Les reactions collectives, A l’occasion de tensions qui se manifesteraient dans des sous-unitbs d’organisation sociale complexe, varieraient en nombre et en caract6re selon le niveau du consensus et la puissance relative des sous- unites. Le niveau de consensus et le degr6 de cohbsion, ce dernier &ant assocfe au consensus, seraient fonction de la differentiation des r6les et d r la frequence des interactions au sein des sous-unites. A mesure que la differentiation dcs r6les et que le t a m des interactions sont determines par la technicite, ce type d’organisation m&me influence les reactions collectives en face de tensions. Des d o n n k recueillies dam une firme industrielle canadienne indiquent que le consensus s’associe 2 la densite des effectifs plut6t qu’8 la differentiation des r6les, alors que le degre de cohesion s’associe plus A la differentia- tion qu’8 la densite des effectifs. Le consensus et la cohesion semblent s’interposer dans les rapports entre la satisfaction au travail et le rendement, l’absenteisme et le roulement du personnel.

Structural Determinants of Consensus and Cohesion in Complex Organizations *

Frank E. Jones, Mcivlaster University

In this article, an attempt is made to explain collective responses to stress in complex organizations. On the basis of a niimber of empirical generalizations which focus on certain aspects of technological organization and which re- veal associations with several relevant aspects of organiza- tional behaviour, an attempt is made to provide a systematic explanation of the relation between technological organiza- tion, consensus and cohesion, and collective responses to stress. Data relevant to certain propositions formulated in the argument are presented and discussed. Although technological organization1 is given sh t : hrift in the major theories of complex organizations 8.; well ,-is in text-book discussions,Z the impact of technological orga- nization on organization structure and functioning is evident in various field studies of organizations. Several of these studizs identify relationships between elements of techno- logical organization and aspects of organizational structure and functioning which are directly relevant to the problems arialysed here. For example, certain studies provide evi- dence of a relationship between technological organization and the presence of stress? among participants in complex organizations. Blauner’s research4 reveals a relationship between an individual’s involvement in his work and the technological conditions of his employment. Distinguishing four kinds of productive process5 (craft, machine-tending,

*A version of this paper was presented at the annual meeting of the Canadian Sociology and Anthropology Association, Calgary, June 1968. I wish to acknowledge financial assistance from McGill University and the Canada Council.

1The term technology refers to the tools, instruments, machines and the different forms of energy utilized in organizational activities. Technological organization refers to arrangements of these elements.

=Such as those proposed by Weber, Simon, Etzioni. Among text- books, an exception to the above generalization is Robert Dubin, The World of Work (Englewood Cliffs, 1958).

3Various concepts such as alienation, satisfaction - dissatisfac- tion and tension are used in the research included in this review which refer to conditions or processes which can be subsumed under the concept stress. The latter concept will be defined later in the discussion.

4Robert Blauner, Alienation and Freedom (Chicago and London, 1964).

5He refers to these as types of technotogy. Although he employs different measures to identify the class of any particular firm, his classification closely parallels that developed by J o a n Walker and discussed below.

219

assembly-line, and continuous-process) and identifying sev- eral dimensions of alienation (powerlessness, meaningless- ness, self-estrangement and social alienation), he shows that alienation varies in relation to the type of technology, with alienation tending to be low in work situations characterized by craft technology, increasing where machine-tending and assembly-line production are utilized, and decreasing in situations employing continuous process technology? The relevance of Blauner’s research to present interests lies not only in its linking of technology and stress but also in its identification of the consequences of technology for worker autonomy and of the relation of the division of labour to social integration? of the work group. Walker and Guests are explicitly concerned with the consequences of assembly line technology for the attitudes of workers and for “the social structure of in-plant society.” They attribute much employee dissatisfaction to repetitiveness in the tasks workers perform and to the lack of worker control of pro- duction processes. They describe how the assembly line, by requiring workers to be stationed in certain positions, influences social participation, limiting it to a loose group of five or six Qperators employed in adjacent work station^.^ The division of labour, and the consequences of technologi- cal organization for worker autonomy and for frequency and patterns in social interaction, then, are prominent in their analysis.

Turner and LawrencelO report that satisfaction is posi- tively and absenteeism negatively related to various aspects of technology including the frequency of interaction and the degree of worker autonomy.ll

6In all of these summaries, it is necessary to omit the authors’ qualifications. Blamer, for example, states that no single $en- eralization can convey an accurate picture of the relationship between technology and alienation since the technology utilized by specific industries may have different effects on the dffPerent components of alienation.

?Social integration is used to refer, in some instances, to the identification of workers with each other, and in others, to the identification of workers with management or the company. In the former use, it is similar to cohesion as used in this analysis.

8Charles R. Walker and Robert H. Guest, The Mun on the Assembly Line (Cambridge, Massachusetts, 1952).

9 A s there is overlapping of such groups, each worker belongs to a group which differs slightly in membership from the group of every other worker.

loArthur N. Turner and Paul L. Lawrence, ZndustriuZ Jobs and the Worker (Boston, 1965).

11The measure, which is referred to as RTA (Required Task Attributes) is based on these six components: 1. variety of ob- jects, tools, work pace: 2. required interaction among workers, 3. op~ional interaction on or off the job, 4. knowledge and slcill, 5. autonomy regarding decisions involving different aspects of the job, 6. respolzsibility.

220

Woodwarc1,’z whose research clearly documents the rela- tionship between technological organization and organiza- tional structure and functioning, is more interested in the consequences of technoIogy for managerial organization*3 than in the variables relevant to the present analysis. How- ever, she does comment that “industrial relations certainly seem to be better in process industry than in large batch and mass production”14 and also shows that interaction patterns of production superintendents changed in three plants changing from unit or small batch to assembly line production.15 To express these findings in terms consistent with the present discussion, technology is associated with different levels of stress and with different levels of inter- action among organizational participants.

