The Significance of Psychology For the Study of Government and Certain Specific Problems Involving Psychology and Politics (Report from the Round Table on Politics and Psychology at the Second National Conference on the Science of Politics, Chicago, September 8 - 12, 1924)
Louis L. Thurstone
1.Purposes of the Round Table. The round table on politics and psychology formulated its objectives in line with those of the conference, which were to discover the technique and the methods by which the study of politics may become a science. Professor Merriam distinguishes between political prudence and political science. By political prudence he means the gathering of opinion, expert opinion if possible, to serve as the basis for intelligent action. By political science, as science, he means the statistical and quantitative analysis of the phenomena of politics so that political action may be based as far as possible on scientific evidence rather than upon personal opinion and judgment.
The study of politics takes its place with economics, sociology, education, psychology, and other social sciences which are in various stages of advancement in scientific technique and method. In this type of
( 111) comparison the study of politics is less developed than economics, for example. Education has established itself as a social science with a vast array of technique and method for quantitative analysis. Sociology is one of the less developed of the social sciences in that it has not yet developed much in the way of technique for scrutinizing the causal relations in the field of its subject-matter. Among the social sciences psychology occupies a middle place as judged by its available technique for scientific inquiry.
The different branches of psychology are not equally well developed. Thus, the experimental psychology of the sense organs has a highly developed technique for exact measurement comparable with the technique of the biological sciences, whereas social psychology is the least developed of the psychological branches in this regard. Unfortunately, considering the development of method, it is largely from social psychology that a contribution may be made to the study of politics and since this branch of psychology is one of its youngest, the contribution must necessarily be meager. On the other hand, psychology has developed, perhaps farther than the other social sciences, the statistical and biometric methods for the study of groups. Since the biometric methods are fundamentally the same in their various applications to biological and social data the psychologist can make this type of contribution to the study of politics. His contribution will be partly in his capacity as a biometrician and partly in his capacity as a psychologist. The psychologist can assist in the formulation of social psychological problems of interest to the political scientist, and the biometrician can assist in the formulation of the statistical and experimental controls by which scientific inquiry may be successfully pursued.
2 What Constitutes a Problem in Social Science? The first session was devoted to a discussion of what constitutes a scientific problem in the social sciences as distinguished from historical inquiry, philosophical speculation, and the expressions of personal opinion and judgment. This distinction is easily made in the exact sciences, but it is frequently confused in the social sciences where the distinction is not often sharply drawn. Every scientific problem is the search for the relation between two or more variables. Before the problem can even be stated or clearly comprehended, each of the variables must be stated separately. Each variable must be so described that it is clear what is meant by more of it and less of it. If this more-and-less aspect of the variable can not be clearly stated, some other basis must be used for dividing the variable into discrete categories each of which is defined. Thus public opinion
( 112) is not a variable because there is no possibility of defining what is meant by much public opinion or by a little public opinion. However, the number of votes cast for or against a proposal is a variable because it has a clear quantitative aspect. Occasionally a factor such as race does not lend itself readily to treatment as a scale with high and low values, but such factors must then be classified into discrete groups so as to make quantitative analysis possible.
Every scientific problem can be stated in the form of the question "What is the relation between A and B?" or "What is the effect of A upon B?" It is a pretty good test of any proposed problem in the social sciences to determine whether it can be phrased in this way. If it can be so stated, it is a bona fide problem. If it can not be so stated, the proposer probably needs to do some more thinking before his problem is ready for scientific inquiry. Then, the units of measurement must be explicitly stated so that another investigator will comprehend it.
When a problem has been stated as above outlined it is ready for the ingenious investigator. The task is then to invent the methods and the technique by which the relation may be established empirically. This phase of the solution of a scientific problem is one which gives unlimited opportunity to scientific ingenuity. The subsequent analysis of the paired observations of the two variables in the problem should be guided by the statistical and biometric methods which constitute essentially a system of logic for the evaluation of mass data.
The several stages in the solution of a scientific problem can be summarized as follows:
1. A felt social need which requires analysis, satisfaction or cure.
2. The phrasing of the need, or perhaps a small part of it, in the form "What is the effect of A upon B?"
3. The definition of the variables A and B, preferably in quantitative terms.
4. The adoption of a unit of measurement for each variable.
5. The experimental arrangement by which paired observations may be obtained for A and B.
6. The statistical analysis of these observations to determine, objectively, the degree of the relation and the nature of the relation between A and B.
