Measurement in Sociology
Miami University, Oxford, Ohio
Communication makes it possible for subjective experience to become objective knowledge. The world of object-events is in continual flux. All knowing is abstract, aspectual, partial. Knowledge is the result of repeated, communicable experience and therefore numerical. If the enumeration is analytic, explicit, statistical, we have natural science knowledge. Dewey seems not to imply that social phenomena cannot be treated statistically, but that so far few of them have been. The only certainties transcending common sense in sociology or any other science are statistical in nature. Dewey emphasizes the idea that physical and social objects are not different kinds of reality but that genuine knowledge of man and society necessarily lags far behind physical knowledge.
Professor House has stated fairly the degree of agreement between the two major methodologies in sociology, the quantitative and non-quantitative or case-study—life-history—attitude—insight and objective-behavioristic. Then, unfortunately, he asks two questions which cannot be answered but must be considered if we are to have a coherent philosophy of science. These two simple questions are: What is science? and How do we know and know that we know? Professor House admits so much in his paper and becomes so neo- or semi-behavioristic that I hope he will sometime write an article on the "Limitations of Subjective Sociology."
Taking the last question first, I venture the dogmatic assertion that no answer can be given that will be completely satisfactory to anyone except the one who answers. At least, it has not yet been done. The only people who know they know are those who have never thought much about the problem of knowing. Most natural scientists pursue their recondite studies with none or only the most naïve common-sense theory of knowledge. Frequently, they are metaphysical or philosophical phobiacs even though they are all unconscious or inarticulate metaphysicians. Their phobia is probably an unconscious defense mechanism, or an equally unconscious father
( 482) or mother hatred, depending upon whether philosophy is the mother or father of science.
At any rate, until scientists become philosophers, and philosophers, scientists, there will be continuous disputes generating more heat than light. These disputes will be of two kinds, terminological and postulational. The first are confusing but unimportant. They are simply symptoms of the disease of language. Sufficient patience and effort may resolve them. But even when the terminology is understood, as it seldom is, postulational differences will remain unresolved even after the postulators have done their best. If there is substantial agreement on postulates, as I think there is in our fraternity, there should be general agreement on methods of getting the knowledge implicit in our postulates. To test this assumption, I must postulate a bit.
First: There are object-events existing and occurring. Man is one of these. He makes certain movements. Among these are total-adjustment movements of approach and avoidance. He also makes movements we call vocalization and memory. Some vocal movements come to stand for other object-events, or functional aspects of them, as actually experienced by the vocalizer. When the same, or recognizably similar, sounds denote, or point out, for two or more persons, the same or similar objects, events or experiences, they are vocal symbols. Thus, vocalization and memory make communication possible; communication makes it possible for subjective experience to become objective knowledge. This is what we mean by knowing. One may "know" in the sense of experiencing, moving internally or externally, but others cannot know he knows unless and until this subjective experience is communicated; this can be done only by symbols, vocal or gestural. All experience is private, but all knowledge is public, i.e., communicated.
Second: The world of object-events, including man, is in continual flux. The rates of change in and between objects are variable; hence, uniformity and stability have meaning only when referred to some standard. Standards have their origin and meaning only because they represent some relatively stable, uniform, and common experience of man. Inescapably, whether we like it or not, we are in a homocentric universe. But we can never know anything about it
( 483) completely and finally because we can experience only limited aspects of object-events at any given instant; we ourselves and all other object-events, indeed the whole universe, are never the same for any two successive instants. Hence, we see in part and know in part. This is as true of acquaintance knowledge, if it is communicable, as it is of knowledge about; they differ only in the degree of objectivity with which the experience is communicated. The world we "know" is an indeterminate, interdependent, interfluent, interactive, and self-active congeries of reciprocal-variable object-events.