Studies of the establishment of automated procedures in various industrial operations and in business offices reveal that, among other consequences, 16 automation is followed by changes in the division of labour in the form of reduced specialization, reduced worker control of the production process, closer supervision of workers, increased physical distance between workers, and changes in the organization of workers from individual line operations to teams of three or four workers. While these studies provide little inter- pretation of the link between automation and such changes, they focus on variables recognized by other researchers as important in describing and explaining the relationship be- tween technical organization and the structure and func- tioning of cornplex organizations.

AIthough all the studies reviewed so far are relevant to this discussion because they concern the relationship of

laJoan Woodward, Industrial Organization: Theory and Practice (London, 1965).

13Characterizing some 100 manufacturing firms in Great Britain according to their mode of production - batch, assembly lint? or continuous-process production - she shows that there are central tendencies associated with each production type in the number of levels of management, ratios of supervisory to non- supervisory personnel and the span of control of first line super- visors. See also W. H. Scott, Technologicat Change in Industrial Relations (Liverpool, 19561, for another study which documents the consequences of technological change for managerial organiza- tion.

14Woodward, Industrial Orgunkation, 55. She uses “industrial relations” and “human relations” interchangeably.

15Zbid., 216.

160ther consequences include changes in job content in terms of increased responsibility, and increased demand for work or attention to the productive process: and changes in supervision in the f o r m of a reduction in the levels of supervision and greater technical competence required of supervisors. See, fo r example, W. A. Faunce, “The AutomobiIe Industry: A Case Study in Automation;” in H. B. Jacobson and J. S. Roucek, eds., Auto- mation and Society (New York, 19591; F. C. Mann and L. R. Hoffmann, Automation and the Worker (New York, 1960); and R. Walker, Toward the Automatic Factory (New Haven, 1957).

221

~~

technological organization to some variable to be included in the analysis, several other studies appear to have greater relevance because they give attention, implicit or explicit, not only to stress but to the conditions which intervene between technology and the perception of and response to stress. For example, Warner and L0w17 see technological change in the form of mechanization as establishing certain conditions which allowed a strike to take place in all the shoe factories of Yankee City. Changes in shoe manufac- turing technology - had drastically reduced occupational differentiation, in terms of skill and prestige, and destroyed a well-structured channel for the realization of status as- pirations, leading to an increase in hostility among workers. This reduction in occupational differentiation also led to an increased consensus among shoe workers about the sources of their difficulties and the appropriate mode of response to them. Technological change, by reducing the workers’ autonomy in the production process, was another source of frustration and consequent hostility among rank and file workers. Moreover, heavy capital expenditures required by the introduction of machine production into shoe manu- facturing caused a shift from commzhnity-based to absentee ownership and imported management. In consequence, labour-management conflict was no longer subject to the moderating influence deriving from non-contractual obliga- tions which had developed when owners and managers were long-time residents of the community and interaction be- tween them and their workers extended beyond the factory to other community rdles and associations.

Technological components are prominent in Kerr and Siegel’s18 explanation of variations in the frequency of strikes among different types of industry. Occupational differentiation is explicitly identified as a critical variable: highly strike-prone industries are characterized by tech- nologies requiring a relatively unskilled labour force which is occupationally specialized but homogeneous, such as the miners, longshoremen, or loggers, while industries whose technologies require a labour force which is non-specialized and heterogeneous occupationally demonstrate a low pro- pensity to strike. Occupational homogeneity implies simi- larity of experience,lg that is, shared definitions of the situation, while occupational specialization can restrict op- portunities for occupational mobility and result in “accumu- lated tensions.” A strike is interpreted as a means of reducing such tensions.20

Although Gouldner does not provide detailed descriptions of technological organization for either gypsum mining or

17W. L. Warner and J. 0. Low, The Bocial System of the Nodern Fuctory (New Haven, 1947).

18Clark Kerr and Abraham J. Siegel, “The Inter-Industry Pro- pensity to Strike - an International Comparison,” in A. Korn- hauser, R. Dubin and A. Ross, eds., Industrial Conflict (New York, 1954).

IgZbid., 110.

*oIbid., 110-112.

222

plaster board manufacturing, he refers to differences in worker autonomy, as reflected in flexibility of production scheduling and in the presence of a conveyor in board manu- facturing and its absence in mining operations, and differ- ences in the extent of technologically required interaction as being among the differences in work environment which he believes explain the lesser influence of the supervisor’s authority, the greater permissiveness in relation to infrac- tions of the rules, and the higher degree of cohesion among gypsum miners as compared to workers engaged in manu- facturing operations in the gypsum plant.21

British studies22 of coal mining operations explicitly relate variations in technical organization to variations in social organization. In the hand-got system which preceded mechanization, coal was obtained by men using picks, work- ing in pairs or trios in various locations along the coal face. The miners, who provided their own tools and who had learned their skills from their fathers, were regarded by themselves and by others in the industry as craftsmen. As mechanization was introduced, changes occurred in work organization and in occupational definition. In the longwall system, where coal was cut by machine but loaded by hand on a conveyor belt, the miners’ status 2s independent crafts- men vanished and instead of working in small teams, the miners were strung out in individual positions along the coal fa~e.~3 The permissiveness which Gouldner associates with gypsum mining is also present in the hand-got system whereas strict supervision and strict demands for confor- mity to the rules are associated with the longwall system. Gouldner also regards technological change as contributing to disturbed management-worker relations where such changes introduce ambiguity in r6le expectations and per- formances, and where technological changes, introduced as part of a drive for greater efficiency, are interpreted by the workers as a form of close supervision.

Sayles’ research24 on grievance behaviour is most relevant to this analysis. He observed that work groups vary in

21A. Gouldner, Patterm of IndwtriaE Bureamacg (New York, 1954).

22See, for example, J. H. Goldthorpe, ‘Technical Organization as a Factor in Supervisor-Worker Conflict,” British Journal of SOC- bZogy, X (19591, 213230, and E. L. Trist and V. Bamforth, “Some Sacial and Psychological Consequences of the Longwall Method of Cd-Getting,” Human Relations, IV (1951), 3-38.

