I- man [A d 0‘ -‘. ‘ o c . 3 . .. WI :'~‘:.L'.;EI'-' v} \ . '73”an 5 . . "' *L'U‘MV- -. - au,o¢mo.-ooo.“-Qv-vap I) .5 3-} {Q‘gxhw’gg II‘\£§#¥.~V: ‘} $5 _.—.‘I., . II II. '. a I . l' .: . ’\ 0‘ \ I - I . ‘ ' _'vl. .' " ' ' :- ’ n I,. II II. ,I -:- l'..| 1 h." “if? ' a» ' I _‘ ' ' I IIIII.I I‘I rs ~ ' =; ., \, aI-m‘ . 5x ._ . , -." :33: «If . ., ? ‘ “‘ I -r w . I I . | J‘I i. I 11 T} 1, 3' .. I " .I . II I ”XIV" H I. a ." I.II.I II IIIIII I" ”II INH‘ ‘ I ‘ '.'.»‘]‘I, O I 'I I " ' .'..'I\I 4 1- .I .‘ . . I I i 1- . I. ‘ III"." I, I b - I, ‘ . ’. l- M in. a") I“ ,I #:III \i‘: O\,\"'IIOHD.I l Ifx ‘I‘. ‘ I‘ '. . (A fi'lu‘} III . I y...“ "H-‘N-H N. f I' -6J| I u ;‘.' I "JIIHI .,, II‘ ‘- ' I‘ I I '. .3, I I UH 1 I‘ , . ". I I I L g ' ..,.. I I, .. , . gm». - . z I.‘ - I .‘.' - . " 'u . " h. 5:4-I'I HI . I‘III . I ‘ II Il‘|‘-‘-IIU ,kIII} II I III : I fu' ' ,‘x ' . I- ‘l-I I I :" '.'l ;I‘I'-~ 'I-I. ' ' , II I I- I I. " -‘ I i 'v . ‘ ~ 'C‘fi. .-:‘-' : "I‘l" ' . V IN "I . Im‘n I’. . I I I J. I l "2' I” I": I II Iy~Iy I .I I -\ I .I“ ‘IIII.";.II‘I'I:‘I1I 'I O l ' I .I " '|;"I'VID"'V..‘VI II ' ' I I n I“ :kzq‘l’I" ."I {‘0 ‘ I . "3‘ ...I;\"-"'Ill III ' I I . I. I:I‘ IV. I I "‘I‘I' '.| .' IIII'I ”II I'. ,.'I'IIIIII \m V‘I’ I I I H) I. IIIIVII I“ S\ j H. 0 “II.- v'rr—v—v-v IV“ cur-- a Y O LIB RA R Y Micizignn State University This is to certify that the thesis entitled A RULES THEORY OF ORGANIZATIONAL COMMUNICATION presented by Robert David McPhee has been accepted towards fulfillment of the requirements for PA. D. degree in We rm '1‘ ®fl~dDiscriminate 4.8 Others' Power to Discriminate 4.9 Relational Power to Discrim- inate 4.10 Outside-Relational Power to Discriminate 4.11 Joint Outside Cooperative Dis— criminatory Power 4.12 General Competitive Interde- pendence A, B C, D R AxB AxC, AxD, AxCxD, BxC, BxD, BxCxD AxBxC, AxBxD, AxBxCxD AxR, BxR CxD, DxR AxBxR AxCxR, AxDxR, AxCxDxR, BxCxR, BxDxR, BxCxDxR AxBxCxR, AxBxDxR AxBxCxDxR aThe effect examples are labeled analogously to ANOVA dimensional effects, simple and general, for a five- dimensional matrix. A, B, C, and D stand for dimen- sions corresponding to choice options for four actors; R stands for a reward dimension, with one level belong- ing to each actor. 58 from the perSpective of the A-B relationship. Second, within each matrix the effects and effect types are con- ceptually and quantitatively independent--we have an exclusive and exhaustive list of components of joint choice relations. (There is some dependence between Sim- ple and overall matrices, but it can be removed by Slight Operational adjustments which, if anything, increase the interpretability of effects.) Third, the effects and interpretations identified here have no necessary rela- tions to the precise rules of game analysis, such as minimax. On the contrary, a person identified as autono- mous and generally powerful may be hurt more than anyone else by a particular joint choice which is optimal by minimax rules--and such paradoxes are easy to construct. So this system is not a new, precise game-analytic method. My argument for the utility of the method is as follows. In real life, especially in ongoing organiza- tions, any game-theoretic representation would have to deal with relationships either as one extremely complex joint choice situation, with each choice representing a complex sequence of actions, or (more likely) as an inter- dependent sequence of joint choice Situations. In general, such representations are possible in principle but in fact just cannot be carried out. But, since such representa- tions are possible i3 principle, we can assert with 59 certainty that the components identified above exist in the situation. Moreover, the components seem much more meaningful and powerful as descriptions of actual situa- tions than the "types" identifiable using game-theoretic distinctions (cf. RapOport and Guyer, 1966, whose list is longer by far than ours for every game situation). And finally, situations which are very complex when considered as choice situations may be stable and relatively clear when considered in the light of the various joint choice effects: that person A is generally more powerful than B in a relationship regardless of Short-term outcomes or relatively minor changes in their situation. My claim that this model is a scientifically use- ful way of representing task relationships rests on three subordinate claims, two of them empirical. First, the components of the model can be used as dimensions for the objective classification of task relationships. This is beyond question, as long as we can find out how people's actions affect their own and others' outcomes. Second, I hypothesize that people use and can deal with these dimen- sions in ordinary language to classify situations and task relationships, so that we can use questionnaires to ascertain the values a task relationship takes on these dimensions. This hypothesis is moderately simple to test--simply present people with games or game—like 6O Situations and see whether their descriptions differ systematically with the model's dimensions. Third, I hypothesize that people reSpond socially to situations according to their values on these dimensions. They react to dependence by attempting to influence; they respond to dependence on cooperation by communication designed to influence Egg coordinate, and so on. Thus, this model has general communication implications. But, more importantly for our purposes, the model can be used to organize our examination of task variables. Variables Describipg Task Interdependence Patternipg of Interdependence An collection of theorists from many different disciplines and using several different conceptual schemes have dealt, more or less directly, with the notion that overall interdependence of tasks is patterned--it is not simply a collection of dyadic interdependence relation- ships. In the area, a key concept is "work-flow pattern," and it is concentrated on most directly by Eliot Chapple and Leonard Sayles in The Measure of Management (1961). Of course, management scientists have examined the concept in exhaustive detail under titles like "critical path analysis," but most of their conceptions have been devel- 0ped to represent industrial design problems, and are relatively irrelevant for our purposes. Much more 61 relevant is the work of the Socio-Technical theorists, eSpecially Eric Miller (1963). These theorists are concerned with the way work flows from one employee to another, and especially with the problem of finding optimal arrangements and natural divisions of the flow, into work-units, departments, and so on. However, the discussion, especially that of Chapple and Syles, focus on work-flow patterns without adducing variables which could describe individual workers, relationships, or (often) work units. From their discus- sions, though, the following variables seem crucial. l. The difference between two task roles. On Miller's analysis, this variable has the additive compon- ents: a. Technological--how different are the operations performed in the two jobs? b. Territoria1--how far apart, physically, do the two workers usually work? c. Temporal--how much of a gap, in raw times, is there between when one worker gets through with working on an item and the other begins on it? Miller argues that boundaries ought to be drawn between workers whose tasks are very different. 62 2. The sensitivity of one task role to another's Operations--if one worker makes a mistake or is absent, how does that affect the other's job? The first two variables were primarily applicable to relations between task roles or work units; the last two are descriptive of particular roles and units. 3. The number pf Sources from which necessary inputs for a task are received-—from how many different people must an employee receive material and informational inputs in order to do his job? 4. The number pf alternative sources for each necessary input. Unfortunately, these variables are often much more relevant to intensity and direction of interde- pendence than to patterning as explained above. The joint choice model clearly contains all the elements necessary to integrate this assemblage of vari- ables. "Patterning" itself clearly implies a group level of analysis, even though it may be interpreted from an individual or relational point of View. For any work group or collection of employees, given a joint choice matrix or simple effects submatrix, we can calculate the degree pf gropp interdependence by summing the General Interdependence (1.5 in Table 3, or 2.6 plus 2.11 in Table 3) and the average interaction effect size for the group. When the general matrix is used, this index 63 measures the extent to which the group is internally interdependent. When measured using a "Simple" submatrix, the index indicates the extent to which the group is an interrelated environment for an individual or subgroup. These indices roughly correspond to the density of con- nections in the "work—flow pattern," and to the "number of sources" of input to an individual, mentioned above. More importantly, they indicate the extent to which the group is relevant to the negotiation of coordination between members. Intensity of Interdgpendence Unlike the variables generally used to deal with patterning of interdependence, intensity of interdepen- dence as dealt with by organizational theorists is direc- tly analogous to the "interdependence" dimensions of the joint choice model identified in Table 3. Therefore, we Shall pay special attention to the operational defini- tions such theorists use, since they may also serve to operationalize the joint choice matrix itself. Intensity of interdependence involves two notions: that the task role of an employee is complex, and that a certain "amount" of this complexity may have an impact on another employee. These two notions relate to three variables which were initially conceptualized by Charles Perrow (1967) to characterize the technology of entire 64 organizations; Andrew Van de Ven and Andre Delbecq have generalized and extended Perrow's conceptual distinctions so that they may be applied to individual task roles, the task of a work group, or the task structure of a larger body. Perrow's initial variable, task variability, is fairly explicitly described and operationalized by Van de Ven and Delbecq: Task Variability. A familiar dimension of work is task variability, which refers to the number of exceptional cases encountered in the work requiring different methods or procedures for doing the work (Perrow, 1967). For Thompson, Hickson and Woodward task variability could be measured as the stability and uniformity of inputs and outputs. Task variability has also been measured as the routinization, repetitive- ness, stability, or rigidity of the work. Index of Task Variability. The index constructed to measure task variability was based on Hall's (1962) and Hage and Aiken's (1967) measures. Respondents were requested to answer the follow- ing questions along a lO-interval scale: (1) How much variety