ALFRED DE GRAZIA
Associate Professor of Political Science and Executive Officer, Committee for Research in Social Sciences, Stanford University.
AN HOUR OF TALK cannot describe all that we know about voters and elections. A recent study of voting in Elmira, New York, listed 209 propositions about voting behavior that several studies contained. The list was a short one and could easily have been multiplied by ten. Much of what is known is familiar to you already. Some of it is unimportant. Much, too, is only known or known within narrow limits of time and space. It is apparent, therefore, that an annotated inventory of the corpus of public opinion and behavior would be inappropriate. My function tonight is perhaps better construed as an explanation of our research behavior than as an inventory of public behavior. How are we studying the public and what major theories are we developing? Hence, the first part of my presentation will summarize six changing techniques of research, and the second part will describe six concepts being subjected to research by means of those techniques.
The Development of Technique
Every technique of research employed today has descended from the techniques used by politicians. Man imitates nature and science imitates man. So refined and polished has many a technique become, however, that it is sometimes incapable of being understood or appreciated in its original sense.
Briefly, when a political scientist wants to know something about voters and elections, he resorts to some versions of six techniques : (1) he describes his own experiences, (2) he interviews exports or informants, (3) he analyzes the content of the press and other documentary materials, (4) he draws upon a sample of people, (5) he analyzes voting figures and census materials, and (6) he performs experiments that resemble "trial balloons."
All of these procedures are, of course, practiced also by politicians. A congressman will use all of these techniques, sometimes in a single day. He sifts and evaluates his work by means of them. He will, (1) reflect on his last visit home for clues to public opinion, (2) ask a fellow-congressman or visitor his opinion of a bill, (3) read his home newspapers and analyze his mail, (4) spend a few dollars in phoning a sample of constituents on a newly arisen issue, (5) glance back over the last election returns for clues about how the new issue might change the next vote, and (6) finally, at the fagged end of the day, call in his assistant and say "John, phone Bill Smith of the Home-Town Herald and tell him I'm not likely to be a candidate for re-election." (I need not specify that this last announcement is a "trial balloon." It could also be termed a "Utopian fiction.")
The scholar is different in one large way: he is specialized in the gathering of facts, whereas the politicians, including the executive government, spends more time doing than studying. Originally, this difference mattered little. Until recently, the observations of American politicians on human affairs were, on the whole, more realistic and penetrating than those of American scholars. But times have changed, and the scholar has developed highly his special concerns. It was not an increase in facts that caused the splitting off and mutation of the scholar; a politician knows as many fats as the scholar, and neither knows as many as a quiz kid. Rather, it was technique that mutated the academic man. Beginning only a generation ago, he began to value highly the possibility of demonstrating with the least possible doubt that, when a vote is cast, its outcome is predetermined by measurable conditions. The logical direction of the mutation involved overdeveloping and improving the traditional modes of knowing politics. The story of how techniques have developed can be illustrated, albeit only briefly.
Politicians, like salesmen and soldiers, swap stories endlessly. Sometimes they write them down, in which event the political scientist analyze them and pass them along to their students. Most of what we know about the relations between politicians and constituents and much of what we know about the behavior of the public come from the stories of men like Farley, Flynn, Voorhis, Hopkins, and Ickes. A recent study by Walter Johnson, How We Drafted Adlai Stevenson, is a good example of that type of story.
Unfortunately, autobiographers ordinarily refuse to let others interpret their stories and often pay often pay more attention to the interpretation than to the stories, Hence one gets too few details and sometimes inaccurate ones. In the long run, it is better to pay less attention to the existing data, and to perfect the techniques of interviewing and sampling. Interviewing by the lengthy methods of psychoanalysis, as developed by H.D.Lasswell, in Psychopathology and Politics (1930) and elsewhere, promises an increasing depth, insight, and fundamental validity. Better sampling techniques can tell us from whom experiences are being extracted and what they represent of the world of experiences in the political process.
Another major hope of improving on the rumination of experience lies in the technique of participant observation. For example William Whyte, a social anthropologist, spent a long time among the young gangs of a metropolitan neighborhood, recording painstakingly their attitudes, way of life, organization, and relations to the political structure. Alexander Leighton, a psychiatrist. contributed similar reflections following experience with Japanese-Americans at their wartime relocation center. The recent development of case studies in administration by Harold Stein should give inspiration to similar studies of the vote decision and grass roots political activity. We could well use accounts of exactly how a politicians spends his days, of what happens in polling places on election day, or of what kinds of people ask questions during or following a political meeting.
"Asking the man who knows" has long been the policy of anyone who is not quite sure of himself in a complicated enterprise. In fact, a century ago, an Englishman, G.C. Lewis, wrote a treatise entitled On the Influence of Authority on Matters of Opinion, which, in substance, held that when one wanted to know something one asked someone who knew it. But the last generation of social scientists has made this simple injunction only a vague and trite introduction to one of the most complicated fields of scientific methodology. The old adage has led to the most strenuous mental and empirical exercises centering around the questions what one wants to find out, who knows the material, how does one know he knows, and how does one find out what he knows.
Those of us who have had experience in treating with clients who have problem, be they politicians, businessmen, or administration, have realized that an expert cannot be asked a proper question when the questions is not sure at all what he wants to find out. The point holds for both pure and applied scientific inquiry. Questions that a generations or two ago would have been considered clear and rational are today considered muddled and impossible to test. A whole literature on the farming of hypotheses has developed, easily rivaling in sophistication the analogous theory of the natural sciences.