23Technological developments in coal-mining have developed be- yond the degree of mechanization present in the longwall system. Goldthorpe describes full mechanization in British mines where coal is cut and loaded by machines. Technological advances in coal-mining in the United States have sharply reduced the em- ployment of miners. In some instances, the miners have been completely replaced by machines. See William Francois, Auto- mation: Indwtrializath Comes of Age (New York, 1964), Chap- ter 1.

24leonard Sayles, The Behavior of Irtdwtriat Work Groups (New York, 1958).

223

frequency and pattern of grievance expression and provided evidence that such variations are associated with differences in technological organization : work groups characterized by a low division of labour, individual, team or short assembly line production, medium-skill jobs, primarily worker-con- trolled production and essential jobs, tend to express grie- vances more frequently than work groups characterized by individualz6 or long assembly line operations, high division of labour and high-skilled or low-skilled jobs.% While Sayles does not provide a systematic explanation for this relation- ship, it is evident that he regards these elements of tech- nological organization as determinants of different levels of consensus and differences of power among work units which, in turn, influence the frequency and patterning of grievance expression.27

Although there is convincing evidence of a link between technological organization and various aspects of complex organizational structure and functioning and considerable agreement on the nature of the critical variables, the sig- nificance of technological elements as independent variables requires more precise statement than can be found in the literature. It is not clear, for example, which technological elements determine which aspects of structure and function- ing, which simply set limits on variation, and which are only regarded as associated with organizational variation. The solutions to such questions, however, are not to be found in research restricted to documenting the existence of a relationship between technological organization and other aspects of complex organizations but through the development of a systematic formulation of the relationship.

This article undertakes a task so far neglected by stu- dents of complex organizations. It aims at providing a systematic explanation of the relationships between selected aspects of technological organization and selected aspects of the structure and functioning of complex organizations. To do so, an attempt will be made to identify certain critical variables of technological organization, whatever the tools and forms of energy utilized, which bear on the development and maintenance of consensus and cohesion. Having done so, propositions relating these variables to each other and, in turn, to collective responses to stress, will be formulated.

z5Individual operations can be assodated with high or low griev- ance expression; the difference depends on the density or the manner in which the workers are spread out as a consequence of their tasks.

26The technical organization distinctions which differentiate pat- tern (that is, the degree of planning of grievance activities) are not as clearly distinguished although differences in power to damage management and in opportunity for interaction appear to be critical variables.

WSayles, The Behavior of Industriul Work G r w p , Chapter 3, especially 42, 59, 70-83, 93.

224

Technology and Social Organization

Technological conditions influencss the way that overall tasks are subdivided and assigned to the employees of an organization. Depending on the nature of the product, the tools and machines available, occupational differentiation or division of labour may vary from simple to complex. Some operations may require a number of persons to per- form the same task, such as assembling components for radio or T.V. sets, while other operations may involve a number of persons who perform different tasks, such as testing and repairing. Technological conditions may also determine how employees are deployed, that is, whether the required tasks are performed by individuals working in relative isolation from others, by individuals working side by side, engaged in non-continuous individual operations, by individuals working on assembly lines, and by individuals working in pairs or small teams. Moreover, the density or concentration of workers in the work area may vary independently of their deployment. Technological condi- tions may also determine variations in the importance of different occupations or of work units29 in the flow of organizational activities. Some operations may be more important than others in that a stoppage in one may have a greater effect on the others. Technological conditions also determine the degree of worker control over the pro- ductive process, varying from complete control to none at all. As technological conditions will determine the mix- ture of occupational skills required by a given production process, such conditions will establish a skill hierarchy

28Although much of the evidence suggests that the organization of the work force is technologically determined, it must be re- membered that most generalizations are empirical rather than theoretical. While it is likely that technological conditions in- creasingly determine rather than infZuence the organization of the work force as there is a shift from human to mechanical forms of energy, the possibility remains that the resulting organ- ization of the work force at any technological level represents management’s judgment as well as the dictates of an inexorable logic of technology. While there is no question, for example, that an assembly line may determine the density of workers in a physical area, density will also be influenced by management’s decisions concerning the use of available space. At any rate, my argument does not require work force organization to be inherently ordered by technological conditions although it is concerned with the consequences for social organization of the use, whether technologically or managerially determined, of specified technological resources. In using the term “determine” I mean that decisions regarding alternative modes of organiza- tion are dictated by technoIogical conditions but as a completely technological determination is unlikely and as the degree of determination will be influenced both by the nature of the tech- nological conditions and the specific variables of work force organization, “determine” should always be understood as “determ- ine in some degree.”

29The operations of complex organizations are frequently mtrib uted among subunits of the organizations such as departments, divisions, sections, etc. The term work units refers to distinttly identified groupings of employees engaged in the different activ- ities of the organization.

