An Attitude on Attitude Research
The development of sociology as a natural science has been hindered by: (1) emphasis upon its normative rather than upon its descriptive aspects; (2) too much attention to subjective factors, such as ideas, ideals, motives, sentiments, wishes, and attitudes, and too little attention to objective, overt behavior; (3) the inaccuracy, indefiniteness, and anarchistic confusion of sociological concepts. A critical examination of the concept "attitude" reveals its scientific shortcomings from all three points of view. It is all things to all men; it is seldom used consistently by any one writer; it is normative, valuative, subjective; it refers to verbal responses, opinions, habits, vegetative processes, tendencies to act, impulses to act, inhibitive impulses, feelings, wishes, values, motor sets, and various combinations of these. The attempt to differentiate "attitudes" and "values" is shown to be impossible in practice. Most so-called "attitude" research is really "opinion" research. The concept is largely invalidated because of its subjective implications.
The constructive part of the analysis results in a definition of attitude as "the relatively stable overt behavior of a person which affects his status." Attitudes which are common to a group are thus social attitudes or "values" in the Thomasian sense. The attitude is the status-fixing behavior. This differentiates it from habit and vegetative processes as such, and totally ignores the hypothetical "subjective states" which have formerly been emphasized.
Investigation of attitudes thus depends upon the observation, quantification, and generalization of overt behavior. The questionary is held to be of little use for attitude research. The life-history, personal interview, and all written documents are little better, except as clues. The best sources are indirect evidences of overt behavior. The final test of an attitude is, "How do persons behave?" Statistical treatment of recorded uniformities is urged as the only valid method of scientific generalization. Examples of this kind of attitude research are cited.
I. IS SOCIOLOGY A SCIENCE?
Sociology is notorious among the sciences for the inaccuracy, indefiniteness, and general confusion of its terms. Most physical scientists regard the "science of sociology" as a pious pretension and "sociologists" as scientific charlatans. Some sociologists admit the allegation and modestly hope that their discipline may some day become a real natural science; others proclaim that sociology is a natural science-it "uses scientific method, and a science is known by its method, not by its subject matter or its results"; still others erect an elaborate defense mechanism, pleading the infancy
( 940) of their science, the complexity and variability of its data, the difficulty of sociological experiment, and so on.
It is safe to say that the sociologist as reformer, uplifter, and prophet of millennial dawn exists only in the popular mind. Among sociologists, however, the ideas of progress, improvement, solution of social problems, good and bad conduct, and the whole field of social ethics plays an ever decreasing part. They are more than anxious to leave these things to social workers, teachers, preachers, publicists, and politicians. Modern sociologists are more and more concerned with finding out how human beings actually behave in groups. They are after objective, quantitative, conceptualized descriptions of group phenomena. This constitutes the only true scientific explanation, and when it is given, prevision and control are possible. If they are successful in this, sociologists are men of science.
Any candid man must admit the truth of many of the charges against sociology. Probably no other scientific discipline spends so much time, energy, and money with so little solid achievement to show for it. A large part of the so-called "research" in sociology results merely in the confirmation of common-sense judgments reached by empirical observation and intuition thousands of years ago, or else in the pretentious elaboration of some subjective preconception. In the opinion of the writer, this largely is due to three things:
1. The conception that sociology is a normative instead of a descriptive science. Too many of us justify the popular notion that we are trying to "help society."
2. The failure to regard our data as "natural." Too much attention is given to consciousness, ideas, ideals, motive, wishes, attitudes, mental and emotional "states," and other subjective phenomena. Sociology will not come of age until we learn to study man in association just as other scientists study atoms, stars, and bacteria, i.e., as energy-units in motion. We must have a frankly behavioristic sociology if we are to have a science. As long as we concern ourselves with man's hypothetical feelings, thoughts, and subjective states instead of with his actual, overt, observable, and
( 942) measurable behavior, we shall not have a science. Perhaps human ecology and behavioristic psychology are pointing the way.
3. The inaccuracy, indefiniteness, and anarchistic confusion of our concepts. This condition is largely the result of the subjectivistic bias just noted. We cannot do valid scientific research until we know what we are talking about. At present all too many sociological terms mean all things to all men, both lay and professional. We have few agreed-upon units, few universal standards of measurement, practically no constants, no universally accepted conclusions, no very accurate prevision, and hence, no science.