The technique of interviewing has developed greatly. As a result, the new political scientist is acutely aware, not only of the necessity of knowing what he wants to find out, but also of the way in which forthright questions may frequently bring him unreliable, false answers. So his craving for specificity and directness is conditioned by new knowledge of the psychology of resistance to being asked questions, whether the subject is a common man or a lofty expert.
Similar development have altered the technique of determining answers to the other queries posed above on who knows, how does one know he known how does one find out what he knows. The works of David Kurtzmann on the control of votes in Philadelphia (1935) and of Harold F. Gosnell on machine politics in Chicago are exemplary of the solid earlier developments in interviewing before World War II. V.O. Key and Alexander Heard could not have written their definitive studies of southern politics without having elaborated a technique of approaching, interviewing and summarizing the opinions of many southern officials. A contemporary example of a most complicated form of interviewing may be cited also. Political scientists of the Center for International Studies at the Massachusetts Institute of Technology are presently analyzing 450 interview of political and business leaders on the developing reciprocal trade issue. A scattered, specialized public is being examined to determine the kind of communication network that exists in the diffuse social space where tariff policy is generated.
Analyzing Qualitative Materials
Politicians have long sought to know how people are thinking by reading the same things that the people read. The late President Roosevelt used to read about 16 newspapers a day. He had a special table from which he fed himself on all kinds of publications on his way to bed at night. He read some fifty letters a day. No doubt he read to learn the truth, and to discover ideas, but like the new political scientist, he read also to learn who saying what to whom with what effect. That is, he, like most politicians, used the printed word as a clue to the behavior of people on whom his course of action depended. When someone says something, politicians want to know what has happened to him or what he is up to. Generally, that is the purpose of content analysis, as it has developed over the past generation: a flow of words is primarily interesting as an expression of attitude.
Several scattered studies in the nineteenth and early twentieth centuries foretold the systematic development of content analysis in the thirties. The pace of development proceeds vigorously today. The contents of newspapers and magazines provide a plentiful supply of raw material to candidates for advanced degrees. The mailbags of a few politicians have furnished samples for scholarly investigation. A Minnesota group has studied the personalities of people who write letters to the editor. Expressions of editorial opinion are correlated with events. For instance, Robert Lane of Yale University has just made a notable application of technique to business publications, discovering that the incidence of protest to new government regulations starts high and then declines steadily and that a new regulations is not regarded as the "straw that breaks the camel's back" but has its own career of protest and acceptance.
Some day the sober prose of Fortune will be connected through columns of figures with the golf scores of its editors. For increased attention is being devoted to translating new kinds of qualitative into quantitative terms. A group of studies in quantitative semantics, in which Harold D. Lasswell is the central figure, might be cited here. Paul Lazarsfeld has contributed a theory of index numbers for the transformation of data into measurable items. At the same time, survey organizations, particularly the Survey Research Center, are using more open ended questions in interviewing, a development that practically necessitates content analysis. For example, the simple question "Is there anything about Eisenhower that might make you want to vote for him," when asked of a national sample in 1952, produced many hundreds of replies. These replies had to be combined into classes to be manageable. The result was eleven categories and over a hundred subcategories into which the replies of all respondents could be fitted so that a complete image of Eisenhower might be constructed, and its parts related to many other characteristics of the respondents.
Polling a Sample
The prominence that the present day affords the science of sampling scarcely needs mention. The aim of sampling is to make possible the description of a whole population by talking about a part of it. To this end, statisticians, psychologists, and sociologists have concocted many complicated procedures in recent years. Although political scientists are not presently in the forefront of sampling technology, it may be recalled that one of the first investigations to employ systematic sampling was the study by Merriam and Gosnell, Non-Voting (1924). Going back much farther, to the debates over virtual representation and proportional representation, one encounters an interest and certainly, at the least, a sociological ingenuity of political scientists in coping with the question: Who represents whom under what circumstances? An early American politician and political scientist, John Adams, in the debates over the Articles of Confederation, called for a legislature in which "the interests within doors should be the mathematical representation of the interests without doors."
Political scientists have also developed the field of sampling in content analysis. Lasswell, Alex George, Ithiel Pool, Nathan Leites, and others should be recognized as pioneers in this important area, which, it is again worthy of note, is preponderantly worked in schools of journalism nowadays. Here, again, is an example of how political of how political science, like bountiful nature, casts out many seeds of new approaches and techniques.
It is more serious that the needs of political scientists are not being met by sampling experts who have been too much concerned with large samples of many hundreds, whereas the political scientist has thousands of problems of sampling that demand samples of a dozen to several hundreds. At this point where the skill of the statistician stops, the training of the political scientist in handling slippery data goes on to good effect.
Progress in the techniques of drawing and working with small samples is slow. Statisticians do not like the problem, I have observed. It is almost hopeless to begin with. The theory of probability will indicate that if even a two-sided relationship has to be generalized from 25 or 50 cases, the margin of error is staggering. The so-called law of averages can hardly begin to work even if the two-sided relationship is represented in a coin. But 25 to 50 cases are all one may have in studying legislative opinion, pressure group opinion, local political organization. grass roots politicians, and so on. Furthermore, the larger the sample, the less possible it is -- in time and money--to ask many questions, and all too many problems of politics can be solved only by intensive interviews. Even the whole public, as Lindsay Rogers has shown, when asked simplistic questions, speaks in "baby talk" rather than expresses public opinion, Samuel Lubell has a deft technique that should be analyzed: neither fish nor fowl vis-a-vis sampling, it combines many brief interviews with a few prolonged conversations in a way that has brought generally credible results. (He bolsters this method, of course, with intensive scrutiny of election returns and public opinion polls.)