225

within and among work units. Technological conditions, then, may divide the participants of an organization, hier- archically and qualitatively, in terms of occupational rdles, work units, or both, and, as will be argued below, these divisions will have consequences for consensus and cohesion among organizational participants. Technological condi- tions may determine the way the work force is deployed and how densely it is located in the work area and conse- quently may determine the frequency of involuntary social contacts and influence the frequency of voluntary social contacts between organizational participants. Technological conditions may determine the relative autonomy of worker from machine or production processes and the location of a worker or work unit in the production process. As these provide differential opportunities to benefit or harm man- agement, technological conditions may be said to influence the distribution of power among workers and work units. Given these consequences of technological conditions, the next step is to consider how r6le differentiation and inter- action frequency are related to collective responses to stress. Stress refers to the individual reactions of participants in a social system to a perceived failure, actual or imaginary, of other participants to conform to r6Ie expectations. To the extent that non-conformity is perceived, a person ex- periences stress, taking the form of feelings, varying in intensity, of dissatisfaction, anxiety, or other negative emo- tional responses. It is assumed that the individual responds by attempting in some way to reduce stress. Collective responses to stress refer to shared or agreedsn stress- reducing responses expressed by group members. Thus, the various forms of grievance expression discussed by Sayles are regarded as collective responses to stress but productivity control, absenteeism, and quitting, although usually considered to be actions of individual workers, may also be regarded as collective responses to stress. Thus, while it is quite possible for an individual worker to control his output, assuming technological conditions allowing the worker rather than a machine to set the pace, sociological research in industry makes it abundantly clear that produc- tivity control is a frequent form of collective action. In some instances, it may be explicit union policy to slow down or work to rule; in others, productivity control may be a result of informal understandings among workers con- cerning output levels and the application of group norms and controls to inhibit productivity beyond levels the work- ers regard as appropriate. Where absenteeism and quitting occur in the context of collective approval or disapproval, it seems reasonable to extend the term collective action to such behaviour.30 3% a study of infantry recruits, the writer observed that going absent without Offidal leave (“taking off,” “going over the hill”) when a soldier became fed up with army life was a response collectively approved by the members of the unit (who shared living quarters). When a man or a group of men announced

[Continuedl

Necessary Conditions for Col- fective Responses to Stress

226

Generalizing from Sayle’s argument, i t may be asserted that collective responses to stress require, for their expres- sion, certain levels of consensus and power. With reference to consensus alone,31 this generalization means that collec- tive responses to stress, whether overtly in some form of grievance expression or covertly as production control, absenteeism or quitting, may be regarded as cooperative behaviour and, as such, facilitated to the degree that there is consensus among the potential c0operators.3~ How- ever, rather than regard consensus merely as a minimal condition for cooperative action, it may be more productive to formulate a co-variant relationship between consensus and the emergence and maintenance of cooperative action. Although it is simple, and justifiable as a working hypo- thesis, to propose that the higher the level of consensus among members of a social system, the greater the likeli- hood of the emergence and maintenance of cooperative action, the form of the relationship may, dep-nclil ; on other conditions, be other than linear and is an appropriate subject for empirical investigation. Given this argxment, the conditions which determine levels of consensus become appropriate areas for theoretical and empirical investiga- tion. Among these conditions are the variations in r6le differentiation and interaction frequency resulting from the technological conditions of the organization. R61e differentiation increases system interdependence but tends to inhibit consensus in any form of social organiza- the intention of going A.W.O.L., others would offer articles of civilian clothing, money and other forms of help to increase the probability of a successful withdrawal.

3lIn this discussion, co12sensus refers to the consensus-dissensus continuum which in turn refers to the variability in shared com- mitment to expectations, cognitions, normative standards and values among the participants in a socia l system. In empirical investigation, the shared elements chosen for study may vary with the problem under consideration. In some instances, values may be the focus of consensual variation; in others, as in the present instance, attention may be restricted to shared definitions of the situation. Although the term consensus may place the emphasis on agreement, it should be understood that the term, as a shorter reference for consensus-dissensus, refers to varia- tions in agreement and disagreement among participants of a soda1 system.

3zDespit.e the apparent rejection of consensus as a critical var- iable by some writers, it is generally agreed that cooperation requires some degree of consensus. This statement reflects the view of “ardef“ theorists such as Talcott Parsons who has stip ulated the existence of shared values and norms as a necessary condition of system boundary-maintenance and of such “conflict” theorists as Dahrendorf, who not only states that some level of consensus is necessary for society to exist but, in his discussion of the process by which a q w i r g r o u p becomes a class, indicates, albeit implicitly, that consensus about interest must occur. For an attempt to reconcile the opposing positions of consensual and conflict theorists, see Robin M. Williams, “Some Further Com- ments on Chronic Controversies,” American Journal of Sociolugy, LXXI, 6 (May 19661, 717-721.

Role Differentiation and Con- setlsus

227

~

t i ~ n . ~ ~ Although dissensus among r6le incumbents partici- pating in a differentiated r6le structure may result from ambiguity concerning r61e expectations,34 the very existence of r61e differentiation has some direct consequences for consensus. R6le differentiation, especially in complex organizations, represents different areas of competence, different sets of tasks which are all related to overall organ- izational objectives but with each directly related to sub- goals or objectives. Incumbents of a specific r6le are, therefore, expected (by co-incumbents and incumbents of other r6les) to devote themselves to their defined tasks and to the achievement of the more limited r6le objectives. Moreover, as they are rewarded for conformity to such expectations, r6le incumbents, acting to maximize rewards, strive to conform to expectations and to develop conditions which facilitate conformity. Under conditions of adequate reward, the incumbent may be expected to identify strongly with the sub-goals and to perceive his tasks and sub-goals, his definitions of what is “important” and what activities and events should have priority, as having primacy over those of incumbents of other rbles. Thus, the greater the r6le differentiation, the greater the differences which may be expected between perceived ranking of tasks and sub- goals.35 Possibly this leads the incumbent to perceive his “rdle field” as co-terminous with the total organization. To the extent that resources are scarce or that requirements of other r6es hinder the r61e incumbent from conforming to r6le expectations and consequently block the achieve- ment of sub-goals, he is likely to regard other role incum- bents as competitors and to insist on his “perception” of the organization as “right.”% It may be said, then, that the greater the rcile differentiation in any complex organi- zation, the lower the consensus among the incumbents of the different rciles.

33This seems to be the import of Durkheim’s argument. See Thar Division of Labor in Society (Glencoe, 19471, 129132. Also see S. M. Upset, PoZiticaZ Man (New York, 19631, for an applica- tion of this idea to account for the level of dissensus and/or conflict in total societies. Studies of complex organizations pro- vide evidence that incumbents of r61es at different levels in the organization perceive the organization, its objectives and its problems differently. See, for example, T. T. Paterson, Mora l s in War and Work: Alz Experiment k the Managment of Man (New York. 1955); W. Caudill, Th.e Psychiutrk Hospital aa a SmaZZ Commwnity (Cambridge, 1958).

