Using Paradox to Build Management Theories PDF

e Academy o/Managemeni fieview, 1989, Vol. 14, No. 4, 562-578 Using Paradox to Build Management and Organization Theories MARSHALL SCOTT POOLE ANDREW H. VAN DE VEN...

e Academy o/Managemeni fieview, 1989, Vol. 14, No. 4, 562-578 Using Paradox to Build Management and Organization Theories MARSHALL SCOTT POOLE ANDREW H. VAN DE VEN University of Minnesota Most contemporary theory construction methodologies attempt to build internally consistent theories of limited scope. Relatively little attention has been paid to the opportunities offered by tensions, op- positions, and contradictions among explanations of the same phe- nomenon. This essay attempts to spell out a set of theory-building strategies to help researchers take advantage of theoretical tensions. Such tensions can be regarded as paradoxes of social theory, and four different modes of working with paradoxes can be distin- guished: (1) accept the paradox and use it constructively; (2) clarify levels of analysis; (3) temporally separate the two levels; and (4) in- troduce new terms to resolve the paradox. These four modes of par- adox resolution are illustrated by application to the action:-.structure paradox in organizational theory. Like most social scientists, organization and will require ways to address paradoxes inherent management theorists are socialized to develop in human beings and their social organizations internally consistent theories. The presence of (Ouinn & Cameron, 1988). contrary or contradictory assumptions, explana- Because organizational theories attempt to tions, or conclusions is often viewed as an indi- capture a multifaceted reality with a finite, in- cator of poor theory building, and theorists are ternally consistent statement, they are essen- encouraged to devote their efforts to carefully tially incomplete. A good theory is, by definition, defined and delimited analyses. The value of a limited and fairly precise picture. It does not rigor and coherence cannot be denied. How- attempt to cover everything and would fail to ever, these qualities are not sufficient to guar- meet the parsimony criterion if it did. Scope con- antee good theories. Ralph Waldo Emerson's ditions are one means of expressing the limita- dictum, "A foolish consistency is the hobgoblin tions of theories. Less evident, but as effective, is of little minds" must also be kept in mind. Social reliance on a limited, carefully prescribed set of science loses an important resource for theory assumptions and explanatory principles. These development if the incompatible or inconsistent assumptions and explanations implicitly state theses which inevitably arise in the study of or- what is relevant and what is not. They deter- ganizations are ignored or are eliminated. mine the operative scope of a theory by speci- There is growing recognition that significant ad- fying what can be explained or understood and vances in management and organization theory what must be regarded as "not of interest for this 562 theory," or as irrelevant. Theories always con- out and work out their relationships. Perrow strain the theorist's field of vision; one of the can- (1986) and Pfeffer (1982) have written excellent ons of good theory construction is to recognize studies in this spirit. these limitations. Contemporary theory construction methods Additionally, researchers are adjured to per- are biased toward consistency. Relatively little fect their theories and to test them. As they do attention has been paid to the resolution of ten- this, there is a tendency for the theory to domi- sions or oppositions. This set of theory-building nate researchers' thinking. The researcher must strategies is proposed to help researchers come focus on the theory, iron out its problems, work to terms with theoretical tensions. out measurement techniques, test the theory Paradox—a term with a long history in phil- and revise it, and defend tests against criticism. osophical and rhetorical studies—is one key to As a result, the theory tends to bind the re- understanding how to work with theoretical con- searcher's judgment. The researcher develops a tradictions and oppositions embedded in com- "trained incapacity" to appreciate aspects not plex traditions. This term has several layers of mentioned in her or his theory. As this progres- meaning, each of which captures some of the sion toward consistency continues, the theory features which make theoretical paradoxes so becomes more and more "perfect," with less interesting and thought provoking. and less correspondence to the multifaceted re- In general parlance, many writers use the ality it seeks to portray. term loosely, as an informal umbrella for inter- An alternative strategy for theory building esting and thought-provoking contradictions of can be proposed: Look for theoretical tensions or all sorts. In this sense, a paradox is something oppositions and use them to stimulate the devel- which grabs our attention, a puzzle needing a opment of more encompassing theories. This solution. strategy requires an exploration of the tradition In rhetorical studies paradox designates a of theoretical debate surrounding important is- trope which presents an opposition between two sues, an identification of alternative or opposing accepted theses. For example, the Elizabethan theories or explanations, and discovery of ways rhetorician Sherry wrote, "He always is an en- of relating, contraposing, or integrating them. emy to his own plans, yet he claims to be a The result will be theories less susceptible to the friend to other men's." Sherry thus questioned limitations of perspective which attend many the trustworthiness of this man by showing a middle range theories. contradiction in his behaviors. The rhetorical This strategy is not a replacement for tradi- paradox is intended to cause the audience to re- tional, univocal theory building, but rather an consider set opinions or to throw into contrast additional arrow for the theorist's quiver. The taken-for-granted presumptions. Its impact researcher consciously and tenaciously pursues stems from its shock value. theoretical inconsistencies, rather than dismiss- In logic, paradox has a narrower, specialized ing them or resigning them to the "theoretical meaning. A logical paradox "consists of two disagreements" category. Rather than regard- contrary or even contradictory propositions to ing each theory as a self-encapsulating whole, which we are led by apparently sound argu- the theorist can play theories off against one an- ments" (van Heigenoort, 1972, p. 45). Taken sin- other, gaining insights from multiple perspec- gly, each proposition is incontestable, but taken tives and comparative analysis. In this view, together they seem to be inconsistent or incom- theories are not statements of some ultimate patible. One famous logical paradox is the Liar, "truth" but rather are alternative cuts of a mul- first studied by the Megaric philosophers around tifaceted reality. Alternative theories give par- 400 B.C. If someone says, "I always lie," how are tial views, and the theorist's task is to sort them we to understand this statement? It seems both 563 true and false. Such great and diverse thinkers doxes. Many such paradoxes have been identi- as Aristotle, Chrysippus, Russell, and Wittgen- fied by Burrell and Morgan (1979), Pfeffer (1982), stein have proposed resolutions for the Liar's Ouinn and Cameron (1988), Smith and Berg Paradox (Neilson, 1967, p. 622; see also Van de (1987), Van de Ven (1983), and Van de Ven and Ven & Poole, 1988). Much effort has been de- Poole (1988). These paradoxes include: the diffi- voted to resolving or understanding paradoxes, culty in reconciling the explanation of behavior because they divulge inconsistencies in our as a function of structural determination with the logic or assumptions. They present opportunities equally strong claim that it is the product of pur- to discover different assumptions, shift perspec- posive action (Burrell & Morgan, 1979; Van de tives, pose problems in fundamentally different Ven & Poole, 1988); good arguments for two in- ways, and focus on different research questions. compatible conceptualizations of organizational All three levels of meaning inform this analy- climate—as an aggregation of individual cli- sis. We are interested in paradox in the lay