Functional Analysis
Functional Analysis
I. Structural-Functional AnalysisMarion J. Levy, Jr.
II. Varieties of Functional AnalysisFrancesca M. Cancian
I STRUCTURAL-FUNCTIONAL ANALYSIS
Few concepts in modern social science history have generated as much discussion as those of structure and function and the type of analysis associated with them. The main difficulty in speaking of structural-functional analysis in general arises from five sources. One, there is the feeling in many quarters that there is something new and special about structural-functional analysis. Two, in general usage elementary procedures in definition have not been observed. The same term has frequently been used for more than one distinct referent. Three, teleology in the sense of scientific fallacy—in this case structural teleology, functional teleology, or both—has frequently been committed in connection with such analysis. Four, the use of stability assumptions in models generally has been both misunderstood and misconstrued. Five, unintentionally, evaluations have been written into the analysis, thereby raising questions about its objectivity.
Structural-functional analysis is not new in either the social or the natural sciences; it has a pedigree that stretches indefinitely back in both fields. The only new aspect of it is its formidable new name, structural-functional analysis. Discussion of it as something new under the sun is the social scientist’s counterpart of M. Jourdain’s discovery that he had been speaking prose. The only possible novelty associated with this form of analysis is the attempt in recent years to be carefully explicit in the use of these concepts and to differentiate special subsidiary forms of the analysis, although none of the latter are substantively new either. Shorn of careless uses of definitions and of teleology, structural-functional analysis is simply a synonym for explicit scientific analysis in general. In scientific fields marked by greater theoretical development and associated applications of mathematical forms of expression, the increased verbal explicitness that has recommended the various concepts of structure and function to many in the biological and social sciences is more cumbersome than available alternatives. Corresponding increases in theoretical development in the biological and social sciences will lead to similar alternatives there.
Simply speaking, structural-functional analysis consists of nothing more complicated than phrasing empirical questions in one of the following several forms or some combination of them: (1) What observable uniformities (or patterns) can be discovered or alleged to exist in the phenomena studied? (2) What conditions (empirical states of affairs) resultant from previous operations can be discovered or alleged to exist in the phenomena studied? (3) When process (or action, i.e., changes in the patterns, conditions, or both, depending on one’s point of view, are discernible between any two or more points in time) can be discovered (or alleged) to take place in terms of observable uniformities, what resultant conditions can be discovered? The first question asks, “What structures are involved?” The second asks, “What functions have resulted (or have been performed)?” And the third asks, “What functions take place in terms of a given structure(s)? Many special forms of these three questions are useful and necessary for the analysis of different types of problems; for example, one might ask about the possibility of adjustment in terms of a system, the normative content of a system, the “necessary” features of a system, the degree of planning or consciousness involved, etc. All of these are variants of the three basic questions or some combination of them. In addition to the most general form of the concepts of structure and function, six subsidiary sets of concepts most generally associated with this form of analysis, either explicitly or implicitly, are defined and discussed below in terms of the special purposes generally associated with them.
Function and structure. These explicit concepts, ostensibly in their most general form, are frequently encountered in the biological and social sciences. In both fields somewhat similar difficulties have been associated with the use of the concepts. Joseph H. Woodger in biology and Robert K. Merton in the social sciences (Woodger 1924, pp. 326-330; Merton 1949, pp. 21-28) have pointed to the profusion of different referents given to the term “function” even in the scientific sense. These different referents have for the most part been left implicit by both authors although the confusion is noted. Perhaps the major difficulty associated with the general concept of function has been the use of a single term to cover several distinctly different referents. This difficulty is much greater with the concept of function than with the associated concept of structure. Most of the general discussions in the literature have been largely preoccupied with the concept of function. The concept of structure has, more often than not, been left undiscussed. To prevent confusion, the most general form of the concepts “structure” and “function” used for purposes of scientific analysis must be defined in such a way that the other uses of the terms represent specific subcategories. The term “function” may be defined as any condition, any state of affairs, resultant from the operation (including in the term “operation” mere persistence) of a unit of the type under consideration in terms of a structure(s). In the case of the biological sciences that unit is usually an organism or a subsystem of an organism. In the case of the social sciences the unit is usually a system of action involving a set of one or more persons (actors). The term “structure” may be de fined as a pattern, i.e., an observable uniformity, in terms of which action (or operation) takes place.
In ordinary usage the term “function” is most generally identified with the term “eufunction” de fined below. Similar confusion of the term “structure” with “eustructure” is not common. The concept of structure in this general form explicitly covers a wide range of possibilities from highly stable uniformities to highly fleeting ones. Any event may contain an element indicative of a structure if it is considered with regard to nonunique aspects or parts.
Much of the interest of scientific social analysis is centered on the structure of societies and other social systems (or the structures of social action in general), that is, on the interrelationships among different kinds, aspects, and parts of structures. The relationship between the concepts of function and structure is close. Structure refers to an aspect of empirical phenomena that can be divorced from time. The patterns of action, qua patterns, do not exist as concrete objects in the same sense that sticks and stones do. The patterns of action in this sense are abstractions from concrete empirical phenomena, and they “exist” and are empirically verifiable in the same sense that the squareness of a box exists and is empirically verifiable. What has been said here of patterns qua patterns does not apply to the patterns when they are considered in operation. Structures in operation (i.e., the exemplifications of particular patterns) are empirical in the same sense as sticks and stones. In this sense the term “structure” in social science is no departure from the usage of the natural sciences.
The concepts of structure and function fall into a peculiar set of concepts. Classification of a referent as a function or a structure depends in part on the point of view from which the phenomena concerned are discussed. What is a function from one point of view is a structure from another. The concepts of consumption and production are more familiar examples of this peculiar set. The manufacture of automobiles is production from the point of view of an automobile user and consumption from the point of view of a steelmaker. Thus functions in this sense are themselves structures (i.e., patterns) or have important structured (i.e., patterned) aspects, and all structures are the results of operations in terms of other structures (i.e., they are functions). The politeness of small children may be considered as a structure of their behavior or as a function of operation in terms of the structures (i.e., patterns) of discipline indulged in by their parents.
Functional and structural requisites. The concepts functional and structural requisites are primarily oriented to the development of systems of analysis for any cases of a particular type of unit. A functional requisite may be defined as a generalized condition necessary for the maintenance of the type of unit under consideration, given the level of generalization of the definition and the most general setting of such a unit. In seeking to discover the functional requisites of a unit one asks the question, “What must be done to maintain the system concerned in its setting on the level under consideration?” A given condition is a functional requisite if its removal (or absence) would result in (a) the total dissolution of the unit, or (b) the change of one of the structural elements of the unit on the level under consideration (i.e., one of the structural requisites).
A structural requisite may be defined as a pattern (or observable uniformity) of action (or operation) necessary for the continued existence of the type of unit under consideration, given the level of generalization of the definition and the most general setting of such a unit. In trying to discover the structural requisites of a unit one seeks an answer to the question, “For a given unit, what structures (i.e., patterns) must be present such that operations in terms of these structures will result in the functional requisites of the unit?” Briefly, functional requisites are answers to the question, “What must be done?” structural requisites are answers to the question “How must what must be done, be done?” Both questions involve the qualifying phrase “if the unit is to persist in its setting on the level of generalization given.” Most of the misunderstanding about stability mentioned above has been associated with this question of persistence. One utilizes an assumption of stability to get at a list of requisites, but this does not imply that any unit analyzed must in fact be stable. To make such units stable by definition is to reduce most discussions of this sort to trivia. To use a stability assumption as an element in a model (i.e., paradigm) does not. Structural-functional requisite analysis includes the following steps in any specific case: (1) define the unit of phenomena to be studied; (2) discover (or hypothesize about) the setting (i.e., those factors determining the limits within which the ranges of variations of the unit concerned take place); (3) discover what general conditions must be met (i.e., functional requisites) if the unit is to persist in its setting without change (i.e., alteration of structures) on the level under consideration; (4) discover what structures must be present in the system, as a minimum, if action in terms of the system is to result in the persistence of the unit in its setting without any change on the level under consideration (i.e., the structural requisites).
Several things should be kept in mind. One, although the definition of the unit is arbitrary, whether empirical referents of such a unit exist is not. Two, the setting of such a unit is not a matter of definition but of discovery. Three, neither structural requisites nor functional requisites can be alleged to exist “because they are requisites.” Such allegations constitute the commission of structural or functional teleology. To assert that requisites exist because they are requisites is to imply that the unit must continue to exist for some preordained reason. Scientifically speaking, nothing can be alleged to exist “because it is a requisite.” The fact that this form of teleology is fallacious is in no way contradicted by the fact that it is frequently a useful element of models of action to assume or allege that the action concerned is oriented to future states of affairs. Four, the determination of the functional requisites of a unit is the determination of the minimum implications of interrelationships between the factors setting the limits of variation of the unit and the unit itself, and in this type of analysis it is never necessary to deal with more than these minimal implications. However, when material is collected in these terms, more than the minimum will always be collected, since such minimal materials never exist neatly separated from all others. Five, there is a systematic test for error in requisite analysis. If a structure is alleged to be a requisite of the unit concerned, and if examination of a particular case of such a unit uncovers no material on this score, one of three or some combination of three things is an explanation of the lack of data. First, the hypothesis that the structure is a requisite of the unit concerned may be an incorrect one. Second, the observer may have misobserved; there may be data he has overlooked. Third, the unit concerned, although it may closely resemble the unit as defined, may in fact not be a case of such a unit.
There is nothing new about requisite analysis. There has never been a time in which people failed completely to ask questions as to whether given conditions or patterns were not in some sense necessary for the continued existence of certain types of units.
Functional and structural prerequisites. The requisite concepts are not in and of themselves oriented to questions of change; the concepts of functional and structural prerequisites are. All questions of change implicitly or explicitly involve comparisons between at least two of at least three possible distinctions with regard to the units under consideration. These distinctions are those of an initial, a transitional, and a resultant stage. Systematic knowledge about any two of these stages makes possible systematic derivation (prediction or postdiction) of knowledge about the third. Requisite analysis can be used for examining any two of these three or more stages in terms of a constant frame of reference. Constants and variables are therefore more easily detected.
A functional prerequisite may be defined as a function that must pre-exist if a given unit is to come into being in a particular setting. Correspondingly, a structural prerequisite may be defined as a structure that must pre-exist if a given unit is to come into being in a particular setting. The closer two stages of a given unit under consideration are in point of time the greater is the probability that the requisites and the prerequisites of a given unit will be identical. To illustrate this, one of the commonest mistakes in trying to understand the problems of development in “underdeveloped areas” is the implicit assumption that requisites and’prerequisites must or do coincide. It is neither obvious nor tenable to take the position that all of the structures that must be maintained if the United States is to continue as a highly modernized society are identical with the structures that have to pre-exist if Nigeria is to become a highly modernized society, or even that they are identical with all the structures that had to pre-exist, say, in the beginning of the nineteenth century, if the United States was to become a highly modernized society.