Moreover, even large, general samples are not large enough for some purposes. For example, let us suppose you want to compare the voting behavior of college-educated children of middle-class parents and college-educated children of working-class parents. Probably not even a national sample of 8,000 persons would provide enough cases of the latter to make reliably the comparison; nor could even a larger sample allow such obvious constants as religion and income to be introduced. Such a group could be reached by more purposive sampling, but then its relations with other aggregates of the general population would remain unknown.
Partly to help solve such problem as the preceding, but also to provide many other forms of aid in studying political behavior, more co-ordination and co-operation of survey agencies is required. What we need, I think, is a Central Sampling Services Agency that can accumulate all the means of drawing samples of all sizes and descriptions, and maintain the materials derived form all such studies so as to relate continually the successive special and general findings.
On some problems of voting, one has facts for the whole population under study. The census and the reports of local election boards and secretaries of state provide most of the information. The very word "statistics," probably meant originally the study of the state. It was narrowed down to the numerical study of the state in the nineteenth century and then widened to include the many numerical studies of nonpolitical aggregates. In research on voting and elections, statistical analysis began with the great debate over apportionment, even as early as the seventeenth century in England and the Colonies. Edmund Burke, for example, compared delegates from "rotten boroughs" with those from populous boroughs and concluded that the quality of representation was not inferior in the one case to the other. Jeremy Bentham, a while later, "proved" quite the opposite.
Again, however, the serious study of social and election data correlations did not burgeon until the twentieth century. In 1894, O.G. Libby showed the geography of the vote on ratifying the constitution, which Charles Beard used in his famed work on the economic interpretation of the constitution. In 1928, Stuart Rice published his Quantitative Methods in Politics, a series of jabs into unknown territory. Many studies followed under James K. Pollock, James Neprash, and others. Harold F. Gosnell and his students developed the statistical analysis of elections to its highest point in Machine Politics: Chicago Model (1937), and Grass Roots Politics (1942).
Where to go from there has been a puzzle for twelve years. Information of election results and the social traits of a population depend on governmental reporting. Only a few characteristics are known for any area in which the two elements are to be related; these are commonly various economic indices, such as median rentals and crop or industrial specializations. Other significant variables are either uniformly distributed over census areas (such as age and sex categories) or have largely unknown distribution (such as religion, ethnic background, and psychological traits and attitudes).
On the other hand, since most samples are designed to represent large populations, questions pertinent to specific jurisdictions cannot be answered from them. General samples do not allow one to compare, say, Nevada and Illinois, or Los Angeles and Dallas. Election statistics may provide the only numerical help in these cases. Few politicians carry a poll in their pockets, but many carry a summary of the election returns, unless they have already memorized them.
Most men follow their every action on other people with attention to the effect produced on the people. The same sequence of behavior occurs when a politicians raises a "trial balloon," save that the motive for the stimulus is not so much to change the behavior as to test the response. President Roosevelt's famous "quarantine of the Axis powers" speech in Chicago in 1937 was partially intended to incite a public reaction that would indicate the limits of foreign policy. Many leaders carry on similar rough tests.
Scientists would call them experiments of the before-and-after type. They have many analogies in public opinion research. Gosnell's early experiment in getting out the vote added a standard feature of more systematic experiments, the control group. He showed that a difference could be produced in the turnout of two similar areas when only one was propagandized in favor of voting. The National Opinion Research Center used a modified design in a study of United Nations propaganda in Cincinnati in 1948. In 1950, Citron, Chein, and Harding staged a drama in a theater and recruited passers-by to see it. On the stage, incidents of bigotry were portrayed, each followed by a different reaction against the bigotry. The audiences were asked which reaction they preferred. The most preferred response to bigotry was a calm, quite argument on behalf of the American tradition. The Minnesota Laboratory of Social Relations has been subjecting individuals of known history and test scores to stresses and shocks, in order to determine the stability of various kinds of political attitude. The variety and ingenuity of experimental design is increasing.
Perhaps there should also be mentioned an equally luxuriant growth of what can be called ex post facto experimental designs. These are essentially retrospective hypotheses: that is, provided the data is at hand in the form of voting results and social data, or as survey materials, one can posit a new theory and go back to reorder the data to test the theory. Properly assembled, a single body of data, such as that collected by the Survey Research Canter in 1952, can be used over and over again for this purpose. The great costs of a large data-gathering enterprise can thus be liquidated over a period of time. Hyman, Sheatsley, Marvick, Eldersveld, Campbell, Davies, Cantril, and Bruner are some who have made new discoveries on the basis of old materials.
Most data of political science have wasted on its collector, no matter how bright his original contribution. The major drawback of Samuel Lubell's analyses, I might suggest, is their dependence on no one but Samuel Lubell. His data cannot be re-worked, re-tested, or re-combined for new discoveries. Incidentally, librarians, who have special professional schools and who should be in the forefront of the task of providing data for scholars, have not an inkling of the possibilities that lie in the accumulation and organization of such social data for use. Most foundations, looking for pie in the sky, are similarly negligent.