To summarize, we have not advanced social knowledge very far beyond the limits of common sense. We still have to deal in "trends," "tendencies," and "general movements." The average politician or business man, by rule of thumb, common sense, "knowledge of human nature," and other non-scientific experiential methods can often outguess the scientific sociologist as to probable group behavior. Our predictions are seldom scientific prevision. They are merely inspired, intuitive guesses, or hopes, or fears, which sometimes come true in the good providence of a merciful God. We cannot tell the seeker with any high degree of probability what will actually happen. If we deal in knowledge, it is mostly on the level of common sense and not of science.
II. WHAT IS AN ATTITUDE?
"Attitude" is a term which has recently come into very general use among sociologists, social psychologists, and writers on education. It is a good example of an ill-defined, or undefined, concept used in a loose, pseudo-scientific manner. The result is a confusion, many times confounded.
A recently published text uses the word in almost every chapter, but the author never tells us what an "attitude" is, or even what
( 943) he means by it. This might be overlooked were the usage merely literary or conventional, but when research articles on "Attitudes" begin to appear, and the word is given technical connotation, it becomes necessary to examine it critically.
The senses in which attitude is used are almost as numerous as the writers. In most cases the meaning is implicit, rather than explicit. P. M. Symonds states that educators use it to refer to the desirable outcome of education. Some sociologists use it in much the same way. It is obvious that so long as "social attitude" means "good" or desirable behavior it can have Do value as a scientific, i.e., a non-normative, objective, descriptive term.
"Attitude" is often used as a synonym for habit. This is usually complicated by references to some hypothetical instinctive, mental, emotional, or feeling concomitant, latent, inhibited, or active in the response. When subjective factors are admitted into the concept of attitude, it means all things to all men---except to scientific men, to whom it means nothing. Regarded as habit it means everything that human beings do, and hence means nothing definite.
Another common usage somewhat like the foregoing indentifies
( 944) attitude with motor or mental set. The motor-set idea is more in accordance with the view of this paper because it minimizes the subjective element and deduces attitude from performance.
Perhaps the most common, and, to the writer, the most indefensible, use of the term identifies it with opinion as revealed by verbal responses. Practically all of the research on "attitudes" takes this point of view.
P. M. Symonds, in the paper referred to, mentions seven ways in which the term attitude is used, viz., (1) great organic drives (motives), (2) muscular set, (3) generalized conduct, (4) neural set. or readiness to adjust, (5) emotional concomitant of action, (6) feeling concomitant of action, (7) accepting or rejecting verbal responses. He ends by deploring the use of the term, saying that we should use habits or skills. But in at least one article he joins the group who use accepting and rejecting verbal responses as evidence of attitudes.
The objections to identifying verbal reactions or opinions with attitudes are apparent. People tend to respond in conven-
( 945) -tional ways when they have to sign their names, and probably when they do not. Most "thinking" is rationalization, or stereotyped reaction. Questions are interpreted differently. "Tabooed questions" are not the same for all persons. Verbation often refutes action, and vice versa. The fact that there is a high correlation between successive verbal responses to the same questions does not alter the situation. The real test is whether the verbal responses correlate with the overt behavior of the subjects. Measures of this relation are very scarce in the literature.
W. I. Thomas was one of the first social psychologists to try to define attitude in a non-normative, objective manner so that it could be used as a tool for scientific research. Attitudes did not play a very large part in the literature of sociology until after the appearance of the monumental monograph on The Polish Peasant, although educational and other psychologists had been using
( 946) the term in a loose and uncritical way for some years. Thomas' view can be summed up in the statement (mine), "An attitude is the subjective reaction to a value." He differentiates social psychology from sociology by assigning to the first the study of the subjective side of culture, attitudes, and to the latter, the study of social uniformities, or values. They have this in common, viz., values draw their reality from attitudes, and hence sociology is subordinate to social psychology. The result of this analysis leads me to conclude that both sociology and social psychology are dependent upon subjective data in the form of individual attitudes, sentiments, feelings, impulses, wishes, and tendencies to action. It is further stated that the same attitude may result in any number of different kinds of action or in none at all, being sometimes a mere tendency to act. Or the same action of ten people may be due to ten different attitudes.