Eufunction, dysfunction; eustructure, dysstructure. These concepts focus attention on questions of adjustment and maladjustment of the units under consideration. Although the term “function” is ordinarily used in several different senses in the social sciences, most often it refers to the concept of eufunction as defined here. With respect to a given unit, “eufunction” may be defined as any function that increases or maintains adaptation or adjustment of the unit to the unit’s setting, thus making for the persistence of the unit as defined in its setting. With respect to a given unit, “dysfunction” may be defined as any function that lessens the adaptation or adjustment of the unit to its setting, thus making for lack of persistence (i.e., change or dissolution) of the unit as defined in its setting. The terms “eustructure” and “dysstructure” are similarly defined, mutatis mutandis. Alternatively, eustructures may be defined as structures in terms of which operations result in eufunctions, and dysstructures may be defined as structures in terms of which operations result in dysfunctions. Eufunctions or dysfunctions and the corresponding variants of structure may exist, as far as a given unit is concerned, as elements of that unit (i.e., internal to it) or as elements of the setting of the unit concerned. Not all eufunctions for a unit are eufunctions of the unit, although ordinarily, when one uses the concept of eufunction, attention is focused on functions associated with the unit itself rather than on functions of operation in terms of other units in that setting.
Implicitly or explicitly, some form of requisite analysis always underlies any form of prerequisite analysis. Teleology must be avoided in uses of the concepts of structural and functional prerequisites as well as in uses of the concepts of structural and functional requisites. To assert that some structure must pre-exist because it is a structural prerequisite of a given unit is to fall into teleological dynamics as distinguished from the teleological statics described in the case of requisites.
The concept of “nonfunction,” in the sense of being neither a eufunction nor a dysfunction, is inutile. Where questions of adjustment are not involved, the general concept of function as used here or another of its special forms will serve. Where questions of adjustment are involved, nothing is less probable than a precisely poised function with no implications for adjustment or maladjustment. Indeed, such considerations define precisely what might be called the category of irrelevant functions, i.e., those lacking any implications for the focus of interest at the time. Such functions would also have to lack any interdependencies with the eufunctions or dysfunctions for or of the systems concerned.
It is in connection with the terms “eufunction” and “dysfunction” (and the corresponding forms of the concept of structure) that most allegations of building evaluations into structural functional analysis are made. In rough terms, a eufunction is a function that tends to preserve the unit as defined and a dysfunction is one that tends to dissolve it. However, loose usage of these concepts results in the use of eufunctional to refer to conditions making for “good adaptations” and dysfunctional to refer to conditions making for “bad adaptations.” Such judgments are a function of one’s evaluation of the unit concerned—if one cares to make such an evaluation. No condition or aspect of a condition is inherently eufunctional or dysfunctional. Without identification of the unit concerned and its setting, no judgment of the eufunctional or dysfunctional character of the condition can be made. The same condition that is eufunctional from one point of view may be dysfunctional from another. For example, conditions that were eufunctional for Meiji Japan were dysfunctional for the continuation of Tokugawa Japan and vice versa.
While these concepts are susceptible to the uncritical inclusion of value judgments, they are also a useful vehicle for the explicit consideration of policy-oriented analyses. Either implicitly or explicitly, in seeking to maximize a given policy goal, one asks the questions, “What conditions make for maximal adjustment of the system concerned to that state of affairs (i.e., the eufunctions for that system in that setting)?” and “What conditions should be avoided as interfering with the maintenance of that system in that setting (i.e., are dysfunctional for it)?” Two other points should be carefully kept in mind. First, teleology is to be avoided in this connection too. No function or structure exists because it is a eufunction or a eustructure, nor is it tenable to hold that any function or structure that persists must be eufunctional or eustructural. Second, a given element with important eufunctional or eustructural implications for a given unit in its setting may also contain aspects with dysfunctional or dysstructural implications. A functional requisite of a given unit is certainly at least in part eufunctional for that unit, but a given function, even though it is a requisite, may also contain dysfunctional implications as well, although presumably in this case they would not be sufficiently pronounced to overcome the requisite nature of the function.
The concepts of eufunction, dysfunction, eustructure, and dysstructure focus on the question of the maintenance or lack of maintenance of a system in its setting. The requisite concepts focus on the question of what a system is like if it is maintained. The prerequisite concepts focus attention on what conditions must pre-exist before a given type of unit can come into being. The requisite concepts are useful primarily for static theories, although dynamic analysis is involved in discovery of the requisites of a given system in its setting. Like the prerequisite concepts, those of eufunction, dysfunction, eustructure, and dysstructure focus attention on dynamic interrelationships—on the implications of the operation of a particular structure or the presence of a particular function for the state of the system concerned at some future point in time.
Latent, manifest, UIR, and IUR functions and structures. This set of terms is adapted from Merton’s suggestion many years ago about latent and manifest functions. Usage here is somewhat changed, since Merton did not envisage their use apart from the concepts of eufunction and dysfunction as defined here, nor did he develop the implications of the fact that his defining conditions could vary independently. Following Merton’s usage an element will be termed “manifest” if it is intended and recognized by the participants in the system of action concerned. It will be termed “latent” if it is neither intended nor recognized. It will be termed “UIR” if it is unintended but recognized and “IUR” if it is intended but unrecognized. These concepts focus attention on the level of explicitness and sensitivity of the members of a given system to the structures in terms of which they operate. Such distinctions are vital to any discussions involving rationality, planning, and so forth. The concept of manifest dysfunction is explicitly included as a possibility here. Failure to consider it explicitly has built implicit evaluations into a great deal of analysis. Social reformers usually concentrate all their efforts on being manifestly dysfunctional for the system they are trying to reform and manifestly eufunctional for what they consider to be a better world.
The concepts of function and structure in general, the requisites, the prerequisites, and eufunction-dysfunction, eustructure-dysstructure, are concepts generally applicable to social and non-social phenomena and to human as well as to non-human phenomena. Concepts of latent, manifest, IUR, and UIR specifically focus attention on the point of view of the actor and involve the assumptions that such actors can and do orient their behavior to future states of affairs about which they are capable of explicit thinking and observation, and that the presence or absence of such explicit thinking, observation, or both has, in turn, implications for their subsequent behavior. Although it is conceivable that such concepts be utilized with regard to nonsocial phenomena, it is not likely that they will prove useful in such a field nor have they generally been so employed. It is quite conceivable that they could be employed with regard to some animate but nonhuman phenomena, but by far their greatest applications are to human social phenomena.
Concrete and analytic structures. The distinction between concrete and analytic structures is oriented to the type of abstraction involved in certain concepts useful for empirical analysis. Concrete structures may be defined as those structures (i.e., patterns) that define the units that are at least in theory capable of physical separation (in time, space, or both) from other units of the same sort. As the term will be applied for social analysis, it refers more specifically to the structures that de fine the membership units involved in social action, i.e., units in regard to which any given individual may be classified as included or excluded, or some combination of the two. Society, as that term is generally used, is a concrete structure considered in operation. So are organizational contexts generally, such as families, business firms, and governments. Concrete structures other than societies are the structures of action that define the membership units within a given society or those relating to two or more societies, i.e., they are all other social systems. All concrete structures are social systems; all societies are social systems; and all social systems other than societies are either subsystems of a given society or the results of interdependencies among two or more societies. The family, as that term is often used, is a concrete structure; its structures define a membership unit. So are business firms. A given individual may at one and the same time be a member of both a given business firm and a given family; nevertheless, the social structures characterizing the business firm are not part of the social structures characterizing the family concern, however complex the interrelationships may be, unless the firm is specifically and completely a family affair. It is at least conceivable that all the members of a given family could be put in one room, that the members of a given business firm could be put into another, and if necessary, those individuals who are members of both could be removed from the two rooms and put into a third.
Analytic structures are defined as structures (i.e., patterns) that define the aspects of units that are not even theoretically capable of concrete separation from other structured aspects. If one defines the economic aspect of action as having to do with the allocation of goods and services and political aspects as having to do with the allocation of power and responsibility, then economic and political structures are analytic structures, since there are no social systems that are totally devoid of either economic or political aspects.
Failure to keep this sort of distinction straight constitutes the fallacy of reification (or misplaced concreteness). Thus the terms “economy” and “pol ity” as generally used cannot occupy the same position in a system of analysis as the term “family.” As these concepts are generally defined, they represent analytic structures, and the concept “family” refers to a concrete one. This distinction, though difficult in the context of social phenomena, is a generally familiar one in the natural sciences. Concepts such as atoms, molecules, cells, organs, and elementary particles are concrete structures. Aspects such as mass, shape, color, temperature, and mitosis are analytic structures.
In actual analysis one always uses some combination of both sorts of concepts. One cannot identify analytic structures without some specification sooner or later of the concrete structures of which they are aspects. Correspondingly, one cannot discuss the nature of concrete structures without sooner or later making reference to aspects that cut across such units. Careful observance of the distinction and the avoidance of the fallacy of reification is the key par excellence to the avoidance of much disciplinary parochialism in the social sciences.
Apart from considerations of methodological elegance there is another reason for being careful about this distinction. One of the most general distinctions between concrete structures or organizations is that between those which are specialized in the predominant orientation of their members to one or another of these analytic structures and those which are not specialized in this way. (Organizations may, of course, be specialized in other ways, e.g., in terms of the product orientations of the members.) Thus, a business firm may be spoken of as a predominantly economically oriented structure (or perhaps more accurately, though more cumbersomely, as a specially economic—analytic-structure-oriented concrete structure). A family unit is not specialized in any one of these aspects in any clear-cut predominance over the others.
Institutions, tradition, and Utopian structures. The three terms “institution, tradition,” and “Utopian structure” refer to different types of structure. The concept “institution” may be denned as any normative pattern, conformity to which is generally expected and failure to conform with which is generally met with moral indignation or some other form of sanction by the individuals who are involved in the same general social system and are aware of the failure. This is the sense of the term employed by Talcott Parsons in his early work. A given structure may be more or less well-institutionalized to the degree to which conformity with the structure is generally to be expected and the degree to which failure to conform to the structure is met by the moral indignation or sanctions of the individuals who are involved in the system and who are aware of the failure. Thus, the two sources of variability with regard to the level of institution-alization may be referred to as conformity aspects and sanction aspects. An institution may be regarded as crucial or more or less strategic. A given institution will be called a crucial institution if it is a structural requisite of the system in which it appears. It may be regarded as a more or less strategic institution to the extent that: (1) it is the institutionalized form of all or a portion of a structural requisite; and (2) the structure concerned may not or may be altered without destroying the structural requisite involved. A completely strategic institution is a crucial institution. The first of the two aspects of the strategic quality of an institution will be called its substantive aspect, and the second will be called its critical aspect. The critical aspect is a special form of consideration of the general problem of functional substitutes, equivalents, or alter natives (i.e., the question of the possibility of a given condition being the result of behavior in terms of one or more alternative structures).