In view of all that has been said thus far, I think it fair to conclude that the origins of our techniques lay in the practices of men, but that only the specially tutored may now work on the frontiers of common sense. Perhaps, in concluding this argument, it should be affirmed that no technique is sufficient in itself to solve a problem and that often all six are employed on the same problem. Furthermore, the high sophistication of one technique used in a problem does not imply that the author is a virtuoso on the other techniques being used therein. Indeed, every technique save one in a given study may be on the level of a wise or not-so-wise practitioner.
The variety of techniques and probably the strictures on them are illustrated in my work, the Western Public, on the one subject of issues of the 1952 campaign. There I injected some observations form my own experiences with political organization in the West and elsewhere; I sought the opinions of experts, made a content analysis of the press and of responses to open-ended questions regarding issues, concerned myself greatly with the possibilities and limits of my modest sample, analyzed the election returns, and liberally employed the retrospective hypothesis in testing new ideas on the fine collection of data with which the Survey Research Center endowed me.
Current Fields of Development
Since the research scholar is a relative and a descendant of the politician, one might expect the politician to express a natural sympathy for the scholar. The scholar, after all, is attempting to improve the techniques for knowing opinion. That this is not case is a matter of common knowledge. In fact, the relation of the politician of public opinion is, in itself, one of the six major areas of research I should like to say something about this evening. However, before discussing that area, I should first mention the other five areas to be treated. Following upon the subject of the relation of politicians to opinion, I shall introduce the problem of identifications of the voters with groups, ideas, and people. The third subject concerns how the voter perceives the world of politics. The fourth concerns his preferences among issues, the fifth is persuasibility, especially the effects campaigns have upon him. the final topic will be the level of political information and activity in America.
Relation of Politicians to Opinion
Frequently, researches into opinion are criticized by politicians in ways that are not germane to their merits. Efforts at having the census gather materials concerning elections and related social variables have been blocked, to cite one example. The politician, to my view, has at least two good reasons, apart from the ever-present reasons of economy, for what may seem sometime to be an obstructive point of view. One is that scholars are politicians and very often are interested mainly in perfecting techniques that may be used by their own party. A second reason is that the politician is somewhat of an expert on public opinion, and the creation of a new group of experts on the subject means simply another group with whom he must consult. Furthermore, he will lose control over the dissemination of knowledge regarding opinion, and although few would dare to argue so publicly, it is a moot question whether information about public affairs, in some cases, is not more dangerous than either lack of information or a questionable kind of information. Perhaps a third, not-so-good reason is the politician's feeling that the political scientist already knows more about politicians than is desirable. Would that this were the case!
We know initially (and numerous studies could be cited to demonstrate this rather commonplace observation) that leadership is a two-way street in which the public influences the politician and the politician, in turn, influences the public. But we do not know much more than that, the street is still lit by gas. Congressional hearings and administrative hearings frequently have a participant official or congressman declaring that he knows opinion in his constituency or in the nation. I do not know how he knows it, in many cases, except that he was elected. Yet one of the conclusions that one must reach from observing politics is that elections are often not a very good test of opinion. Furthermore, a man may be popular but not know the reason why.
Or again, a politician may really know opinion, but we cannot very well tell that he does. Doctor Claude E. Robinson once collected the private predictions of politicians in sixteen states on the outcome of the 1928 presidential election. Among the Democrats the average error in judging pluralities was 18 per cent. Among the Republicans the correct pluralities were misjudged by an average of 7 per cent. Some errors topped 41 per cent. Actually, most errors went in the direction of the wishes of the politician.
In two later studies, in 1944 and 1950, a total 42 county chairmen of both parties again predicted their country votes. Their average error was 6.7 per cent. If the vote had been predicted to be exactly what it was in the preceding election, the average error would have been 5.2 per cent, 1.5 per cent less than the average error of the 42 chairman.
However, a canvas of the vote intention by party workers can be highly accurate. In Elmira, New York, in 1948, the Republican party workers predicted the vote with an average city-wide error of 1.2 per cent under the actual Republican vote. The highest error in a given ward was 5.8 per cent. One might say that the Republican leaders had good intelligence.
Generally, politicians preserve a wise caution with respect to the polls. A study of senators and representatives by George F. Lewis, Jr., in 1940 with a 40 per cent response to his questionnaire, reported that 13 per cent thought that public opinion polls correctly portrayed public opinion, whereas 72 per cent believed that they did so only in part. Nine per cent said that the results of the polls helped them in deciding on the desires of their constituents; 30 per cent said they did so in part. In 1944, a poll of 52 representatives and senators by Martin Kriesberg indicated that the preferred methods of testing public opinion were, in the order stated, personal mail, visits to the public, newspapers, visits from the public, and the polls. Forty per cent of the group thought the polls were usually accurate; almost as many that they were accurate about half the time,; and barely a quarter declared they were seldom accurate. Administrators in the executive branch of government rated polls first as a method of ascertaining public opinion, and ranked personal mail and visits by the public fourth and fifth.
It is strange and true that we know very little about the influence of opinion on leaders.The nineteenth century studies of Dicey and Thompson and the recent studies of Morton Grodzins and Stephen Bailey are among the few that attempt in detail to trace the impact of public opinion on policy. The study by Gosnell, Democracy: Threshold of Freedom (Chap.8), contains perhaps the only theoretical model of the investigation of constituent-representative relations. It should be noted, regarding the barrenness of this field of knowledge, that the relation is an old problem of representative government that has been debated over centuries and in millions of words. But it is of particular interest to political scientists; and many social psychologists who have interested themselves in problems of voting behavior have neither shared that interest nor provided the research techniques.