To the writer, such statements invalidate the concept of attitude as a tool for scientific research. They hold that a particular act is no proof of a particular attitude, and an attitude "felt," or verbally expressed, may result in a great variety of actions or in none at all. Now it is a generally accepted axiom of science that a scientific fact must be the same to all competent observers. This means that only phenomena observable by the senses can come within the purview of science. Hence, to assert that an action may be indicative of several attitudes removes attitude from the category of science and definitely places it in the realm of imagination and speculation. All that science can deal with are acts, behavior. Therefore, since the only way we can judge of the existence of a subjective "state" is to observe action, to say the action betrays this or that subjective state is a mere tautology, so far as science is concerned. The scientific fact must deal with a generalization of sense-observable phenomena. When these phenomena fall into relatively stable and repetitive patterns, it is possible to deduce a sci-
( 947) -entific fact from them, and not otherwise. Science cannot. deal with hypothetical subjective states, but only with sense data which are the same to all competent men using the tools of scientific observation and measurement: instruments, logic, and mathematics.
Thomas implicitly admits this deficiency in the concept of attitude when he says that the bond between attitude and value is activity, in the now widely known principle, "The cause of a value or an attitude is never an attitude or a value alone, but always a combination of an attitude and a value," and in his often indistinguishable usage of the two terms in the latter pages of the "Methodological Note" and throughout the Introduction. What can an "abnormal attitude" mean, other than a way of acting that is contrary to a "value"? If a number of people act in this manner, a so-called "abnormal attitude" is a "value" to them. Hence, in addition to the ethical implications of the term "abnormal attitude," an attitude is, or may be, a value (p. 80). I find attitude used with such apparent meanings as a tendency to act, desire to act, possibility of activity (p. 80), verbal disapproval (pp. 103-7), "Attitude of the group" (value) (pp. 116-32), ideals, standards (p. 132), opinions of, treatment of, understanding of (p. 138), judgment (p. 187), practice or act (p. 198), ritualized acts (p. 202). In most of these cases, opinion or value could be substituted for attitude without changing the meaning one iota.
Hence, it appears to the writer that the elaborate attempt to differentiate attitude from value ends in a tautology, a mere verbalism. It is immaterial which term we use, since both seem to refer to the same phenomena a good part of the time. The attempt to differentiate them seems to be based upon the old antithesis between individual and group, while the interchangeable usage results from an implicit recognition of the fundamental organic unity of individual and group.
It should be remarked in passing that most of the research on attitudes in The Polish Peasant is based upon verbal reactions as recorded in letters, diaries, and life-histories, and so identifies attitudes with opinions, judgments, ideals, rationalizations, and the like. The deduction of attitudes thus depends upon what the investigators thought the subjective states of writers were. If it be argued that the attitudes were deduced from the actions of the writers, we must reply that the investigators had no data other than what the writers said they had done. However, in most cases the attitudes are deduced from what the writers say they thought, felt, or believed. There is no attempt to quantify the opinions of the subjects, which would be useless anyway from the Thomasian point of view, since the same act or opinion may be clue to any number of attitudes. The investigator must deduce attitudes by an exercise of his imagination, or interpretative genius, upon the inaccurate, half-forgotten, rationalized, distorted elaborations of the subject's verbal responses. It is not denied that such a method, used by a man of broad experience and sympathetic insight, may throw considerable light upon the behavior of the subject. He can uncover many lies, rationalizations, and half-truths by seeking other sources of information and by critically examining the inner contradictions of the subject's report. This is essentially the method of the case worker and the psychoanalyst. However, the usual "attitude research" would not pass muster with a modern trained case worker, and is certainly a poor substitute for objective scientific research.
Thomas' attitude toward attitudes has been somewhat modified and elaborated by his followers, although none of them have
( 949) clearly broken with his subjectivistic point of view. None of them have dispensed with the "tendency and impulse to act." Commonly, they have conceived of attitudes as combinations of wishes, impulses, sentiments, and other hypothetical mental, emotional, and un-get-at-able "entities." For example, Park and Burgess say, "The wishes enter into attitudes as components," and then state that the wish for recognition may be expressed as boasting or humility. The wish is the same, but the attitudes are different. It would seem to follow that attitudes also enter into the wish as components. The wish gives rise to various attitudes, and yet an attitude is made up of wishes. Thus it becomes impossible to tell what is wish and what is attitude, as it is impossible to differentiate the attitude of a group from a value.