A great deal of nonsense is talked and written as a result of misunderstanding the problem of functional substitutability. Many use this nonsense to argue the severe limitation of what can be done in these terms. Much of the confusion results from initially posing a question implicitly on one level of generality and in the course of the discussion proceeding, still implicitly, on a different and less general level. If the argument is correct that role differentiation on the basis of absolute age is a requisite of all societies, there can be no functional substitutability for that. The fact that the members of one society handle role differentiation in terms of five absolute age distinctions whereas those of another handle it in terms of ten simply indicates the fact that a particular solution to absolute age distinctions is not determinant for all societies. For a particular society, however, either five or ten distinctions may be just as much a requisite as some sort of role differentiation on the basis of absolute age is for all societies. Some discussions of functional substitutability imply that the structures involved can be changed without other changes on the level of generality on which the treatment has been focused. There are, of course, instances in which a given function may result from operation in terms of various alternative structures on specific lower levels of generality without changes on the most general level under scrutiny, but that question is by no means settled by the analyst’s simply being able to conceive of another structure from action in terms of which the function in question could result—and so asserting his conclusions.
The term “tradition” may be defined as an institution whose perpetuation is institutionalized—that is to say, as a special form of institution. An institution may be considered more or less traditional, or more or less traditionalized, to the extent that its perpetuation is institutionalized without regard to changes of the functional implications of operations in terms of it—whether these be eufunctional or dysfunctional implications. In this sense, mo nogamous marriage in the United States would seem to be much more highly traditional and traditionalized than the structure of driving on the right-hand side of the road. A tradition in this sense is a double institutionalization: (1) the structure concerned is an institution; and (2) the perpetuation of the structure is also an institution. Traditions may vary in at least two radically different ways. One, they may, of course, vary with regard to the institution that is traditionalized, although implicit in the concept of any institution in its conformity aspects is some degree of traditionalization. More important, traditions—like other institutions—may vary with regard to their combinations of conformity and sanction aspects. For example, the tradition of driving on the right-hand side of the road in the United States has very high conformity aspects and relatively minor sanction aspects, whereas the tradition of the incest taboo has very high conformity aspects and very high sanction aspects.
Utopian structures may be defined as those particular sets of normative patterns which, though not institutionalized, do require allegiance to them as institutionalized ideals. Allegiance to them as ideals is highly traditionalized in both conformity and sanction aspects. Thus, one does not in fact expect conformity to the principle of “Love they neighbor as thyself,” but expression of it as an ideal is certainly institutionalized in some social contexts, and its perpetuation is institutionalized. In the social systems in which they are found, Utopian structures often have the function of making it easier to teach the actual institutional structures of the system and of setting the framework in terms of which actual conformity to less extreme expectations is, in fact, expected.
Ideal and actual structures. Ideal structures may be defined as those structures that the members of any given system (or systems) feel should be the structures of their behavior or those of others. Actual structures may be denned as the structures (i.e., patterns) in terms of which the members of the system in fact behave as they would be described by an observer with theoretically perfect scientific knowledge. The terms “ideal” and “actual” in these respects may be applied in other connections, but they are most generally applied to the concepts of structure. The following generalizations in terms of them are relevant in almost any analysis. One, there are no peoples who do not make some distinction between ideal and actual structures. Two, ideal and actual patterns of a given system never coincide perfectly. Three, the members of the systems concerned are always to some extent aware of the fact that the ideal and actual structures do not coincide perfectly. Four, some sources of stress and strain in any social system inhere in the fact that the ideal and actual structures do not coincide perfectly. Five, not paradoxical relative to the fourth generalization, some of the possibilities of integration in terms of any social system inhere in the fact that the ideal and actual structures do not coincide perfectly. Six, failure of the ideal and actual structures to coincide perfectly is never explicable solely in terms of hypocrisy. Seven, perfect coincidence of the ideal and actual structures would require both perfect knowledge and perfect motivation on the part of all of the members of the system concerned in all possible situations. This would (a) overload any known set of cognitive capacities of individuals; and (b) were it possible to have perfect co incidence of the ideal and actual structures, the resultant system would be totally brittle, since any change of setting factors necessitating or causing any change internal to the system would require complete resetting of all of the cognitive capacities of the members of the system.
These concepts can be further elaborated; e.g., there may be ideal ideal structures (i.e., Utopian structures) as well as many pseudo or pretended ideal structures (i.e., hypocrisies which are ideal structures which are not actual ideal structures).
Structural-functional analysis in the most general sense, shorn of confusion of terminology, misuses of stability assumptions, teleology, and implicit evaluations, is synonymous with scientific analysis in general. The special forms of structural-functional analysis are not new, although the attempt to be explicit about them in general is recent. Of the various special forms of structural-functional analysis, there is one sense in which requisite analysis underlies the rest. It is oriented to the derivation of a system of analysis for any unit of the sort under discussion, and all of the other forms of structural-functional analysis, whether in terms of prerequisites, manifest, latent, UIR, and IUR elements, eufunction-dysfunction, eustructure-dysstructure, etc., presuppose two or more states of some sort of unit(s) which presumably can be analyzed systematically. Of the various special forms of structural-functional analysis, those having to do with the latent, manifest, UIR, and IUR elements, the concepts of institutions, traditions, and Utopian structures, and the ideal and actual structures are the most specifically oriented to the analysis of human social phenonema. The other special forms are much more generally oriented.
Marion J. Levy, Jr.
BIBLIOGRAPHY
Davis, Kingsley 1959 The Myth of Functional Analysis as a Special Method in Sociology and Anthropology. American Sociological Review 24:757–772.
Firth, Raymond 1955 Function. Yearbook of Anthropology [1955]: 237–258.
Hempel, Carl G. 1959 The Logic of Functional Analysis. Pages 271-307 in Llewellyn Gross (editor), Symposium on Sociological Theory. New York: Harper.
Levy, Marion J. Jr. 1952 The Structure of Society. Princeton Univ. Press.
Levy, Marion J. Jr. 1954 Comparative Analysis of Societies in Terms of Structural-Functional Requisites. Civilisations 4, no. 2:191–197.
Malinowski, Bronislaw 1944 A Scientific Theory of Culture and Other Essays. Chapel Hill: Univ. of North Carolina Press. → A paperback edition was published in 1960 by Oxford Univ. Press.
Malinowski, Bronislaw 1945 The Dynamics of Culture Change: An Inquiry Into Race Relations in Africa. New Haven: Yale Univ, Press. → A paperback edition was published in 1961.
Merton, Robert K. (1949) 1957 Social Theory and Social Structure: Toward the Codification of Theory and Research. Rev. ed. Glencoe, III.: Free Press.
Nadel, Siegfried F. 1951 The Foundations of Social Anthropology. London: Cohen & West: Glencoe, III.: Free Press.
Nadel, Siegfried F. 1957 The Theory of Social Structure. London: Cohen & West; Glencoe, III.: Free Press. → Published posthumously.
Nagel, Ernest 1957 Logic Without Metaphysics, and Other Essays in the Philosophy of Science. Glencoe, III.: Free Press.
Nagel, Ernest 1961 The Structure of Science: Problems in the Logic of Scientific Explanation. New York: Harcourt.
Parsons, Talcott 1937 The Structure of Social Action: A Study in Social Theory With Special Reference to a Group of Recent European Writers. New York: McGraw-Hill.
Radcliffe-Brown, A. R. 1952 Structure and Function in Primitive Society: Essays and Addresses. London: Cohen & West; Glencoe, III.: Free Press.
Woodger, Joseph H. (1924)1948 Biological Principles: A Critical Study. London: Routledge.
II VARIETIES OF FUNCTIONAL ANALYSIS
For the past few decades functional analysis has been a major approach to understanding the organization of society. But at the same time it has become the target of very serious criticism and has been attacked as illogical, value-laden, and incapable of explaining anything. As one would expect, there is a serious problem of definition.
The term “functional analysis” or “structural-functionalism” has been applied to a great variety of approaches (Merton 1949; Firth 1955) that share only one common element: an interest in relating one part of a society or social system to an other part or to some aspect of the whole. Within this variety, three aspects or types of functionalism can be distinguished: the first is based on the concepts and assumptions of sociology; the second, on the theoretical orientation that all major social patterns operate to maintain the integration or adaptation of the larger social system; the third, on a model of self-regulating or equilibrating systems. Most functional analyses contain all three aspects, but a clear description and evaluation of functionalism requires their separate consideration.
A nondistinctive type. The first type of functionalism consists in stressing sociological analyses as opposed to psychological or historical analyses. It is based on the assumption that the social traits existing in a society at a given time are interrelated in a systematic way and that ordered relationships can be discovered among “social facts” or social institutions without necessarily bringing in psychological or historical factors. This guiding assumption is accompanied by an implicit consensus on the important elements of social structure and by a developing set of hypotheses on their interrelations. The crucial elements of social structure include: the composition and formation of groups, the organization of economics and religion, and the distribution of authority and prestige. These are the now-traditional concerns of sociology and social anthropology, the elements stressed by the major theorists from Émile Durkheim and Max Weber to A. R. Radcliffe-Brown, Talcott Parsons, and George C. Homans.
This type of functional analysis is most characteristic of the monographs of British social anthropologists and their adherents in the United States, and of the essays of sociologists, especially Kingsley Davis and Talcott Parsons, on some aspect of modern societies. The anthropologists typically present a careful description of a primitive society. The sociologists begin with a more impressionistic account of some institution in their own society. Then both go on to show how the various parts of the society or institution are interrelated, how the whole system “hangs together” despite some strains and inconsistencies. This type of work has led to many valuable hypotheses, such as those relating the isolated nuclear family and the modern economic system, or matrilocal residence and divorce rates.
Careful description and a search for general patterns are obviously necessary to social science, and, as Davis (1959) points out, functionalism has been an effective banner under which to fight for these qualities. In anthropology, functionalism success fully opposed the diffusionists and the empiricists, who tended to ignore general patterns, and the evolutionists and the rigid monocausalists, who tended to ignore the necessity of careful description.