mate perceptions or as a macrolevel system sense—in the interesting tensions, oppositions, property (Glick, 1985, 1988; James, Joyce, & and contradictions between theories which cre- Slocum, 1988); the question of whether social or- ate conceptual difficulties. Because theory ganizations are fundamentally stable orders or building is a discursive enterprise, rhetorical continuously changing emergents (Burrell & strategies of handling paradox effectively are a Morgan, 1979; Pfeffer, 1982; Weick, 1979); and central concern. Communication and rhetorical the trade-off between the need to establish indi- studies have a long tradition of thought con- vidual identity in groups and the collective na- cerned with the creation of knowledge through ture of group action (Smith & Berg, 1987). Each discourse and with effective expression. Four "side" of these tensions has been advocated or discursive strategies, which will be discussed, emphasized by different theorists, but together are open to theorists interested in paradox. The they form a tradition of theoretical discourse paradoxes in management are not, strictly which is potentially richer than either theory by speaking, logical paradoxes. However, an un- itself. The problem is how best to mine this rich derstanding of paradoxes and ways to work vein of insight. with them is greatly enriched by the philosoph- This is not a trivial problem, as the example of ical tradition. By far the largest body of work on the stability: :change tension illustrates. It is evi- paradoxes can be attributed to logicians, and dent that organizations are admixtures of stabil- the philosophical treatment of paradoxes will be ity and change: Organizations are relatively sta- used as a touchstone for this analysis. ble, enduring features of life, yet when we look Four general methods to address paradox in closely they do not appear stable at all. They are management and organization theories are continuously changing, continuously being pro- suggested. These four methods for addressing duced and renewed by member activities. Nev- paradox are illustrated by applying them to the ertheless, an argument can be made that stabil- action: :structure paradox in organizational the- ity is primary; any change is observable only in ory. contrast to some stable state. Organizational change also can be explained as aberrations from the stable state, as sudden upheavals Four Ways to Address Paradox which disrupt organizational stability. The in- in Organization and corporation of stability and continuous change Management Theories in the same theory poses a paradox, because each is defined as the opposite of the other. Organizational and management theories in-... Hernes (1976) argued that adequate theories volve a special type of paradox—social para must explain stability and change in the same 564 terms. Generally, however, organizational it. The types of paradoxes concerned with here theories have emphasized either stability or are tensions and oppositions between well- change, slighting the other term. As Van de Ven founded, well-reasoned, and well-supported al- and Poole (1988) showed, most organizational ternative explanations of the same phenome- theorists have attempted to incorporate both sta- non. When juxtaposed, they present a puzzle for bility and change, but as specific theories de- the theorist, because each side seems valid, yet velop and are refined, there seems to be pres- they are in some sense incompatible or hard to sure to take one as the primary term and to sub- reconcile. ordinate the other. How can both faces of As these examples show, paradoxes in social organizations be encompassed in the same theories are not strictly logical paradoxes. Social framework? scientific paradoxes tend to be looser: The op- Theorists can attempt to use these and other posing terms are often somewhat vague, and tensions in several ways. They can identify in instead of logical contradictions, tensions and which side of a tension their current work is an- oppositions between incompatible positions chored and then expand their perspective by must be considered. Further, whereas logical addressing criticisms lodged by the other side. paradoxes exist in timeless, abstract thought, so- For example, researchers who assume a rela- cial paradoxes are about a real world, subject to tively stable organizational backdrop for their its temporal and spatial constraints. This opens work might attempt to enlarge their theories by the possibility of dealing with social paradoxes taking into account the criticisms implied by not only through logical resolutions, but through Weick's (1979) perspective. Alternatively, theo- taking into account the temporal or spatial na- rists can start at the concrete level, identify ture of the social world. Thus, methods of coping anomalies in their object of study, and locate with logical paradox must be transformed when tensions that could account for these anomalies. social scientific paradoxes are considered. The organizational climate literature, for exam- ple, is replete with inconsistent findings, which Four Methods for Working with Paradox may be traceable to inconsistent definitions of climate as a psychological or organizational We propose four generic ways in which two construct, respectively. Once the tension has opposing theses, A and B, might be related: (1) been identified, it might be possible to move to- We can keep A and B separate and their con- ward some synthesis, or theorists can start with trasts appreciated; (2) We can situate A and B at abstract oppositions themselves and carry them two different levels or locations in the social into a problem area. This is a case of theoretical world (e.g., micro and macro levels, respec- interests generating particular applications. A tively); (3) We can separate A and B temporally good example is Allison's (1971) Essence of De- in the same location; or (4) We can find some cision, which contrasted three decision models new perspective which eliminates the opposi- applied to the Cuban missile crisis. tion between A and B. (The second and fourth Some examples of nonparadoxical situations modes of paradox resolution were suggested in theory building are also useful, because not by Nielson, 1967.) Stated schematically, the all inconsistencies are paradoxes. The tension four relations correspond to opposition, spatial between empirical observation and theoretical separation, temporal separation, and synthesis, propositions is not a paradox, because valid em- respectively. They represent a logically ex- pirical observations can correct a theory. Nor haustive set of relationships opposing terms can are most contradictions within a single theory take in the social world. Each of the four meth- paradoxes; most often, these stem from faulty ods represents a different way of transforming reasoning and can be eliminated by correcting our theories and ways of thinking. 565 Opposition: Accept the Paradox and Use search. This approach assumes that one horn of It Constructively the paradox operates at one level of analysis (e.g., macro), while the other horn operates at a The first response is to accept the paradox and different level (micro). To utilize this strategy suc- learn to live with it. This response does not mean cessfully, it is necessary to specify as precisely that the paradox is ignored. Rather, the impli- as possible how the levels interrelate. cations of the paradox are pursued actively and Researchers who take this approach must used to stimulate theory development. A great grapple with difficult and important theoretical deal can be learned from juxtaposing contradic- problems. Despite much research on the aggre- tory propositions and assumptions, even if they gation of individual acts, attitudes, and prefer- are incompatible. Theorists may feel a strain to- ences into social actions, climates, and choices ward cognitive consistency, but that does not (e.g., Coleman, 1986), there is still no completely mean that their theories must fit together neatly. satisfactory solution (Arrow, 1970). The same Paradoxes remind theorists of this inconsistency can be said for other level distinctions. Given the and enable them to study the dialectic between difficulty of spelling out interlevel relations, opposing levels and forces which are captured many researchers have let them stand, while