34That is, the incumbents of two different r8les may disagree about the legitimacy of the specific allocation of tasks to their respective r8les, about the relative rewards for conformity to r8le-expectations, or about the extent that either incumbent achieves conformity with r61e expectations.

35For further discussion of this problem, see James G. March and . Herbert H. Simon, Organizations (New York, 19581, Chapter 3,

especially 40-44,6!5-70.

36This seems to explain the low levels of consensus among in- cumbents of staff and line positions in certain organizations.

228

Moreover, these consequences of rde differentiation may be heightened to the extent that rdle structure reduces or minimizes interaction between participants or permits inter- action to occur in formal situations such as interdepart- mental meetings.37 Where technological conditions require the involvement of participants who occupy different rdles and where rewards are directed to inter-rGle cooperation, the effects of rdle differentiation on consensus would be minimal.

The assertion that the way rtjle differentiation structures interaction has a bearing on consensus is based on the assumption that some level of interaction is necessary for a person to learn or to adopt the perceptions, feelings and evaluations of another, whether they are incumbents of the same or different r6les. It may be further asserted, therefore, that levels of consensus in social groups will be related to the frequency of interaction among members. While this relationship is undoubtedly interdependent, whether consensus or interaction are increasing or decreas- ing functions of each other depends on the influence of other variables. Basically, the direction of the relationship is determined by the gratificational significance of the interaction to the actors. An actor will seek to act or avoid acting toward another depending on whether he experiences such interaction as gratifying or depriving. Thus, two persons interacting communicate or exchange ideas, per- ceptions, values, norms, and evaluations. To the extent that these communicated elements help to solve either person’s problems, provide support for his meaning system, or otherwise reduce demands on his energy, the person is constrained to seek interaction with the other and, as relevant, accept the other’s ideas, perceptions, and values. To the extent that interaction has effects opposite to those cited, either person will seek to avoid the other and, as relevant, reject the other’s perceptions, etc. It has been reported38 that where a member of a group deviates from expected behaviour, the other members increase interaction with the deviant in an effort to increase or maintain gratifi- cation. However, when the deviant fails to conform follow- ing such a period of increased interaction, the members withdraw or introduce some form of punishing action, that is, where the exertion of energy increases deprivation in- stead of gratification, interaction decreases. In certain situations, the cost of maintaining gratifying interaction may be borne unlaterally, requiring the adoption of the views - the perceptions, feelings, evaluations, the norms and expectations - of the other; or it may be borne bi-

interaction and Consensus

Wtrong identification with sub-goals arising from r61e dmerentia- tion explains the failure to achieve a required level of coopera- tion among the personnel in the air-force base studied by Pater- son. He achieved the required level by promoting interaction among incumbents of dif€erent r6les and developing consensus concerning the identification of the “real enemy,” the weather.

385. Schachter, “Deviation, Rejection, and Communication,” Jour- nal of Abnormal and Social Psychology, XLVI (1951),190207.

229

laterally, in the sense that the norms and expectations of the actors converge to a common position. Where either of these conditions obtains (and the relevant actors regard the interaction as gratifying to them) consensus and inter- action will be positively associated. However, where the interaction is not seen as gratifying? the relevant actors will attempt withdrawal and avoid pressures to consensus. If withdrawal is not possible, as in conditions required by r61e demands in organizations, dissensus may be the ex- pected result of continued interaction. To introduce the idea of reward-significance of interaction is to introduce the idea of cohesion39 into the discussion for, following Homan~,4o it may be asserted that under appro- priate conditions, interaction and cohesion are positively associated. In the case where the interaction of two persons is mutually gratifying, in the sense outlined in the preceding section, each may generalize such gratification to the other as an object-source of gratification; that is, each likes the other because the other gratifies him.41 In short, if one actor gratifies another, the other likes him, is positively emotionally committed to him, and will seek interaction with him: the higher the cohesion? the higher the inter- action. However, the more these two actors interact, the more either obtains gratification, and the more likely the positive emotional commitment will be maintained or further increased; the more the interaction, the higher the cohesion. It follows, therefore, that interaction may be linked to consensus through cohesion. Moreover, consensus and co- hesion may be regarded as interdependent, mutually rein- forcing, and together regarded as basic conditions for the development and maintenance of cooperative behaviour.

As technological conditions which determine the deploy- ment and density of workers provide variable opportunities for interaction, especially voluntary interaction, consensus and cohesion in work groups may be expected to vary as a consequence of these technological conditions. It is now possible to consider the relevance of consensus and cohesion as conditions for the expression of stress-re- ducing action in organizations.

It is assumed that collective responses to stress require collective perceptions of stress, that is, consensus concerning the source of stress and consensus about appropriate stress-

39Cohssiolt, along with the adjectives high and low, refers to variability in interpersonal commitment among the participants of a social system. Although this use may suggest that erne tional commitments are always positive, it also refers to situa- tions where people are indifyemnt to each other or experience negative feelings for each other.

doGeorge C. Homans, Th% Humaa C r o u p (New York, 1950, 43; and Social Behaviour: Its EZementary F o r m (New York, 19611, Chapter 10.

41See Talcott Parsons’ formulation of t he process of socialization as it occurs within a developing system of mother-child Idles. Talcott Parsons and R. F. Bales, Family, b&Zization, and hter- adion Process (New York, 1955).