However, as Davis also argues, this type of functionalism is essentially equivalent to sociological analysis, and it is misleading to distinguish it by a special name. The functionalist position that societies should be viewed as systems having definite structure or organization means primarily that the parts of a society are interrelated in some nonrandom way; i.e., that there are regular patterns and relationships, and, therefore, it is reasonable to try to construct a science of society. This position underlies all sociological analysis, although there is considerable disagreement on whether all parts of a society are interrelated or whether there are relatively autonomous subsystems. The elements of society that these functionalists consider are also not distinctive, nor are the kinds of relationships among elements. Functionalists do stress reciprocal relationships, but the typical monograph or essay also includes relationships based on similarity, one-way causation, fulfillment of a psychological or social need, etc. The problems and accomplishments of this kind of functional analysis are the same as the problems and accomplishments of sociology and social anthropology in general. Therefore, the discussion below will concern only the other two types of functionalism, both of which are distinct forms of analysis.
Two distinctive types. The first of the two remaining types of functional analysis will be called traditional functional analysis because it is the most widely used type. It is based on the theoretical orientation that all major social patterns operate to maintain the integration or adaptation of the larger system, and it is further distinguished by two crucial attributes, both of which make it very difficult to construct adequate explanations. First, a social pattern is explained by the effects or consequences of the pattern, and, second, these consequences must be beneficial and necessary to the proper functioning of society. The traditional approach includes those functionalists who focus on a few aspects of a society at a time and attempt to link one social pattern with one need and there by “explain” the pattern. It also includes those who deal with a complex system of many elements and try to show how these elements are interrelated so as to form an adaptive and consistent system.
Traditional functional analysis can be contrasted with the second type: formal functionalism. This approach is concerned with homeostatic, or equilibrating, systems, and so with feedback and self-regulation. It abandons the two distinctive attributes of the traditional approach. The effects of a trait for some part of the system are used to explain that part and not to explain the trait; and there is no restriction on the kinds of consequences to be considered. They may or may not be beneficial or necessary to society. This approach is called “formal” because it does not include a theoretical orientation or a substantive hypothesis about empirical events. It is a model of kinds of relationships between elements, like a mathematical model.
The distinctiveness of traditional functionalism from other forms of analysis can be readily illustrated. For example, a traditional functionalist explanation of adolescent rebellion in the United States might point to the positive effects or functions of rebellion in securing independence from parents, which is crucial in our society. A non-functionalist explanation might focus on causes of rebellion, such as parental ambivalence about the desirability of being stable and mature. Formal functionalism would produce an interpretation different from both of the above, since it is concerned with equilibrating or feedback systems and not with relationships of one-way effect or cause. It should be noted again that these different types of analysis are usually combined in practice. Thus Parsons’ discussion of adolescence (1942) includes both of the arguments in the above example, as well as many more.
Functional analysis clearly is not distinct in some ways. It does not have any necessary monopoly on the empirical problems of analyzing total societies or comparing institutions cross-culturally, even though historically these fields have been dominated by functionalists. It is also not the case that complex, formal models of functional systems include some logically special types of propositions. Any model that can be communicated to social scientists can be reduced to a combination of simple statements such as: “if a then b “ and ” a = (f)b.” Nor does traditional functional analysis involve a distinctive form of explanation; rather, it must be evaluated according to the general standards of adequate scientific explanations.
Traditional functional analysis
Scientific explanation
The criteria of an adequate scientific explanation are extremely difficult to state precisely, and the philosophers of science continue to struggle with the problem. However, Nagel (1961), Hempel (1965), and other philosophers have stated the criteria in a way that seems close to the actual practice of scientists and that is based on the idea that explaining a statement necesarily involves deducing it from some more general statement or statements. This is the view of explanation that will be used here.
A scientific explanation consists of deducing the proposition to be explained (the explicandum) from some true or plausible general assumptions in conjunction with a statement of initial or antecedent conditions. The statement of initial conditions connects the explicandum to the assumptions. For example, if one wanted to explain the existence of witchcraft among the Navajo, using the two assumptions: (1) “if a society is subjected to externally imposed change then there will be a high level of aggression within the society,” and (2) “if high aggression then witchcraft” then the appropriate statement of initial conditions would be “imposed change has occurred among the Navajo.”
The minimal requirements for an adequate explanation are (1) all the propositions must be clearly stated, that is, stated clearly enough to enable recognition of a negative case; (2) the assumptions must not be empirically false; (3) the statement of initial conditions must be true; and (4) the explicandum must be derivable from the assumptions and initial conditions according to the rules of logic. Formulating an explanation is equivalent to constructing a theory, for if the assumptions are general it will be possible to deduce from them many verifiable propositions and not only the explicandum. In other words, the logical structure of explanation is identical with prediction if the prediction is based on a theory. In both cases one demonstrates that the phenomenon in question had to occur, given the assumptions. The only difference is that in prediction one starts with the assumption and proceeds to deduce testable propositions; in explanation one begins with a tested proposition and searches for some appropriate assumptions from which it can be deduced. Explanation, no less than prediction, requires precise propositions with clear empirical implications.
Traditional functional “explanations” usually fail to meet most of the minimal requirements of an adequate explanation. It is, of course, difficult to construct adequate explanations or theories in social science, regardless of the type of analysis that is used. However, the concepts and strategy of traditional functionalism create formidable problems, and the problems are especially tenacious. Many contemporary functional analyses contain the same errors that have been pointed out for decades in numerous papers (for an excellent presentation of the problems of functional analysis, see Durkheim 1895, chapter 5).
Vacuous functional explanations
The problems of functional explanation are illustrated in Kluckhohn’s sophisticated and complex interpretation of Navajo witchcraft (1944). He uses the three major types of functional arguments: (1) a vacuous explanation in terms of various unrelated, adaptive effects of witchcraft; (2) a more adequate explanation in terms of a functional prerequisite theory; and (3) an interpretation of witchcraft as part of an equilibrating system, using the model of a functional system.
Kluckhohn first explains Navajo witchcraft pat terns simply by demonstrating that they have a variety of beneficial effects. He states that the major question (explicandum) is why witchcraft patterns have survived through a given period of time and that his basic postulate is “that no cultural forms survive unless they constitute responses which are adjustive or adaptive, in some sense, for the members of the society or for the society considered as a perduring unit” (1944, p. 46). This postulate is very similar to Merton’s statement of the three basic postulates of functional analysis: “Substantially, these postulates hold first, that standardized social activities or cultural items are functional for the entire social or cultural system; second, that all such social and cultural items fulfill sociological functions; and third, that these items are consequently indispensable” ([1949] 1957, p. 25). These postulates have been criticized for leading to a static, conservative orientation (Dahrendorf 1958); for if every pattern helps to maintain every other pattern, then the system will not generate conflict or change; and if every social pattern is indispensable, then it is clearly unwarranted and probably dangerous to change any pattern.
The postulates also lead to vacuous explanations like Kluckhohn’s. After stating the major question and basic postulate, Kluckhohn proceeds to de scribe many adjustive and adaptive consequences of witchcraft. They include (1) expressing aggression outside the group, since accused witches are generally outsiders; (2) retarding the accumulation of wealth, shi’-e rich people are frequently accused; and (3) preventing adultery, since fear of witches keeps people from leaving their homes at night.
The form of this type of explanation is:
(1) Social patterns persist if and only if they are adaptive (or functional or fulfill needs);
Such explanations are vacuous because the term “adaptive” or “needs” or “functional” is so loosely defined that all social patterns can be viewed as adaptive (or maladaptive). “Adaptive” is used to include anything that is beneficial to some groups or system. With a little imagination, one can show that any social pattern has adaptive consequences for some states of some systems and maladaptive consequences for some states of the same or other systems. Therefore, the presence or absence of any social pattern can be “explained” by this assumption, and any social pattern can be used to confirm or falsify the assumption. For example, one could focus on the maladaptive consequences of witchcraft and argue that since witchcraft has persisted and is maladaptive, the assumption is falsified.
This problem has been pointed out clearly in many critiques of functional analysis (Merton 1949; Nagel 1957); it is usually stated as the necessity for clear system reference or for defining “functional” or “adaptive” relative to specific states of specific systems. Kluckhohn is aware of the problem, and Merton discusses it explicitly and at length in his famous paper “Manifest and Latent Functions” (1949). Nonetheless, Kluckhohn, Merton in his explanation of city bosses (1949), and many less competent functionalists often point to the functions or beneficial effects of the phenomenon being considered and write as if this constituted an explanation of the phenomenon. These interpretations often contain valuable hypotheses about how the phenomenon is related to other parts of the society, but they typically ignore the maladaptive consequences of the phenomenon and hinder the development of theory by implying that the task of explanation has already been partially accomplished.
The form of functional explanations
Kluckhohn’s second line of argument in explaining Navajo witchcraft is based on the beginnings of a theory of functional prerequisites which limits the relevant consequences of witchcraft to the effective management of anxiety and hostility. The outline of his argument, leaving out many complications, is (1) if a society is to survive as an integrated system, it must provide a relatively non-disruptive way of expressing or dealing with hostility and anxiety; (2) hostility and anxiety can be managed through witchcraft and related scape-goating behavior or through withdrawal, passivity, sublimation, or other functionally equivalent patterns; (3) because of various conditions of Navajo society, the functional equivalents of witchcraft cannot be the main ways of dealing with anxiety and hostility; and finally (4) Navajo society has survived. From these assumptions (1 and 2) and initial conditions (3 and 4), Kluckhohn derives his explicandum: Navajo witchcraft patterns have survived.
The form of the argument is represented in the following scheme:
(1) If a society survives (A), then hostility is managed (B);
(2) If hostility is managed (B), then witchcraft (C) or a functional equivalent (D) is present;
(3) The functional equivalents of witchcraft (D) are not present in Navajo society;
Kluckhohn’s argument here is one of the most sophisticated and precise attempts to construct a general functional explanation. It also illustrates one form of an adequate functional explanation, which henceforth will be referred to as the paradigm of functional explanations: (1) if a system survives or is integrated, then a functional prerequisite is fulfilled; (2) if this prerequisite is fulfilled, then pattern X or its equivalent is present in the system; (3) the functional equivalents are not present in system S; (4) S has survived or is integrated, and therefore X is present in S. There are other forms of explanation that use the basic concepts of functionalism, but this form seems to be implicit in most traditional functional analyses.
Let us first consider the simplest question: the logic of the explanation. Substituting symbols for the phrases in Kluckhohn’s argument, it is apparent that the logic is sound: (1) if A then B; (2) if B then C or D; (3) D is false; (4) A is true, and therefore C is true. Hempel (1959) has argued that in the typical functional explanation the second assumption is usually stated in reverse. Ignoring the functional equivalent term D for purposes of clarity, this means changing assumption (2) to “if C then B.” The argument is then logically incorrect and contains the fallacy of affirming the consequent: if A then B; if C then B and A is true; therefore C is true. However, Hempel has misinterpreted the functionalists, which is easy to do given the discursive-essay style of most functional explanations.
