An attempt can be made to use polls to influence rather than to reflect public opinion. Polls can be manipulated to give a false picture of public opinion. Moreover, there is evidence that since polls are believed to be reliable and useful, the public could be misled by unreliable surveys.

What influence have they?

But the major polling organizations argue that the polls exert an influence on the public in much the same manner that any book, any set of facts, or discussion of public affairs does. Opinion surveys can, doubtless, help the public by stimulating discussion of current problems.

The public itself—if we are to judge by a poll on the subject—has a lot of confidence in opinion sampling. Asked if they think the polls “are a good thing or bad thing in our country,” 73 percent said, “a good thing,” while 21 percent admitted they didn’t know.

Early in 1945 the Gallup Poll released results of a survey which showed that a majority of voters favored a “work or fight” bill rather than attempts to get people into war jobs by voluntary methods. The division of opinion was reported as follows:

Favor keeping voluntary methods                                    39%

Favor drafting people                                                     53%

Uncertain                                                                      8%

This subject was being widely discussed and debated at the time and the results of the poll intensified the discussion.

Was it good or bad that this evidence was made public? Could the figures influence public opinion? If so, was it a bad influence? Should influences on public opinion be restricted to radio talks, newspaper and magazine articles, pamphlets, public speakers, and other means of reaching the public and not include the results of public opinion surveys? Can you support the argument that it is against public interest to know what a cross section of the population says it-believes about an urgent public question?

The modern polls are designed to report—they do not usually pretend to solve problems. They try to record, not to form opinion or solve highly technical problems. They may exercise the same indirect power on policymakers and the rank and file that any published studies exert.

Do they help load the bandwagon?

Whether the public is actually swayed by the results of opinion polls is hard to say. One test is provided by the election polls. If opinion surveys exert an important influence on the public, then the division of opinion during an election campaign should be in the same direction as the polling results. The leading candidate should gain in strength as the campaign proceeds. Voters who hadn’t made up their minds or who had favored the opposing candidate should be found climbing on the leading candidate’s bandwagon. Can we find any evidence that there is such a trend?

In 1936 the Literary Digest poll showed Landon winning by a landslide. Landon was badly beaten in the election. In the 1940 election one major poll showed that Willkie was gaining strength in the final stages of the campaign. Perhaps he was, but he lost the election.

During the presidential election campaign of 1944 the Gallup Poll published figures showing that 71 percent of a cross section of all voters thought Roosevelt would win the election, 17 percent thought Dewey would win, and 12 percent were undecided. But the civilian vote in the election ran about 53 to 47 in favor of Roosevelt.

Consequently major polling organizations argue that the “bandwagon theory” has not been supported by election data. In general, the public appeared to vote for its candidates even if the odds were against them. The people did not seem to be swung in significant numbers one way or another by opinion survey data.

What do studies reveal?

Studies in local areas have indicated, however, that the bandwagon appeal actually does operate during a political campaign, although a real effect on the outcome of elections has not been proved. A study of the 1940 presidential election in Erie County, Ohio, for instance, showed that some persons who changed their intention to vote for a candidate during the campaign said that they had been influenced by the polls. The number was small. Nevertheless, the study showed that there may have been some bandwagon influence in this case, however small the final effects on the election.

Is it dangerous for the public to follow the polls as a measure of public opinion? No such danger has been proved. On the contrary, disinterested opinion leaders have not hesitated to study data resulting from opinion surveys and to use them freely in public discussions. Polling results are published widely in reputable periodicals and in articles by conscientious students of public affairs. Results of opinion surveys are included in studies such as the Foreign Policy Reports, not as conclusive evidence, but as contributions to an understanding of public attitudes. This would indicate that many students of public affairs take the results of polls seriously.

From EM 4: Are Opinion Polls Useful? (1946)