in cases, claims, or things do you generally encounter in your working day? (2) Regardless of the variety of cases, claims, or clients, to what extent are the activities or methods you follow in your work about the same for dealing with categories of cases, claims or clients? (3) To what extent would you say your work is routine? For the next four questions please indicate the extent to which each of the following statements are true or false: (4) People in this unit do about the same job in the same way most of the time; (5) Basically, unit members perform repetitive activities in doing their jobs; 65 (6) In doing their jobs from day to day, unit members generally have to adopt different methods or procedures to do their work (Reverse Scoring); (7) There are different types of kinds of work to do every day in this job (Reverse Scor- ing). (1974, pp. 183, 196) Here they depart from Perrow mainly in that he conceives of variability solely in terms of exceptional cases which require hierarchical consultation with work super- visors or even upper-level management. This difference becomes important in reference to the second major vari— able characterizing organizational tasks: task analyz- ability or difficulty. For Perrow this was the ability of management to deal with exceptional cases--are there standard procedures or not? Van de Ven and Delbecq include the entire task in their conception of task dif- ficulty. Task Difficulty. Task Difficulty refers to the gpalyzability pf the work itself and to the extent to whiCh there is a known procedure that specifies the sequence of steps to be following in performing the task. For Perrow (1967), who largely developed this concept, task difficulty refers to the degree of complexity of the search process in performing the task, the amount of thinking time required to solve work-related problems, and the body of knowledge that pro- vides guidelines for performing the tasks. Index of Task Difficulty. BaSed upon Perrow's (1970) measure, the following index was con- ducted to measure task difficulty. Respondents were requested to answer the following questions along a lO-interval scale: (1) To what extent is there a clearly defined body of knowledge or subject matter which can guide you in doing your work? 66 (2) To what extent is there an understandable sequence of steps than can be followed in doing your work? (3) During the course of your work, how often do you come across specific difficult pro- blems that you don't know how to solve immediately; (4) In general, how much actual thinking time do you usually spend trying to solve such specific problems? (5) If there is something that you don't know how to handle in your work to what extent can you go to someone else for an answer to the problem? (6) In some jobs things are fairly predictable. In others, you are often not sure what the outcome will be. What percent of the time are you generally sure what the results of your efforts will be? (7) On the average how long is it before you know whether your work effort is successful? (1974, pp. 183, 196) For Perrow, as for Van de Ven and Delbecq, these dimensions of task were conceptually independent, although (probably) statistically interrelated--if a task is difficult it is usually seen as variable, too. In their later work, Van de Ven pp 21. (1976) bor- row the concept of interdependence from James Thompson (1967). Thompson conceived of interdependence as involv- ing primarily the work flow between workers or units. He conceived of an ordered array of modes of interdependence: pooled (no inter-unit flow, but common dependence on resources), sequential (one-way flow), and reciprocal.(two- way flow). Van de Ven pp 31. add a fourth mode—-team interdependence (joint work on the same operation)--and used the array to characterize work-unit tasks. IItshould 67 be noted that this array is, in principle, unidimen- sional--the more channels for flow exist between units, the more interdependent they are. This array of variables is probably best adapted to handle the work group level of abstraction, but may be extended to larger units. It would be hard to use the first two variables to characterize relationships, or the third to characterize a task role. However, relating the variables to the joint-choice model is fairly straight- forward. Variability and difficulty are rough measures of the number of choices an employee must make in doing his job a job with high variability and difficulty requires many choices involving extended thought. The index of interdependence measures the extent to which any other employee or group depends on this difficult task. Therefore, a multiplicative combination of the three mea- sures itself provides an indicator for how intense the interdependence between the workers is. Note that, since the interdependence measure applies to groups larger than dyads, this measure can be used, roughly, to operation- alize interaction effects in joint choice matrices of higher dimensionality than two. Direction of Interdependence The direction of interdependence between two employees is fairly simple to explain in a joint choice 68 context. Three types of dimensions are involved: depen- dence (or power), antonomy, and outside interdependence. Interdependence is directional to the extent that one employee can and must respond adaptively to another's choices, in order to achieve a satisfactory task outcome. Directionality, that is, implies unilateral rather than mutual adaptation. The value or need of this type of adaptation is measured in each simple matrix as Selective Interdependence (1.3 in Table 3). The difference between these values for two individuals is thus an initial mea- sure of the directionality of their interdependence. It is only an initial measure because direction- ality of interdependence also depends on the freedom of the dependent person to choose, and this freedom can be limited to two ways. First, an employee's "autonomy" (1.1 in Table 3) can be understood as constraint--there are restrictions or requirements, internal to A's job, that prevent him from adapting to B's choice. Second, an employee can be interdependent with third parties who constrain his ability to adapt to any single other. The clearest measure of such constraints is, in the termi— nology of Table 3, the Benefit of Outside Relations (3.5% Thus, a better measure of directionality of interdepen- dence is the absolute difference in dependence, minus the 69 autonomy of the weaker person, minus his benefit from outside relations. In sum, our examination of the joint choice model provides us with independent conceptual definitions of variables measuring the patterning, intensity, and direc- tion of interdependence in an organization. In the next chapter, we shall examine the communication implications of these variables. CHAPTER IV THE IMPACT OF INTERDEPENDENCE ON RULE SYSTEMS In this chapter I introduce several variables descriptive of systems of organizational communication rules, and explain how they are related to the variables describing task interdependence which were presented in the last chapter. Variables Describing Rule Systems In this section I will describe a set of variables which may be used to characterize organizational communi- cation rules. In accord with the analysis of the earlier chapters, these variables do not deal with the existence or content of specific communication rules--there are many functionally equivalent systems of rules that might develop for almost any task. But if a system of communi— cation rules is used to coordinate task activity, the rules involved must have certain necessary features: They must allow expectations to be formed, either about the subject of communication or about later communication itself, which have a certain degree of precision and cer- tainty. These variables all characterize the ground for 70 71 drawing precise and certain inferences about communica- tion--consensus on rules. An initial perspective on the quantitative description of systems of communication rules is provided by Scheff (1967a) in an article precisely about consensus. Scheff notes the common-sense notion of consensus as mere agreement--consensus on some proposition X would just be the percentage of peOple in the group of concern who agreed with X (1967a, p. 33). But he contrasts this with the interactionist concept, which develops a notion of consensus as coorientation_in the context of role-taking. The advocates of this position argued that social agree- ment about some norm will not necessarily affect behavior, if no one knows that such a state of agreement exists. Scheff cites the phenomenon of pluralistic ignorance, where all peOple act or form expectations in accord with a norm that no one believes in, simply because they think everyone gl§g_believes in it. More importantly, an agree- ment definition of consensus about communication rules is futile. The point of communication is to convey meaning from one person to another; and understanding by the second person is necessarily prior to his evaluating or being persuaded or informed by the communication in a way that is likely to enhance coordination. Scheff sug- gests that consensus as coorientation necessarily 72 involves multiple levels: at a minimum it involves belief in norm X, plus belief that others believe (or do not believe) in X; but this second belief may also become the subject of consensus--I can believe that others think I believe in X, just as I can think others believe in X, and so on. Scheff adapts the system developed by R. D. Laing and his colleagues to characterize these different levels of consensus in a dyad. They have developed a technique for measuring three levels of co-orientation between mar— ried couples, which they call agreement, understanding, and realization. At the first level the pair, interviewed sepa- rately, simply give the same reSponse to an issue. For example, the issue may be the fol- lowing statement: "Mary is dependent on John." If Mary and John express agreement with this statement independently, they agree. The second question is to ask each how the other would answer the question. If John answers that Mary will agree, and in fact she does agree, John understands Mary on this issue. If John's guess does not agree with Mary's actual answer, then he misunderstands Mary. The third and final level is called realization. The operational index for realization is contained in the question: How will Mary think you have answered this question? If John correctly judged how Mary thinks he has answered the question, then John realizes that he is under- stood by Mary. If he does not, then he fails pg realize that he is understood. (1967a, p. 37) These three variables, agreement, understanding, and real- ization, are generalized by Scheff to other units of 73 analysis. In particular, he distinguishes between group agreement on an issue (i.e., whether or not a majority agrees on one particular stand (or rule)), group under- standing (i.e., whether or not a majority understands