In logic, the statement “if A then B” or “A ⊃ B” is identical with the statement “if not-B then not-A” (Kemeny, Snell & Thompson 1956). But when a functionalist says, “witchcraft results in the effective management of hostility,” he does not mean “if witchcraft, then management of hostility.” This would be identical to saying, “if hostility is not managed, then there will not be witchcraft,” and most functionalists would not want to say that, since they would hypothesize that managing hostility depends on several factors in addition to witchcraft (and its functional equivalents). What they mean to say is “if witchcraft (or its equivalent) is not present, then hostility is not managed,” which is logically identical with assumption (2) in the above paradigm. In other words, a given social pattern and its equivalents are typically interpreted as necessary but not sufficient conditions for fulfilling a functional prerequisite.
The major problems in Kluckhohn’s explanation concern the basic concepts of functional theory: adaptation or survival, integration, functional prerequisites, and functional equivalents. An adequate theory or explanation requires clear definition of these concepts and direct or indirect verification of the assumptions in which they are interrelated.
However, this requirement is so difficult that almost all functionalists, including Kluckhohn, fail to meet it. It is so difficult that at this stage of the development of social science it might be advisable to abandon the attempt to construct functional theories or explanations at the total society level.
System reference
In Kluckhohn’s first assumption, the crucial terms are survival and management of hostility. Neither term is defined precisely enough to allow discrimination between societies that have survived or have managed hostility and those that have not. Some of the problems involved in conceptualizing survival, integration, and functional prerequisites will be discussed in later sections.
The remaining term in the first assumption is “society,” and here again we run into difficulties. Any theory must specify its scope—the units to which it applies and under what conditions, and “society” is a unit that is notoriously difficult to define. Kluckhohn himself states that it is questionable whether the Navajo as a whole constitute an integrated society, and a crucial part of his interpretation refers not to Navajo society, but to the consumption unit of several families. Witchcraft accusations are generally made outside the consumption unit but within Navajo society; there fore the statement that witchcraft manages hostility by directing aggression outside the group is true only of the consumption groups. Yet Kluckhohn’s first assumption refers to a different unit— society. This shift invalidates his argument. His assumptions here are essentially: if a society survives (A), then hostility is directed outside of the society (B); if hostility is directed outside consumption groups (C), then witchcraft is present (D). It is impossible to deduce anything from these assumptions since B and C are not the same.
Functional equivalents
In Kluckhohn’s second assumption we come to the most admirable part of his argument, for, unlike most functionalists, Kluckhohn explicitly faces the issue of functional equivalents. His inability to handle it adequately indicates the difficulty of the task.
The concept of functional equivalents is based on the idea that several social patterns can fill the same function or have the same consequence; to put it colloquially, “there’s more than one way to skin a cat.” The concept has been useful in suggesting the classification of social patterns on the basis of common function. Such classification often reveals important similarities among apparently different social patterns, for example, the similarities between witchcraft and racial prejudice when both are viewed as ways of focusing aggression outside the group. However, the concept leads to great difficulties in functional explanations of a particular social pattern.
Kluckhohn’s second assumption states that witchcraft or its functional equivalent is a necessary condition of managing hostility. The assumption is meaningful (directly or indirectly testable) only if it is possible to distinguish between patterns that are and are not equivalents of witchcraft. This requires listing all functional equivalents, or categorizing them, or stating their distinctive attributes. This must be done in such a way that the assumption is falsifiable—in this instance, so that it is logically possible for a society to manage hostility and not contain witchcraft or any equivalent pattern.
Kluckhohn’s approach is to list the functional equivalents of witchcraft, and he states: “Withdrawal, passivity, sublimation, conciliation, flight and other responses” are alternative adjustive responses to hostile feelings (1944, p. 51). One problem with this and all lists is that there is no reason not to add more items, thereby making assumption (2) virtually unfalsifiable. Kluckhohn also attempts to identify the distinctive attributes of all the patterns on this list: they are all adjustive responses to hostility. But this only makes assumption (2) a tautology: if hostility is man aged, then there will be witchcraft or some other way of managing hostility.
The need for theory. After defining the functional equivalents of a pattern, another task remains. In order to explain why a particular social pattern is present instead of another, functionally equivalent pattern, it is necessary to develop a set of assumptions or a theory that specifies which pattern will occur under which conditions. Kluckhohn implicitly states such theories as he eliminates the functional equivalents on his list on the basis of specific conditions of Navajo society or its environment. For example, he says that withdrawal is of limited effectiveness because Navajo “types of shelter and the needs for co-operation for economic ends sharply limit the privacy of individuals” (1944, p. 52). Kluckhohn is moving in the right direction here, but he is clearly a long way from formulating a clear and explicit theory about the conditions determining the occurrence of a particular social pattern.
It should be noted that if such theories are developed, functional explanations would be expanded to include the cause of origin of social patterns and the cause of their persistence (Bredemeier 1955). Most traditional functional explanations of a social pattern X refer to the causes of X only if there is a reciprocal relationship in which X causes itself, that is, if X has certain effects on Y that in turn cause X. In the paradigm of functional explanations, social pattern X is explained by its effects and not by its causes, “cause” here meaning the necessary or sufficient antecedent conditions for pattern X. But the paradigm can be expanded to include assumptions about the conditions determining the presence of a particular pattern, following Kluckhohn’s procedure, and these assumptions would probably take the form of stating the causes of the origin and persistence of pattern X.
Such an expansion has the additional advantage of making it easier to include social change in traditional functional analyses. If one excepts the concept of functional differentiation (Parsons 1964), which has limited applicability, it must be admitted that traditional functional analysis is not very useful in dealing with change. However, with the specification of antecedent conditions determining particular social patterns, it would be possible to predict the change from one social pattern to another, functionally equivalent pattern.
Kluckhohn’s explanation of Navajo witchcraft illustrates two major problems in traditional functional explanations: consistent system reference and functional equivalents. Other traditional explanations will be considered in the following discussion of the concepts of functional prerequisites, adaptation, and integration. We will return to Kluckhohn in the discussion of models of functional systems.
Functional prerequisites
The remaining problems in traditional functional explanations concern the definition of the concepts in the first assumption of such explanations: if a system survives or is integrated, then a certain functional prerequisite is fulfilled.
The best-known attempt to formulate a list of needs or prerequisites is that of Aberle and his co-workers (1950). They start with a definition of the survival of a society, which includes biological survival of its members and the absence of general apathy, and then state nine functional prerequisites, such as regulating affect and con trolling disruptive behavior. The list is presented as a tentative set of cross-cultural categories, an improvement on the standard chapter titles in monographs.
This type of list is not very helpful in constructing functional explanations, because it does not limit the domain of functional prerequisites. The needs are so broad that almost any social process can be seen as contributing to at least one of them, and, being a list, there are few constraints on adding more needs. But a limited set of prerequisites is essential to functional explanation; without it, an imaginative investigator can “explain” any social pattern merely by describing some consequence or effect of the pattern and arguing that this consequence is a functional prerequisite.
There are two main approaches to arriving at a limited set of prerequisites. One is to construct a set of prerequisites and empirically test whether they have been fulfilled in surviving societies, while other conceivable prerequisites have not. The second approach is to devise a theory that includes an assumption specifying social prerequisites. The assumption would not be directly testable. It would have to be combined with other propositions stating the empirical implications of (1) asserting that all societies must meet these prerequisites or (2) classifying social patterns according to the pre requisites they meet.
Parsons (1961) is using the second approach when he asserts that all systems must meet the prerequisites of latent pattern maintenance, integration, goal attainment, and adaptation. He is not making an empirical statement but is proposing that it would be fruitful to categorize and interpret social structures in terms of their contributions to these functions. His system is not directly useful in explaining social patterns because, as yet, he has not developed a clear set of propositions that specify the empirical relevance of placing a structure in one of the four categories as opposed to another. This crucial step was made in Bales’s general theory of functional prerequisites (1950). He specifies two basic functions—adaptation and integration—and supplements this nonempirical assertion with a complex set of propositions about the empirical processes that accompany each function. For example, adaptation is associated with increasing division of labor, stratification, and formality, while the opposite characterizes integration. This suggests specific predictions about the attributes of organizations oriented to adaptation, or of task leaders as opposed to expressive leaders, or of small groups in the adaptive phase. Although Bales’s theory still awaits precise formulation, it is a rare example of a theory that is both general and functional and that leads to adequate explanations or precise predictions [seeInteraction, article onInteraction Process Analysis].
Adaptation and integration
The difficulty in defining the prerequisites for the adaptation or integration of social systems is based in part on the ambiguity of the concepts “adaptation” (or “survival”) and “integration.” These concepts are crucial to the first assumption in the paradigm of functional explanations: if a system survives or is integrated, then certain functional prerequisites are fulfilled. The assumption requires a definition that will discriminate between systems that are adapted or integrated and those that are not; and this has proved to be difficult. The concept of adaptation is further complicated by its association with evolutionary theory.
The difficulty in defining adaptation has long been recognized (Radcliffe-Brown 1952). Adaptation may be equated with physical survival, but this runs into the problem that societies, unlike organisms, rarely fail to survive in the sense of keeping enough of their members alive and together. Most recorded cases of the physical death of societies are the direct result of hostile action by overwhelmingly powerful outsiders who are new to the environment, and this is not the kind of situation with which functionalists are usually concerned. Since almost all societies survive, barring sudden catastrophe, it is impossible to test a statement of the form “X is necessary for survival” if X is universal. It is also impossible to test a statement of the form “X1 is necessary for the survival of social system S1” if X1 has always been present in S1.
Adaptation or survival may also be defined as maintenance of the social organization or structural continuity. Survival in this sense is by no means universal; in fact it might be difficult to find a society that has survived over the last century, i.e., one that has not undergone basic change. This leads to new problems, because functional explanations of the persistence of social patterns require the assumption that the system has survived (statement 3 in the paradigm). Precise definitions of “stability” or “structural continuity” would have to be worked out, and such explanations would be limited to societies that fit the definition of “stable.” Alternatively, adaptation can be defined as ensuring high reproduction rates following the biologists’ revision of Darwin (Simpson 1949). This definition has the advantage of discriminating among societies, being easy to measure, and making it possible to distinguish different degrees of adaptation, as opposed to the all-or-nothing criteria of physical survival. These two later definitions or others probably could be formulated precisely and could be used in specifying the prerequisites of adaptation. If the problem of functional equivalents could be solved, it would then be possible to explain social patterns according to the paradigm of functional explanations.