Third: From the foregoing, we see the significance of Dewey's statement that data are "takens" as well as "givens," i.e., they are abstractions from experienced objects or events. But for the same reasons, "objects" are also "takens." They are known objectively only by symbolic communication and enumeration of their abstracted characteristics as experienced by subjects. The expectancy we have of recurrent object-behavior depends upon the degree of agreement between our own experience and that of others who have communicated their experience to us. The frequent occurrence of such agreement is the only "proof" we have that "givens" as well as "takens" actually exist. This is as true of cultural objects or events as it is of physical or biological. The same radical abstraction of private experience must be made in sufficient number, be symbolized, and be communicated, before knowledge is attained. Most such radical abstractions from all kinds of objects and events are useless as data for scientific knowledge. They merely provide vocal exercise. Some of them are data that give useful common-sense knowledge, some furnish data for normative, and a few, for natural, science. The main point in this postulate is that all knowing, even purely subjective knowing, is abstract, aspectual, partial; all communicable knowing, i.e., knowledge, is possible only by numerative symbolical abstraction. The "givens" from which the "takens" are taken, as well as the "givens" in the subject which determine what "taken" he takes, must be sufficiently stable to permit the repetition of relatively similar experiences. If these conditions are met in two or more subjects, vocal symbolization of the experiences becomes possible and, with it, communication and objective knowledge. Repetition and, hence, numeration are indispensable to knowing.
If we agree on these postulates about the nature of nature, of knowers and knowing, as I think we do, or must, sooner or later, we should have little difficulty with the nature of science and its relation to measurement—even in sociology. We are considering natural science which is concerned only with prediction of the occurrence of natural phenomena. Such prediction is possible because there are relatively stable uniformities in the "givens" and hence in the "takens," both of objects and knowers. By our second postulate, it is also true that the universe is indeterminate even though all object-events are determined or conditioned by antecedent and contemporary factors. Natural science is concerned solely with going beyond the generalizations of common sense in scope and accuracy, and in making new generalizations quite unknown to common sense. Like all science, natural science strives to harmonize and systematize its findings, although this is not its fundamental objective, of course. If it can find and generalize relatively stable and hence predictable types of natural phenomena, however limited the generalization may be, it has done its task.
Natural science data, all being "takens," are limited by the nature of the "givens" which, by our second postulate, are instable variables. Consequently, there can be no natural science generalizations that are universal, final, absolute, and at the same time realistic. They can be valid only within the abstraction-delimited field of the "takens." For example, the laws of gravitation have no validity for the interactions of the human object-events in a room. According to the third postulate, all objective knowledge is numerical as well as symbolical and aspectual. Now if there is relative stability in the object-events and the knowers, the recurrent behavior of the latter with reference to the former is relatively uniform and hence is subject-matter that can become object-matter through symbolic reference and thus become communicable objective knowledge, and not otherwise. The nice point is that all such knowledge is the result of repeated, communicable experience, and is therefore numerical. If the enumeration is informal, implicit, unanalyzed, uncritical, we have common-sense knowledge; if it is analytic, explicit, statistical, we have natural science knowledge if the data are such that verification is possible according to the tested methods of natural science.
Four kinds of knowing have been mentioned: (1) subjective, uncommunicated, or incommunicable; (2) common sense (implicitly numerical); (3) normative; (4) natural science. All four kinds of knowing are reputable and valuable, but only the last three, aspectual, symbolical, communicable, are knowledge. Most of our subjective experiences we never even try to communicate—just having them is enough; some we do try to communicate on the common-sense level, with varying success. We frequently communicate value-judgments and are often understood even when our values are flouted; but natural science communication compels all reasonable men to accept the facts found by legitimate scientific methods. The meaning of the facts, their value-significance, is another matter and is no concern of the natural scientist as such.
Professor House's interpretation of Dewey's position in regard to measurement both surprises and shocks me. Either I do not understand Dewey, or Professor House does not, or Dewey is inconsistent (which would be a sore scandal in the chaste temple of philosophy), or it is a case of the devil quoting scripture. Two quotations from The Quest for Certainty are given to show the limitations of measurement. The first is from page 124. On page 125 is a very clear statement that measurement is of the very essence of natural science. Page 216 is also quoted, but on page 221 occur these words: "Nevertheless, in the end thinkers in all lines are dependent upon the mathematician and physical enquirer for perfecting the tools employed in their respective callings." On page 249, Dewey says: "No mechanically exact science of an individual is possible. An individual is a history unique in character. But constituents of an individual are known when they are regarded not as qualitative but as statistical constants derived from a series of operations." This is in accord with our second and third postulates. Similar expressions are found on pages 57, 87, 88, 92, 127, 133, 152, 206, 240, 241, and elsewhere. There can be no question that Dewey regards statistical measurement as the most important method of getting dependable scientific knowledge. He would probably almost subscribe to Kant's dictum that a study can be called scientific only in so far as it is mathematical.
