Polls and the Science of Public Opinion
Floyd Henry Allport
After discussing the bandwagon tendency and the function of polls in correcting "pluralistic ignorance," Syracuse University's social psychologist examines the nature of public opinion, on the basis of which he then criticizes the polls for ignoring several important dimensions of opinion measurement, and for framing questions in collective phraseology rather than in terms of the individual's own experience and expectations.
ONE OF the most significant of modern advances toward realism in the study of political and social problems is the rise of polls of public opinion. As with any new and basic method, however, it is expected that in their earlier stages certain crudities will inevitably appear. Since the objective in establishing the polls was one not only of fact finding hut also of reporting and publicizing the facts found, the techniques, on the side of content and phraseology, have savored strongly of the journalistic approach. In spite of the wholesome progress toward accuracy in sampling and the use of statistical methods, the content of the questions themselves rests not so much upon scientific precision as upon the clichés and stereotypes of thinking which newspaper editors are prone to regard as the mental equipment of the "man in the street." Now that clouds of mistrust regarding the polls are arising in some quarters, it is a strategic time to subject the procedure to a more careful methodological scrutiny. This paper is concerned with only a small part of this broader undertaking, namely, a discussion of opinion polling as viewed in the perspective of a scientific description of the public opinion situation, and with the representative character and significance of the questions asked. There is space for only the briefest outline of the problems which emerge.
Among the criticisms of the practice of publishing poll results, one of the sharpest is that which is based
(250) upon the well-known "bandwagon" tendency. We tend to behave as we see others behave, or, if we are feeling or acting in a given direction of our own accord, we tend to be swayed the more strongly if we see or hear a large number of others acting in the same direction. A published poll result, showing clearly where the majority lies, will therefore be one of the most potent methods of encouraging this tendency, and of piling up, deliberately or unconsciously, large modalities which are based more upon social influence and suggestion than upon rational decision. The writer long ago described these tendencies as basic principles of social psychology under the terms of "attitude of conformity" and "social facilitation."' It is interesting that political leaders should have been unconcerned about their existence until brought face to face with them in situations where their effects are hound to tread on certain toes. Having discovered that public opinion is a term to conjure with, especially if one has the data proving that a certain position is public opinion, the guardians of our legislative halls are likely to be "touchy" as to whose side the conjuring advantage is on, and to question therefore whether every Tom, Dick or Harry should be allowed to conjure. h might even be better, in their opinion, to investigate, and perhaps to suppress, the entire procedure.
This criticism, however, seems to rest upon the assumption that conjuring of this sort is a new thing. As a matter of fact it is probably as old as the game of politics. There is scarcely a campaign or lobby which does not draw to its close without the emphatic announcement that "public opinion"is overwhelmingly on the side of the speaker or the lobbyist, and that his victory, or that of his cause, is predicted by a large majority. The only difference is that, whereas the sense of universality was previously only an "impression," capable of being distorted to suit the publicist's convenience, it is now, through the polling procedure, capable of being based upon substantial fact.
Correcting Pluralistic Ignorance
There is good reason indeed to turn our attention from public knowledge of citizens]1] reactions to the reverse condition, namely public, or (as the writer has elsewhere termed it) "pluralistic" ignorance.
The lack of awareness, rather than the awareness, of how others think and feel is the true danger sign for democratic processes. In such a situation suggestion and propaganda are of the greatest influence, since it is easy, without any check on facts, to make citizens believe that a certain view is the majority, if not the universal, opinion (impression of universality). Another well established principle of social psychology is one which we may term social projection. Where the attitudinal composition of the field is unknown, individuals tend to "project" into it, usually un-
(251) -consciously, their own views, prejudices, or desires—in short to think that others are behaving or thinking as they are.
If an individual has been convinced unknowingly (as most of us are) by some appeal, based upon inadequate data, that a certain proposition is true, he will then tend to project the acceptance of this same proposition into others. By this projection the principles of conformity and facilitation are thus secondarily involved to clinch and strengthen the belief. It is here that the real menace to democratic government is to be found. An accurate knowledge of how others actually think and feel is not a peril, but a corrective of this danger of "pluralistic ignorance." This knowledge of the opinions of others must, of course, be accurate, or else this corrective advantage will not be gained; hence the importance of a scientific conduct of polling organizations.