Identification of Voters with Groups, Ideas, and People
The second important field of research, one that is much more productive today, is that of identifications. It is not unrelated to the topic just mentioned, for H.D.Lasswell carried into politics the Freudian psychoanalytic technique in order to establish connections between personal, often unconscious, motives and identifications with public objects people, power, and issues. By an identification is meant the feeling of oneness and unity that a person has with a thing, group, person, or idea. Many concepts have been constructed by social scientists to gain entrance into a theory of the public. For example, Marxism is based generally on the concept of economic determinism of human behavior; and innumerable consequences flow an investigation and extension of the concept. The concept of identification or roles, in non-Marxian. It is neutral with respect to what kind of motive is the most important determinant of human organization and conduct. It says that the most important primitive idea underlying human action is man's sense of belonging. Man is essentially defined by his roles. From infancy onwards he thinks through his identifications. Some of them are his family, his school, his locality, his factory, his army company, his political party, his church, and his country. His personality can be considered as a welter of overlapping affiliations, complementary to one another in most cases, conflicting in other cases.
The most interesting identification in the specific study of voters and elections may be the identification of a person with a political party. To what extent does a person feel at one with the symbol of the Democratic or the Republican party? In 1952, the Survey Research Center developed a seven-degree measure of the extent and kind of party identification. Depending on how they responded to the question regarding their loyalty to a specific party, or to no party at all, people were grouped as Strong Democrats, Weak Democrats, Independent Democrats, Independents, Independent Republicans, Weak Republicans, or Strong Republicans. The distribution of a national sample of Americans among these types was as follows: 22 per cent, Strong Democrats; 25 per cent, Weak Democrats; 10 per cent, Independent Democrats;5 per cent, Independents; 7 per cent Independent Republicans; 14 per cent, Weak Republicans; 13 per cent, Strong Republicans; and 4 per cent, None, of Minor Party, or Not Ascertained.
The advantages of such a division of the population became immediately apparent. The number of independents became accurately known and was surprisingly small. The variations in the intensity of party affiliation became known. Numerous cross-relationships might now be attempted. Different kinds of education, family background, and voting behavior were seen to characterize the different degrees of party affiliation. Attitudes toward straight-ticket voting, reactions to the presence of unattractive candidates on the party ticket, and many other facets of attitude and behavior were seen to depend on the degree of party identification. The inheritance of party loyalty could be subjected to a much closer and accurate scrutiny than was ever before possible.
On the one hand, the discovery and location of degrees of party identification were sure enough so that the degrees could be connected with a number of principles presently being developed in experiments in group dynamics on a highly abstract level. On the other hand, the measured concept has, to my way of thinking, rendered useless many statements based on dividing the populations into Democrats, Republicans, and Independents; that is, differences between any two of the seven categories are likely to be so great on the usual questions that are asked about people's party behavior, that lumping the two groups together in a statement is likely to render the statement false or ambiguous.
I believe that the general concept of identification will soon reorient research on voters and elections as well as researches into elites and into other political problems, and that measures of party identification will constitute hereafter a fundamental instrument for the analysis of politics. A thorough going theory of identification can avoid a rigid and static dimension of the group theory of politics that is presently being put forward by David Truman, Earl Latham, and others. For the concept can be developed to encompass identification with a personality (such as that of Eisenhower in 1952), with an idea or issue, with any of the groups to which a person belongs, and also with the nation of a common interest or a general will. It is this last type of identification, an identification with the general interest, that at the present time requires formulation more than any other. Transforming the brick-and-mortar theory of groups into an intensity-continuum-role theory of groups will, in my opinion, reconcile the instinctive doubts that we have a against regarding society as a collection of groups and the patent fact of group power in political life. Once developed, also, it will reassure the politician, and those who criticize him, that he is not crazy when he believes he is getting at something many people want, though no organized pressure signifies the want.
Perception of Voters
Perhaps the question should first be asked why perception did not precede identification. Does one not perceive and only then attach oneself to the object of one's perception? Perhaps there is something of the old problem of the chicken and the egg here, but the weight of psychological investigation at this time favors the theory that perception is dependent on interest. Descartes, several centuries ago, said "Cogito, ergo sum." I would propose a modern version of this: Cogito at sum. As man is, so he perceives.
Identification precedes and determines perception. Identification is, a source of interest. Perhaps the infant would never surmount infancy if he did not discover, through shocking experience, that there is something outside of himself and thus learn to perceive aspects of that something. After wards, he holds on to that something as part of himself if it sufficiently rewards him. His interest then sets up and frames demands. In seeking satisfaction of the demands, the person incurs a reinforced identification because he has built an ever greater investment. He has put a lot into the role, so that he must get a return on his capital.
A poll analysis by Herbert Hyman and Paul Sheatsley in 1947 indicated that people who are already interested in a general public issue know about a related specific issue and that their interest precedes their knowing, that is, precedes their perception of facts. This is nothing but the adult and politically relevant version of the infant's preoccupation with his bottle. He can pick it out of the surrounding better than he can perceive other things; I have been reminded myself, numerous times, of the presence of a bottle in the room by such a potential consumer. If it had been a bottle of spirits, our roles might have been reversed. Yet, how many times must we point out these simple facts of life to our students and our colleagues, and to the general public as well, who persist in a kind of rationalistic, logical interpretation of public opinion and public information, holding that some different set of motives must be operating on political questions, or perhaps no motives at all, but that the topics are somehow graded by divine power in order of interest, and that any fool should be able to deduce what that order of interest is.