in different theories. advancing only partial and tentative solutions. However, living with paradox has its costs as However, many insights have resulted from at- well. Usually, contrary perspectives are em- tempts to sort out levels and their relationships. braced by different researchers. This may pro- Ethnomethodology (Garfinkel, 1968; Cicourel, duce specialized versions of the two theories 1971), Parsonian sociology, and exchange the- which retard the recognition of relationships ory (Blau, 1964) all involved efforts to distinguish and generate diatribes between proponents of levels of analysis. the "correct" horn of a dilemma. Nor is it always Of the same general type as the level distinc- clear just what sort of relationship "tensions be- tion is spatial separation of paradoxical expla- tween opposing positions" constitute. This ambi- nations. One horn of the paradox is assumed to guity can result in sloppy analysis. Notwith- operate in one physical or social locus, while the standing, to accept a paradox is a positive other operates in a different locus. For example. stance. It is to acknowledge that theorists need Explanation A might hold for the upper echelons not be completely consistent; that seemingly op- of an organization, while Explanation B holds posed viewpoints can inform one another; that for line workers. Some treatments of motivation, models are, after all, just models, incapable of for instance, implicitly assume that top execu- fully capturing the "buzzing, booming confu- tives must be understood in different terms than sion," no matter how strongly logical arrogance workers or lower-level management. tries to convince theorists otherwise. The remaining three strategies attempt to re- Temporal Separation: Take Time into Account solve paradox by spelling out the nature of the tensions between contrary positions. A third approach takes into account the role of time. In this resolution, one horn of the paradox Spatial Separation: Clarify Levels oi Analysis is assumed to hold during one time period and the other during a different time period. The two The second response resolves paradoxes by contrary assumptions or processes exert sepa- clarifying levels of reference and the connec- rate influence, and each may influence the tions among them. Level distinctions such as other through its prior action. Several types of part-whole, micro-macro, or individual-society temporal relationships may exist among con- have proven extremely useful for social re- trary forces: 566  One side of the paradox may influence the con- Combinations and Reflections ditions under which the other will operate, as in Reese and Overton's (1973) formulation of Although analytically distinct, the four ap- cognitive development, in which behavioral proaches can be combined in practice. For ex- learning sets the stage for cognitive acquisi- ample, paradoxical terms may be related both tions; temporally and by level. Moreover, the first ap- one side may create the conditions necessary proach, which accepts paradox and tries to for the existence of the other, as in Smelser's work out its implications, can serve as a prelim- (1962) theory of collective action, in which in- dividual activities may be tiie "precipitating inary step to the other three. By accepting par- event" for collective beliefs to develop; and adoxes, the researcher undertakes to examine his or her theories for hidden tensions and to there may also be mutual influence over time, with swings between one side and the other, search for opposing or contrary positions. Once as in Buckley's (1968) morphogenetic theory of identified, relationships among the positions social systems. can be explored with the three remaining meth- The difficulty of achieving a clear temporal ods. separation of contrary assumptions, theories, or No matter which of the four responses are cho- processes remains to be solved. When does be- sen, working with paradoxes is challenging. It is havioral theory stop holding and cognitive the- difficult to manage the dialectic between posi- ory begin? At what point does individual moti- tions if theorists choose to live with paradoxes. If vation leave off and collective action begin? theorists attempt to resolve paradoxes, they face Most attempts at temporal resolution have formidable obstacles. Nevertheless, the recogni- glossed over the issue of transition points and tion of paradoxes forces theorists to ask different focused instead on the periods of relatively pure questions and to come up with answers that action on either side of the paradox. stretch the bounds of current thinking. The re- Synthesis: Introduce New Terms to Resolve sulting formulations are likely to be of interest the Paradox not just to organizational scholars, but more widely as well. The resolution of paradoxes by level distinc- In the next section, the four methods will be tions or temporal analysis leaves each set of as- applied to address the action::structure para- sumptions or processes basically intact. Both dox. Historically, this paradox has presented a sides of the paradox are assumed to be funda- major barrier to achieving a general theory of mentally sound, and the paradox is resolved by action. Only one specific theory will be used to separating them and specifying how one side exemplify each method, as, obviously, the four influences the other. However, it is also possible methods are amenable to many other theories that the paradox may stem from conceptual lim- which cannot be covered here. itations or flaws in theory or assumptions. To overcome these limitations it is necessary to in- Organizational Structure and Action troduce new concepts or a new perspective. However, such advances are hard-won, and The Action::Structure Paradox many apparent resolutions may lead to dead Most theorists view organizations as social ac- ends. When perspectives radically shift, theo- tion systems constructed by people who use rists may lose as well as gain. The new perspec- them as arenas in which to achieve their goals tive may oversimplify some issues or ignore the and ambitions. Consequently, the central ob- problems that originally gave rise to previous jects of study are the structural properties of a positions. These risks are the price theorists pay social system, the purposive actions of people, for theoretical advances. and the relationship of system and action. Indi- 567 viduals are viewed as purposeful and goal cause it connects the individual to society, links directed, guided by interest or values, and by "positive social theory and normative social the rewards and constraints which are both cre- philosophy," and implies how action might ated by individuals and imposed upon them by change society (Coleman, 1986, p. 1310). the social environment. Individuals interact with This form of methodological individualism others to achieve their self- and other-regarding continues to be a firmly entrenched assump- interests. These interactions are structured by tion in most theories of organization and man- social relationships consisting of interrelated agement. Even those who take a more pessi- and recurrent patterns of social ordering. One mistic view that individuals are merely products important source of social ordering is the orga- of their environments do not deny the role of nization itself, whose structural properties chan- individual motives and actions. Accordingly, nel individual activity and shape members' per- the functioning of an organization as well as spectives and perceived options. the source of organizational change must be These statements are deceptively simple and grounded in the purposive actions of individu- satisfying. However, within them contrary as- als. sumptions regarding action and structure can The problem, however, is that social and or- be discovered. Individuals are free, purposeful ganizational theorists have not been successful actors, in control of their own behavior; yet, or- in developing a theory of action which connects ganizational or institutional structures constrain individual interests with social structure (Cole- action, even shaping individuals' desires and man, 1986). In part, this is because of the diffi- purposes. Individual actors create and maintain culty posed by the action::structure tension, as structures, but an organization is a powerful so- Coleman's (1986) historical account suggests. cial institution with a life of its own. Parsons abandoned his attempt to ground social Attempts to encompass these tensions in a sin- theory in purposive action because he could not gle, coherent theory often have gone awry. The make the link between intentions of persons with path of least resistance has been to emphasize macrosocial consequences. Parsons ushered in either system structure or individual action, and a period of structural-functionalism by focusing many theories originally intended to deal with on the characteristics and equilibrium states of both terms gradually move in this direction. For requisite functions of social systems, "which had example, Coleman (1986) describes how Par- no place for individuals (except as deviants from sons (1937) attempted to balance action and norms) and no place for social change except by structure in his early work. However, as he elab- theoretical fiat" (Coleman, 1986, p. 1311). By orated and defended his theory. Parsons (1951, showing the functional and dysfunctional sides 1964) increasingly emphasized the role of struc- of social systems, Merton (1948) redirected atten- ture. Action was largely determined by struc- tion to individuals' purposive actions. However, ture in Parsons' later work (cf. Garfinkel, 1968; ironically, his analysis ultimately removed the Giddens, 1976). individual from the equation, because he em- Traditionally, action has been conceptualized phasized explaining a social system not by prox- as a micro-level phenomenon, while social imate causes, but by its consequences. Homans structure has been construed at the macro level. (1958) challenged Parsons most explicitly by in- Historically, making the theoretical linkage be- troducing purposive action, but never moved tween micro motives and actions of individuals beyond the social-psychological level of small and the macro structure of social systems has group analysis. The other major challenge to been an important concern of social theorists, functionalism, conflict theory, remained at the from Hobbes through Weber to Parsons, be- collective level and thus failed to provide a the- 568 ory grounded in purposive action of individuals. understood as a result of intersubjective pro- This whole series of great and influential theo- cesses of intentionality and practical reasoning. ries is haunted by the difficulty of mediating ac- The third paradoxical aspect derives from tion and structural elements of social life. how action and structure enter into social scien- To work with any paradox, it must first be un- tific explanations. Traditionally, structural em- derstood. Three basic aspects are part of the ac- phasis has resulted in a deterministic explana- tion: :structure paradox. First, there is ambiguity tion. Organizational structure is conceptualized surrounding the genesis of action and structure. and measured as variables (formalization, deci- Most theories of action view individual purpose sion centralization, complexity), and research- and action as the source of organizational struc- ers study how structure accounts for variance in ture: Organizations evolve from recurrent pat- behavior (Hall, 1963; Hage & Aiken, 1967; Pugh, terns of individual interaction that gradually be- 1981). This approach leaves little room for pur- come formalized (Blau, 1964). However, the posive action, because it emphasizes cause- script is different for structuralists focusing on is- effect connections in which meaning is reduced sues of power and the way coordination is to variables measuring subjects' perceptions or achieved (Parsons, 1951). For these writers, ac- attitudes about the structure (Starbuck, 1981). tion is impossible without resources such as Explanations couched in terms of action theory authority, shared rules, and information— emphasize reconstructing the meaningful con- resources that stem from organizational struc- nections underlying individual choices. This ture. Explanations for coordinated action de- reconstruction relies on interpretive methods. pend on common goals and structural features Seldom can it be represented in variance- which facilitate coordination. Thus, action re- explained terms, because each action ensues quires structure, yet structure only exists through a process of contextually bound practi- through action. While it is easy to assert the cal reasoning and choice making which is con- common sense solution that action and structure text bound (Garfinkel, 1968; Lindblom, 1981). are reciprocals of each other, it has been difficult To work with this paradox, the genetic, onto- to incorporate in an explicit theory. logical, and explanatory tensions must be ad- A second aspect of this paradox derives from dressed. Four methods for doing so have been contrary ontological assumptions about struc- defined. ture and action. Organizational structures are generally assumed to be concrete and measur- Resolving the Action::Structure able. They seem to be tangible objects, because Paradox their traces reside in organizational records, Method 1: Accept the Paradox and Use rules, buildings, and outputs, and because It Constructively members treat them as objects. On the other One way to address a paradox is to acknowl- hand, action is more subjective and ephemeral. edge it and use it as a theory-building resource. To document actions it is necessary to identify Burrell and Morgan (1979), Van de Ven and Ast- motives or purposes, and many scholars have ley (1981), and Pfeffer (1982) adopted this ap- acknowledged the difficulties inherent in grasp- proach in their assessment of different organiza- ing the meaningful connections actors perceive tion theories. They argued that many organiza- and the practical reasoning that precedes be- tional problems can be pursued in a productive havior (Mills, 1940; Weber, 1947; Parsons, 1951; way by keeping organizational theories distinct, March, 1981). There is a tension here: Organi- as opposed to collapsing them into a unified per- zational phenomena are studied in terms of con- spective. Considering the tensions and conflicts crete, measurable properties, yet they are best among theories sensitizes researchers to critical 569  issues and divulges weak points in theories. In systemic problems of selecting, socializing, and addition, key dynamics can often be explained controlling individuals for roles and positions in and understood at one level of organization as a the structure on the one hand, and on the other result of processes occurring at another. Oues- hand, examining how the purposive actions of tions problematic for one theory can often be people over time restructure and renegotiate addressed by another. Insights gained from one these roles and positions. At the group or de- position should supplement and balance those partmental level, there is the issue of how the from others. structural division and integration of labor and For example. Van de Ven and Astley (1981) resources among subunits both influence and and Astley and Van de Ven (1983) distinguished are influenced by the emergence of collective between deterministic structural forms and vol- norms, interaction patterns, conflict, and power untaristic personnel actions at macro and micro relations within and between groups. At the or- levels of organizational analysis. As Figure 1 il- ganizational level, consideration of the tension lustrates, the action: :structure tension can exist reveals the problem of how system structure in- at multiple levels of organizational analysis. As fluences and reflects environmental shifts and an illustration of how Figure 1 might be used, strategic choices of powerful individuals within the insights that can be gained by considering and outside the organization. Finally, at the the tensions or oppositions inherent in (a) the population or interorganizational level, ques- horizontal relationships between structure and tions are asked about the way organizational action, and (b) the vertical relationships be- niches or market structure are both the product tween parts and wholes of structure or the self and constraint of collective working rules ar- versus collective orientations of personnel action rived at through a series of political contests and can be discussed. bargains among pluralistic interest groups. Ac^ion-Sfrucfure Tensions. The tensions be- These