The Nature of Public Opinion
What we have just said points further to the need of clarifying some naïve, but common, assumptions about the nature of the public opinion. It is frequently believed that public opinion is like a great voice, or force; and that as a new common need arises and citizens gradually make up their minds about it, this voice swells and gathers momentum until it compels the attention of law makers or executives and forces them to act. Such a conception, probably more applicable to an earlier period of our history, is now outmoded. The dynamics of the public opinion process must now be seen as lying within the framework of collective systems of events. Collective activity is actually the end result of a kind of circular process in which individuals, in their many riles or capacities of workers, clerks, farmers, merchants, financiers, and executives, all cooperate. Each plays his part, reacting to others who are in rôles similar to or different from his own, in the well-known manner prescribed by the division of industry, transportation, communication, and employee and managerial function. Frequently such organization takes the outlet of collective conflict; each individual doing his part toward an end result, formulated by a leader, and conceived in the interest of those involved upon the one side or the other. Political processes, whether the voting and controls are exercised within a political, an industrial, art ecclesiastical or other organized structure, are merely the popular mandates by which power is conferred upon officials to guide this organized system, or authority given for certain specifications regarding its operation. Societal "event-systems" of this character are enormously complex; and their full description and understanding is beyond the range even of the expert.
Opinion As Energy
Space does not allow us to describe this process in detail. It can only be said that it is this collective event-system framework which, under leadership, determines the character of the solution of an issue and the alternative courses which may be pursued, and not the attitudes of the citizens who operate within it. Attitudes and opinions are neither the initiative nor the directive force of these systems, but are merely the energies (in the form of morale, self-interest, loyalty, etc.) which make them operate. The energies of "public opinion" are thus dynamically like the energy afforded by the spring of a watch. They do not dictate that the outcome will be of a particular character (movement of the hands), but only whether or not the end dictated by the system and its executives will be accomplished. If the energy level is too low (i.e. "public opinion" is too weak or sparse) the end event will not take place. To make it occur, that is, to make the event-system operate, a certain threshold of extensity and intensity must be crossed; and this means that a more widespread appeal must be used, or the conditioning by propaganda of the more fundamental "groundwork" attitudes of citizens to the symbols of control must be effected.
Public opinion, therefore, is likely to be regarded from the standpoint of the directors of the event-system not as something graded or measurable, like all scientific variables upon a continuum, but as an "all or none" affair. It is either for a thing or against it. Voting procedures, whether official or in non-official opinion polls, are really testing devices by which those directing the event-systems can know whether the popular energies for realizing the objectives of those systems will be forthcoming. It is this dichotomous view of public opinion, used for event-system operation, which is responsible for most of the distortions which creep into the current polling methods. From this interpretation we see that the notion of public opinion as that qualitative self-expression of every citizen, upon which a democracy operates, must be drastically modified to fit the modern scene. Public opinion dynamics are no longer of this simple and straight-forward sort. The "man in the street" has very little direct knowledge upon which to base his judgment about either national or international policies. His "public" opinion is therefore not of or about himself in a direct relation to his government, but is third-personized; it is about some proposed law, plan, or controlling party the operation of which is beyond his power or understanding, and about which he gets all his information second-hand.
The dynamic of the "swelling tide" of popular support is itself a circular process, a process of interstimulation between the publicists and the people. The output of editors, news writers, and broadcasters arouses widespread emotion or agreement; the signs of this agreement further encourage the writers and broadcasters, and behind them the
(253) directors of political event-systems, so that they redouble and strengthen their publicity efforts. This increase of publicity in turn raises the energy level of the masses still higher upon the issue, and so on by a process of circular reinforcement. Meanwhile a further circular process of reinforcement is going on through facilitation, attitude of conformity, impression of universality, and projection, among the citizens themselves. The very evidences of this heightened influence (energies) are used by the leaders, under the slogan of "Public Opinion," for the purpose of further accelerating the process. Public opinion thus generates itself, or rather the term "public opinion" is used as a control by which attitudes of citizens of the public are progressively modified in a certain direction. "Public opinion" is thus a tool of collective system control, quite as much as it is an expression of any individual's attitude or feeling.