7. The interpretation which consists in reading causality into the observed concomitance of the two variables.
8. The formulation of more problems which arise from doubts in the interpretation and from which the cycle repeats itself.
3. Pre-scientific Studies. Before quantitative work can be begun in the social sciences it is perhaps characteristic to find a period of speculation and historical inquiry relating to the subject. Out of much study come the hypotheses that can, at a later time, be subjected to quantitative scrutiny. It is frequently possible to discuss these hypotheses gleaned from historical and other informal evidence without seeing the immediate possibility of experimentally verifying them. Such theories, doctrines, and hypotheses break up gradually into groups of more specific questions that may be studied with the quantitative controls of scientific method. This will probably be the case with such questions as the influence of Nordic or Mediterranean nationalities on American civilization. The question can not itself, and in such a form, be subjected to scientific study, but it will in time break up into many specific questions that do lend themselves to rather well-controlled scientific study. For this reason we must be content with the realization that much of the important subject-matter of a social science will necessarily he speculative during the earlier phases of its development, and we must make the attempt so to phrase our speculation that it looks ultimately toward specific and quantitative verification.
4. Proposed Psychological Problems in the Science of Politics. In his essays on Politics and Psychology, Professor W. H. R. Rivers has suggested a number of problems some of which could probably be attacked with profit even at the present time. One of his suggestions concerns the relative effectiveness of committees for advisory and for executive functions. His guess is that the committee organized for advisory functions is serviceable, whereas the committee which is organized for an executive function is of questionable value. A case study of pooled experience on this problem might yield empirical data of considerable value. The presence of a single dominating figure in a committee and the presence of indifferent committee members and their effect on the committee's work are further problems on which a profitable case study might be made.
Professor Rivers has ventured the assertion that the successful leader is one who makes his appeal primarily to the emotions and not to intelligence. It would seem possible to verify or refute such an hypothesis by the study of many successful and unsuccessful leaders and the relative degree to which they depended on emotional and intellectual appeal for their support.
Another of his many suggestions is contained in the following quotation (page 72) : "I have suggested that the social counterpart of the
( 114) nightmare is the revolution; and if the effects natural to the experience of social wrongs are not allowed to find expression in such a way as will lead to the recognition of the wrongs and to the measures which follow upon this recognition, there will sooner or later be violent and unregulated all-or-none manifestations comparable with those of the nightmare". This is a problem in which the historian, political scientist, and the psychologist could cooperate with profit. The study should be made by the case method and the analysis of each case should follow informally a common plan or schedule. It was Professor Rivers' thought that much legislation which is aimed to cure a social wrong accomplishes little more than the palliatory measures of a physician who relieves the symptom without diagnosing its deeper cause. Just as the mental hygiene movement is now giving attention to preventive measures by which mental evil may be anticipated and avoided, so the political scientist should attempt so to diagnose political evil that legislation may be preventive rather than curative in character. This is an informal type of inquiry out of which will come ultimately more specific questions that can be solved by quantitative empirical methods.
5. Methods for Investigating Changes of Opinion as Expressed by the Ballot. The projects discussed in the conference divided themselves into two large classes, namely, those which concern the change of opinion and those which concern leadership. The opinions which are recorded in a ballot are probably affected by a number of factors some of which may be experimentally analyzed. The following problems were discussed by the conference:
The effect of reading persuasive material on the ballot. In this type of experiment a ballot would be taken on some question such as prohibition, war, or any issue on which public opinion varies. The subjects in the experiment would then be asked to read persuasive material for the issue, other subjects would be asked to read persuasive material against the issue. The effect of the reading would be ascertained by a second ballot.
The effect on the voter of announcing the majority opinion. In this experiment a ballot would be taken on any suitable issue after which the vote of the majority in the group is announced. A second ballot is then taken in order to determine this effect.
The effect of announcing the opinion of alleged experts on the ballot. This is again similar to the previous experiment except that between the first and second ballot is inserted an announcement of the opinions of experts to determine the effect on the balloting.
The relative effect of oral and written presentation of persuasive material. This type of experiment can be arranged according to the same general plan as the previous ones but it is more difficult to control so that the conditions can be reproduced by other experimenters.
Such experiments can be varied by introducing material which is neutral as far as the subjects in the experiment are concerned, but which has in reality a calculated effect on the subjects. Thus in an experiment of the first type the subjects were asked to ballot on their attitude toward war which was phrased in a number of statements. The reading material introduced between the first and second ballot related to children and the readers probably assumed this material to be neutral as far as the peace and war proposals were concerned. It was found experimentally, however, that the effect was to increase the pacifistic opinions as shown on the second ballot. We have here in miniature and in experimental form the effect of propaganda in which opinion is influenced by presentations that are thought of as neutral by the readers hut which are intended to have a definite effect on their opinions. This type of experimentation is very suggestive for further work of a similar kind.
