatorial opponent in the Idaho Mormon belt.' The poll proved to be right— or at least when the time came youthful virtue was triumphant.
The investigation Senator Gore has called for may never come to pass. If his party leader, Senator Kennedy, proves error in the trend re- ported by Dr. Gallup on August 17, he will not be too eager to have his lieutenants in Congress explore this subject in any depth, for Senator Kennedy, more than any other candidate of recent times, has relied on the work of the poll- takers. He has for some time retained one of the leading profeisionals. Louis Harris, and he planned his strategy in every primary election on the basis of Mr. Harris's findings. He is said to have spent more than a quarter of a million dollars testing his strength here, there, and everywhere, and Mr. Harris's work would pro- vide the best study materials for any investiga- tion. Furthermore, it is difficult to imagine any legal measures that would lessen the evils of the practice in the event that it was found to be associated with any evils. It is a free country, and any man has a right to ask any other what he thinks, and to publish. in the free press, any findings that do not violate the laws of libel or those protecting privacy. If dangers •exist, we shall have to live with them.
Previous page