While Thomas mentioned it, other writers have placed more emphasis upon action, or overt behavior, as the mark of an attitude, and they have clearly differentiated attitude from opinion, which Thomas fails to do. But most all of the writers still speak of tendencies to act, emotions, sentiments, wishes, desires, and other subjective states as constituent parts of attitudes.
Enough has been said to show the confused, contradictory, and unscientific usage of the term "attitude." The principal aspects of this are the confusion between opinions and attitudes, the impossible attempt to separate attitudes from values, and the identification of attitudes with hypothetical subjective states of consciousness. Finally, it has been pointed out that some writers have elaborated Thomas' suggestion that attitudes are primarily concerned with ac-
( 950) -tivities. This leads to another confusion, however, in making attitudes equivalent to habits.
But this emphasis on action gives us our cue for a redefinition of attitude as a concept for sociological research. If sociology is to become a natural science, it must use the methods of natural science. This means that the whole traditional clutter of conscious states and subjective concepts must be thrown overboard; or, if retained, must be redefined in terms of movement. "Subjective" must mean merely the movements of a subject and carry none of the present implications of so-called mental, conscious, unconscious, and emotional states. Feelings, sentiments, tendencies and impulses to act, wishes, attitudes, and so on, mean nothing, and worse than nothing, unless they are interpreted as overt behavior of some kind. Every man of science will agree with Lotka that "We must wait till the thing is present to our senses before we name it." In other words, we cannot speak of the existence of attitudes or wishes or sentiments or any other phenomena of consciousness except as they are manifested in overt behavior.
Furthermore, if attitude is to be a sociological term, it must refer to the overt behavior of human beings in relation to other human beings. This will prevent the use of attitude to describe vegetative processes and individual habits, if there are any, which have no social significance. As Weiss has well said, "The study of human behavior is concerned primarily with those movements made by man which establish the social status of the individual in some . . . . social group." The attitude of a man, conceived as a status-getting action, implies a certain stereotyped and repetitive type of behavior. If this relative permanence is not present, of course, no scientific study is possible.
So we may say, "An attitude is the relatively stable overt behavior of a person which affects his status." The definition of Park and Burgess would be consistent with this point of view if the "tendency" aspect were left out, if the attitude were not regarded as composed of wishes, and if the affective, or sentiment, element
( 951) were not insisted upon. Their definition would then read, "An attitude is the positive or negative reaction of a person to a total situation." However, this neglects the social-value aspect of attitude which Thomas has so rightly insisted upon. To define attitude as a mere behavior pattern would make it equivalent to habit and vegetative processes. There is a habitual element in attitude as I have defined it, but it is social habit, value habit, status-fixing habit. Putting on my left shoe first is habit, and behavior pattern, but it is not an attitude. Wearing shoes to banquets, however, is an attitude, because that affects my status. It is immaterial to the scientist what so-called "subjective" motives or wishes or desires induce people to wear shoes to banquets. It may be pride in small feet, to please wives, to keep feet dry, or what not. The scientist is concerned only with the fact that they do it. For him, whatever the subjective states may be, attitudes are the same when behavior is the same. The subjective motives are no more a part of his concern than the hypothetical consciousness, desires, and wishes of an atom are concerns of the physicist. It is needless to say that a man who regularly goes barefooted to banquets also exhibits an attitude because such behavior affects his status.
I have included the word "overt" in my definition to emphasize the fact that an attitude designates the total response of a person to a total situation. Since thinking and verbation are forms of physical action, and may be important in fixing status, we might properly speak of "verbal attitudes." But in "opinion" we already have a good term for these partial responses; and since verbation does not always pass over into overt action, attitude should be reserved for the total status-fixing responses of a person. The same criticism lies against using postural movements, gestures, tones, and other partial and preparatory movements as synonymous with attitudes as Bernard would have us do.