However, many interpretations in terms of adaptation do not follow the paradigm. They often use the form of argument discussed in the section on vacuous explanations (patterns persist only if they are adaptive; pattern X is adaptive; there fore X persists). This type of argument is not vacuous if “adaptive” is precisely defined, and some investigators seem to be moving toward a fairly precise definition which equates “adaptation” with “maintaining adequate levels of nutrition and health for most members of the society.” Harris’ interpretation of the sacred cows of India (1965) and other recent papers on economics and ecology in primitive societies (Leeds & Vayda 1965; Suttles 1960) are examples of this kind of argument, in which a social pattern is interpreted by showing how it helps to ensure an adequate food supply. These interpretations are unobjectionable if they are viewed as illustrations of a specific hypothesis, such as Harris’ hypothesis that here is a “cross-cultural tendency to maximize production of food” (1959, p. 194). However, the interpretations are often presented as if they were explaining a social pattern in terms of the assumption that “social patterns persist only if they are adaptive.” Adequate explanations require plausible assumptions, and the plausibility of this assumption rests on shaky grounds.
First, the assumption is plausible if one accepts the idea that there is a process of selection of the fittest social patterns or societies. But I doubt that many social scientists imagine, for example, that some early societies developed the family and therefore survived, while others did not and there fore died out, or that some Northwest Coast Indians developed the potlatch and survived, while the others became extinct. One alternative and somewhat more plausible notion is that some groups developed distinct methods of socialization, or the potlatch, and these groups reproduced faster than the others and eventually dominated the society. However, I know of no data that clearly demonstrate that either of these assumed processes ever occurred except in the few situations where extinction through disease or military aggression was imminent. Another alternative idea is that adaptive social traits are more widespread than less adaptive traits because the weak tend to imitate the strong (Dore 1961). Social patterns may diffuse to weaker groups from groups with more resources and power, that is, with patterns that are more adaptive.
The assumption that only adaptive traits persist is also plausible if restricted to the manifest functions of social patterns, that is, to the effects of the pattern that are recognized and intended by members of the society (Merton 1949). If the decision makers of a society are aware of the adaptive value of a social pattern, this will probably influence whether they want to maintain the pattern. However, the crucial variables here are the decision makers’ beliefs about the adaptive or integrative consequences of the pattern. These events are a subset of the class of goal-oriented behavior; people behave so as to maximize certain goals and values, and one of these goals is usually successful adaptation to the environment. In these cases, it would be better to abandon functional terminology for a more cognitive, actor-oriented approach.
The concept of integration is just as problematic as that of survival. Integration is defined in a wide variety of ways and often masks value judgments or unnecessarily complicates a clear nonfunctional statement. This section will focus on the difficulties in defining integration so that it can be used in traditional functional explanations. Many analyses involving integration or adaptation supplement or replace the traditional approach with the use of models of self-regulating systems, and this aspect of functionalism will be considered later in the discussion of formal functionalism.
There are two major approaches to defining integration. One draws on system theory and defines integration in terms of interdependence and frequent interaction among the members of a bounded unit. Although the approach leads to a relatively precise and manageable definition, it excludes many aspects of the intuitive notion of integration. The second approach draws on the ideas of congruency and fit or the absence of conflict, strain, or anxiety. The notion of fit is extremely vague. The statement that pattern X fits with pattern Y means only that there is some relationship between X and Y. The type of relationship is often left unspecified.
The typical functionalist approach to the concept of integration is to equate fit with lack of conflict. Four kinds of definitions in terms of absence of conflict can be distinguished. The first defines integration as logical-meaningful consistency among social patterns (Sorokin [1937-1941] 1962, vol. 1, pp. 15, 23). Integration means that similar themes, values, or principles are used in all spheres of social life. One ambiguity in this definition is the meaning of absence of integration. It could mean conflict within the cultural system; however, many functionalists seem to mean conflict within individuals, and they make the dubious assumption that logical inconsistency among the beliefs will produce psychological conflict. The second definition of integration focuses on the absence of aggression and exploitation of the powerless among members of a social system. An integrated system is marked by cooperation, consensus on mutual rights and duties, and loyalties that crosscut the subgroups within the system. The third definition focuses on the absence of conflicting demands on individuals as to the appropriate behavior or feelings in a specific situation; and the fourth definition concerns the fit between an individual’s motivation and the behavior required by the social system.
The obvious general problem in using any of these definitions is that a functional explanation must assume that the society being considered is integrated. This is proposition (3) in the argument: (1) if the system is integrated, then need N is filled; (2) if N, then pattern X; and (3) the system is integrated; therefore X. This limitation is implicitly recognized by functionalists, for they tend to shift to nonfunctional explanations when interpreting social patterns that result in increased conflict. One example of such a shift is Parsons’ discussion of the female role in the United States (1942).
It seems very unlikely that any society or large social system is perfectly integrated, regardless of which definition is used. Therefore, integration needs to be defined in a relative way: if a system has Y amount of integration, then N will be fulfilled. This is not an impossible task but a difficult one. It requires extensive theoretical and empirical work on the distribution and implications of different levels of integration, the prerequisites for a given level of integration, and the functionally equivalent social patterns that fulfill each prerequisite. Until some such specification of integration is made, the term cannot be used in explanations [seeIntegration, article onSocial Integration].
Alternative approaches
Traditional functionalism has been defined as those analyses that use certain concepts (adaptation, integration, functional prerequisites, and equivalents) and that explicitly or implicitly use the paradigm of functional explanation in which social patterns are explained by their effects. Because of the many problems discussed above, few functionalists succeed in formulating adequate explanations or theories. If some of the concepts are abandoned and the paradigm is changed, then alternative types of analysis are revealed that have a better prospect of leading to adequate explanations. Although some of these alternatives are frequently used, they are not labeled as “functional analysis” as often as the traditional approach.
One frequently used alternative is to turn functional explanations around, that is, to reverse the explicans and the explicandum in the paradigm of explanation so that phenomena are explained by their causes, not by their effects or functions. Instead of explaining a social pattern by deducing it from the assumption that the society has survived, one explains the survival of the society by the patterns whose effect is to ensure survival. For example, the statement that “witchcraft lessens in-group aggression” is used to explain rates of aggression, not witchcraft. This approach, combined with a focus on systems other than total societies, includes studies of the prerequisites for the survival of marriages, Utopias, associations, or other social systems. It also includes studies of the prerequisites for the structural continuity of social systems, that is, of the conditions for maintaining a certain type of social organization in a specified environment. This is the approach described by Levy in his outline of “structural-functional requisite analysis” (1952, pp. 34 ff.).
A second possible change of traditional functionalism, as mentioned above, is to consider smaller systems under specified conditions rather than total societies in general. One great advantage in this procedure is that systems such as therapy groups, unions, or lineages often fail to survive or seem to have low levels of integration; therefore, statements about the prerequisites for survival or integration can be tested. In addition, the possible functional prerequisites and equivalents are probably more limited, and, finally, there would be less difficulty in maintaining consistent system reference because there are fewer subsystems than in a society.
A more radical change is to abandon the focus on adaptation and integration and to consider the prerequisites of any state of a system—such states as high suicide rates, for instance, or high or low levels of in-group aggression. With this revision, we have eliminated the distinguishing attributes of functional analysis: explaining patterns by their beneficial (adaptive or integrative) effects. Functionalism so defined includes all analyses of the prerequisites, conditions, or determinants of any attribute or state of human behavior. Very little information is transmitted by labeling all these analyses as “functional,” and very few social scientists use the term in this way.
However, there is a type of functionalism that does not involve explaining patterns by their beneficial effects and is a distinct form of analysis. This is formal functionalism, which is distinguished by the use of models of self-regulating or functional systems. Traditional functional analyses contain many examples of such systems: of reciprocal relationships, of one pattern varying so as to maintain another within a given range, and of complex feedbacks and causal chains that involve many parts of a system. But some traditional functional explanations do not utilize this type of model. For example, the explanation of the incest taboo in terms of its function of creating alliances among families, or the explanation of rites of passage by their function of reducing conflict between old and new roles, does not involve reciprocal or self-regulating relationships. Therefore, the theoretical orientation of traditional functionalism and the model of self-regulating or functional systems should be kept distinct.
The assumption that models of functional systems fit society is the basis of some of the concepts and procedures of traditional functional analysis. The stress on latent functions—effects that are unintended and unrecognized—is based in part on the assumption that models of complex, self-regulating systems apply to many social systems. Therefore, a change in one social pattern may affect seemingly unrelated patterns or, because of compensating changes in other social patterns, may have effects that are opposite from those intended —or no effects at all. Another instance is the functionalist postulate that all social patterns “are functional for the entire social or cultural system” (Merton [1949] 1957, p. 25). This is another way of saying that the whole social system forms some sort of self-regulating system. These are premature judgments about the extent to which models of functional or self-regulating systems accurately represent the relationships found in social systems. The applicability of the models is an empirical question for which the relevant data have not been collected.
Formal functionalism
Formal functionalism consists in the attempt to construct models that describe how the units of a system are interrelated so as to maintain each other or some other state of the system. In contrast to traditional functionalism, it contains no theoretical orientation and has no empirical content. The models must be supplemented by theory and research that specify the empirical units or variables that correspond to the abstract elements in the models.
As an illustration of the use of such models, Kluckhohn’s analysis of Navajo witchcraft is useful again. Kluckhohn supplements his structural analysis with an examination of changes over time. He describes the deprivation and anxiety of the Navajo during their confinement at Fort Sumner and several decades later, during a period of lesser stress that was caused by the livestock reduction program. He notes that warfare was impossible for the Navajo after Fort Sumner, that witchcraft accusations and executions were most frequent shortly after Fort Sumner, and that accusations also increased after the inception of the livestock program. In his interpretation, Kluckhohn assumes, first, that deprivation and frustration cause increased hostile feelings, and, second, that these impulses will cause in-group aggression un less the hostility is directed outside the group; warfare and witchcraft are viewed as social pat terns that direct aggression to outsiders. Given these assumptions, the changes in frustration, hostility, aggression, and witchcraft can be accurately represented as a self-regulating system directed to maintaining a low level of in-group aggression. When the environment (the United States government) imposes increased frustration, hostile impulses increase, but this is followed by increased witchcraft accusations, given the impossibility of warfare, and this aggression to outsiders reduces hostile impulses back to a level where they no longer threaten the maintenance of low in-group aggression.
Nagel’s model
The model used in Kluckhohn’s analysis has been precisely described by Nagel (1957). Starting with Merton’s article “Manifest and Latent Functions” (1949), Nagel rigorously describes the model that seems to be implicit in functional analysis in the social sciences.