Dimensions of Opinion Measurement
We can see the influence of the collective event-systems of modern social organization upon the nature of the questions which are asked by the directors of the opinion polls. To show this let us consider the dimensions which opinion measurement might take. Without prejudging the actual number of possible dimensions, we may here name three, viz., the societal-logical, the a affective or intensity variable, and the relic. By the societal-logical dimension we mean a continuum upon which a series of political plans or policies might be stated, all related to each other as fulfilling in different degrees the purpose of some course of action lying between two definable extremes; for example, a series of plans for extension of government control of industry, varying from no control to the maximal possible control. It is noteworthy that, with few exceptions, this dimension, which is so necessary to discover all phases of opinion on a vital question, has been neglected in the polling procedures. Instead of evoking replies all along a dimension of this sort, the interviewees usually have been asked to state their opinions only upon a single item of such a possible continuum, as for example upon the particular policies of the New Deal Administration. The reason for this takes us back immediately to the event-system and to the administrative dichotomies of energy upon which event-systems operate. Political leaders and newspaper men are not interested in the whole gamut of possible policies, for the debated
(254) issue lies at only one paint of the scale. The question for them is not what possible alternatives there are, and how these fit in with the lay of the attitudes of the citizens, but whether a particular plan in which they are interested can be put into operation. The finer shades of individual choices are thus concealed.
Meanwhile it is possible to bring to bear upon the one, often oversimplified, solution all the weight of rhetoric and propaganda, and to employ toward this objective the mechanisms of impression of universality, conformity, and projection. A good example of this concealing effect is shown in the pollings revealing the trend of favor toward President Roosevelt's administration. With this one position on the continuum alone, we cannot know for certain whether these polls indicate the trend of concern of the people about war-danger, or some other reason connected with the administration, or perhaps a fluctuating regard for Mr. Roosevelt himself. If the Roosevelt administration however were made one step in a number of societal-logical continua, that is, one step through which a number of such continua intersect (e.g. social security, desire to improve business conditions, trend toward economy, or aversion to war), we might be able to tell what this poll-trend really indicates. To be sure we have concurrent samplings on these other questions, but these also do not admit of clear interpretation without a picture of the distribution upon a continuum of which they are a part.
The second dimension, that of affect or intensity, measures not the logical position of the choice of the citizen, but the degree of intensity of feeling with which he clings to the choice which he does hold. It applies to one position on the societal-logical continuum alone. Suppose, for example, we take sit-down strikes as one method of a possible range of methods for attempting to settle industrial disputes. The affect continuum here is one of the entire gamut of feeling one might have about this matter, from the most intense opposition to the most intense approval. But here again, the polling procedure, with some exceptions, has been one of simple dichotomy. The question as put has been merely "do you approve or disapprove"?
Here again the reason for the administrative dichotomy lies in the extensity threshold of the event-system. A vote is a vote; it is of no consequence whether the feeling with which it is given is strong or weak, so long as it is on the "approval" side. This is, of course, a short-sighted position, both for democratic ideals and for the prediction of possible future opinion trends, a matter with which every political leader must be deeply concerned. The writer would suggest that an opinion index of true dynamic significance might be afforded by discovering, through polling, that societal-logical continuum on which there would be
(255) an increasing skewing of the distribution on the intensity dimension of each succeeding step.
The Telic Dimension
There is, finally, the telic dimension. The question here raised is not where the interviewee's choice belongs on a societal-logical scale, nor how intensely he feels about it, but how effectively will he act toward its
realization. The effectiveness of his action depends somewhat upon the energy he puts forth, but more especially upon the role he plays, and is able to play, in the operation of the event-system itself. The technique of this dimension would consist of questions as to what the individual will do in certain circumstances, and a method of evaluating his replies.