The experimental measurement of nationality, race, sex, schooling, age, economic status, occupation, religion, and other factors as determinants of opinion. Such experiments would have for their object the determination of group differences in opinion on stated issues.
The effect of the factors listed above and others on the willingness to change opinion on public issues. This type of experimentation could be carried out by submitting to people a list of proposals for changes in the established order with or without some reasons for the proposed changes. It would be illuminating to study the group differences in willingness to consider the feasibility of changes in the established order. It would probably be found that some of the groups as defined above would be so set in their opinions that they would not even consider the proposals as even possible while other groups would be more open-minded in expressing their willingness to consider revisions in the customary social habits and traditions.
A parallel experimental study could be made of the group differences in the effect of different types of material on changes in opinion. Such studies might reveal that certain groups are more easily persuaded to change their opinions than other groups. In order that such experimentation may be at all trustworthy it would be necessary sooner or later to carry out the experiments on a sufficient range of issues to establish general principles.
Experiments could be carried out by asking the subjects to assign their reasons for voting as they do on the first ballot. The persuasive reading material would then be presented. On the second ballot there would be changes of opinion by some of the subjects. It would then be possible to determine empirically whether those who assign certain typical reasons are more or less easily persuaded to modify them than those who give other types of reason. If such findings should reveal any general principles they would be useful in predicting the change ability of opinion in terms of the reasons assigned for them.
The ballots should be arranged so that the voter expresses not only what his opinion is but also in some manner the strength of his conviction in voting. If the subject is prohibition the voters might check various statements that they endorse. These statements should include extreme temperance statements, statements favoring a moderate government control, and statements endorsing complete license. The relative strength of the conviction of the voter could then be ascertained in his ballot and these relative degrees of conviction could be studied with reference to the variables above mentioned.
A variation of these experiments would be to ascertain the effect of secret and open ballots on the total vote. The experiment might be arranged either in the form of a "yes" and "no" ballot or in the form of graded statements which the subjects endorse or refuse. Still another factor is the comprehension of the reading material. Its effect on the final ballot might be ascertained by comparing the graded ballot with the degree of comprehension of the reading matter, which should of course be determined by objective tests.
An important qualitative study would be the psychological analysis of the personalities that react negatively to the suggestions in the reading material. It was found in the preliminary experiments of Mr. Sturgis that some of the subjects modified their voting in a direction opposite to that of the reading material which was inserted between the first and second ballots. If negative suggestibility can be experimentally verified as a characteristic of some individuals it should be an important consideration in any problem affecting the change of personal opinion. It would be rather easy to determine experimentally the relation between the intelligence of the subjects and their changeability in voting on several issues with two ballots and persuasive material between the two ballots.
A more or less qualitative inquiry would be a study of the relative effectiveness of emotional, dogmatic, and logical appeals on the chang-
( 117) -ing of personal opinion. The effectiveness of each of these appeals may in turn be studied for each of the groups defined by sex, nationality, age, education, religion, and race. It is quite possible that these different types of appeal may be entirely different in their effectiveness for these different groups. Such studies cannot at present be reduced to an entirely quantitative form. However, enough can be done experimentally even with variables of this kind to remove the problem from pure speculation to the realm of empirical study.
6. Plan for the Study of the Distribution of Opinion in any Given Population. Professor Allport presented in outline form a plan for the study of the distribution of opinion in any given population. If prohibition be the issue chosen for experimental study the solution of the issue would be represented by a series of statements ranging from extreme abolition on the one hand to extreme individual license on the other hand. The statements numbering a dozen, more or less, would be arranged on a scale at the center of which would be a statement representing a middle ground, more or less neutral and not colored by prejudice or violence of opinion. It would then be possible to obtain a graded ballot from a given population so that the number of individuals in each population sanctioning each statement could be represented graphically. The highest ordinate on such a chart would represent the general tendency of public opinion in the given population. The shape of the distribution chart would itself be of considerable scientific interest. If the shape of that curve were studied before and after the presentation of persuasive material leaning toward one extreme, the effect of the reading could be ascertained in the modified shape of the distribution of opinion on the issue. It could then be ascertained whether it is the people in the general middle range or the people at the extremes who can be persuaded to shift their opinions by any specified amount as a result of the persuasive reading.