A functional system, according to Nagel’s definition, is made up of two types of variables: G’s and state coordinates. G is the property of the system that is maintained or is stable. State coordinates determine the presence or absence of G and may include parts of the system’s environment. The values of the state coordinates vary to such an extent that the maintenance of G is threatened, but when one exceeds the “safe” limits for G, the other(s) compensates and G is maintained. Such a system of G and state coordinates may be called functional with respect to G, and the state coordinates may be described as having the function of maintaining G.
For example, a small task-oriented group could be treated as a functional system (Bales 1950). Let G be the solution of the group’s task or problem. Let the state coordinates be task-oriented activity and emotionally supportive activity. If these three variables can be usefully treated as a functional system, then: (1) problem solution is dependent on task-oriented activity and emotionally supportive activity; (2) at certain times, there will be such a preponderance of task-oriented activity that problem solution will be threatened because of decreased motivation or resentment over following others’ suggestions—at these times emotionally supportive activity will increase, and problem solution will no longer be threatened; (3) at certain times, there will be such a preponderance of emotionally supportive activity that problem solution will be threatened. At these times task-oriented activity will increase to maintain problem solution or G.
By definition, more than one state of the system leads to maintenance of G. Thus, in the preceding example, eventual problem solution might result both from initially high task-oriented activity and low supportive activity, followed by increased supportive activity, and from initially low task-oriented activity and high supportive activity, followed by increased task activity. In a functional system there is more than one combination of the values of certain parts of the system which will result in the same trait or will have the same consequences (maintenance of G). This is one way of stating the notion of functional equivalents.
The other crucial elements in Nagel’s model concern the limitations of the variation of state coordinates. Three kinds of limits are relevant. First, there are the limits dictated by physical reality. For example, the amount of interaction in a given period cannot exceed the maximum number of acts possible in that time period. Second, within the limits of physical reality there are limits determined by the definition of the system under consideration. If a property is used to define a system, one cannot analyze conditions under which this property disappears. The third and most important set of limits are those beyond which compensation is impossible and G ceases to exist. The potential stability of a given G in a specific functional system can be expressed in terms of the discrepancy between two of these sets of limits: the possible variation for each state coordinate and the limits of variation within which compensation by other state coordinates will occur. G becomes less stable as the discrepancy between these two ranges increases.
It should be noted that, contrary to the strategy of traditional functionalism, stability of G is not assumed. The traditional approach is to assume the stability of G (integration or survival) and explain state coordinates in terms of their efficiency in maintaining G; states of the system are explained by their effects. In Nagel’s model, the stability or disappearance of G is explained by the extent to which state coordinates compensate for each other’s variation; states of the system are explained by their causes. A rough definition of “cause” is being used here: A causes B if the occurrence of B depends on the previous occurrence of A.
Some functionalists have suggested that the notion of cause is not applicable to complex social systems, but this seems to mean only that social patterns are often caused by a complex set of interrelated factors and not by one simple factor. For instance, in the small group example, task solution depends on: (1) a certain amount of task-oriented activity; (2) emotionally supportive activity; (3) mechanisms which cause a switch from one type of activity to the other when one type is dangerously low; and perhaps additionally on (4) some reciprocal relationship whereby group perception of progress in solving the task is necessary to maintain the two types of activity. This is a complex system, but these four factors can be isolated as the determinants or causes of task solution. The unwillingness to use the term “cause” often indicates only that the analysis is incomplete and that it is not yet possible to clearly define the relevant parts of the system and to specify the direction and magnitude of the relationships among these parts.
To summarize Nagel’s model, a functional system is one that satisfies the following conditions: (1) the system can be analyzed into a set of inter dependent variables or parts; (2) the values of these variables determine whether or not a certain property G will occur in the system; (3) there are limits on the variation of state coordinates such that variation within the limits is followed by compensating variation of other state coordinates, and G is maintained; (4) variation beyond these limits is not compensated for by other state coordinates, and G disappears.
The type of analysis suggested by Nagel’s formalization would probably avoid many problems of traditional functional analysis. As in any formal model, this one specifies the entities that must be clearly and consistently defined: the system, goal state, state coordinates, and the limits on their variation. Any state of any system might be viewed as a G state in a functional system; therefore the problematic concepts of survival and integration are not necessary, although they could be used as the G state (see, for example, the series of analyses in Leeds & Vayda 1965, in which the G state is the survival of the population in the face of variation in the environment). The model also encompasses various kinds of change: (1) goal states that are moving equilibria; (2) compensating changes in state coordinates; (3) compensating changes in subsystems that are treated as state coordinates of a higher level system; and (4) disappearance of G resulting from state coordinates exceeding the limits of compensation (Cancian 1960-1961).
The contribution of formal functionalism to the explanation of social patterns is quite different from that of traditional functionalism, because a formal model lacks substantive propositions and therefore, by itself, cannot explain anything. If the correspondence between the model and the parts of a specific system are demonstrated, this would constitute, at the least, a precise description of the system. The description might also be called a “little theory” (Cancian 1965, chapter 13)—a “theory” because it would explain and predict how variance in one part of the system would affect other parts of the system, and a “little theory” because the propositions would be limited to one specific system. In order to construct a general theory or explanation, it would be necessary to add an assumption stating that the model fit a certain class of systems, of which the specific system was a member.
Specification of functional equivalents. The major problem left unresolved in this model is that of functional equivalents. An example illustrates the difficulty. Assume that one is investigating the system of relationships between North American college students and their parents and that one attempts to explain independence from parents (G) by the state coordinates of peer-organized rebellion against authority, communication with parents, and other variables. How would one deal with the different types of organized rebellion, such as panty raids, burning draft cards, or picketing for civil rights? The approach suggested by the model would be to consider only the primary function of these activities for the goal state of independence. Panty raids, burning draft cards, etc., would be ranked according to the degree of rebellion involved, and all other differences among the activities would be ignored. The advantage of this method is that it results in a general explanation that includes college students participating in various types of peer-organized rebellion. Note, however, that the idea of functional equivalents then has no special meaning; it merely indicates that a general variable takes various empirical forms. For example, reinforcement can take the form of food or water.
As it is usually used, “functional equivalents” does have a special meaning. There is often an implicit assumption that if one social pattern disappeared, a functionally equivalent pattern would appear or would be more likely to appear. This aspect of functional equivalents can be dealt with by treating the equivalent patterns as a subsystem, with the G of fulfilling the function of the patterns. In this subsystem, as one rebellious activity decreases, some other activity increases so as to maintain peer-organized rebellion, if the subsystem is a functional system with respect to rebellion. An examination of this subsystem would indicate the conditions (limits of variance) under which a functional equivalent would take the place of a disappearing activity. The problem of predicting whether a specific rebellious activity would be maintained within certain limits could be handled by considering the subsystem in which the G state is that activity and the state coordinates are the variables that maintain the activity.
Alternative models
Every functional system must contain certain minimal attributes: (1) variable A that is regulated or maintained within a range; (2) variable B that does the regulating; (3) a mechanism that communicates the variation of A to B. These minimal requirements are exemplified by the famous system: room temperature-furnace-thermostat, and by various systems in the social science area (Lennard & Bernstein 1960). Note that the existence of feedback, or reciprocal or mutual causation between A and B, is not enough. The feedback must be of a special kind; it must operate so as to maintain at least one of the variables with in a given range. For example, according to many observers, the system of “juvenile crime-reform school” is not a functional system with respect to maintaining low crime rates. There is a reciprocal relationship; reform school does feed back to crime rates, but its effect appears to increase crime.
This minimal functional system can be elaborated in many ways, resulting in alternative functional models. The primary alternative to Nagel’s model is based on changing his notion of a G state. Nagel’s system is hierarchical, that is, it contains several variables that interact so as to maintain some state of the system within a given range (G), and the effect of G on these variables is not considered. In the alternative, horizontal model, the effects of the variation in G on the state coordinates are included in the analysis. Both G and the state co ordinates vary so as to maintain each other within a certain range; therefore there is no reason to discriminate between the two types of variables. There are many illustrations of such systems in the area of complementary roles or interpersonal relations. Thus, an authoritarian professor and a dependent student or a sadistic-masochistic couple often maintain each other in their respective roles.
Within the general framework of a hierarchical or horizontal model, many specific varieties of functional systems can be constructed. The system may be simple and include only the minimal attributes of a functional system, or it may be very complex and contain many parts and subsystems, like Mc-Cleery’s description of a prison system (Barton & Anderson 1961) or Parsons’ model of the social system (1961).
The system may operate so as to maintain one, or several, or all of its components within a given range. Systems also differ in the amount of variation necessary to set off compensating processes, in the limits beyond which compensation is impossible, and in the time required for compensation. When it does occur, the compensation may be too much or too little or out of phase (Scientific American 1955). Compensation for the variance of one component may take the form of pushing back that component within a range that is less threatening to the system, or it may consist of a proportional change in other components. There are many possible complexities and variations in functional systems, some of which are described in the literature, especially in the fields of economics, cybernetics, and systems engineering.
Using models
There are two major procedures for utilizing functional models in constructing theories or general explanations. One approach is to begin with a careful investigation of a single system. After describing the major components of the system and how they are interrelated, the model implicit in the description can be examined. Some previously developed model may fit the description and produce new statements about the system, or it may be necessary to construct a model. The next step is to verify the “fit” by examining the particular system over time, observing variations in the components, and seeing if there is, in fact, compensation that maintains some components within a given range. The final step is to try to generalize this “little theory” to other systems by examining the boundary conditions or parameters of the systems and by restating the components in more general terms. An example of this approach is Sweet’s analysis of how Bedouin camping units maintain a certain size camel herd by raiding, migration, splitting up the social unit, and other means (1965). Her description could probably be restated as a theory of the bedouin camel complex, that is, as a limited set of propositions about herd size, raiding frequency, etc., and the relations between them. This theory could be tested against the historical record of bedouin behavior. If verified, it might be possible to restate the “little theory” in more general terms, so that, given certain conditions, it applies to camel herders in general, or pastoral peoples, or people dependent on material goods whose availability fluctuates widely.
Another approach is to start with an abstract theory about a general class of phenomena that are interrelated so as to form a functional system. This procedure is exemplified in the research on balance theory in social psychology (Heider 1958). The basic idea in the theory is that individuals will vary their sentiments and evaluations so as to maintain a balanced state. Thus, if two individuals like each other and one of them likes a third person but the other does not, then the individuals will move into a balanced state, that is, they will both come to like or dislike the third person, or they will change their feelings about each other. The relationships among components in this theory correspond to the relationships in mathematical graph theory, and the application of this mathematical model has made the theory more precise and has yielded new propositions (Berger et al. 1962).