that group agreement exists), and group realization (i.e., whether or not a majority realizes that group understand- ing exists). From here the extension to intergroup and group-individual consensus is straightforward. Of course, these three level variables may be conceived of quantitatively rather than dichotomouSly, in two distinct ways. First, rather than asking whether a majority agrees, etc., we can ask for the percent who agree, etc. Obviously this extension is possible only for group and larger units of analysis. An alternative way to quantify these variables, applicable to relationships involving either individuals or groups, is to ask, not "DO they agree?" but "How far apart are they?" This approach to quantifying similarity of opinion about an issue has been developed most thoroughly by Jay Jackson (1975), although he does not consider the levels of consensus mentioned above, but sticks to agreement. For any norm or rule of action (I neglect here the full complexity of the return potential model), he asks mem- bers of a group how intensely they favor or disapprove of the norm. The average value of group answers (disregarding 74 the positivity or negativity of opinion expressed) is the intensity of expectations about the norm. Of course, even if everyone in a group feels intensely about a norm, Opinions about it may differ, some members being extremely positive and some extremely negative, while in other cases some group members will feel more intensely than others. The variable crystallization is designed to reflect this presence or absence of group agreement in quantitative terms. The crystallization dimension of the RPM is used to represent the degree of consensus among others' potential sanctions for actor's conduct in a given situation. Crystallization is measured in terms of the variance among expectations and varies between zero and a maximum value dependent on the level of inten- sity. . . . when others are indifferent to actor's conduct, intensity is zero, there is no variance among expectations, and crystal- lization must be perfect: a state of vacuous consensus. When intensity is at a maximal level, however, crystallization can vary between maximum and zero. . . (1975, p. 245) Jackson conceives of intensity, for an individual, as a certain active concern or energy the individual devotes to the norm, and he wishes to investigate the expenditure or effect of this energy. But he recognizes that, since intensity does not imply consensus, he is aware that intensity in favor of the norm may simply be devoted to conflict with someone who is equally unfavor— able to the norm. But if the group has both consensus and intensity, the energy involved should be available to 75 influence behavior and warrant certainty of expectations. Jackson breaks the group's store of intensity, its score on the intensity variable, into two parts. The first part is called normative power, and represents intensity channelized by crystallization. The second part is called conflict pptential, and represents that part of intensity which is vitiated by lack of agreement or crystallization and channelized, if anywhere, into intra-group conflict and inconsistent expectations. Jackson's two variables, normative power and con- flict potential, are extremely useful, not only in reflec- ting the harnessing or dissipation of intensity in a group, but also in investigating the implications of consensus. Since normative power and conflict potential do not depend on the size of the groups involved, we can conceive of two subgroups, polarized and oppositely oriented, which each possess high normative power which turns into pure con- flict potential when they are considered together as a group. Alternately, if the issue at stake is the proper behavior of some focal individual, we may locate groups with high normative power but which may have trouble bringing it to bear due to the existence of rival groups. Given these general discussions, we may isolate three types of variables descriptive of rule systems. 76 First, for any set of issues (such as a system of com- munication rules), we can determine the "amount" of agreement, understanding, and realization that exist on that set within any dyad (or larger social group). We can also calculate the normative power and conflict poten- tial existing in the dyad for that rule system. All five of these variables are to be thought of as attributes of rule systems, given particular social contexts. I shall label these variables collectively as "the intensity of consensus" on the rule system of interest. Second, norma- tive power, conflict potential, and the levels of consen- sus can also be thought of as attributes of various social units or groups: the percentage of agreeing, understand- ing, and realizing individuals, or their internal vari- ance, and the intensity of their agreement of disagreement. The variables so interpreted I shall refer to as the "pat— terning of consensus." Third, as Scheff (1967a, b) notes, consensus can be imbalanced or "asymmetric" as well as general and symmetric. Indicators of this characteristic of social relationships include three absolute differences: between the levels of understanding, of realization, and of intensity, between actors. These derived variables I will term asymmetry of understanding, of realization, and of intensity; collectively they will be referred to using the label "asymmetry of consensus." 77 The three classes of variables just presented deal almost exclusively with consensus about communication rules, per pg; the next and final class I will discuss iS more descriptive ofuthe internal nature of the system of ____,._.l. rules. This final class consists of the variables dif- ferentiation and integration of the system of rules. Per- haps the two most relevant sources of these concepts are George Kelly (1955), and Harold Schroder and his col- leagues (1967). For Kelly, differentiation stood for the number of construct dimensions a man possesses to discrim- inate among objects and to avail himself of "channels of choice" (Kelly, 1962, p. 86). The more constructs a per- son has, the more flexibly he can deal with the world, since he can conceive of it more Specifically and with attention to a broader range of information. But, the broader the range of information a man can consider, the more likely it is that his information will include incon- sistencies or unclear implications, and the more he is in need of the ability to iptegrate his information. Devel— oped by Schroder, Heinz Werner, and other authors, inte- gration as a cognitive trait characterizes a man's ability to deal with initially incoherent, inconsistent, and unsystematically related information in ways that progres- sively take into account more and more of the total body of information and its implications. (For a more complete 78 summary and analysis of the literature surrounding and underpinning these concepts, see Phillips and Thompson, 1976, Chapter 3.) Schroder and his colleagues (1967) and Walter Crockett and his colleagues (1974) have detailed schemes for picking out conceptual indicators of differentiation and integration from written descriptions or accounts of persons: constructs in the case of differentiation, and integrative rules or procedures in the case of integra- tion; and these schemes have, for content-analytic methods, remarkably high reliability. As Mischel (1964) has insightfully pointed out, in their valid implications the concepts of "construct" and "rule" are identical; constructs differ from communication rules as they have been described here primarily in that communication rules are consensual, have practical force, and are function- ally related to task performance. Thus, I prOpose to use the same methods on written descriptions of task roles and unitary work operations. For example, the Kingdon's (1973) analysis of conflicts between engineers and pro- grammers, he found that each group was able to describe relations between the two roles, and that the two groups worked together on fairly well defined projects, on which their cooperative activity went through a number of stages, for which they were able to articulate prescriptions 79 about responsibility and task initiation. In Chapter II, three different kipds pf rules vital to organizational .1--4—— ,“_______ m - communication were identified. Some related to the orga— nizational role of an employee—-his possession of infor- mation and decision powers, in particular. Another set dealt with the relations between two employees, and their obligations to inform and consult with one another. The final kind of rule was the decision premise, which deter- mines how much force a fact has to influence a joint decision. It is clear that the descriptions Kingdon elicited quite definitely articulated sets of organiza- tional communication rules. If organization members in general can articulate as clearly the role characteristics and operational characteristics of their immediate orga- nizational environment, then the content-analytic schemes mentioned above can be used as the first step in an operational procedure to evaluate the consensus, and dif- ferentiation and integration, of communication rule sys- tems. Once possible rules have been identified using content analysis, they can be restated as propositions and submitted to both the original subject and the other organization members with whom he must coorient. The variables describing level of consensus, and normative power and conflict potential, can be assessed for the pro- positions, both to identify them as rules and to assess 80 the whole range of communication-rule-system related var- iables. In sum, then, we have isolated four classes of variables descriptive of communication rule systems. The class "intensity of consensus" included agreement, under- standing, realization, normative power, and conflict potential for a rule or rule system. The class "pattern- ing the consensus" included the same five variables mea- sured as attributes of social units or groups. The class "asymmetry of consensus" included asymmetry of understand- ing, of realization, and of intensity--of course, these are attributes of the relationship between pairs of actors or groups. Finally, the class "complexity of a rule- system" included its degree of differentiation and inte- gration. Propositions and Rationales In general, we can find broad relationships link- ing classes of variables descriptive of communication rule systems, to the variables descriptive of task inter- dependence relations outlined in the third chapter. "Intensity" g: interdependence brings about "intensity" and complexity of consensus on communication rules, due to the increased demand for precision of communication and power of influence when interdependence is complex. Patterning pf interdependence brings about a Similar 81 patterning of consensus, due to the need to simulta- neously "differentiate" and "integrate" the organization, in Lawrence and Lorsch's sense (to be explained below). Finally, the direction pf interdependence has its pri— mary impact on the asymmetry of consensus. Basically, this is true because dependent employees need to be able to understand and anticipate the behavior of those on whom they depend. 