issues lead to useful insights because tween action and structure can be identified by they (1) admit to the existence of both determin- moving along the x-axis for each pair of terms. istic and voluntaristic aspects of social systems; Interesting questions and problems turn on how (2) juxtapose these aspects by reciprocally relat- structural forms and personnel actions at each ing structural forms and personnel actions at level of analysis interrelate and produce con- comparable levels of analyses; and (3) focus on structive tensions. how these relationships unfold over time in com- For example, by considering this tension at plementary and contradictory ways (Van de the individual level, theorists can discern the Ven & Astley, 1981). Macro Level (Whole, Collec- industry/community structure industry/interorganizational tive Orientation) collective action organization design board of directors executive committee Micro Level (Pan, Self Orien- department/division task force management tation) committee roles and positions individual Structural Forms Personnel Actions Figure 1. Illustration of structural forms and personnel actions at micro to macro levels of orga- nizations. This diagram is based on the discussion in Van de Ven and Astley (1981). A similar diagram appeared in Van de Ven and Poole (1988). 570  Tensions in Part-Whole Relations. Consider- macro levels of organizational analysis. Orga- ing the vertical relationships among micro and nizations can be understood in terms of the ten- macro dimensions in Figure 1 divulges the con- sions between (1) personnel action and struc- flicts and tensions inherent in part-whole or self tural forms at each level of organizational anal- versus collective frames of reference. Defining ysis and (2) the forces of conflict, coercion, and tensions between these levels is useful, because disruption at one level of organization, and many problems and solutions apparent at one forces of consensus, unity, and integration at level of organization manifest themselves in dif- another level—forces which can be seen as ferent and contradictory ways at other levels. both prerequisites and reciprocals of each other. For example, relying on the concept of requi- One of the most useful products of the first ap- site variety, Weick (1979) argued that with in- proach is discovery of underlying tensions and creasing environmental complexity, uncer- inconsistencies, and a heightened awareness of tainty, and variety, the overall structure of the their potential. organization must become more complex, loosely coupled, decentralized, particularistic, Method 2: Clarify Connections between and anarchistic. At the same time the structure Organizational Levels of the individual parts or groups within the or- A second way to address the action::structure ganization become more simple, tightly cou- paradox is to distinguish levels of reference in pled, hierarchical, universalistic, and cohesive. the contrary propositions and to spell out the The whole tries to become more adaptive, but connections between them. Theories of social this results in the parts exhibiting characteristics action by Arrow (1970) and Coleman (1973, 1986) which lead to nonadaptiveness, narrowness, are attempts to apply this method. Basically, and "group think." This unintended conse- their approach assumes individuals can act, but quence is the result of a basic principle of con- organizations cannot. They then attempt to tradictory part-whole relationships established specify models by which individual actions can by Georg Simmel: "The elements of differenti- combine to create collective outcomes. These ated social circles are undifferentiated, those of system-level outcomes may, in turn, impose undifferentiated ones are differentiated" (Blau constraints on individual actions (Coleman, translation, 1964, p. 284). 1986). Coleman (1986) suggested that such a Another part-whole contrast can also be dis- model should consist of (a) a set of assumptions covered: Many macro theories of order and con- about how the interests and goals of actors stand sensus include micro theories of conflict and co- in relation to one another vis-d-vis the collective ercion, and vice versa. For example, structural- action system, and (b) a set of combinatorial functional theories have been attacked for their rules for merging individual actions into collec- inability to explain change because of the em- tive acts. (Potentially there could also be map- phasis on order, consensus, and unity (Silver- ping rules that specify how collective structures' man, 1970). While this is true at the macro- "feedback" affects individual action.) organizational level, it is only possible because With regard to component (a) of the model, of coercion, domination, and control of disrup- Coleman (1986) discussed three types of interde- tive tendencies at the micro level. If this were not pendent relationships among individual inter- so, there would be no need for rules, indoctri- ests, which lead to different types of social orga- nation, socialization, and control mechanisms— nizations: concepts which are central in structuralists' views of organizations. A pure market is a configuration in which there Much can be learned by explicitly accepting are "independent actors, each with differing the action::structure paradox at micro and private interests and goals and each with re- 571  sources that can aid others' realization of inter- ory. Hernes (1976) provides a dynamic logic for ests. The actions that purposive actors will en- structure-action relationships between macro gage in when this configuration of interests and and micro levels. resources exists is social exchange, and wiien a number of these exchange processes are inter- Third, Coleman's model tends to oversimplify dependent, we describe the whole set as a mar- the actor. His model makes global assumptions ket institution." (p. 1324) about the types of relations that hold among ac- A hierarchy is a set of relations "in which one tors, which amounts to an ideal type of actor. actor's actions are carried out under the control For example, Coleman's (1986) description of of another and advance the other's interests." market relations assumes that actors are inde- The associated institution is the formal organi- pendent, have resources that can meet each zation or authority structure, (pp. 1324-1325) other's needs, and will form stable exchange re- A federation [our term] is a set of independent lationships. This ideal-type actor facilitates pre- actors linked by common interests. They are dictions about the outputs of social action pro- connected by a constitution embodying a set of cesses in the aggregate. Given the standard norms regarding rights and obligations, (p. 1326) model, there is no need to look at action per se; theorists need merely to apply the ideal type, With regard to component (b). Arrow (1970), given certain levels of inputs and contextual Coleman (1973), and others have proposed sets constraints. Whether the model captures the of mathematical or logical rules for combining subtleties and many variations of action and ac- individual interests and actions into collective tors is open to question. action and macro-structural forms. In principle, an aggregated model with a non- Coleman's theory is a promising develop- typal theory of action could be generated, but ment, but several problems illustrate the difficul- this would result in multiple outcomes, and ties inherent in building multi-level theories. therefore multiple ideal-type actors. This pro- First, his approach involves the restrictive as- cess would make aggregation procedures very sumption that only individuals can have pur- complex, and the resulting models would be dif- poses. Individuals are assumed to be the prime ficult to solve. Maintaining a balanced empha- motivational force behind any organizational sis on both levels of analysis is a major chal- activity. This assumption leads to an overem- lenge for this mode of theory building. phasis on the micro (action) to macro (structure) This approach can be used to discover as well relation, and an underemphasis on the macro to as resolve theoretical tensions. For any given micro relation. The approach devotes much at- theory at one level of analysis, the researcher tention to combinatorial rules, but much less to can search for