ion-and, on the other hand, the attempt to isolate key variables of social reality within the situation that may be universal in their application. This is essentially the distinction between science in the naturalhistory sense and science in the hypotheticodeductive sense.4 An application of this distinction to the analysis of anthropological literature is Watson's polarity of "developmental" and "causal-correlational" theoretical interests.5 These two bases of distinction-(1) the extent of the range in time and space of the data to be investigated, and (2) the degree of abstraction in the process of investigation -together offer suggestive divisions for the subject matter of sociology. A formal application of these criteria would lead, by dichotomizing each, to four ideal types of sociology with perhaps such names as "concrete-microcosmic," "concrete-macrocosmic," "abstract-microcosmic," and "abstract-macrocosmic." However, if interest is less in creating a formal model than in clarification of the observed varieties of sociology, it would seem that the concrete-abstract dichotomy would not be particularly useful in clarifying the "macroscopic" range. In the first place, a considerable amount of abstraction is made necessary by the importance of selection from the wide range of phenomena studied. That is, it is extremely difficult to represent the broader features of cultures and societies in thoroughly concrete terms. In the second place, careful building and precise testing of theoretical models of abstract social science can better be done with studies of limited situations than with broad historical and cultural studies. Thus the broader ranges would also be less readily applicable to the development of a systematic experimental science. By thus eliminating the abstract-concrete distinction with regard to the broader ranges of subject matter, one could consider sociology as divisable into the remaining three types. For convenience we shall identify these types as (1) historical and cultural sociology, (2) concrete and clinical sociology, and (3) logico-experimental sociology. Historical and cultural sociology typically has societies and cultures or some broad phase of a complex society as its area of endeavor. One may deal with trends in a society or civilization (the historical approach) or follow lines of cross-cultural comparison (the anthropological approach), but in either case the unit of study is "society" in a rather broad sense. This range of sociology is devoted to describing main trends or chief features of complex societies. Data are typically a result of second-hand observation, based on a careful evaluation of the reliability and representativeness of sources. Analysis is commonly in terms of either explicit or implicit "ideal types," and the process of analysis demands a great degree of judgment on the part of the sociologist to separate the significant from the irrelevant. The approach thus must be largely qualitative, though quantitative data may have a vital place in the factual underpinnings. Clinical or concrete sociology 6 deals not with trends of the broad social order, but rather with a total situation within restricted limits of time and space. The limits are usually imposed by the group or organization under consideration, and the aim is to understand the situation in its entirety or the "system" as a functioning unit. The concepts that prove useful for this range of study are those that give the investigator a general sense of reference in approaching concrete situations. Blumer has aptly labeled these as "sensitizing" concepts, distinguishing them from more "definitive" concepts.7 A variety 4For a formal development of this difference, see F. S. C. Northrop, The Logic of the Sciences and the Humanities, New York: Macmillan, 1947, especially pp. 35-76. 5 Watson, op. cit., pp. 139-144. 6 Two alternate terms are here used because either one alone may be misinterpreted. By "clinical" is meant a general and diagnostic mode of analysis. This does not necessarily imply specific value judgments or dedication to solving social problems-although sociologists have sometimes used the term with such implications [for example, Alfred McClung Lee, "The Clinical Study of Society," American Sociological Review, 20 (December, 1955), pp. 648-653]. Common usage may also make the concept of "concrete" open to misinterpretation. However, by opposing it to the process of abstraction, we have tried to indicate above what is here implied by the term. 7 Blumer defends the role of such "sensitizing" concepts as the dominant concepts of sociology. In so doing, however, he finds it necessary to identify "the natural social world of everyday experience" -what we have here called the concrete or clinical This content downloaded from 207.46.13.33 on Wed, 12 Oct 2016 05:05:19 UTC All use subject to http://about.jstor.org/terms 662 AMERICAN SOCIOLOGICAL REVIEW of specific methods may be used in concrete sociology, but in any case skilled observation must play the major role. Indeed, the area of concrete sociology can be delineated as the area in which "participant observation" may be used as a key method of research. Logico-experimental sociology deals with relationships of variables abstracted out of concrete phenomena. The aim is to discover those aspects of social phenomena that may be seen as keys in determining other phenomena, and to describe the kind and measure the degree of the relationships between such "independent" and "dependent" variables. This is the part of sociology that most thoroughly shares the aims and methods of certain physical sciences. for here the emphasis is not on concrete analysis or naturalhistory description but rather on testing hypotheses and establishing a predictive hypothetico-deductive system. Concepts of a truly experimental sociology must be definitive-that is, the concepts must refer "precisely to what is common to a class of objects, by the aid of a clear definition in terms of attributes or fixed bench-marks." 8 And above this, if a successful hypothetico-deductive system is to be developed, the concepts must bear logical relationship within a theoretical system at the same time that they bear specific relationship to observable facts. Methods of logico-experimental sociology usually reserve a key place for exact measurement; but quantification and measures of significance do not preclude the necessity of a creative imagination for perceiving key variables and forming hypotheses concerning their relationships.