Interaction, Cohesion and Consensus

Relations between Consensus Cohesion and Productivity, Absenteeism and Turnover

230

reducing action. Thus, assuming that stress is likely to be present in organizational role relationships, not only as intrinsic to role differentiation but because of other factors which induce nonconformity on the part of r6le incumbents, work groups, depending on their technological organization, may be expected to vary in consensus concerning stress and concerning appropriate stress-reducing action. Moreover, where members of work groups possess facilities which they regard as effective in a conflict with other groups or with management, they are likely to express the conflict by overt action directed against their opponents, for example, by the expression of formal grievances or by a strike. How- ever, where the group regards itself as weak relative to its opponents, stress-reducing action is likely to be covert, consisting mainly of passive withdrawal or a seeming lack of collective or cooperative endeavor. It is in such in- stances that stress reduction may be expected to take the form of productivity control, absenteeism and quitting. Con- sequently, under such conditions i t may be expected that the greater the consensus regarding stress and the need for stress-reducing action, the lower the productivity and the higher the absenteeism and turn-over. In effect, tech- nological conditions which determine the extent that work groups control production and which determine their posi- tion in the flow of production will consequently determine whether stress-reducing action will be overtly or covertly expressed. In either case, technological conditions will determine the frequency or intensity of the expression of stress-reducing action. It is argued that the frequency of collective responses to stress in units of complex organizations is determined by the level of consensus and the power position of these organizational units. Consensus and cohesion, which is seen to be positively associated with consensus, are deter- mined by the degree of r6le differentiation and the fre- quency of interaction, especially where interaction is volun- tary, in the organizational units. R61e differentiation and frequency of interaction are determined by specified con- ditions established by the technological organization of the units. Levels of consensus in organizational units, there- fore, are determined by structural requirements rooted in technological and social organizations.

Summary of the Theoretical Argument

Research Findings An adequate test of the propositions advanced in the argu- ment requires a complex research design providing for distinct differences between work units in terms of the specified variables and allowing control of the effects of the specified as well as for extraneous variables. The research reported here falls short of these requirements. Only seven departments (see Table I) employed a sufficient number of workers to make statistical analysis worthwhile; and although there were differences between these depart- ments in terms of the critical variables, these differences were not pronounced. Moreover, it was not possible to control for the effects of extraneous variables, such as age or sex, and, as can be seen from Table I, it was necessary to compare departments which differed technologically and

231

Tab

le I

T

echn

olog

ical

Cha

ract

eris

tics

of

the

Seve

n D

epar

tmen

ts

Dep

t. T

ask

N

Dep

loym

ent

M

F

Den

sity

D

ivis

ion

of

Lab

our

Ran

k O

rder

Hom

ogen

eity

R

ank

Inde

x**

Ord

er

A

B

C

D

E

F

G,

Pic

ture

Tu

be M

anuf

actu

ring

27

0

Tea

m a

nd A

ssem

bly Line

7 .2

15

chas

sis

com

pone

nts

and

prin

ted

circ

uits

Too

l R

oom

: di

e an

d to

ol

25

0 In

divi

dual

5

.246

m

akin

g, a

nd

mod

el

cons

truc

tion

Rad

io

Cha

ssis

A

ssem

bly

7 28

Lon

g A

ssem

bly

Lin

e 2

.477

Tes

t an

d R

epai

r

T.V

. an

d R

adio

Pri

nted

Boa

rd

6 49

Short

Ass

embl

y L

ine

3

.a7

Ass

embl

y

T.V

. F

inal

Ass

embl

y 24

0

Tea

m

1

.482

Car

Rad

io C

hass

is A

ssem

bly

16

105

Long

Ass

embl

y L

ine

4 .5

25

Mac

hine

Sho

p:

prod

uces

31

7 In

divi

dual

and

Tea

m

6 .207

*The

sev

en d

epar

tmen

ts w

ere units

of a

com

pany

, loc

ated

in

Can

ada,

man

ufac

turi

ng a

nd a

ssem

blin

g el

ectr

onic

pro

duct

s.

**H

omog

enei

ty i

ndex

is

dete

rmin

ed b

y th

e fo

llow

ing

form

ula:

H

.I. =

xfz

w

ith

f be

ing

the

freq

uenc

y of

em

ploy

ees

in a

giv

en O

CCU

patiO

Ml

clas

s.

(x f)

2

~ ~~~ ~ -

in terms of one or more extraneous variables. Fortunately, these extraneous variables did not, on the whole, affect the response distributions. Analysis reveals no significant differences in responses to job or authority items but there were differences in responses to items concerning the company in general associated with occupation, age, sex, and years employed with the company.42 Despite these liimitations, the findings43 are relevant to discussion and direct attention to necessary modifications of the theoreti- cal argument.

Proposition 1. The greater the r 6 e differentiation, the lower the consensus among employees of work units.

Proposition 2. The higher the density, the higher the consensus among employees of work units.

The degree of r6k differentiation was determined by examining the occupational “mix” in each department as represented in job descriptions provided by the company and by the company’s occupational classification which coded both occupational differences and skill levels. Density refers to the concentration of workers in a given work area. As measurements of density were not available, departments were ranked on the basis of ratings by managerial personnel and field observation.44 Consensus in the departments was determined in terms of the distribution of responses to questionnaire items directed to assessing worker satisfac- tion with his job and working conditions, with supervision, with the company generally and with the available rewards controlled by the company. Consensus index values45 for 42However, the rank ordering of the Departments according t o consensus on authority was random with respect to sex distribu- tion. 43The results are presented in the terms of the propositions for which relevant data are available. I gratefully acknowledge Professor Franklin Henry’s programming assistance and the contribution made by Keith Tucker, my research assistant, to the analysis. 44Although a simple classification of team, assembly lime, and individual operations was used to categorize departments in terms of deployment, a reliable ordering was not obtained and data relevant to propositions related to deployment are not reported. Difeculties in ordering arae because some departments contain mixed deployment categories and because it was difficult to assign ranks to departments categorized in terms of individual operations without taking account of density. While it could be assumed that opportunities for interaction in assembly line departments would be negatively associated with the length of the line, there was no empirical evidence to evaluate this assump- tion or to establish critical differences in the lengths of the lines. 45The consensus index is determined by the following formula,

z f 2

(z f ) Z C.I. = k (-) - 1

k - 1 where k equals the number of response categories and f the frequency of responses in the ith category. The formula was devised by George Ferguson, Department of Psychology, McGill University, and modified by Franklin J. Henry, Department of