Every study that has, in these last several years, endeavored to examine the perception of voters, concludes that perception is selective and distorted. It is distorted by the same causes that are the causes of its initial perception. That is, we perceive because we have an interest, and we distort the perception because the interest will not be controlled. To date our best material on political perception has come from the Elmira studies of Lazarsfeld, Berelson, and McPhee. They find, for example, that people tend to believe that others who belong to their groups will vote as they do. They tend to believe that people whom they dislike or wish to be dissociated from will vote for the other party or candidate. A prejudiced Protestant voter in Elmira was more likely to assign Negroes, Catholics, or Jews to a party other than his own. Also, people of one party sincerely perceive many groups as beings as being more of their party than the groups actually are. And finally, they sincerely believe that their party has a better future than the election day results may indicate.
The voters also tend to perceive party differences to be greater than they actually are, as measured by the beliefs of other voters or the majority of the voters of the party or by the views of the candidate or even the party platform. In Elmira in 1948, only about one third of the voters accurately perceived the position of the two presidential candidates on the four outstanding issues of price control, the Taft-Hartley law, public housing, and policy towards the U.S.S.R. Some distortions of perception were striking. Among Republicans who favored Dewey but who also favored price controls, 70 per cent thought Dewey favored price controls. Among Democrats who favored Truman but were for the Taft-Hartley Act, 40 per cent thought Truman was for Taft-Hartley Act, but among Democrats favoring Truman who were against the Taft-Harley law, only 10 per cent thought Truman was for the Taft-Hartley law.
When asked the position of a candidate on an issue, there were a certain number of "don't-know"; the people who themselves took a different position from their own candidate on the issue "didn't know" that their candidate had a position on the issue more often than others "didn't know." Cross-pressures in this case seemed to produce non-perception, reminding us of a new meaning to the old saying the "what you don't know won't hurt you." This last finding is related to a study by Kriesberg that showed apathy to be the result sometimes not of ignorance but of conflict from two opposing pressures. The apathetic voter should be watched therefore. What seems to be apathy may be internal turmoil, and a sudden unlocking of the conflicts by a new force of personality, issues, or a series of events may produce a reversal of behavior and a high level of activity.
Preferences of Issues
Every year bring many new surveys of the preferences of Americans on issues of all kinds. Obviously, one would be foolish to undertake a summary of all that has ever been discovered about the attitudes of Americans on issues, for the huge work of Strunk and Cantril listing the bare findings of several large polling organizations between the 1930's and 1945 contains 1191 pages. Recent research in the field, however, seems to me to raise several problems. One concerns the point at which an issue exists; a second concerns the clarity with which an issue is present in the public mind; a third, the range of issues that may be before the public at any time; and a fourth, the consistency that various clusters of issues possess in the individual mind. Finally, something should be said about the force of issues on attitudes, as opposed to the force of the personality of candidates and of the affiliations with the party organization.
The ideal kind of data with which to begin an analysis of issues would be a record of all the political impressions and thoughts crossing the minds of a sample of American voters over a period of time. Then one could begin to classify and interrelate them, ultimately ending with accurate tallies of the numbers, quality, and intensity of issues. Unfortunately, nothing approaches this kind of data. The most valuable collection of information on issues is contained in the materials of the Survey Research Center on the 1952 election. The responses of the voters to very general questions about each candidate and each party are recorded in detail. The Elmira study asked voters what they though were the most important issues of the 1948 campaign, and the study of the Survey Research Center in 1952 also asked the voters about seven specific issues.
One thing seems to me to be fairly sure; unless what is worrying people generally in an election campaign is known, how attitudes on issues will affect the vote cannot well be known. But free-association answers, especially when working with large numbers of voters, create an enormous problem of analysis, not to mention interviewing and recording. Yet, whenever certain issues are chosen to ask the voters about, a great initial bias is surely introduced into any subsequent judgment on how issues have affected the election. A smooth way of coping with this dilemma has not been found. We have no film with which we can easily photograph the condition of issues.
I believe there is little point in disputing, from what is known of issues and the public, that issues range greatly in their degree of specificity. Many Democrats are still loyally voting against what they call the Hoover depression. Others are voting on a specific issue of FEPC legislation or the Dixon-Yates contract with Atomic Energy Commission. No shepherd can drive them into the logical analysis of a set of uniformly specific issues, on the basis of which they will subsequently vote.
To the wide range in degree of specificity must be added a wide range of substance. Whereas one voter is influenced by a defeat in Korea, another is moved by the prices of his groceries. Many dozens of such issues were exposed in the responses to the open-ended questions of the survey of 1952 by the Survey Research Center.
So when the pattern that the manifold of events impresses on the public mind is disclosed, it shows an enormous diffuseness. How do the scholar and the politician move into this field? Probably the first job is to segregate those people who are unmovable by any technique of campaigning. (Of course, this re-defines the issues as any perception and attitude that is alterable by the forces of the campaign.) That means isolating the traditional voters. It also means isolating the personality-oriented voters, those who are deciding their vote almost entirely on the attractiveness of the candidates. That these two groups can be segregated from the issue-oriented voters is, I believe, shown by the work of Campbell, Gurin, and Miller of the Survey Research Center. The measures they used and those I used in the study of the western public would seem to indicate that about a third of the electorate builds its vote primarily out of concern for issues.