1. Intensity of interdependence was the name we gave to one dimension of the joint choice model, but also to a multiplicative index of the three variables task variability, task difficulty, and task interdepen- dence, as interpreted by James Thompson in terms of directionality of workflow connections. It Should be noted that each of these variables has an impact on, e.g., organizational structure because it conditions the impact problematic cases have on relatively inter- dependent workers. High task variability increases the probability that new and problematic cases will crop up; high task interdependence increases the likelihood that any given problematic case will have a harmful impact on the work of the other employees; and high task dif— ficulty implies that, when a problematic case does arise, it will not be easily or quickly dealt with. 82 As intensity of interdependence increases, the system of rules governing the relationship between inter- dependent workers must, to remain functional, increase in intensity and complexity of consensus: that is, the number of levels on which consensus exists must increase; understanding and realization, in particular, must in- crease; and the normative power (Jackson) of the rules must increase. Consensus at higher levels must increase because each worker needs to be able to predict, e.g., when the other wants to be informed of a contingency, or to participate in decision-making, and to know when the other is mistaken in his impressions about these matters. Normative power must increase because the power of these rules, either to produce efficient coordination or to reveal tacit conflicts, must increase to avoid unneces- sary interruptions in work. In addition, increasing intensity of interdepen- dence also requires increasing complexity of the system of communication rules. This is essentially because exceptions function essentially as uncertainpy does in Galbraith's (1977) theory of information processing Struc- tures in organizations. Galbraith, one of the leading contingency theorists, argues that organizational uncer- tainty implies that the planning of task activity must be decentralized and take place more during the actual 83 execution of work. Therefore, Galbraith argues, more information processing structures must be added inside the organization, to cope with the additional processing demands. Just so, differentiation and integration of the communication rule system must increase because both modes of complexity are instrumental in the processing of information. Foster (1975) describes an example of a construc- tion company owner (himself) who, working as the firm's general manager, sought to increase profitability by taking on more and more kinds of jobs, and employing more of the time of his subordinates. As this change in the task structure took place, he and (e.g.) his carpenters began to learn each other's precise demands for informa— tion and consultation because the slack time and resources that allowed for Slow and painstaking adjust- ment to contingencies was no longer present. In partic- ular, he had to learn the limits of his knowledge about carpentry, so that he could consult the carpenters about scheduling their work precisely wnen necessary. In other words, bnderstanding, realization, and éomplexity of the foreman's communication rule set increased as a result of higher intensity of interdependence. Moreover, the normative power of the communication rules in the company increased with more intense interdependence. 84 Employees p§g_to report and discuss their progress and problems on jobs--otherwise, unexpected delays would dis- organize the scheduling, not just of their own work, but of the work of many other employees. In sum, we have supplied a rationale for the fol- lowing propositions: a. More intensity of interdependence leads to interdependence leads to more agreement on rules. b. More intensity of interdependence leads to more understanding of others' perspectives on rules. c. More intensity of interdependence leads to more realization. d. More intensity of interdependence leads to more normative power of rule systems. e. More intensity of interdependence leads to more differentiation of rule systems. f. More intensity of interdependence leads to more integration of rule systems. 2. Patterning of interdependence was character- ized in terms of the degree of group interdependence, from two points of view: the point of view of the whole group, and the point of View of selected subgroups and individuals. This question of the "texture" of 85 interdependence has been the object of insightful research by Lawrence and Lorsch (1967), who note that the basic division of labor common in many firms, into research, production, and sales division, produces in many indus— tries highly sensitive dependencies among these divisions. Lawrence and Lorsch argue that competitive excellence requires at once differentiation and integration of these departments. "Differentiation" requires that cognitive, emotional, and temporal orientations in the different divisions should be idiosyncratically adjusted to the task they perform, and in particular be different in dif- ferent divisions. The examples they cite indicate that this is precisely a difference in decision criteria, and the most functional arrangement is a precise patterning of consensus on those criteria along lines of task differ- ence: high normative power for temporal and other decision criteria within divisions, high conflict poten- tial between divisions. "Integration," in this usage, means that the organizational structure must be so supple- mented so that, despite their differentiation, the divi- sions can cooperate intensely and responsively. Lawrence and Lorsch suggest structural mechanisms for accomplish- ing this endeavor, one of which is particularly relevant here. They suggest setting up special integrating roles and departments with orientations midway between those 86 of the contending divisions, so that conflicts can be mediated and seeming conflicts can be clarified by people who can understand both sides. In other words, units are constructed from whose viewpoint the dissimilar depart- ments are very highly interdependent; workers in these units then undertake to construct communication rules which allow them to deal with this interdependence. Lawrence and Lorsch also argue that just one division, determined by the industry of the firm in ques- tion, must deal with the dominant competitive issue in that industry, and that it is particularly dependent on responsive adaptation from the other two divisions. They also find that maximal efficiency in coordination requires that integration mechanisms be biased in their orientations in favor of the division dealing with the dominant competi- tive issue. Once again, the patterning of interdependence, particularly the number of necessary inputs from other divisions, has an impact on the patterning of consensus. This argument provides a rationale for the following pro- positions. a. The larger the group for which interdepen- dence is high, the larger the group which agrees on rules governing communication about interdependent activity. 87 b. The larger the group for which interde- pendence is high, the larger the group which understands communication rules. c. The larger the group for which interde- pendence is high, the larger the group which realizes about communication rules. d. The larger the group for which interde- pendence is high, the larger the group for which normative power on communica— tion rules is high. e. The higher the interdependence for a group from a subgroup's point of View, the more consensus that subgroup has with the group's communication rules. f. The higher the interdependence of a group from a subgroup's point of View, the higher the normative power of that group's rules within that subgroup. 3. Direction gf ipterdependence refers to a variable index which is the total difference in dependence between two task roles, controlling for autonomy and inter- dependence with third parties. This variable has its major impact on communication because it represents the extent to which interdependence can lead an employee to need and 88 be able to adjust to another's activity, more than the reverse. One underlying principle guides our analysis: if high selective dependence by one party on another's activ- ity exists, the first will adjust his activity to the second. This principle is directly supported by advocates of a "Strategic Contingencies" theory of power in orga- nizations (Hickson pp 31., 1971). Working from an exchange-theoretic proposition relating power and depen- dence, they argue that the department dealing with a strategic contingency--an ongoing, crucial problematic situation facing and affecting the organization as a whole-—will exercise the most influence over decisions in the organization. They argue that this is true because the key department, by coping with a key problem, does a service to other departments which results in increased power for it, when compared to the lesser ser- vices they perform for it. That is, the difference in dependence favors the key department. This argument is incomplete, though, in two ways. First, it neglects the fact that the contingency is a major constraint on activity by the key department. Its members cannot as easily adjust their activity and still 89 do a good job. Second, it neglects the fact of central- ity--a strategic contingency leads the other departments to organize around the key department, so that its interdependence on third parties is greater than that faced by other departments. Both these factors lead to increased adjustment to the key department, and both are provided for in my index of directionality. If dependent units require modification of activ- ities relative to the units they are dependent upon, then a functional communication system would facilitate such modifications. In particular, we can expect (a) that the dependent unit would need much more accuracy in order to adjust as required, so that an asymmetry of consensus level would arise; and (b) that decision premises which support the unit depended on will have high normative power within the dependent unit. Thus, the type of task interdependence will exert its chief impact on the asymmetry of consensus on communication rules. These considerations provide support for the fol- lowing propositions. a. The greater the directionality of inter- dependence, the greater the asymmetry of accuracy. b. The greater the directionality of inter- dependence, the greater the asymmetry of 90 realization. c. The greater the directionality of inter- dependence, the greater the asymmetry of intensity. This concludes our discussion of rules systems and the principles directly governing