-

233

Table II ~ -~ ~~ ~~

Distribution of Average Consensus Index Values for Different Satisfaction Dimensions

Department Job Authority Company

A .18 .l3 .I2 €5 .17 .18 .15 C .26 .l3 .zl D 30 .40 .22

F .17 27 .l2 G .24 .l2 .l5

E .n .15 .15

Theoretical range of the C.I. 0.00 - 1.00 Table III Rank Order Correlations+ between conSensus and Technological

Characteristics

Job Company Authority

Role DUl’erentiation -0.22 0.01 0.00 Density 0.28 0.19 0.58

*Rank order correlations were computed by Spearman’s formula As consensus levels among departments scarcely differed at all on items assessing satisfaction with rewards, the relationships indicated by the relatively high rank order coeffidents were judged to be spurioua and were excluded from the table.

each of the four dimensions of satisfaction were computed for each department.

Although levels of consensus were not strikingly high in any department and differences between departments were not pronounced (see Table II) , the direction of the results as shown in Table IXI, is supportive of both propositions. How- ever, Proposition 2 is more strongly supported as all co- eff icients expressing the relationship between consensus and density are positive as expected while only one coefficient expressing the relationship between r6le differentiation and consensus is negative.

The data f i t the theoretical argument better if rale differ- entiation and density are considered together in relation to consensus.4~ Taking Propositions 1 and 2 together, high consensus would be expected where r6le differentiation is low and density is high while low consensus would be expected where r6le differentiation is high and density is low. As can be seen from Table IV, all departments with the exception of Department C, conform to these expecta- tions. As Department C consists of highly-skilled employ- ees who are somewhat segregated from other Departments, these conditions may have overcome the effects of relative- ly high file differentiation and relatively low density. De- partments E and G, with middle ranks on one or both variables, fall about middle rank on consensus. Sociology and Anthropology, McMaster University. (See W. E. Lambert and Otto Klehberg, ChildTm’zt Ykws of F&g% Peopk: a CTOaS-NUtiOnUZ Study (New York, 1966). 46Moreover, dichotomous variables are more appropriate, given the distribution of differences on the variables.

234

Table I V Distribution of Departments by Rank on Role Differentiation, Density and Consensus

Density

Role DitXn.

LQW Consensus 3 F 3.5

'Cell entries designate the Department by letter and rank on consensus on all satisfaction items, 1 = highest consensus.

Proposition 4. The greater rvile differentiation in a work unit, the lower the cohesion in that unit.

Proposition 5. The greater the density in a work unit, the higher the cohesion in that work unit.

Cohesion was assessed in two ways: 1) Actual cohesion was assessed, on the assumption that

voluntary interaction implies interpersonal commitment, by determining the ratio of in-group to out-group participants in voluntary social contacts as these were revealed by asking workers to name those with whom they associated during breaks, lunch hours, and after work. Rank orders were identical for interaction during breaks and lunch and these were highly correlated with social contacts after working hours.

Perceived cohesion was assessed by asking how work group members perceived their groups in terms of mutual help, pride in accomplishment, attractiveness of the group,

2)

Table V Rank Order Correlation between Cbhesion and Technological Characteristics

Actual Cohesion Perceived Cohesion

Breaks & Lunch- 6 Hours Ri3le DiEerentiatlon Density

039 0.61 -0.07 -0.n -0.46 -0.25

etc.47 Contrary to the theoretical argument, actual cohesion is (see Table V) positively associated with r6le differentia- tion and negatively associated with the density of the work group. Taking the two variables together, as shown in Table VI, actual cohesion is close to expected levels only in Depart- ments B, D and F. In terms of perceived cohesion, there is a similar lack of support for the propositions although 471 acknowledge Stanley E. Seashore's permission to use several items which he used to assess cohesiveness among work groups in a factory. See Stanley E. Seashore, Group Cohesivsnsss in the Industrial Work Group (Ann Arbor, 1954).

235

~

Table VI The Relationship between Rble Differentiation, Density and Actual Cohesion

Density

E 6

Expect High Acoh

D 4 F 3

High

Rble Differentiation

LOW

Expect Low Acoh

A 2 B 5 c 1

G ?

E 7

Expect High Perceived Cohesion

D 3 F 6

Table VII* The Relationship between RBle Differentiation, Density and Perceived Cohesion

Expect Low Perceived Cohesion

A 5 B 4 c 1

G 2

Density

High

Mle Diffn.

LOW

*These cell entries designate the Department by letter and by rank (1 = highest P/coh) on perceived cohesiveness.

the relationship to r6le differentiation is slightly in the predicted direction. Taking the two variables together, as shown in Table VII, perceived cohesion is close to expected levels only in Departments A, B and D.

In general, then, workers were more likely to associate voluntarily with each other in production units where r6le differentiation was relatively high and opportunities for interaction relatively low. Although workers also showed some tendency to perceive their groups as cohesive where interaction opportunities were low, r81e differentiation seemed to have little relationship to the perception of mhesion.48

Although it is difficult to understand why r6le differen- tiation is positively associated with actual cohesion, it is 48The data do not support a predicted positive association between density and actual cohesion, when the reward significance of interaction as assessed by perceived Cohesion is controlled.

236

possible that perception of rble differentiation is not congru- ent with observed rble differentiation or that choices con- centrated on incumbents of higher-ranking rbles where r6Ie differentiation was relatively high but were randomly dis- tributed among all workers where r6le differestiation was low. No data are available to test these possibilities.

It is also possible that conditions other than r81e differen- tiation operate to increase actual cohesion. For example, Department A (highest on actual cohesion), although rela- tively highly differentiated and relatively low in density, was set apart physically from other production units so that interaction with “out-group” members could be diffi- cult. Moreover, the technical organization of the workers took the form of a combination of team and line operations and team work may have increased interaction, i.e., actual cohesion. Department C (next highest on actual cohesion), while relatively high on r6le differentiation and low on density, consisted of the highest-skilled workers in the plant and prestige considerations may have acted as a constraint on out-group choices. On the other hand, Department F (rank 3 on actual cohesion) with the low differentiation and high density and Department B (rank 5 on actual co- hesion) with high rde differentiation and low density provide examples where rank on actual cohesion is close to expectation.