One of the fundamental problems in the study of distribution of public opinion would be the determination of the units of measurement that are to be used on the base line of the graphical representation. There is a logical analogy here with the steps in the educational scales that have been developed for use in measuring proficiency in handwriting and in other psychological and educational performances. It might be possible to use the principle of just noticeable differences as the unit of measurement by which the several statements that are used in the scale may be located at numerically assigned points on the scale.
7.Study of the Relative Effectiveness of Different Methods of Propaganda. Perhaps one of the most interesting of the larger psychological problems in the science of politics is the analysis of the methods that have proved successful in propaganda, including war propaganda, drives for benevolent causes, and the propaganda by which large corporations have successfully changed public opinion about them. The first step in such an inquiry would probably be to collect a considerable number of descriptions of successful and unsuccessful propaganda methods. The specific instances should be collected so that the material is historically accurate. A free and informal psychological analysis may then be made of the methods that seem to be successful and out of such a study there should come a number of psychological hypotheses regarding effective propaganda methods. Some of these principles might later he subjected to empirical verification by controlled experiments. Such a study would make interesting reading and it offers the opportunity for ingenuity in extracting the common psychological principles. Although such studies are not now quantitative there must first be some hypotheses or problems before any definite experimentation can be begun.
A slightly more specific study would be to analyze the propaganda by which public opinion is stirred up to war and also the rate at which the war hatred returns to a normal attitude between the two countries. The former problem is the more difficult although it can be studied qualitatively. The rate at which two warring countries return to a normal attitude toward each other after peace has been declared can be studied in a roughly quantitative way. For example, after peace is declared an analysis could be made of the number of articles or editorials, or the amount of editorial space, given to antagonistic and to friendly comment about the other nation. Such tabulations could be made for weekly or monthly periods. This would show at least roughly the rate at which the return to normal international attitudes is approached. If the shapes of these curves should prove to be similar for several modern wars some illuminating psychological generalizations might be drawn from them.
8.Possible Significance of the Study of Modern Methods of Publicity. It was proposed that the scientific study of propaganda might be facilitated by a consideration of the methods that are used in advertising. Since advertising has been subjected to quantitative experimental study it might be possible to transfer tentatively some of the psychological principles to the field of politics in the experimental study of propaganda.
Such a consideration of the scientific study of advertising methods might at least yield hypotheses for experimental inquiry in the field of politics. A more specific project of this kind would be the study of the methods of newspaper publicity that have been employed by successful and unsuccessful candidates for political office.
It has been suggested that the votes on constitutional amendments in some states tend to correlate with each other when presented simultaneously. If public opinion is against one of the amendments it tends to draw a negative vote on any other issue which is presented at the same time. This would seem to be a rather simple problem to analyze if the data can be provided. It would consist in obtaining the inter-correlation of the votes on several simultaneously presented amendments. If these correlations are positive and clearly above the limits of chance these indices would practically prove that pronounced public judgment on one issue may drag with it the same judgment on other issues presented at the same time even though the several issues do not necessarily have anything in common.
In an entirely different experimental setting, but involving nevertheless the same fundamental problem of conditions under which public opinion can be modified, was the suggestion that the applause in a theater be loaded at different points in the play and that the suggestiveness of the audience be studied under these conditions. It would probably be found that the audience will take the suggestion of the loaded applause more readily at some points than at others. The results might be analyzed with the purpose of discovering the psychological principles by which the points of applause might be predicted. Such principles, if experimentally verified so as to establish their universality, would be of great commercial, political, and psychological interest. These experimental findings could be compared with the practices of loading the applause and popular approval at political conventions.
9. Study of the Methods of Stimulating Voting. Of more immediate significance are the studies of methods for stimulating voting. In these studies the end result which is measurable is the number of people who turn out to register and to vote. The independent factors which partly determine this end result include such measurable facts as the giving of information about the time and place of registration and voting, the necessity for registration, information about absent-voting laws, information about ballots and the details of voting, mailing persuasive literature, having competitions between small areas or groups for larger registration. The results would be measured in terms of the registration and the votes.
An experimental study might be made by means of a survey to determine the relation between the intensity of people's opinions and the percentage who go to the polls.
It might be possible to study the effects of different sorts of news-paper publicity on elections. The election results might be studied in relation to such factors as amount of newspaper space for and against, number of papers for and against as well as their circulation, amount of logical and of emotional writing, and other factors. These studies would be complicated by the fact that newspapers sometimes support the candidate or the issue which in their opinion will win the election in order to enhance their own prestige. This effect would have to be ruled out in some experimental procedure before the conclusions could be of fundamental significance.