oppositions at different levels. rules that show how structure influences action. Once found, such oppositions can be used to Second, Coleman's approach confounds the generate more encompassing theories. action-structure dimension with the macro- micro dimension. He only admits to purposive Method 3: Use Time to Relate Structure action at the micro level and structural forms at and Action the macro level, but purposive action can also exist at the macro level in collective decision- Time is a third resource for resolving the ac- making bodies or committees, and structural tion: :structure paradox. Action and structure forms are also present at the micro level in the can be related through an alternating temporal roles and positions individuals occupy. Thus, a order. Buckley's (1967; Archer, 1982) morphoge- more complex scheme, like that illustrated in netic systems theory follows this strategy to ar- Figure 1, might promote a better-rounded the- ticulate a coherent relation between action and 572 structure. The theory of morphogenesis pro- Action is not left out of this theory, but its oper- poses that action and structure influence each ations remain opaque. Also, while processes of other, but in alternating cycles over time. The organizational convergence and equilibrium cycles are composed of three phases: (1) a pe- are richly described by Tushman and Romanelli riod in which pre-existing structures dominate (1985) in the punctuated equilibrium model, the behavior; (2) a period when action begins to ar- punctuation process itself remains underdevel- ticulate alternative arrangements, which sets oped (Van de Ven & Poole, 1988). To be effective, the stage for structural change; and (3) a period these theories have to perform a balancing act, of structural elaboration, in which changes in and close attention must be given to articulating structure are institutionalized. Then the cycle the relation between structure and action over can start over again. time. Tushman and Romanelli's (1985) punctuated Finally, while relations between structure and equilibrium model of organizational evolution action may be chronological, they may or may follows a pattern similar to Buckley's theory. not be causal. Tushman and Romanelli (1985) They posit alternating cycles of convergence, posited that reorientations may result from inter- "which elaborate structures, systems, controls, nal inconsistencies during periods of conver- and resources toward increased coalignment," gence, major changes in the environment, or and reorientation, "periods of discontinuous transformative leadership. Careful longitudinal change where strategies, power, structure, and observations of specific cases are needed to in- systems are fundamentally transformed towards fer whether temporal relations between periods a new basis of alignment" (p. 173). Convergence might be causal. Further, prediction of future seems to be predominantly influenced by struc- changes, as opposed to explanation of past ture, whereas reorientation is driven by purpo- changes, may be difficult with a punctuated sive actions of executive leaders. equilibrium model. Like its counterparts in biol- The key problems which must be resolved in ogy, the organizational punctuated equilibrium taking this approach relate to the boundaries model permits a wide variety of options which it between temporal periods and to a tendency to- cannot precisely predict (though it does provide ward structural bias. The problem of establish- predictions regarding how effective an organi- ing boundaries between periods is not a trivial zation will be, based on how it deals with con- one. A crisis that upsets existing structural ar- vergence and reorientation). rangements presents a relatively clear bound- Temporal analysis can be used to uncover in- ary. However, problems also can develop grad- teresting tensions as well. The researcher can ually, and incremental adjustments can cumu- look for anomalies, points in time when the the- late to produce qualitative change (Lindblom, ory does not seem to fit. In some cases this may 1981). When structural and action influences be fertile ground for the interplay of contrary shade together, the resulting approach is not theories. Another tactic is to take a general the- particularly informative, because it relies on ory of temporal alternation, such as Buckley's, temporal sequencing. and use it as an analytical lens to identify peri- In these models, structure is easier to portray ods when one or the other horn of the dilemma than action because structure is more easily ob- prevails. served than individual motives and behavior. Method 4: Advance a New Conception of the As a consequence, these theories tend to reflect Structure-Action Relationship a bias for structure. For example, Buckley ac- corded much more space to an analysis of struc- A final attack on the action: :structure paradox tures and their effects on action than vice versa. involves developing a wholly new conception of 573 the action-structure relationship. This novel con- manager might use a union contract as a norm ceptualization dissolves or supercedes the oppo- to justify her claim that overtime should be given sition. Giddens (1976, 1979, 1985) advanced a to certain workers. In so doing, the individual, theory of structuration that exemplifies such a according to her own motives and within the new approach. (See also, Berger and Luck- limits of her interaction skills, is using the con- mann, 1966; Barthes, 1979; Touraine, 1977; Ran- tract as an institution in action. She is producing son, Hinings, and Greenwood, 1980; Bartunek, and reproducing a mode of contractual interpre- 1984; Poole, Seibold, and McPhee, 1985, 1986; tation. Hence, institutions determine the struc- Barley, 1986.) Structuration refers to the process tural features available for appropriation and of production and reproduction of social systems therefore limit the possible ways in which an via members' application of rules and re- individual can act. However, the individual re- sources. Implicit in this definition is a distinction shapes the institution as a particular modality of between system and structure. Structure refers structuration. to the rules and resources people use in acting In drawing on structures, individuals repro- and interacting. A system is the outcome of the duce the structures. The structures they draw on application of rules and resources, the observ- more often become more important, whereas able patterns of relations between people and those they use less decay. In addition, how peo- groups. For example, the status hierarchy of an ple use structures determines how they are re- organization is a system. The structure underly- produced and whether they remain stable or ing this system includes rules, such as norms change. However, despite the central role indi- of superior-subordinate interaction, and re- viduals play in the production and reproduction sources, such as formal authority or superior of structures, the complexity of social systems knowledge. The system exists because of its means that people do not wholly control the pro- structuring; members use rules and resources to cess. In complex systems, apparently straight- create and maintain status relationships, and forward actions by individuals trying to control the hierarchical pattern can be explained by dif- the system may lead to unintended conse- ferences in the rules and resources available to quences. different individuals. Barley's (1986) study of CT technology imple- The theory of structuration assumes structures mentation provides a good example of the mul- have a dual nature: They are both the medium tiple routes action may follow. In two similar and outcome of action. Structures make action, hospitals, with similar team compositions and and hence the existence of social systems, pos- similar power distributions, the CT technology sible. Nevertheless, structures only exist as they led to quite different organizational structures. are continuously produced and reproduced in In one case, the new technology led to a hierar- interaction. Thus, structure and action mutually chical structure dominated by physicians— entail each other. Giddens' insight explains how essentially a reproduction and reinforcing of sta- this entailment comes about and what its signif- tus quo relationships. In