them. In the next chapter we turn our attention to the generating mechan- isms supporting consensus on rule systems, and inquire into the principles governing their nature. CHAPTER V THE IMPACT OF INTERDEPENDENCE ON GENERATING MECHANISMS In this chapter I deal with various types of gen- erating mechanisms which regulate consensus on the rules governing organizational communication. The first sec- tion contains a presentation of the types of mechanisms that can exist in organizations. The focus of discussion is on each mechanism as the entering employee is social- ized to understand and act within it; that focal direc- tion makes it easier to explain the range of rules that can fall under each mechanism and the relations of the mechanisms to one another. The second section introduces and argues for propositions relating presence and domi- nance of the generating mechanisms to task interdepen— dence variables. Types of Generatipg Mechanisms At any point in time an organization has a nomi- nal and a real task structure. The nominal structure is partially and formally Specified; it consists of a set of stipulated task roles, each of which contributes to a formally stated organizational task goal. The real 91 92 structure consists of the task roles members actually perform, which may or may not "add up" to achieving that same overall task goal. As members enter the organiza- tion, they begin by learning the formally stipulated task role, along with whatever extra knowledge is necessary for fundamental job performance. But as they gain exper- ience, they learn more about the freedom they have in the organization. This freedom allows them both to perform their job in a way that they choose and to negotiate and collaborate with others to change or remove parameters that would otherwise restrict their range of choice. As their knowledge of the organization increases in breadth, their autonomous activities in concert with other members come less and less to affect merely their own jobs and more and more to affect the operations and the effective goal of the whole organization. Members join an organization in return for compen- sation of some sort in return. They contribute effortful work to the organization. As their rank and experience increase, compensation comes to be organization—centered-- they care less about external rewards and more about the ability to influence the operation of the organization p35 pg (cf. Burns and Stalker, 1961). AS their influence increases and they become able to identify with organiza- tional acts, their own work seems less of a cost to them 93 and more of an activity valuable in its own right. As their range of choice in an organization increases, they become more effective in maximizing the value of their "compensation," while minimizing the "cost" of their con- tributions. I argue that in order to increase their range of choice, organization members enter into a succession of social units with whose members they cooperate in the pursuit of some task goal. These units are generating mechanisms for social behavior in that they have the power to generate and enforce behavioral and communication rules. The types of social units I concentrate on are task struc- ture, hierarchical control structure, and association structure. These types are successively ordered in a num- ber of reSpectS. They constitute a logical sequence—- socialization into later mechanisms requires prior social- ization into earlier mechanisms. They are sequenced in terms of autonomy--the task goals are successively more flexible and open to the exercise of choice by members. Collaterally, the mechanism types are sequenced by the degree to which their task goals require unit self-control; later mechanisms are focused less on sheer performance and more on self-control of performance. 94 Mechanism I: Task Structure When a member enters an organization, the first thing he must learn is the task he is to carry out. All tasks can be characterized as the performance of some operation on some input, producing some output. Either the new member's superior or a coworker tells him how to obtain the input, what operations to perform on it, and what to do with the output. Usually a worker's "entrance ritual" includes introductions to other employees whose jobs are directly interdependent with his. The communica- tion and behavior rules guiding a worker at this stage are determined by his task and by the traditions he is taught. A variety of writers (Mead (1934), Lauer and Boardman (1971)) have noted that a person's learning of a set of rules can be characterized in terms of a number of stages. A sequence of stages particularly interesting to organizational researchers because it focuses on rule-sets organized by some dominant task operation, is the follow- ing: (1) A worker begins by Simply carrying out (almost mechanically) the particular set of operations described to him. Following Mead, we shall call this stage imipg— Eiyg. (2) After some experience carrying out the set of rules, a worker comes to understand the operation as a unity, to understand its logic. He can spontaneously adapt it to some narrow range of exceptional inputs, and 95 he can recognize when something goes seriously wrong in the course of the operation. This stage I shall label insight. (3) After a person understands the logic of his own role, he can begin to understand the logic of roles associated with his own. He first grasps the logic of each associated role from its own point of View, since that is inevitably the way it is presented and described to him. He understands and can adapt to its demands on his own behavior, so I call this the adaptive stage. Given a point of View, a member can understand the inter- action of two or more roles from that point of view, but he cannot yet establish his own viewpoint. (4) Here the organization member separates out his own role from the set he has come to understand and uses it to establish a persPective on the others. Just as insight was his inter- pretation of his own task operation and goal, so in this active stage he establishes an interpretation of the organizational task goal, at least so far as it is rele- vant to him. From this emergent perspective he can initiate activity and make demands on other role—holders in the name of his own task needs. This ability is the first step in the process of learning and establishing one's autonomy in the organization. It is necessary for an employee to learn his own task role in order to earn the compensation the 96 organization offers for his services. It is not neces- sary that a worker proceed to the active stage of role learning, but insofar as he cannot effectively operate at that stage, he will be forever subject to the blandish- ments of his colleagues, and will be relatively ineffi- cient in performing his own task. Mechanism II: Hierarchical Control Structure The Sheer logic of his task might lead a personnel department employee to spend weeks writing a perfect report on turnover and absenteeism among workers, but on the job he soon finds that perfection is not necessary, expected or rewarded. Only certain attributes of any employee's work are ever examined or evaluated, and he soon learns to adapt his activity to the control structure which does the evaluating. The control structure for any employee consists essentially of his formal superior, but includes any other organizational members who have the formally stipulated power to evaluate his work. The task goal of this struc- ture is the control of job performance. But the complete control of performance is impossible, so the control structure always allows more flexibility to task perform- ance than would a complete set of task rules. In the course of interacting with his superior and, more or less directly, with other members of his 97 control structure, a member comes to understand the con- trol structure. He learns the controlling limits on his performance, what the ordinary consequences of crossing those limits are, and what excuses are acceptable for crossing those limits. He also learns what information his superior needs and gets about his performance and how much he can distort that information. Since a superior is responsible for the perform- ance of his subordinates, he must make some choice about the degree of autonomy to grant them. Complete autonomy would probably be disastrous; complete control is impos- sible. Given that initial choice by the superior, his subordinate is left with two ranges of choices. First, since the subordinate has at least §gmg_autonomy, he can make undictated choices inside that range. Second, he can try to interact with his superior so as to increase or decrease his range of autonomy. (Interaction here includes the whole gamut of relational strategies avail- able to a subordinate.) Of course, the superior's choice is not really prior to or independent of this interaction. Employees go through the various stages in devel- oping a relationship with a superior (probably the crucial relationship in the hierarchichal control structure). Since each new employee is learning a system of control rules, the prOgression begins with an imitative stage, 98 during which he learns the particular mechanisms and the attributes they control by trial and error, by observa- tion, and by talking with coworkers. In the final, active stage, he has learned the nature of the control system governing his own job and has also acquired enough information about his superior's job to under- stand what control data he receives, uses, needs and what his reactions typically are. In particular, he knows what kinds of deviations from standards his superiors find out about and care about, and how to control or cope with the control system when deviation is inevitable. Entry and socialization into this system is a nec- essary step for a jobholder because the control system determines one's retention on the job and discretionary rewards like raises and advancement. But, more impor- tantly in terms of our analysis, knowledge of the hierar- chical control system increases the freedom of choice of an organizational member in the performance of his job: he knows not only what he should do by stipulation, but also what he must do and what he can do and get away with. Mechanism III: Associative Structure Up to this point, I have concentrated on mechanisms which determined or limited the task role of individual organization members. While such mechanisms do not com- pletely determine task behavior or communication, each 99 member exercises autonomy within limits stipulated for his task role. But the organization also contains a set of mechanisms which allow additional autonomy for employees, while delimiting their behavioral options in new ways. Such mechanisms constitute the associational structure of the organization. Very often, such mechanisms are not formally rec- ognized in the organization. Nonetheless, they are nec- essary results of two organizational phenomena: the decentering of goals and indirect interdependence. As an employee learns his task role, he develops a conception of his role as functional--as making an understandable contribution to a goal more abstractly specified. When he learns his role and place in the control structure thoroughly, he understands the contribution his job makes, not just to other employees who depend on him, but also to the task stimpulated for his immediate work-group, pos- sibly his department, and so on. To some extent, he becomes interested in performing in accord with these more abstract task goals. Correlatively, he becomes aware of ways that the more abstract task goal could be more effective performed, possibly at less cost to him- self. Some of these ways require changes in his task role that affect no one, and these he may simply adopt. 