The methodological reason for denying that attitude shall connote subjective tendencies, partial, inhibited responses, and be made up of simpler conscious states is clearly set forth by Thomas when he says "To explain this `why' we should have to know the whole past of the individual, of the society, and of the universe."  This
( 952) is precisely the case with any scientific explanation that tries to tell "why" in the sense of an ultimate and final causal explanation. That quest always ends in the dark, cold waters of metaphysics. It is enough to say, "This is the fact; the action is the attitude; this is how a person behaves at this moment under these conditions." That is all any scientific explanation can do. The generalization of the observed and measured uniformity is the scientific fact; the real causes are all the antecedent factors from the beginning of time. If the cause is demanded, we may say the quantitative generalization is the cause, as when we say gravitation causes the object to fall the way it does. But there are usually a great number of factors affecting the fall, many of them incommensurable.
Thus, an attitude is an arbitrarily chosen end result. If one says, "Well, why does a person act this way, have this attitude?" the scientist must reply that be does not know-and neither does the actor, for that matter. That is a problem for further research, and one for which we have very little scientific technique at present, although we do a lot of speculating about it, Freudians, case workers, preachers, fortune-tellers, men on the street, and the rest of us. Scientifically, we cannot go behind the data present to our senses, except to make probable inferences and logical deductions on the basis of past tested and tried sense-experience generalized into scientific facts.
III. HOW SHALL WE STUDY ATTITUDES?
The rigid behavioristic definition of attitude given above carries with it some interesting implications for research. In the first place, it would restrict the use of the questionary in this type of investigation. For those who have tried to use questionaries and for those who have been bedeviled continually with requests to fill them out, this result would doubtless be bailed with joy. Of course, the questionary can legitimately be used to determine "verbal attitudes," but everyone knows that the scientific value of these is almost nil when we are trying to study overt behavior. At least, we can never be sure that the subjects did act, do act, or will act as their replies indicate, because they lie, forget, rationalize, and mis-
( 953) understand the questions. If their verbal attitudes do correlate with their overt actions, that does not add anything to our knowledge of behavior as such.
Even if our questionary is of the so-called "concrete type," asking what you do, or did, instead of what you think, feel, or will do, such as "Do you smoke? One cigarette a day? Five? Ten?" our subject may lie; or wonder whether we mean smoke 'em close, or only a few drags; or say, "Some days I do and some clays I don't"; or, "It depends"; or, "It's none of your business." If we send these questions to one thousand people, those who are "for" and "against" it will tend to answer and those who are indifferent will not reply. What is the attitude of the group? However, if our group is representative and fills out our schedule, we may be able to determine social attitudes by the overt-action type of questionary, unless the actions we ask about are tabooed. If we inquire about tabooed behavior, the subjects will lie or rationalize, and tabooed acts are not the same for all men. By the proper kind of research we can determine many attitudes of a group without asking anyone anything. True enough, we will not discover individual attitudes by this indirect method, but, if necessary, individual behavior can also be determined in many cases without relying wholly on verbal responses. The questionary is easier than the indirect, objective method, but armchair speculation is easier still. The value of scientific research appears to be almost directly proportional to the difficulty and drudgery of doing it.
Some people, despairing of the questionary and recognizing the contradictions between overt and verbal attitudes, have advocated the case method, personal interviews, life-histories, etc., as the best way of getting at real attitudes. Many of the objections to the questionary are valid here, but we may check the subject's story and so find out how he really acted. Nevertheless, he will tend to justify his action, forget, rationalize, and distort. This method is valuable for getting clues for more objective investigation, but can have no value for scientific generalization unless we collect enough similar cases to warrant statistical treatment. Thus, we might dis-
( 926) -cover the attitude of applicants for relief, or inmates of jails, toward family desertion by proper case study, accurate records, and statistical treatment of the data. However, each case would have to be checked by objective data regarding the act of desertion, not by taking the client's word for it.
I have Mr. Lundberg's permission to cite our own investigation of attitudes as a terrible example of how not to do it. We were under no illusions regarding the shortcomings of the questionary method. We tried to obviate them by the means described in my article. But instead of finding attitudes in the sense of this paper, we found opinions, and these of very doubtful scientific value. We asked students to tell us what they thought they thought. about questions which many of them did not understand. Many of the questions were ambiguous; some of them were taboo questions; some referred to things the students had no information about and had had no experience with; and yet they were told to give a positive or negative reaction to the questions. They did it. We tabulated the results, made averages and distributions, drew "conclusions," and published these studies in reputable journals as pieces of scholarly research! I justify the title of this article in the light of my definition of attitude by the fact that I shall never make a similar investigation and brand it "attitude research."