the other case, the CT icance is. technology provided lower level participants In the terminology of Figure 1, Giddens intro- with a way to gain some control, resulting in a duced a new intervening column between struc- flatter structure in which physicians and lab tural forms and personnel actions, which he technicians were more nearly equal in power. termed modalities of structuration. A modality of In this latter case traditions were not preserved, structuration represents the individual actor's but rather the innovation altered system pro- appropriation of structure for use in a particular cesses. action context. For example, in a conflict, a As this case suggests, an account of the struc- 574 turation of stability and change in systems re- the theory of structuration on precisely these quires researchers to focus on the effects of ac- grounds. She argued that separation of action tion on structure, and the influences of structure and structure permits them to be studied as sep- on action. This reciprocal relationship is medi- arate features of social life; structuration, how- ated by modalities of structuration. This raises a ever, is not susceptible to empirical research, new research question: What forces shape the because modalities are complex constructions way people use institutions and structures avail- with obscure empirical references. Giddens able to them? If the actor feels driven or con- (1979) offered few recommendations for research strained by the structure, there would be simple on structuration, beyond an artificial separation reproduction of the structure. If the actor per- of "the analysis of strategic action" from "the ceives that he or she is able to construct and analysis of social structure." Although proce- adapt the structure, instability and change may dures have been developed for the study of result, if actors so choose and the system per- structuration, the book is not closed on this issue. mits it. The fourth method of paradox resolution also The second issue to consider is whether an can be used to uncover paradoxes. If theoretical enacted social structure takes hold. As structur- tensions can be characterized, criteria for a new ation theory suggests, the reproduction of the construct can be defined. Conscious compari- system and structure determine whether inno- sons of existing theories can stimulate new vations persist or are extinguished. For exam- ideas. ple, it might be interesting to go back to Barley's hospitals two years later to assess the persis- Discussion tence of the structures he described. Were con- The complexity of organizations guarantees ditions such that structures were reproduced? that theories cannot give a complete represen- In developing a structurational theory, the en- tation. Nevertheless, m pursuit of an elusive tire configuration of the action system must be consistency, researchers may create self- considered, including the inclinations of individ- encapsulating theories which may freeze think- uals, the constraints the system puts on them, ing. There is great potential to enliven current and the nature of the macrolevel process. theory and to develop new insights if theorists Hence, the theory of structuration cuts through search for and work with inconsistencies, con- the action::structure paradox. It posits that struc- tradictions, and tensions in their theories, and in ture and action coexist in a mutual process of the relationships between them. production and reproduction. Action draws on Much of contemporary organizational theory structure; structure only exists in action; and is still struggling to live with paradoxes. A con- they connect in modalities of structuration. structive approach to living with paradox is to In its very boldness, this fourth method of par- adopt the first method, that is, to juxtapose and adox resolution may present special problems compare how contrasting theories deal with the for theory construction. New concepts bring with same organizational problems. This strategy them unknown and uncertain entailments. Intro- requires theorists to engage in comparative duction of the idea of structuration solves several analysis of theories cast on both sides of the par- conceptual problems, but it may also introduce adox and at different levels of analysis. In addi- other problems, such as the unclear conceptual tion to generating insights from divergent per- status of modalities of structuration (Poole, 1985). spectives, this approach helps the researcher New theories, advanced to correct problems in become aware of tensions and oppositions extensively studied areas, may be difficult to which can be addressed by the other three study in their own right. Archer (1982) criticized methods. 575  If theorists attack a paradox by clarifying lev- has been to advance and advocate an addi- els of analysis, they are led to formulations tional level of theoretical reflection. Rather than which specify how theories operating at differ- simply recognizing incompatible positions and ent levels or in different sectors of the organiza- stopping, theorists should attempt to specify the tion or society interrelate. This is exemplified by relationships among them. Sometimes this spec- Coleman's efforts to show how individual ac- ification can lead to a creative integration that tions combine into collective structures. These greatly enhances understanding. combinatorial models describe how structure The four modes of paradox resolution can be determines the impact individual actions have regarded as a set of topoi for theory develop- on collective processes; they also specify how ment. Topoi was the term used by classical structure limits individuals' options. To carry off Greek thinkers to designate standard forms of these models of linkages between action and argument that served as models for the inven- structure, assumptions about an ideal-type actor tion of arguments for specific cases. These four had to be made. This move enabled Coleman to modes of paradox resolution can serve as topoi characterize individual actions more or less uni- for the generation of organizational and man- formly, in such a way that they can serve as agement theories; they give us a range of theo- inputs to the models that map one level onto retical possibilities to work with. another. The hallmark of this approach is care- Although the focus has been on methods for ful specification of levels and of the way one resolving or capitalizing on paradoxes, para- level maps onto another, and vice versa. doxes should not be "eliminated." Indeed, If temporal sequence is used to resolve a par- rather than using paradoxes to build theory, re- adox, researchers must work out a formulation searchers can build theories about paradoxes, with alternating cycles of two or more theoretical which is what Smith and Berg (1987) have done. explanations. It is especially important to dis- Nor is it clear that researchers can ever avoid cuss, as does Buckley (1968), how the transitions paradoxes. One challenge is the possibility that between the various phases are accomplished. the resolution of one paradox may inadvertently Finally, if paradoxes are addressed by ad- create another. The complexity and interdepen- vancing new concepts or distinctions, a new dence of individuals and organizations typically conception must be developed. This is exempli- exceed researchers' capabilities to describe or fied by the reconceptualization of the action- explain them with coherent and consistent theo- structure relationship advanced by the theory of ries. Resolutions of paradox in one aspect of a structuration. When the concept of modality of theory often create inconsistencies in another structuration is added, this approach attempts to part of the theory. Thus, it seems unlikely that bridge action and structure and to show their theorists can ever escape or resolve theoretical intimate connections. In this approach, any dis- paradoxes completely. It has been suggested tinction between action and structure is for an- that at the heart of any theory that solves a par- alytical purposes only; retaining it as an as- adox is another, different paradox, waiting to be sumption about the nature of organizations un- discovered. If this is true, there is one consola- necessarily bifurcates a holistic phenomenon. tion: Tackling the same old, well-known para- The theory tries to provide an analytical vocab- doxes, researchers may uncover as yet un- ulary that would enable researchers to study ac- known ones that can move social inquiry in new tion and structure in a fundamentally new way. directions. This also raises the issue of whether There is no single best way to address para- the world is consistent or whether it is actually dox, and each of the four methods suggested inherently paradoxical. The latter possibility is here has both benefits and costs. The objective intriguing, but it is a subject for another essay. 