Now given this third of the vote, and the great variations in generality and substance among issues, the problem of saying what issues are the most important is eased, for at least its boundaries are known. Within the area, one then looks for issue clusters--forms or shapes or patterns of opinion that hang together and move in a certain way. The Korean issue in 1952 was the nucleus of such a cluster. Many other issues hung on it and were shaped to fit it. I suppose the astute politician will grope for these nuclear issues that affect a cluster and develop the vote-getting potential of the cluster.
There may be only one issue cluster or several in a campaign. For some years, from 1930 until 1947, a general issue cluster seemed to be forming that divided the two parties. Correlations began to show up between economic status and voting behavior. Many said class politics had arrived in America. It was heralded too soon, I believe, or it may already have passed away. Conventional socialist theory had its day in the data of the thirties. The puzzling lack of association among traditionally associated variables has, I think, an explanation in the rise of foreign policy to paramountcy among issues. Today the domestic issue clusters are not associated with the foreign issue clusters and some traditional correlations are no longer strong. I see no end to this. Future, consistent issue clusters, containing both national and international issues, will be based more on ideological-psychological grounds. My findings about a cluster of negativistic and pessimistic attitudes among some Republicans and a cluster of positivistic and optimistic attitudes among some Democrats in the 1952 elections are a step toward abstracting what may be the new developments in American opinion.
Persuasibility of Voters
A fifth area of research into voting behavior concerns the extent to which people may be moved by politicians. This is part of the general problem of persuasibility. Certain obvious limits are set to the problem by the preoccupation of most research with political campaigns, so the question becomes: What is the effect of campaigning? Perhaps this is too limited a perspective for political science, though it is certainly broad enough for this lecture. For research in persuasion occupies a good many hundreds of professors and many thousands of politicians, businessmen, and husbands and wives. The limits of persuasibility are, you might say, the limits of orderly change. Suppose, for example, once committed to public power, people cannot be changed into private power advocates; then our political system has to be pictured as not fully self-determining. Also, should it be shown that only a negligible change can be made in attitudes in a short time, then those who desire to change them must resort to methods that may have little to do with our frequent elections. No one has yet attempted systematically to develop this general problem and its philosophical implications. There have been a number of polls on the same questions over periods of time, but not even this superficial data has been analyzed. Perhaps Hadley Cantril's study of changes in American opinion with reference to World War II is grist for this mill. A recent pamphlet of the Princeton Center for Research on World Political Institutions, on changes in American attitudes toward an international police force, is also pertinent. Some experiments in changing attitudes by propaganda, especially those of Carl Hovland, also deserve attention.
However, the best analyses are concerned with changes during a political campaign. Only several elections and settings have been studied in this context. Probably the most remarkable are the 1948 Elmira study and the 1952 Survey Research Center study. The Elmira survey used four successive interviews of the same people (or most of them) from June to November.
A few general conclusions regrading this complicated problem can be offered. One is that most people who ultimately vote have their minds made up before a presidential campaign beings. However, those who make up their minds after the nominating convention, under ordinary circumstances, number enough votes to win an election, despite the "premature" deciders. This is especially true if there is a heavy turnout on election day, because the nonvoters are more neutral and nonpartisan in attitude than the rest, on the whole. In 1948, Truman probably won by the votes decided in the last two weeks of the campaign. In 1952, Eisenhower won with the votes he had when nominated. Since there can be no control over such an election study, it cannot be said what kinds of issues or tactics might have changed enough minds for Stevenson to win. He was boxed in rather badly.
Moderate or weak party supporters are more likely to change during the campaign than strong partisans. Hence, perhaps nonpartisan appeals and personalities count especially strongly toward the end of a campaign, party loyalty in the beginning. Also, as the late deciders seem to be more "Democratic" by sentiment, and less interested and instructed on issues, perhaps the vote-getting campaign should end on a mild "liberal" note that eschews specific discussion of issues.
Level of Political Information and Activity
Information follows interest. Few people are interested in most issues. Hence, information is at an absolutely low level on political issues. This holds true especially for between-election times, but also during the campaigns. Hyman and Sheatsley, for example, once found that only 12 per cent of a national sample could identify correctly five major even tissues. Consumption of information rises during a campaign, but perhaps a third of the population does not feel any interest in even a presidential election (many who vote, do so quite mechanically).
Since information follows interest, and activity does as well, information level is highly related to activity level. In absolute terms, the political activity level of the population is low. No more than 10 per cent of the adult population exercise the minimum activity of voting regularly, trying to persuade other People of their political opinions, and taking a tiny, sporadic, and passive part in a politically oriented group or in a political campaign.
Only 2 per cent to 3 per cent. or one out of 35 to 50, adult Americans is a modest activist or participating in a fairly steady way in politics. The total number is perhaps two and a half millions. Politics, in what may be the most free political system in the world, is the work of a few people. There are about as active citizens as they are active criminals in the United States. (Lest this comparison seem depressing one can console himself with the knowledge that in most historical societies, the proportion of criminals has been much larger.) The rareness of political activity, it needs scarcely be pointed out, is a contradiction of the Jeffersonian ideal of general public participation. Indeed, considering the small number of the politists, it is a wonder that the ideal flourishes. But it does, and probably it also increases the number of politists beyond what could be expected if Americans did not continually exhort one another to political participation.