The real question is: can social phenomena be treated this way?
(486) Professor House seems to think that Dewey thinks they cannot. I think Dewey thinks only that as yet few of them have been so treated. Both of us, or the three of us, may be wrong. It is obvious that the social and physical sciences do part company at this point. It is so true that most physical scientists say that sociologists are not natural scientists at all—and never can be. Perhaps they are right. They certainly are if Professor House's tentative acceptance of Werner Sombart's view is correct—if that is really Sombart's considered view. If there are no social data that can be treated by the methods of natural science, certainly we can never build a natural science.
It is true that little so-called sociological research has been done either from the point of view, or by the methods, of natural science. It is also true that many social phenomena do not lend themselves readily to natural science investigation—but the same is true of all other kinds of natural phenomena. Many of them cannot be studied scientifically; many that can be, have not been, and probably never will be. The same is true of social phenomena. But some have already been quantified and more will be. The only certainties transcending common sense in sociology or in any other science are statistical in nature. The degree to which such methodology can be applied to social data will determine whether sociology is to become a natural science or remain forever a bastard discipline sired and dammed by common sense and normative knowledge; whether it is to be a natural science or a hodgepodge of pretentious words, random observations, speculations, opinions, pious hopes and fears, attitudes, wishes, sophistical logic, and literary purple patches. Should a sociologist be a Zola or a Quetelet?
On pages 186–200 and 270–86 of The Quest, Dewey discusses at length the necessity for applying the "experimental method" to social relations. He emphasizes the idea that physical and social objects are not different kinds of reality (page 217) but that genuine knowledge of man and society necessarily lags far behind physical knowledge (page 271). If genuine knowledge and experimental method in Dewey's sense of the terms are applied to human affairs, both of necessity must be based upon scientific knowledge derived by statistical methods. This is the implication of his whole book and
( 487) the note upon which it ends. Of course, we may be "experimental" on the basis of common-sense knowledge, and often of necessity must be, but it is frequently a tragic and costly business and would never be resorted to except by fools and knaves if scientific knowledge were available or could be derived quickly enough to meet the demand for practical action of some kind. Three short paragraphs and we close.
First: Attitudes. Ignored; because everything I have to say on this subject for the present is in print, or in this paper by implication. Dangerous; because Professor House has made me say by "implication" what I have qualified by explication in several papers. His general reference is correct, but does not give my fundamental objection to most so-called attitude studies, viz., though frequently statistical in form, they also frequently violate the canons of sound natural science method. However, even though the data are doubtful, such studies may develop techniques applicable to the study of valid scientific data. It may even be useful for some purposes to know what people think they think, or say they think they think, or wish or hope or fear they will do or may do or ought to do, to say nothing of the past tense of all these possibilities. So my attitude toward the attitude enthusiasts, case study students, and life historians is charitable. They may be right. They certainly should not be prohibited, but they should certainly be soundly and frequently criticized.
Second: Insight. I have little more insight into insight than Professor House has implicit vocalization of neuromuscular, but to me it seems suspiciously similar to acquaintance knowledge. Both, when communicated, are common-sense or value-judgment knowledge, never natural science knowledge. Both may furnish data which may be treated scientifically; or they may suggest problems for investigation by scientific methods; or they may be used to interpret the meaning and significance of scientific facts. These are the only possible relations they can have with natural science.
Third: To conclude on the same note of agreement with which we began, I think almost everyone would answer Professor House's third issue as follows: By all means let all who can get the money and have the inclination make all the case studies they can—the
( 488) more the better. If they study enough cases, they will be forced to use statistical methods and so, by this declension, may eventually arrive at some dependable scientific knowledge. They will necessarily abstract some behavior aspects from their cases some of which may be characteristic of other similar cases. Thus, their studies may become objective enough to be repeated, and so become subject to scientific verification. In any case, case studies may suggest problems which can be approached statistically, may suggest fruitful hypotheses, and thus become indirectly an influence for considerable advancement of sociology as a natural science.