576  References Allison, G. (1971) Essence of decision. Boston: Little, Brown. Hage, ]., & Aiken, M. (1967) Relationship of centralization to Archer, M. (1982) Morphogenesis versus structuration. Brit- other structural properties. Administrative Science Quar- ish Journal of Sociology, 33, 455-483. terly, 12, 72-92. Arrow, K. (1970) Social choice and individual values. 2nd Hall, R. H. (1963) The concept of bureaucracy: An empirical ed. New Haven: Yale University Press. assessment. American Journal of Sociology, 69, 32-40. Astley, W. G., & Van de Ven, A. H. (1983) Central perspec- Heigenoort, J. van. (1972) Logical paradoxes. In Paul Ed- tives and debates in organization theory. Adminisfrafive wards (Ed.), Encyclopedia of philosophy {pp. 45-51). New Science Quarterly, 28, 245-273. York: Macmillan. Barley, S. (1986) Technology as an occasion for structuring: Heigenoort, J. van. (1958) Social behavior as exchange. Evidence from observations of CT scanners and the social American Journal of Sociology, 63, 597-606. order of radiology departments. Administrative Science Hernes, G. (1976) Structural change in social processes. Quarterly, 31, 78-108. American Journal of Sociology, 82, 513-545. Barthes, R. (1979) S/Z. New York: Hill and Wang. Homans, G. (1958) Social behavior as exchange. American Bartunek, ]. M. (1984) Changing interpretive schemes and Journal of Sociology, 63, 597-606. organizational restructuring: The example of a religious James, L., Joyce, W., & Slocum, J. W., Jr. (1988) Comment: order. Adminisfrafive Science Quarterly, 29, 355-372. Organizations do not cognize. Academy of Management Berger, P. L., & Luckmann, T. (1966) The social construction Review, 13, 129-132. of reality. Garden City, NY: Doubleday. Lindblom, C. (1981) Comments on decisions in organiza- Blau, P. M. (1964) Exchange and power in sociai life. New tions. In A. Van de Ven & W. Joyce (Eds.), Perspectives on York: Wiley. organization design and behavior (pp. 245-248). New Buckley, W. S. (1968) Sociology and modern systems theory. York: Wiley. Englewood Cliffs, N]: Prentice-Hall. March, J. G. (1981) Decisions in organizations and theories of Burrell, G., & Morgan, G. (1979) Socioiogicai paradigms and choice. In A. Van de Ven & W. Joyce (Eds.), Perspectives organizafionai analysis. London: Heinemann Educational on organizafion design and behavior (pp. 205-244). New Books. York: Wiley. Cicourel, A. V. (1971) Cognitive sociology. New York: Free March, J. G., & Simon, H. A. (1958) Organizafions. New Press. York: Wiley. Coleman, J. S. (1973) The mathematics of collective action. Merton, R. K. (1948) Sociai fheory and social structure. New Chicago: Aldine. York: Eree Press. Coleman, J. S. (1986) Social theory, social research, and a Mills, C. W. (1940) Situational actions and vocabularies of theory of action. American Journal of Sociology, 16, 1309- motives. American Sociological Review, 5, 904-913. 1335. Nielsen, H. A. (1967) Antinomies. New Catholic Encyclope- Garfinkel, H. (1968) Sfudies in efhnomefhodoiogy. Engle- dia (pp. 621-623). New York: McGraw-Hill. wood Cliffs, N]: Prentice-Hall. Parsons, T. (1937) The structure of social action. New York: Giddens, A. (1976) New rules of sociological method. New McGraw-Hill. York: Basic Books. Parsons, T. (1951) The social system. New York: Eree Press. Giddens, A. (1979) Cenfraiprobiems in social theory. Berke- ley, CA: University of California Press. Parsons, T. (1964) The social system (2nd ed.). New York: Eree Press. Giddens, A. (1985) The constitution of society. Berkeley, CA: University of California Press. Perrow, C. (1986) Complex organizations: A critical essay (3rd ed.). Glenview, IL: Scott, Eoresman. Glick, W. (1985) Conceptualizing and measuring organiza- tional and psychological climate: Pitfalls in multilevel re- Pfeffer, J. (1982) Organizafions and organization theory. search. Academy o/Managemenf fleview, 10, 601-616. Marshfield, MA: Pitman. Glick, W. (1988) Response: Organizations are not central ten- Poole, M. S., Seibold, D. R., & McPhee, R. D. (1985) Group dencies: Shadowboxing in the dark, round 2. Academy of decision making as a structurational process. Ouarferiy Management Review, 13, 133-137. Journal of Speech, 71, 74-102. 577 Poole, M. S., Seibold, D. R., & McPhee, R. D. (1986) A struc- W. Joyce (Eds.), Perspectives on organization design and turational approach to theory-building in group decision- behavior (pp. 167-199). New York: Wiley. making research. In R. Hirokawa & M. S. Poole (Eds.), Starbuck, W. H. (1983) Organizations as action generators. Communication and group decision making. Beverly American Sociological Review, 48, 91-102. Hills: Sage. Touraine, A. (1977) The self-production of society. (D. Colt- Pugh, D. (1981) The Aston program of research: Retrospect man, Trans.) Chicago: University of Chicago Press. and prospect. In A. Van de Ven & W. Joyce (Eds.), Per- spectives on organization design and behavior (pp. Tushman, M. L., & Romanelli, E. (1985) Organizational ev- olution: A metamorphosis model of convergence and re- 135-167). New York: Wiley. orientation. In B. Staw & L. Cummings (Eds.), Research in Oumn, R. E., & Cameron, K. S. (Eds.) (1988) Paradox and organizational behavior [Vol. 7, pp. 171-222). Greenwich, transformation: Toward a theory of change in organiza- CT: JAI Press. tion and management. Cambridge, MA: Ballinger. Van de Ven, A. H. (1983) Review of In Search of Excellence. Ranson, S., Hinings, B., & Greenwood, R. (1980) The struc- Administrative Science Quarterly, 28, 621-624. turing of organizational structures. Administrative Science Van de Ven, A. H. & Astley, W. G. (1981) Mapping the field Quarterly, 25, 1-17. to create a dynamic perspective on organization design Reese, H., & Overton, W. F. (1973) Models of development and behavior. In A. Van de Ven & W. Joyce (Eds.), Per- and theories of development. In J. R. Nessleroade & H. W. spectives on organization design and behavior (pp. 427- Reese (Eds.), Life-span developmental psychology: Meth- 468). New York: Wiley. odological issues. New York: Academic Press. Van de Ven, A. H., & Poole, M. S. (1988) Paradoxical re- Silverman, D. (1970) The theory of organizations. Exeter, NH: quirements for a theory of organizational change. In R. Heinemann. Ouinn & K. Cameron (Eds.), Paradox and transformation: Toward a theory of change in organization and manage- Smelser, N. (1962) Theory of collective behavior. New York: ment (pp. 19-63). Cambridge, MA: Ballinger. Eree Press. Weber, M. (1947) The theory of social and economic organi- Smith, K. K., & Berg, D. N. (1987) Paradoxes of group life. zations. (A. M. Henderson k T. Parsons, Trans.) New York: San Francisco: Jossey-Bass. Free Press. Starbuck, W. H. (1981) A trip to view the elephants and rat- Weick, K. (1979) The social psychology of organizing (2nd tlesnakes in the garden of Aston. In A. H. Van de Ven & ed.). Reading, MA: Addison-Wesley. Marshall Scott Poole is Associate Professor of Speech- Communication in the College of Liberal Arts, Univer- sity of Minnesota. Andrew H. Van de Ven is 3M Professor of Human Systems Management in the School of Management, University of Minnesota. An earlier version of this paper was presented in the symposium on Theory Building in Organization and Management Sciences at the Academy of Manage- ment Conference in Anaheim, California, August, 1988. A related extension of the ideas presented here is available in A. H. Van de Ven and M. S. Poole, Paradoxical requirements for a theory of organiza- tional change. In fl. Ouinn and K. Cameron (Eds.), Paradox and transformation: Toward a theory of change in organization and management (pp. 19- 63), Cambridge, MA: Ballinger, 1988. We would like to thank Robert D. McPhee and several AMR reviewers for their constructive comments on an earlier draft of this essay. 578