Whether the proportion of politists has increased or diminished over the last generation or century in America cannot be stated now, because of the lack of research on the subject. It has taken us a long time /9for no good reason) to discover even a few essential facts about the quantity of political activity.
Barring the unlikely future discovery that the quantity of politists has double or halved, the problem of changes in quality may be of greater importance to the future of the American system of government. There can be little doubt that important changes in quality or composition have come about. A number of activities that are surrogates to general political activity have developed. More people are active in voluntary associations, pressure groups, and labor unions. More people are active in community chest and other civic function. In addition, many specialized functions--regarded as social, technical, or economic mow that the "cure" of politics has been removed--have grown up. Noteworthy are social service work and career administration. The "depolitizing" of these tasks has eliminated from the ranks of the politists many who formerly constituted the bulk of general political workers. They have left politics, having lost their combined material-spiritual motives for participating, or have become professionalized and "cleansed" of politics.
The privilege of being a politist is being restricted to fewer and fewer Americans. A large proportion of the population is cast out of politics by civil service and military rules, by working on government contracts, by working for large corporations that must appear "nonpolitical," and by the increased social pressure and preference for participating in civic "non-political" organizations. Security regulations prevent access of the politists to materials that formerly provided them with food for their functions. The development of "nonpartisan" elections to local offices also limits the scope of participation for a number of people. Perhaps two out of seven Americans encounter serious occupational disabilities if they have general political interests. (And I need not dwell on the absolute limitations to participation that are imposed by physical mobility, personal unsuitability, minority disabilities, and other relevant factors.)
A great specialization has occurred, and specialization must always delimit a road in order to speed the journey. In the present case, it is wreaking havoc on the fundamental doctrine of the general public.
Strange to say, the generally active public today doe4s not bear on its face the ravages just described. For whenever I make these points, I am told that the politists of today are better-informed and educated than those of yesterday. And I know this to be true. Hugh Bone's study of precinct workers in Seattle reveals a group that is occupationally and educationally "superior" to the precinct workers surveyed earlier by Gosnell, Forthal, Kurtzmann, and Mosher in Chicano, Philadelphia, and upstate New York. Ithiel Pool, Julian Woodward, and Elmo roper have also given us facts about the politists in 1948. Of the two and a half million American politists, about 40 per cent are urban males in professional, managerial, official, technical, and proprietary occupations. In these occupations, incidentally, are only 6 percent of the adult population. Among the politists, men outnumber women two to one. The average age is older than that of the adult population. More Republicans than Democrats are among them. One sixth of them are urban males who hold college degrees (only 2- per cent of the adult population falls into this category). Eighteen per cent of the politists came from the top economic level of the four levels into which the sample was divided. Sixteen per cent were housewives, most of them, of course, reinforcing the male occupational pattern. Therefore, although the politists today are unrepresentative of the whole population by such criteria--more so than the old politists--they would appear to be intellectually and occupationally suited for performing the functions of a general public.
But the appearance may be deceiving. It is may opinion that this middle level leadership group of the American political system is now more colored by psychological and ideological traits. The aggregate has lost many of its humdrum and steady members and laid open the field to the greater in fluence of those who are generally active because of an ideal or an urge. It has more outside critics and fewer people functionally integrated into the government. To the politician, who may himself be changing consciously or unconsciously in response to this new situation, the new politists are more erratic, impatient, demanding, undependable, and unassimilated to the conservative and complicated American government structure. I believe these traits occur despite (and also because of) the more apparent and considerable increase in the occupational, economic, and educational levels of the new politists.
This is one side of the development of political participation in America in mid-century. The other side is equally striking. Change in the quantity and quality of the politists occurs in relation to a variable scope of the power of government. governments are increasing their activities. No year goes by without bequeathing to us a new function of and more regulations by government. Hardly a voice can be raised against the intrinsic merits of many of such activities. The extrinsic and more subtle effects are generally ignored; these may be summed up in the observation the society is being more and more politized (or totalitarian in a strict and quantitative sense of the term). Yet we see that this increased transformation of nonpolitical subjects into political ones is being accompanied by an absolute decrease in the freedom to "politick" and in the operational posture of the politists. The conclusion that suggests itself is ominous. Traditional democracy is being slowly crushed in the gigantic pincers of depolitization and totalitarianism.
How can this come about so as not to seem utterly in congruous to those who are watching it and are committed to the old system? It comes about through the age-old human way of preventing the left hand from knowing what the right hand is doing. Each special case of a new activity or new extraction from the public is judged narrowly on this merits, rather than on its merits to more general principles. Furthermore, the massive development has marvelous semantic cloak: what is being depolitized is being taken out of politics! But power is never destroyed; it is only transformed and redivided. The poser of the old politics is being channeled into the new nonpolitics that grows apace.
The foregoing essay was contained in RESEARCH FRONTIERS
IN POLITICS AND GOVERNMENT, a series of lectures conducted by and published by the Brookings Institution of Washington D.C. in 1955.
Individual authors of the various essays were:
Stephen K. Bailey
Herbert A. Simon
Robert A. Dahl
Richard C. Snyder
Alfred de Grazia
Paul T. David
David B. Truman