100 Some require task role changes that affect others with whom he is directly interdependent, and he may negotiate such changes with the interested others. But in many cases role changes affect others with whom he is indirec- Ely_interdependent. (Indirect interdependence will be more fully characterized below; for now it may be thought of as similar to "pooled" interdependence as defined by Thompson (1967) and discussed above.) Groupings of indirectly interdependent workers are what I shall call associations. Associations as generative mechanisms sup- port action and communication rules which allow changes in lower level task goals and roles in more efficient pursuit of higher order goals. There are numerous exam- ples of such associations, of which the most well known is the work-group. A man on a basketball team playing man-to-man defense will sometimes commit himself to steal the ball and get out of defense position. In general, he will (or should) do so only with the tacit assumption that the other players on his team can adjust to the act--they must be able to adjust their own roles fairly well to pick up his burden. He can make that assumption on the basis of practice and long experience together. A work-group in an organization is a group of employees who have a chance to develop that kind of mutual experience so that they 101 can develop supportive and collaborative relationships. Emery and Trist state this requirement clearly: Grouping produces its main psychological effect when it leads to a system of work roles such that the workers are primarily related to each other by way of the requirements of task per- formance and task interdependence. When this task orientation is established the worker should find that he has an adequate range of mutually supportive roles (mutually supportive with respect to performance and to carrying stress and that arises from the task). (1960, p. 91) They mention that supportive relationships require under- standing of the group task as well as the individual role, based on experience and not necessarily on any close friendship relations. Examples of work groups which have performed, as associations a generative function are certainly not rare. Homans (1950) cites the case of the Bank Wiring Observation Room group, where workers traded jobs for short times, in part to allow lagging workers not to fall too far behind the standard activity rate, although such "trades" were officially prohibited. Blau (1956) cites a law enforcement agency where workers exchanged advice about difficult cases, contrary to prescribed procedure. Sayles (1958) cites the case of a large work group depen- dent on the activity of another prior group whose task activity rate is variable. Since direct communication to the prior group was impossible when their activity was 102 interrupted, employees would come gp_mg§§g to pressure their superior to exert pressure on the 95235 group's superior to adjust his production rate. (Apparently employees found mass pressure to be more effective than Single reports or complaints.) Not all workers can develop work-group relations of the sort described here--for instance, Miller and Rice (1967) describe the salesman's role as one which typically excludes meaningful membership in a work group-- so that participation in this type of generative social unit is not a necessary condition of organizational life as participation in the previous two social structures has been. Socio-technical theorists argue that the group is a most powerful mechanism when the group is composed of interdependent but distinct roles and performs a group task that is an autonomous unit with respect to the rest of the organization. In such structures, and in struc- tures that depart more or less from this ideal, joint group experience underlies COOperative activities that may go beyond stipulated task roles. Once again, a member learns the rules governing group collaboration and support in stages: he progresses from the imitative stage where he learns by imitation and observation the range of collaborative and supportive behaviors typically carried out in the group to the active 103 stage, where he understands and acts on both the logic of the group norm system and the unique relations his role and skills allow him to provide or request. He thus develops knowledge of a set of extra choice options for performing at his job--new sources of information, advice, support and collaboration. He also grasps a new perspec- tive on organizational activity——that of the work group-- and reevaluates task endeavors in that light. What task underlies the work group as a genera- tive mechanism? In general, the logic of departmental- ization arranges individual roles into groups in such a way that one can speak of a group operation (composed of a collection or complex of individual operations) on input, yielding a group output. The group operation,so conceived, is at least part of the group task. But another part is best described as "earning compensation for the group." Such compensation includes compensation for group members, plus greater individual autonomy due to group support; it also includes group preservation and autonomy to be earned by effective group performance. So group membership entails a range of compensations for individual activity-- rewards from organization to individual, from organization to group and from group to individual. The last two sets of rewards are mainly instrumental, as we have conceived of work groups--they entail a wider range of options for 104 individual task performance. This perspective may seem like an atypical per- spective on groups in organizations. Work groups are often seen as restricting member behavior by group norms, status relations, etc.--as requiring and sanctioning con- formity. What is usually not noted is that conformity to norms is like conformity to formal task rules--it allows coordinated activity without communication. And the same criticisms and limitations hold for the View of groups as norm-determined systems as for organizations as formal systems--such views imply a degree of static univocal consensus which rarely exists, and such views become less and less tenable as the group task becomes more complex. Another example of an association as generative mechanism is the coalition. Coalitions exist only in relation to formal or informal decision-making centers in an organization, and have no formal existence. They con- sist of sets of employees who have agreed to seek a desired decision and who attempt to determine or influence that decision (or set of decisions). Probably the best discussion of coalitions in organizations is that by Cyert and March (1959). They mention two prime requirements for considering a group of employees a coalition: 105 . . . That through bargaining and side-payments the participants in the organization enter into a coalition agreement for purposes of the (decision to be influenced). This agreement specifies a joint preference-ordering (or orga- nizational objective) for the coalition. . . . That thereafter the coalition can be treated as a single strategist, entrepeneur, or what have you. (p. 78) Here the indirect interdependence of the coalition members lies in the fact that their joint action is more likely to be effective than uncoordinated influence attempts. Just as some workers could not or did not partic- ipate in work groups, so some employees may never be mem- bers of a coalition--their task role may be too void of autonomous power for them to be included. Alternately, members of autonomous work groups are rarely members of coalitions simply because there is no task activity which requires coordination with members outside the work group. In this sense, there is probably a trade-off between work group membership and coalition membership. Other rele— vant but less structured associations include communica- tion networks and peer groupings. For instance, Gross (1953) reports on a "grapevine" communication network that was regularly used to communicate orders to a work group when formal channels were clogged. Such an informal channel by no means merely carries social gossip-—it is an organizationally functional association. Again, Jacobson (1976) has cited the case of a peer association in an auto 106 body factory. Because the various first-line supervisors in the plant face common problems in dealing with super— iors and subordinates, they communicate to exchange advice, solutions, and support, even though such communication is not officially mandated. In each case the task goal toward which cooperative activity was directed was a col— lective goal, with which various members had become famil- iar due to their experience in the organization. This is not to say that such collective goals exist ready-made; on the contrary, they sometimes require formation through complex negotiation, as in the case of a coalition. And in every case they involve a process of interpretation-- there is no overall univocal perspective according to which the collective goal is completely determined. But interpretation runs in two directions—-the collective goal allows a reinterpretation of the employee's individ— ual task role and goals, and allows him to see new Options for activity and communication. When will various associational forms develop, and who will they include? In part this is a question about the nature of the necessity with which task goals call forth associational generating mechanisms. Due to the incomplete specification of tasks and goals at all orga- nizational levels, it is necessary that some associational structure develop in the organization to interpret task 107 and goal assignments. But because the task goal is incompletely specified, the Specific associational struc- ture that arises is incompletely determined. At present I can cite only three non-task antecedents that condition associational structure. First is the form of the hierarchical control system. Individual workers cannot easily deviate from their roles if they are constantly observed and sanctioned for deviations. Even coalitions at high levels in the firm are limited by budgetary and departmental constants. Employees will interpret orga- nizational goals and form associations