These two approaches start with a specific or general empirical system and then examine the model or set of relationships in the system. An alternative would be to start with a model and then search for some empirical systems that fit it. This approach would be much less likely to succeed than the first two because the investigator might not encounter any phenomenon of interest that fits the model, and then the effort of constructing the model would be completely wasted. However, there are some guidelines for locating empirical systems that operate like functional systems.
A primary indicator of the applicability of a functional model is the persistence of a social pattern despite forces that tend to destroy the pattern. For example, the apparent persistence of European gypsies suggests the presence of compensating processes, as does the persistence, despite reform movements, of deviant patterns such as city bosses or prostitution. Of course, it may be that such patterns persist because of some constant, powerful factor or factors, but it would probably be fruitful to try to analyze the persistence in terms of a functional model. Another relevant attribute is the presence of inherently runaway processes that are kept within limits. Some of the social processes that might increase indefinitely, but often do not, are the exploitation of followers by powerful leaders or the rejection of competing loyalties by a sexually attracted couple (Slater 1963). Still another attribute is the presence of reciprocal relationships—that is, of regular sequences of social patterns that return to the point of origin. The most obvious class of phenomena in this category is self-other patterns, or complementary roles, but reciprocal relationships also occur in large social systems, as shown by Leach’s analysis of cyclic change between two political systems in Burma (1954). These various attributes provide clues as to the applicability of a functional model, but in every case the appropriateness of a functional model and the fit of different types of models are questions that must be empirically investigated.
The usefulness of models in general cannot be disputed, and the importance of constructing models will surely increase as social scientists attempt to deal with complex systems of variables in a precise manner and as they shift from monocausal theories to the notion of interdependent systems. The usefulness of models of functional systems in particular depends on the extent to which the phenomena investigated by social scientists exhibit equilibrating or self-maintaining processes. Formal functionalism is based on the most valuable part of the research strategy of functional analysis: an attempt to view behavior patterns as mutually reinforcing parts of an equilibrating system. Whether or not this strategy will succeed in yielding general and verified theories is an empirical question that will not be settled until this strategy is given a thorough trial.
Francesca M. Cancian
[Directly related are the entriesIntegration; Interaction, article onInteraction Process Analysis; Systems Analysis. Other relevant material may be found inModels, Mathematical; Simulation; Social Structure; Sociology, article onThe Development of Sociological thought; and in the biographies ofDurkheim; Henderson; Malinow-ski; Radcliffe-Brown; Weber, Max.]
BIBLIOGRAPHY
Aberle, David F. et al. 1950 The Functional Prerequisites of a Society. Ethics 60:100-III.
Bales, Robert F. 1950 Interaction Process Analysis: A Method for the Study of Small Groups. Reading, Mass.: Addison-Wesley.
Barton, Allen H.; and Anderson,, Bo 1961 Change in an Organizational System: Formalization of a Qualitative Study. Pages 400-418 in Amitai Etzioni (editor), Complex Organizations: A Sociological Reader. New York: Holt.
Berger, Joseph et al. 1962 Types of Formalization in Small-group Research. Boston: Houghton Mifflin.
Black, Max (editor) 1962 The Social Theories of Talcott Parsons: A Critical Examination. Englewood Cliffs, N.J.: Prentice-Hall.
Bredemeier, Harry C. 1955 The Methodology of Functionalism. American Sociological Review 20:173–180.
Buckley, Walter 1957 Structural-Functional Analysis in Modern Sociology. Pages 236-259 in Howard Becker and Alvin Boskoff (editors), Modern Sociological Theory in Continuity and Change. New York: Dryden.
Cancian, Francesca M. 1960-1961 Functional Analysis of Change. American Sociological Review 25:818-827; 26:930–931.
Cancian, Frank 1965 Economics and Prestige in a Maya Community: The Religious Cargo System in Zinacantdn. Stanford Univ. Press.
Collins, Paul W. 1965 Functional Analyses in the Symposium Man, Culture, and Animals. Pages 271-282 in Anthony Leeds and Andrew P. Vayda (editors), Man, Culture, and Animals: The Role of Animals in Human Ecological Adjustments. Washington: American Association for the Advancement of Science.
Dahrendorf, Ralf 1958 Out of Utopia: Toward a Reorientation of Sociological Analysis. American Journal of Sociology 64:115–127.
Davis, Kingsley 1959 The Myth of Functional Analysis as a Special Method in Sociology and Anthropology. American Sociological Review 24:757–772.
Davis, Kingsley; and Moore, Wilbert E. 1945 Some Principles of Stratification. American Sociological Review 10:242–249.
Dore, Ronald P. 1961 Function and Cause. American Sociological Review 26:843–853.
Durkheim, Émile (1895)1958 The Rules of Sociological Method. 8th ed. Edited by George E. G. Catlin. Glencoe, III.: Free Press. → First published as Les règles de la méthode sociologique.
Fallding, Harold 1963 Functional Analysis in Sociology. American Sociological Review 28:5–13.
Firth, Raymond 1955 Function. Yearbook of Anthropology [19551:237–258.
Goldschmidt, Walter 1966 Comparative Functionalism. Berkeley and Los Angeles: Univ. of California Press.
Gouldner, Alvin W. 1960 The Norm of Reciprocity: A Preliminary Statement. American Sociological Review 25:161–178.
Harris, Marvin 1959 The Economy Has No Surplus? American Anthropologist New Series 61:185–199.
Harris, Marvin 1965 The Myth of the Sacred Cow. Pages 217-228 in Anthony Leeds and Andrew P. Vayda (editors), Man, Culture, and Animals: The Role of Animals in Human Ecological Adjustments. Washington: American Association for the Advancement of Science.
Heider, Fritz 1958 The Psychology of Interpersonal Relations. New York: Wiley.
Hempel, Carl G. 1959 The Logic of Functional Analysis. Pages 271-307 in Llewellyn Gross (editor), Symposium on Sociological Theory. Evanston, III.: Row, Peterson.
Hempel, Carl G. 1965 Aspects of Scientific Explanation, and Other Essays in the Philosophy of Science. New York: Free Press.
Homans, George C. 1964 Bringing Men Back In. American Sociological Review 29:809–818.
Johnson, Harry M. 1960 Sociology: A Systematic Introduction. New York: Harcourt.
Kemeny, John G.; Snell, J. Laurie; and Thompson, Gerald L. (1957) 1962 Introduction to Finite Mathematics. Englewood Cliffs, N.J.: Prentice-Hall.
Kluckhohn, Clyde 1944 Navaho Witchcraft. Harvard University, Peabody Museum of American Archaeology and Ethnology, Papers, Vol. 22, No. 2. Cambridge, Mass.: The Museum.
Leach, Edmund R. 1954 Political Systems of Highland Burma: A Study of Kachin Social Structure. London School of Economics and Political Science; Cambridge, Mass.: Harvard Univ. Press.
Leeds, Anthony; and Vayda, Andrew P. (editors) 1965 Man, Culture, and Animals: The Role of Animals in Human Ecological Adjustments. Publication No. 78. Washington: American Association for the Advancement of Science.
Lennard, Henry L.; and Bernstein, Arnold 1960 The Anatomy of Psychotherapy: Systems of Communication and Expectation. New York: Columbia Univ. Press.
Levy, Marion J. 1952 The Structure of Society. Prince ton Univ. Press.
Malinowski, Bronislaw 1931 Culture. Volume 4, pages 621-645 in Encyclopaedia of the Social Sciences. New York: Macmillan.
Martindale, Don (editor) 1965 Functionalism in the Social Sciences: The Strength and Limits of Functionalism in Anthropology, Economics, Political Science, and Sociology. Philadelphia: American Academy of Political and Social Science.
Merton, Robert K. (1949) 1957 Manifest and Latent Functions. Pages 21-84 in Robert K. Merton, Social Theory and Social Structure. Rev. & enl. ed. Glencoe, III.: Free Press.
Mills, Theodore M. 1959 Equilibrium and the Processes of Deviance and Control. American Sociological Review 24:671–679.
Nagel, Ernest 1957 A Formalization of Functionalism. Pages 247-283 in Ernest Nagel, Logic Without Metaphysics, and Other Essays in the Philosophy of Science. Glencoe, III.: Free Press.
Nagel, Ernest 1961 The Structure of Science: Problems in the Logic of Scientific Explanation. New York: Harcourt.
Parsons, Talcott 1942 Age and Sex in the Social Structure of the United States. American Sociological Review 7:604–616.
Parsons, Talcott (1961) 1965 An Outline of the Social System. Volume 1, pages 30-79 in Talcott Parsons et al. (editors), Theories of Society: Foundations of Modern Sociological Theory. New York: Free Press.
Parsons, Talcott 1964 Evolutionary Universals in Society. American Sociological Review 29:339–357.
Radcliffe-Brown, A. R. (1952) 1961 Structure and Function in Primitive Society: Essays and Addresses. London: Cohen & West; Glencoe, III.: Free Press.
Radcliffe-Brown, A. R. 1957 A Natural Science of Society. Glencoe, III.: Free Press.
Scientific American 1955 Automatic Control. New York: Simon & Schuster.
Simpson, George G. (1949)1961 The Meaning of Evolution: A Study of the History of Life and Its Significance for Man. New Haven: Yale Univ. Press.
Slater, Philip E. 1963 On Social Regression. American Sociological Review 28:339–364.
Sorokin, Pitirim A. (1937–1941)1962 Social and Cultural Dynamics. 4 vols. Englewood Cliffs, N.J.: Bed-minister Press. → Volume 1: Fluctuation of Forms of Art. Volume 2: Fluctuation of Systems of Truth, Ethics, and Law. Volume 3: Fluctuation of Social Relationships, War, and Revolution. Volume 4: Basic Problems, Principles, and Methods.
Suttles, Wayne 1960 Affinal Ties, Subsistence, and Prestige Among the Coast Salish. American Anthropologist New Series 62:296–305.
Sweet, Louise E. 1965 Camel Pastoralism in North Arabia and the Minimal Camping Unit. Pages 129-152 in Anthony Leeds and Andrew P. Vayda (editors), Man, Culture, and Animals: The Role of Animals in Human Ecological Adjustments. Washington: American Association for the Advancement of Science.
Tumin, Melvin M. 1953 Some Principles of Stratification: A Critical Analysis. American Sociological Review 18:387–394. → See the debate in this issue and in Volume 28, 1963.
Van Den Berghe, Pierre L. 1963 Dialectic and Functionalism: Toward a Theoretical Synthesis. American Sociological Review 28:695–705.