This dimension explores the extent to which citizens can or cannot become agents in controlling the systems in which they operate. Hence it may have possibilities for helping us toward the democratic ideal of true individual self-expression in a society cast, as is ours at present, in the mould of event-systems of collective action. The writer knows of no instance thus far of the use of the relic dimension by any of the American polling organizations.
The consideration of what citizens actually will be willing to do in a situation brings us to a broader criticism of the type of polling questions usually employed. Public opinion dynamically is an energy level which will make a system of operations work. This fact slants political planning, and therefore opinion interrogatories, in the direction of collective phraseology and broad governmental or social policies, rather than directly toward the more intimate concerns of individuals. Partly this error is inherent in the representative legislative process. Officials must think in somewhat abstract terms of plans for the country as a whole. According to democratic principle they must go back to the citizens for a mandate with reference to these plans. For citizens to give their opinion upon such matters is therefore eminently proper. If, however, we rely mainly upon the presentation of questions of this type, we shall fail to approach realistically the true problem of the meaning of government to individual citizens. The "man in the street,"generally speaking, is little informed upon these matters, and no available information would make him competent to decide the merits of the issue as a national or state policy. Even experts are often at sea about such matters.
Consider, for example, questions dealing with such issues as a mea-
(256) -sure to reorganize the Supreme Court, to alter neutrality restrictions, to authorize greater Congressional control over industry and agriculture, or to authorize a federal loan for the recovery program. In these questions the average citizen can think of the proposal only as something applying to an abstraction such as the State of the Nation. It is vague in its implications, and apart from himself. Sometimes the national entity is almost personified, as in such a question as "should the United States increase the size of its army"? By asking questions of this type we take the issue away from the individual and "third-personize" it. In order to discover how the individual stands in the most dynamic and pragmatic sense, ,we must bring the problem back for him into the first person. The things which the citizen does know, and the questions which he, and he alone, is competent to answer, are how he is faring under the plan now in operation, what he wants for himself (not for the nation), to what degree he is realizing it, and what he is willing personally to do or to sacrifice in order to get it.
An example of concealment of meanings through third-personization is seen in such interrogatories as "which party would you like to see win the Presidential election of 1940"? Many of those interviewed probably never get beyond the notion of a great Being, the Party, and all the emotional conditioning which they have attached to the stereotypes "Republican" and "Democrat" will determine their answer. Let us suppose, however, that the interviewee goes beyond personification, and takes the question as referring to individual Republicans or Democrats who will be elected as leaders. In this case the question, to have any paint beyond party labelling, must be stated more specifically to call attention to the particular candidates involved; for it is only the background and merits of these actual candidates which has any realistic bearing upon the problem. Men do not acquire any special character by virtue of a label of party affiliation. Or again, it may be the difference of platforms or policies. In this case, similarly, the question should involve this discrimination, and should provide, as usual, some test as to whether the interviewee knows the difference between the platforms of the two parties.
For many to say they "do not know" in response to this more specific questioning might be less exciting for journalistic purposes; but it certainly would be more revealing than the "group fallacy" procedure of asking preference for a party entity. The disclosure of such fields of lack of knowledge would, in itself, be one of the most valuable contributions of the polling procedure. Finally, the third-personized form of this question should be further resolved, and still more illumination produced, by asking the individual interviewed not merely which men or which party policies he prefers, but just what it would mean to him personally if they are, or are not, accepted at the coming election. Such a modification of poll-
(257) -ing procedure would, moreover, go far toward eliminating the effects of emotional propaganda, impression of universality, and attitude of conformity which are characteristic of bandwagon movements and other evils of current opinion situations. It would be one step further in the direction of breaking through the event-system of collective action and Finding the individual.
Some of the polling questions used are, indeed, couched in these first-personized terms; but they seem to he far outnumbered by questions of the third-personized variety. A preliminary count of 198 of the questions asked over the period from 1937 to 1940, in the American Institute poll shows that the latter stand in a ratio to the former of three to one.