The polls can find out what people are thinking about important public questions. According to their champions, the polls are fact-finding devices which help to keep the public well informed. The polls can keep the public in touch with important shifts in public opinion, it is said, and therefore in the shifts of forces which decide where the political, economic, and social power in this country lies.

Critics, however, point out that polling, in addition to being restricted by the small size of the sample, is subject to the bias of interviewers and those who analyze the data, and that in some instances also the questions may not be well or carefully phrased. Some discount in the reliability of opinion surveys must therefore be made.

Despite the skepticism of polling practices voiced by some sociologists and economists, polls are being used more and more to gather facts and also opinions. The Bureau of the Census uses the sampling method, for example, to gather currently factual information about the labor supply in the United States. Other organizations try to find out the opinions of people on any of the day’s significant issues. Information which the polling organizations attempt to rather can be illustrated by the following questions:

1. What are the wants and wishes of the public?

2. What are the people thinking about?

3. What is troubling the public?

4. What are the opinions of people on current problems?

5. To what extent and in what respects are people badly informed about current issues?

6. What are the voting habits of people, in general, and of special groups within the population?

7. What are the differences of opinion among people in various geographic areas in the country, among political groups, among groups representing different social and economic levels?

The comment is often made that elections do not answer these questions fully. They provide no means of testing the public will between elections. They do not show clearly what the public thinks of current issues when it votes for the candidates. Their results can be misinterpreted. Users of polling methods say that opinion surveys can fill this gap in the information which reaches the public.

What value are pre-election polls?

Election predictions are dramatic tests of polling methods and the reputation of the pollers has come to be based largely on them. Elections offer a severe test of the poller’s methods. He must make two predictions: which people will vote and how they intend to vote. He cannot merely sample a cross section of the adult population. He must attempt to find out the opinions of those who will vote at the election. He must attempt to overcome the influences resulting from weather conditions on election day and other factors which will affect the turnout. He must study the influences exerted by political machines, by eleventh hour campaigning, and the possibility of interest or lack of interest in the election among important parts of the voting population. If his predictions forecast the viewpoints of the adult population rather than the preference of the people who actual ally voted, perhaps more accurately than the election itself, and if they vary widely from the election returns, he will have failed in the eyes of the public. And whenever the election is closely contested, an error of a few percentage points might cause him to predict the winner wrongly. The main purpose of polling organizations is not, however, to make election forecasts.

To be sure, they offer a testing ground for finding out why people vote as they do, when and on what basis they make up their minds, what relationships exist between their votes for certain candidates and current issues, and what differences of political opinion exist among various groups and types of people. The elections also give pollers a chance to study to what extent the voters appear to be affected by political platforms, speeches, and various other forms of publicity.

It would be difficult to show that the 164 election forecasts made by one polling organization during nine years up to the end of 1944 had in themselves served an important public purpose other than to provide a public test of polling techniques. If a poller can forecast elections correctly time after time within a few percentage points, can he be sure that the accuracy of his public opinion polls conducted with the same methods on social, political, and economic questions will be high?

The answer is not simple. The polling of voting behavior is relatively easy and not subject to question. It is like counting a show of hands at a meeting. Polling people on current issues involves complications. Asking people, for example, whether they favor a hard or soft peace for Japan requires them to think about the matter before giving an answer. What does a hard peace mean? A soft peace? Many of those interviewed will answer only to be obliging even though they may not have enough information to give an answer. You will notice that polls often report a percentage who “don’t know” or “haven’t any opinion” about the question asked. This percentage may be a key to the meaning of the poll’s results.

Trust not the populace; the crowd is many-minded. Attributed to Phocylides, 560 B.C.

Are polls reaching new fields?

Surveys based on current issues which are not necessarily connected with a forthcoming election are, then, the particular field which the polls are cultivating. The area of public health is an example. Are people poorly informed about public health? In what respect is more information likely to result in better standards of health and the saving of lives? In what localities are the needs peculiar? What public health measures is the public willing to support? To answer these questions requires a survey among the people. Questions must be asked and answers recorded.

The people have the right and duty to decide on matters of policy. Government officials need advice from the people on questions of policy. They can carry on their work more confidently and intelligently if they know the public attitude. Whether we should have a social security program may well be the subject of a public opinion poll. On the other hand, the detail of just how the social security policy shall be administered is not likely to be a suitable subject for an opinion survey. The public cannot be expected to serve as a congress of experts for considering matters which it doesn’t understand or which can be understood only by persons who have special knowledge of the problem. Technical details and the means for carrying out public policies will have to be left to experts or persons who have had specialized training.

The pressure of public opinion is like the pressure of the atmosphere; you can’t see it—but, all the same, it is sixteen pounds to the square inch. James Russell Lowell

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