It was suggested that straw votes be taken on a selected group of individuals at intervals of three months between March and November in order to ascertain the changeability of public opinion in specified voting groups.
Professor Shephard has recently published an article an the relation between the two-party system and the co÷perative forms of sports. It is suggestive of the application of scientific method to social and political phenomena.
10.Survey of Classical and Modern Political Writers with Reference to Psychological Principles. One of the informal and qualitative studies of great importance in the scientific work toward a psychological under-standing of politics would be a careful survey of the classical and modern writers on politics with special reference to the psychological principles which they either imply or state explicitly. Such a study should include a wide range of writers and the effort should be made to state the psychological generalizations that are involved. By comparing the generalizations so derived many hypotheses would appear that could be subjected to experimental or historical inquiry to establish their validity. This topic may be suggested as especially suitable for a seminar in the psychology of politics.
11.Various Projects for the Psychological Study of Leadership. The subject of leadership and its psychological analysis was the second large subject considered by the round table. A sub-committee made the following recommendations regarding projects for the psychological study of leadership :
1. The analysis of political leaders of the past in order to set up a provisional list of traits that characterize them as leaders.
2. The analysis of living leaders and non leaders by means of a personality questionnaire similar to that of Professor Allport of Syracuse University.
3. The analysis of thirty living leaders of recognized prominence by means of mental and physical examinations in which should be included various psychological and physiological tests that may be regarded as having possible significance.
4. The study of the relation of leadership in the adolescent age to leadership in the adult years.
5. The study of the academic associations of men who attained leadership later in life.
6. The study of the relation between leadership in childhood and the same trait in the adolescent age.
7. Exploratory studies of an intensive order of a small group of per-sons in order to determine if possible the psychological and environmental factors that make a leader.
Professor Yoakum suggested the possibility of what would amount to a job analysis of the political leader. He suggested a possible study of precinct leaders by means of a questionnaire or schedule which would bring out their essential characteristics. Three types of procedure might be used in case studies, namely: (1) the large group method in which several thousand precinct leaders would be studied by a schedule designed to get extensive information; (2) the selection of a sample of leaders from different types of groups; (3) the intensive study of five or six local political leaders.
The many ideas that appeared during the conference will probably be suggestive of the lines along which the problems of political science, or small parts of them, may be isolated for objective study. The contribution of the conference consists mainly in pointing out further the lines of progress in which statistical methods and psychological methods may be applied with profit in the scientific study of politics. Even if only a few of these suggestions should lead to studies of a scientific order the conference will have been distinctly worth while.
12. Recommendation that Courses in Statistics and Psychology be Established. In view of the fact that the students who are now majoring in political science are preparing themselves for productive work, and since the study of statistical theory and of psychology are essential for the mastery of the working tools by which the problems of the conference may be solved, recommendations were submitted to the general conference regarding the encouragement of students of political science to study these two subjects. The recommendations were as follows:
In order to provide courses in statistics and psychology which are suitable for students of government and for students in the other social sciences, and in order to secure proper recognition for such courses, it seems wise to address a communication to three organizations which can contribute in various ways to the end in view. It is, therefore, recommended that a communication be sent to the American Political Science Association, one to the Social Science Research Council, and one to the American Psychological Association. The three formulas recommended for these three associations are as follows:
First, to the American Political Science Association,— It is recommended that suitable courses in psychology and statistics be provided and required as parts of the early and advanced training of students in government.
Second, to the Social Science Research Council,— It is recommended that suitable courses in psychology and statistics be a part of the requirement imposed upon all students who are carrying on studies in the social sciences. A consideration of the type of courses in the subjects mentioned would be a suitable topic for a special committee to be created by the Social Science Research Council. To this end it is recommended that the American Psychological Association be invited to appoint representatives to a joint committee which shall include also representatives from the various social sciences. The joint committee thus constituted shall take up in detail the content and administration of courses which may properly be recommended as fundamental to the study of all the social sciences.
Third, to the American Psychological Association,— It is recommended in view of the wide-spread demand for suitable courses in psychology and statistics to serve as the foundation for studies in the social sciences, that the American Psychological Association appoint a committee to take this matter into consideration and co÷perate with representatives of the social sciences to the end that a complete definition be arrived at of the topics and method of treatment of psychology suitable for students of the social sciences.
L. L. THURSTONE.