Actual and Perceived Coh sion and Consensus Proposition 6. The more cohesive a work group, the

higher the consensus in that group. Proposition 7. The higher the reward-significance of

interaction among members of a work group, the more cohesive the work group, and the higher the consensus in that work group.

Contrary to expectations, the data indicate a weak nega- tive association between consensus and actual cohesion, (r8 = -0.10, - 0.01, 0.10, for Job, Authority and Company, respectively), measured by voluntary contacts during work- ing hours, and a moderately strong negative association with actual cohesion, measured by voluntary contacts after working hours (r, = -0.40, -0.38, -0.08). In work groups, therefore, where agreement tends to be high, there is a slight tendency to choose out-group members for informal social interaction during working hours and this tendency becomes more pronounced where social contacts after work are concerned. In short, workers do not tend to choose as friends, persons with whom they share attitudes concerning the work situation. Looked at from the viewpoint of con- sensus, it may be said that a work group’s level of consensus does not appear to be a consequence of a high level of actual group cohesion. It may also be noted that the relationship between consensus and perceived cohesion varied in relation to the different dimensions of the work situation. Thus, there was a strong positive relation between perceived cohesion and consensus about the company (r, = 0.631, a weaker positive relationship in relation to consensus about job satisfaction (r, = 0.22) and a moderate negative rela-

237

tionship in relation to concensus concerning authority (r, = 1-0.34).49 As there is only a weak relation between perceived and

actual cohesion (rs = 0.18 and -0.14 for working hours and after work respectively), the data (see Table VIII) provide little S U D P O ~ ~ for Promsition 7.

High

Perceived Cohesion

LOW

Table Vm*

4

Expect High consensus

c 1 D 1 G 4

Expect Low Consensus

A 5 B 6.5 F 65 E 2

~~~ ~

Relationship between Consensus and Actual cohesion in Relation to Perceived Cohesion

Actual Cohesion

Proposition 8. If stress is experienced by members of work units in their relationships with management, the higher the consensus and/or cohesion, the lower the pro- ductivity of the work unit and/or the higher the absentee- ism and/or labour turnover.

The presence of stress was assessed by the levels of satisfaction in the departments as revealed by responses to the questionnaire i tems.61 As levels of either consensus or cohesion might be expected to strengthen or weaken opin- ions held by members of a work unit, the relation between satisfaction and the performance variables might be in- creased or diminished depending on whether consensus or cohesion were high or low, e.g., high satisfaction along with high consensus would be associated with low productivity. On the whole, there is little evidence to suggest that con- sensus operated that way in relation to productivity in these groups. However, since productivity could be in- flueiiced by various extraneous factors such as the avail- ability of materials and since departments varied in worker/ machine control of the productive process, i t is a somewhat unreliable measure of group performance. Absenteeism and turnover are more reliable indicators of stress-reducing action and, in relation to these, consensus does appear to

The Relationship of Satisfaction to Absenteeism and Turnover in Relation to Consensus Levels

Department Satisfaction Consensus* Actual Cohesion Absenteeism Turnover

F H B H C H D M G L E L A L

L H M L H H H M L L H L L I3

L H M L L L H L H M H H L H

*Based on average rank on Job, Authority, and Company scales.

Watisfaction has the clearest relationship to productivity, absent- eeism and turnover. The strength of the relationships varies according to satisfaction scales: the strongest being between absenteeism and job satisfaction, 4.85, the weakest being between authority and absenteeism, 0.11, with most values varying around 1-03. In contrast, productivity is both positively and nega- tively related to consensus (0.07, 0.18, -0.371, and negatively related to actual cohesion. Consensus and absenteeism are positively re- lated, (,, = 0.67, 0.51 and 0.06 for job, company and authority respectively) but contrary to expectations consensus and turn- over are negatively related. Actual cohesion, also contrary to expectations, is negatively related to absenteeism and turnover. If relationships between the technical variables and consensus were clearly established, strong relationships would be expected between the technological and performance variables. However, only weak relationships, mostly in the wrong direction, are indicated by the rank correlations.

239

Summary

have a mediating effect on the relationship between satis- faction and absenteeism or turnover. Thus Department C which is high on consensus and satisfaction and Department E which is low on satisfaction and high on consensus are, respectively, low-ranking and high-ranking on absenteeism and turnover; departments G and A which are low-ranking on both variables are respectively high and low on absent- eeism, and medium ranking and high on turnover, sug- gesting that low satisfaction and low consensus both contribute to stress and lead to withdrawal as a response to stress; Department F which is high-rank- ing on satisfaction but low on consensus is low on absenteeism but high on turnover, and Department B which is high-ranking on satisfaction but medium rank on consensus is medium on absenteeism and low on turnover. Actual cohesion seems to have a more pronounced mediating effect than consensus. Departments F and C which are high-ranking on satisfaction and cohesion were both low on absenteeism with one low and one high on turnover; G and E which are low-ranking on satisfaction and cohesion were high on absenteeism and medium and high on turnover while Department A which was 16w on satisfaction but high on cohesion was high on absenteeism and turn-over. Somewhat similar results were observed in relation to pro- ductivity. There is some support for the position that technological conditions, as these influence rdle differentiation and inter- action opportunities, are associated with consensus in work groups. However, a negative association was found for actual cohesion and consensus. It was suggested that other constraints, imposed by such technological requirements as physical isolation and levels of skill, account for the con- siderable difference between the observed events and the

. model of the relationship between actual cohesion and consensus. The data also indicate that consensus and actual cohesion influence the direction of the association between satisfaction and stress-reducing actions as absenteeism and turnover.

240