How, then, would we investigate the religious attitudes of college students according to my proposed definition of attitude? First of all, we would mot ask thetas what they believe, think, or feel. We would ignore entirely the subjective side of their religious behavior. Their attitudes are what they do. The investigation would concern itself wholly with the religious activities of students. How can these be determined?
In the second place, if a questionary were used, it would consist wholly of queries regarding actions. Such questions as "Are you a church member? What church? When did you join? Do you read the Bible? How much? How often? Do you pray? What do you
( 955) say in your prayers? Give examples. I Do you have family worship? How often? What part do you play in it ? 1)o you contribute to the church? Do you do church work? What? and so on. Interviews with students might supplement this material, but it would have no scientific value unless sufficient cases for statistical treatment were collected.
A better method would be to get evidence of actual behavior from indirect sources. For example, church membership could be determined from matriculation cards, church attendance by stationing student counters in the various churches, or by counting students in campus religious meetings. This method would require greater ingenuity and more work than the questionary method, but the results would be more objective, and hence more valuable.
Whether it is possible by this method to study the attitudes which we tried to discover by such questions as "Is God a Person?" "Is God an Impersonal Force?" I do not know. Perhaps some basis for inference might be gained by studying the church-going, church-working, and church-supporting habits of a particular community or group. If a church emphasizes belief in a personal God, the supporters of the church may possibly be said to have a personal-God attitude. But there are many other factors than belief in a personal God which contribute to the church-supporting actions of such people. We can only be sure of this: church-supporting is a status-raising action in some communities, and is therefore an attitude. All we can say is that a certain percentage of the people in a given community have church-supporting, church-going, church-working attitudes, and so on. We cannot deal with subjective motivations scientifically. We can deal with verbal attitudes, opinions, "reasons," and the like, but they do mot give a scientific explanation of overt behavior. Only the behavior itself can do that. In other words, if we find people acting in such ways as to promote belief in the virgin birth, whether their action is social-status-fixing verbal statements, the boycotting-legislating against, burning at the stake, or otherwise overtly attacking nonbelievers, perhaps we may say that such people have a virgin-birth attitude. And so on for all the subjective so-called attitudes which we attempted to discover in our investigation.
It is evident that this method will make it very much more difficult to do attitude research. It may be impossible to find out many things we should like to know; but that is true in all fields of research. The scientific value of the method proposed is that we shall not be deluding ourselves with the semblance of scientific knowledge, the substance of which we do not possess. Sociology as a natural science must devote itself to the sense-observable, status-fixing activities of persons when it undertakes to study attitudes.
As an example of attitude research which appears to be theoretically sound in the light of this discussion, I will cite May and Hartshorne's study of the attitudes of students toward cheating. According to their definition an attitude is a permanent set, as opposed to desire, which is a temporary set (p. 148) ; but both are based on behavior. Their distinction is admittedly verbal, since both attitude and desire are determined by the amount of cheating actually done. Theoretically, the scale would range from a set of conditions under which everyone would cheat to one under which no one would cheat. Under conditions most favorable to cheating, they found that 85 per cent actually did so. But the theoretical importance of their method is its indirection. The students never knew they were being tested for cheating. Thus, status-affecting behavior patterns under rigidly controlled conditions were determined without. any attention being given to subjective factors as such. Their analysis makes very clear the difficulty of ascertaining why a given group of students have certain measurable cheating attitudes under one set of conditions and different attitudes under other conditions. In this study we have an excellent example of an objective scientific quantitative study of attitudes.
The foregoing analysis may be summarized as follows: (1) There is a general confusion in the meaning of the term attitude. (2) This arises from emphasis upon hypothetical subjective factors, from attempting to differentiate attitudes and values, and from identifying opinions with attitudes. (3) I have contended that an
( 957) attitude is the relatively stable overt action of a person which affects his status in groups. (4) The only way to determine attitudes is by observation and statistical treatment of behavior in social situations. (5) The questionary is of doubtful value in this type of research. (6) The case method, life-history, and interview are little better, but have some advantages over the questionary. (7) The best general method is the statistical treatment of indirect evidences of overt behavior in carefully defined or experimentally controlled situations. (8) The feasibility of such research has been demonstrated.