grounded in behaviors over which they have some level of control. Second is the requirement of joint experience. Employees working different shifts or in different plants (Miller, 1959), or employees whose experience gives them basically different orientations, find it hard even to cooperate, let alone to form a mutual overriding goal (Lawrence and Lorsch, 1973; Kingdon, 1973). But not only similar experience is necessary; conjoint experience is required. Only if two members know each other's jobs will they know that membership in an association might be useful. (Of course, this does not imply that gyggy pair of mem- bers have conjoint experience.) Finally, members must be indirectly interdependent. Direct interdependence requires coordination and communication as a condition of 108 doing one's job. Indirect interdependence requires coordination, and communication, as a condition for altering the parameters that govern one's job. (Of course employees can also be, and nearly always are, indirectly interdependent if they are directly interde- pendent.) Our three mechanisms can now be arrayed and examined. In the task structure, a member learns the concrete and the organizational logic of the task he performs. In the hierarchy he simplifies his image of his role, learning what is required as activity. In the associational structure he learns what parameters of his task role can be altered, and how such alterations can be accomplished, in the interests of the achievement of a broadly conceived task goal. So arrayed, the three types of generating mechan- isms resemble the three stages of joint activity in which coordination may be sought. In the task structure, an employee learns to coordinate his execution of a task with interdependent others, and does not question the plgp_which constitutes his work role. In the hierarchical structure, the employee learns to question the ordinary plan, learns the extent to which he can plan his own activity, but does not question the overall gggl and its correlative control mechanisms which bound his autonomy. 109 In the associational structure he learns to interpret or posit broad task goals, bounded by overall requirements of organizational coordination, and to cooperate with other employees in redesigning activities to achieve the "new" goal. In addition, I believe that participation in the later types of social units requires more complex role- taking and coordination skills, though the coordination to be accomplished is often on much simpler tasks (exchange of tasks, voting in a meeting). But the mech- anisms are sequential in another respect, as well: the socialization routines of the mechanisms are logically (though not always temporally) sequenced. Until a worker has a clear conception of the range of operations he can perform, he can't be very clear about the structure of rules involved in controlling those outputs-~he can't even clearly know which Operations are controlled. Under- standing the logic of his task is necessarily prior to understanding why his task is controlled the way it is. Similarly, until a member understands the control struc- ture that determines his retention in the organization, he can't understandingly participate in associational activities which seek to circumvent that structure. His participation might be unsafe for him--he might unknow- ingly violate some strongly enforced organizational rule. 110 His participation is not even safe for other group mem- bers--until he establishes a clear relationship with the control structure, they cannot be sure where he stands. I believe, but will not try to demonstrate, that a somewhat stronger statement about mechanism sequencing could be made. The sets of rules composing the mechan— isms are learned in stages, described above. I believe that no stage for a later mechanism can be achieved, until the same stage has been achieved for earlier mechanisms. This means, empirically, that cognitive knowledge of organizational roles interdependent with one's own and coordinating facility in dealing with other members, are achieved first for interdependent roles in the task structure, then in the hierarchy, then in the work-group or coalition. While this proposition is not vital to the logic of the analysis presented above, it is interesting and testable in its own right. This completes our analysis of social generating mechanisms. We have pointed out the tasks and knowledge states that distinguish the mechanisms and pointed out the developmental sequence followed by an incoming member of an organization. Propositions and Rationale In this section influences on the occurrence and dominance of various generating mechanisms are 111 investigated. The pattern of justification here is once again functional: a certain mechanism exists or predomi- nates because it is necessary or best adapted to achieve task coordination. In what fashion do task variables influence the presence or dominance of generating mechanisms? That question is hard to answer without an answer to a prior one: what do we mean, operationally, by the "presence" or "dominance" of a generating mechanism? Two answers are available to this second question. First, we can examine the rationales given by organization members for their various communicative acts. If we ask, for instance, why an employee passes along information to various others in the organization, he may reSpond in one of three ways. First, he may make reference to his task role and its requirements, or to the goal of the organization: "I let them know the client's sick so they can make a house call to check on him." Second, he may make reference to the expectations of his superiors in the organization: "My boss blows up if we don't let him know about every little problem." Third, he may make reference to an arrangement he has with some other organization member: "Sam and I keep each other posted on anything that crops up." Of course, each of these modes of reference covers many concrete 112 reasons and several distinct modes of expression, which are deducible from the discussion in the last section. Clearly, the continued or extended use of rationales in one of these three categories constitutes operational evidence of the presence of a generating mechanism regu- lating communicative practices. For another source of Operational indications of the dominance of a generating mechanism, I am indebted to Van de Ven, Delbecq, and Koening (1976), for setting forth a typology of coordinating mechanisms. Several of the mechanisms the mention clearly fall within the category of task structure: "formally or informally understood polic— ies and procedures," "predetermined work plans," and "a standing committee . . . to plan and coordinate work." Others are equally clearly related to a hierarchical con- trol structure: "the unit supervisor," "an assistant unit supervisor," "a formally designated work coordinator." Two final mechanisms are more dubiously connected to the third generating mechanism type, the associational struc- tures: "informal communication channels," and "unsched— uled group meetings" (1976, p. 327). Organization members, by indicating which of these mechanisms they use to coordinate work, produce some evidence of the existence and dominance of various generating mechanisms. The evi- dence in this case is not fully valid, since the 113 "coordinating mechanisms" of Van de Ven pp pp. may be interpreted as rules, rather than as the generating units underlying those rules--after all, unscheduled group meet- ings may be a task-role requirement for some members, and a superior may easily require the use of predetermined work plans or even horizontal communication channels. Van de Ven pp pp. produce some evidence of the content validity of their mechanisms in research which bears directly on our focal question: what factors influence the dominance of generating mechanisms? AS in Chapter IV, it is probably simplest to array our reason- ing and propositions according to their independent vari- ables, those describing task interdependence. 1. Intensity of interdependence indicates the extext to which contingencies of one job must be Specifi- cally adjusted to by people doing other jobs. Since our conception of interdependence takes into account the variability and intractibility of component tasks, the argument of Galbraith (1973) explained in the last chapter, once again holds: task role rules and especially the efforts of superiors are too limited to cope with the high uncertainty represented by high interdependence. There- fore, the coping must be supported primarily by associa- tional mechanisms, and the proposition follows that greater intensity of interdependence leads to greater 114 dominance by associational structures and reduced domin- ance by hierarchical structures. This reasoning receives some indirect support in the research of Hage (1974) and Van de Ven pp pp. (l976)--greater task interdependence and uncertainty does lead to increased reliance on infor- mal, especially horizontal, communication as a coordina- tive strategy. 2. The patterning of interdependence primarily indicates the size and "shape" of highly interdependent groups. But the larger the group is, the harder it is to use associational mechanisms, due to the requirement of conjoint experience; large but cohesive work groups and alliances are rare indeed. Thus, the patterning of interdependence mediates the influence of the intensity of interdependence: the larger an interdependent group is, the less the dominance of associational mechanisms, and the greater the dominance of task role mechanisms. This proposition receives firmer support from Van de Ven pp pp. (1976): as the size of the work units they studied increased, the dominance of formal coordinative mechanisms did as well. 3. Directionality of interdependence indicates the extent to which interdependence is actually dependence. The logic of associational mechanisms makes it unlikely that they will come into play in situations where one 115 party is dominant in the coordinative relationship--the exchange of information, orders, and activity is usually as unidirectional as the interdependence. Moreover, it is usually most efficient to formally link the dependent job to the independent job. The following proposition seems plausible: the greater the directionality of interdependence, the greater the dominance of task struc- ture as a generating mechanism. This proposition too receives some support from the study by Van de Ven pp pp.: in work units with sequential interdependence--the most directional form distinguished by Thompson (l967)—-the occurrence of formal coordination devices like plans and standard procedures reached its apogee. This completes our discussion and justification of propositions characterizing generating mechanisms as func- tionally related to task coordination requirements. CHAPTER VI CONCLUSION AND CAVEATS The structure or gridwork of a rules theory of organizational communication has now been presented. The variables of central concern have been defined concep- tually, and operational procedures for their measurement have been indicated. A network of propositions linking these variables, rationally derived from and justified by the principle that organizational communication arrange- ments are functionally adapted to task requirements, has been enunciated, and the limited available empirical evi- dence bearing on the theory has been marshalled. Several caveats are in order, however. First, precise and complete operational definitions of the vari- ables have not been provided. In part, this is because their conceptualizations are still in the process of develOpment; in many ways, this thesis is an evolutionary milestone rather than a completed product. But more importantly, the operational definitions of many key variables must be organization-relative. Organizations have a wide variety of tasks and conditions, and whatever 116 117 questions we ask of subjects must be adapted to those con- ditions. My variables, like my theory, are abstract and require specific interpretations for concrete contexts. A similar caveat applies to the theory as a whole. In the first chapter I mentioned several functional alternatives to communication; as intensity of interde- pendence increases they become less and less efficient, and thus of rare importance; but still they limit the necessity of my propositions. Other factors, too, may intervene: Mohr (1971), for instance, cites the noise level in a factory as a variable which sometimes criti- cally affects the communication there. The precise nature range of the necessity of these prOpositions will have to be determined by empirical research. Finally, there is the question long ago raised by Hempel (1965): how much explanatory force has a func- tional prOposition? I have argued for the necessity of my prOpositions by referring to the "rationality" of the organization, but I think a somewhat stronger ground exists. Although an organization may not achieve its goal, it nearly always carries out some kind of process. Given descriptions by employees of the kind of process they carry out from day to day, I have tried to character— ize the kind of communication rules that must have been present, for them to even think they were effectively 118 carrying out a process. My propositions, then, have at least the necessity Hempel granted to functional eXplana- tions: for activity of a certain order of complexity to take place, communication rules of a certain sort have to have been available to actors. More precise design of my propositions, to reflect the kind of adaptation of communication to task that actually occurs in organiza- tions, must await a research effort. LIST OF REFERENCES LIST OF REFERENCES Amarel, S. On representations of problems of reasoning about actions. In D. Michi (Ed.), Machine intelligence 3. London: Oliver and Boyd, 1968, 131—171. Ajzen, I. Attitudinal and normative messages: An investigation into differential effects of per- suasive communication on behavior. Sociometpy, 1971, 34, 263-280. Blau, P. Bureaucragy in modern society. New York: Ran- dom House, 1956. Bonini, C., Jaedicke, R., and Wagner, H. Management con- trols. New York: McGraw-Hill, 1964. Burns, T., and Stalker, G. The management of innovation. London: Tavistock, 1961. Chapple, E., and Sayles, L. The measure of management. New York: Macmillan, 1961. Crockett, W., Press, A., Delia, J., and Kenny, C. Struc- tural analysis of the organization of written impressions. University of Kansas, 1974. Cushman, D. Alternative theoretical bases for the study of human communication: The rules perspective. Paper presented at the annual convention of the Speech Communication Association, 1975. Cushman, D., and Craig, R. Communication systems: Inter- personal implications. In G. Miller (Ed.), Explorations in interpersonal communication. Beverly Hills, Ca.: Sage, 1976, 37-58. Cushman, D., and Florence, B. The development of inter- personal communication theory. Today's Speech, 1974, 22, 1-15. Cushman, D., and Pearce, W. Generality and necessity in three types of communication theory, with Special attention to rules theory. Communication Year- book, 1977, 1, 173-182. 119 120 Cushman, D., and Whiting, G. An approach to communication theory: Toward consensus on rules. Journal of Communication, 1972, 22, 217-238. Cyert, R., and March, J. A behavioral theory of organiza- tional objectives. In M. Haire (Ed.), Modern organizational theory. New York: Wiley, 1959, 76-90. Dewhirst, H. Influence of perceived information-sharing norms on communication channel utilization. Academy of Management Journal, 1971, 14, 305-315. Drucker, P. Management: Tasks, responsibilities, prac- tices. New York: Harper and Row, 1969. Emery, F., and Trist, E. Socio-technical systems. In C. Churchman (Ed.), Management sciences, models and techniques. London: Pergamon Press, 1960, 83-97. Farace, R., and Albrecht, T. Network analysis biblio- graphy. Unpublished manuscript, Michigan State University, 1975. Farace, R., Monge, P., and Russell, H. Communicating and organizipg. Reading, Ma.: Addison-Wesley, 1977. Galbraith, J. Designing complex organizations. Reading, Ma.: Addison-Wesley, 1973. Galbraith, J. Organizational desigp. Reading, Ma.: Addison-Wesley, 1977. Goofman, E. Strategic interaction. New York: Ballantine Books, 1969. Gross, E. Some functional consequences of primary con- trols in formal work organizations. American Sociological Review, 1953, 18, 370-376. Hage, J. Communication and organizational control. New York: Academic Press, 1974. Hage, J., and Aiken, M. Relationship of centralization to other structural properties. Administrative Science Quarterly, 1967, 12, 72-92. 121 Hall, R. Intraorganizational Structural variation: Application of a bureaucratic model. Adminis- trative Science Quarterly, 1962, 7, 295-308. Hall, R. Organizations: Structure and process. Eagle- wood Cliffs, N.J.: Prentice-Hall, 1971. Hempel, C. The logic of functional analysis. In Aspects of scientific explanation. New York: Free Press, 1965, 297-330. Homans, G. The human group. New York: Harcourt, Brace, and World, 1950. Jackson, J. Normative power and conflict potential. Sociological Methods and Research, 1975, 4, 237- 263. Jacobson, E. Personal communication, 1976. Katz, D., and Kahn, R. The social ppychology of organiza- tions. New York: Wiley, 1965. Katz, F. Organization and autonomy. New York: Random House, 1968. Kelly, G. The psychology of personal constructs. New York: Norton, 1955. Kelly, G. Europe's matrix of decision. Nebraska gym- posium on Motivation, 1962, 9, 83-125. Kepner, C., and Tregoe, B. The rational manager. New York: McGraw-Hill, 1964. Kingdon, D. Matrix organization. London: Tavistock, 1973. Korman, A. "Consideration," "initiating structure," and organizational criteria--A review. Personnel Psychology, 1966, 19, 349-361. Lauer, R., and Boardman, L. Role-taking: Theory, typology, and propositions. Sociology and Social Research, 1971, 23, 137-147. Lawrence, P., and Lorsch, J. Organization and environ- ment. Boston: Harvard Business School, 1967. 122 Lawrence, P., and Lorsch, J. Developing organizations: Diagnosis and action. Reading, Ma.: Addison- Wesley, 1973. Leavitt, H. Managerialgpsychology. Chicago: University of Chicago Press, 1958. Lee, Wayne, Decision theory and human behavior. New York: Wiley, 1970. Mead, G. Mind, self, and society. Chicago: University of Chicago Press, 1934. Michaels, J., and Wiggins, J. Effects of mutual depen- dency and dependency asymmetry on social exchange. Sociometry, 1976, 39, 368-376. Miller, E. Technology: territory, and time. Human Rela- tions, 1959, 12, 243-272. Miller, E., and Rice, A. Systems of organization. London: Tavistock, 1967. O'Reilly, C., and Roberts, K. Information filtration in organizations: Three experiments. Organizational Behavior and Human Performance, 1974, 11, 353-365. Pearce, W., and Cushman, D. Research employing the con- struct communication rules: A test of the pro- cess of explanation and verification. Unpub- lished manuscript, 1977. Perrow, C. A framework for the comparative analysis of organizations. American Sociological Review, 1967, 32, 194-208. Perrow, C. Complex organizations: A critical essay. Glenview, I1}: Scott, Foresman, and Company, 1970. Rapaport, A., and Guyer, M. A taxonomy of 2 x 2 games. General Systems, 1966, 11, 203-213. Roethlisberger, F., and Dickson, W. Management and the worker. Cambridge, Ma.: Harvard University press, 1948. Rubin, Z., and Brown, B. The social_psychology of bar- gaining and negotiation. New York: Academic Press, 1975. 123 Sayles, L. Behavior of industrial work groups. New York: Wiley, 1958. Scheff, T. Toward a sociological model of consensus. American Sociological Review, 1967a, 32, 32-46. Scheff, T. A theory of coordination applicable to mixed- motive games. Sociometry, 1967b, 30, 215-234. Schelling, T. The strategy of conflict. New York: Oxford University Press, 1960. Schroder, H., Driver, M., and Streufert, S. Human infor- mation processing. New York: Holt, Rinehart, and Winston, 1967. Schwartz, S. Normative explanations of helping behavior: A critique, proposal, and empirical test. Journal of Experimental Social Psychology, 1973, 9, 349- 364. Searle, J. Speech acts. New York: Cambridge University Press, 1969. Simon, H. Administrative behavior. New York: Macmillan, 1948. Simon, H. Some strategic considerations in the construc- tion of social science models. In P. Lazarsfeld (Ed.), Mathematical thinkingyin the social sci- ences. Glencoe, Il.: Free Press, 1954, 388-412. Simon, H. A behavioral model of rational choice. In Models of man. New York: Wiley, 1957a, pp. 241- 260. Simon, H. Rationality and administrative decision making. In Models of man. New York: Wiley, 1957b, 196- 206. Thibant J., and Kelley, H. The social psychology of groups. New York: Wiley, 1957. Thompson, J. Organizations in action. New York: McGraw- Hill, 1967. Toulmin, S. Reason in ethics. New York: Oxford Univer- sity Press, 1947. 124 Toulmin, S. The uses of argument. New York: Oxford University Press, 1962. Toulmin, S. Reasons and causes. In R. Borger and F. Cioffi (Eds.), Explanation in the behavioral sciences. Cambridge: Cambridge UniverSity Press, 1969a, 1-26. Toulmin, S. Concepts and the explanation of human ‘ action. In T. Mischel (Ed.), Human action. New York: Academic Press, 1969b, 71-104. Turner, R. Role—taking: Process versus conformity. In A. Rose (Ed.), Human interaction and social process. Boston: Houghton Mifflin, 1962, 21-40. Van de Ven, A., and Delbecq, A. A task-contingent model of work-unit structure. Administrative Science Quarterly, 1974, 19, 183-197. Van de Ven, A., Delbecq, A., and Koenig, R. Determinants of coordination processes within organizations. American Sociological Review, 1976, 41, 322-338. Veron, E. Pragmatics of meaning. Semiotica, 1968, 2, 37-74 0 Walter, B. Internal control relations in organizational Rierarchies. Administrative Science Quarterly, 1966, 11, 179-206. von Wright, G. Explanation and understandipg. Ithaca, N.Y.: Cornell UniverSity Press, 1972. “William