AN EMPIRICAL EXAMINATION OF THE REFERENTIAI. FUNCTION OF INFORMAL COMMUNICATION GROUPS Thesis for the Degree of M. A. MICHIGAN STATE UNIVERSITY IAMES ALLEN TAYLOR 1976 ' , v—v——v"'—"“ \IIIIIIIIII M I I I TIMI II III :rfi ABSTRACT AN EMPIRICAL EXAMINATION OF THE REFERENTIAL FUNCTION OF INFORMAL COMMUNICATION GROUPS BY James Allen Taylor It has been argued that the informal relations between members of an organization significantly affect the attitudes and work behaviors of participants. This research reports an attempt to examine hypotheses de~ rived from this proposition. Derivation of hypothesis is made through linear force aggregation theory. This theory specifies that a person's perception of a given attitude object is a linear function of the amount and valence of all messages received about the object. It is argued, therefore, that members of a communication network who are highly embedded into a subset of the network (e.g., a communication group) will have a high degree of similarity on certain attitude items because of the fre- quency of contact. Additionally, it is hypothesized that the degree to which an individual is integrated into the group will predict the degree of similarity. Hypotheses are examined by using network analysis to determine the underlying informal structure, and by using items derived from the Insti- tute for Social Research Interorganizational Survey Instruments. Results strongly support the notion that informal groups do determine the atti- tudes of members on work related topics; however, predicted linearity did not emerge. AN EMPIRICAL EXAMINATION OF THE REFERENTIAL FUNCTION OF INFORMAL COMMUNICATION GROUPS BY James Allen Taylor A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Communication 1976 Accepted by the faculty of the Department of Communication, College of Communication Arts and Sciences, Michigan State University, in par- tial fulfillment of the requirements for the Master of Arts degree. Committee Chairman, 7 fikcémfl U FW CNN? 014%..in in, ii ACKNOWLEDGEMENTS The author wishes to acknowledge the valuable contribution of Dr. R. Vincent Farace to the preparation of this thesis. He alone knows the inertia it took to realize this product. Thanks also to Dr. Bettinghaus, Dr. Woelfel, Kathryn Clyde, Dr. Rolf Wigand, and Dr. James Danowski. The author would also like to offer special acknowledgement to Dr. George Barnett fOr the comments and criticisms, and Dr. Michael Moore for his guidance. Finally, thanks to my wife Judy. iii Chapter INTRODUCTION......... TABLE OF CONTENTS 1. STATEMENT OF THE PROBLBMOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO. SUMMARYOOOOOOOO 2. THEORY AND HYPOTHESES..... HYPOTHESES 1-5..... SUMMARY......... 3. METHODS AND RESULTS.... DESCRIPTION OF DATA SOURCE.............................. RESULTS......... DISCUSSION..... CONCLUSIONS.......... 0..0.0000000000IOOOOOOOOOOOOO IMPLICATIONS FOR MANAGEMENT THEORY....... ....... ........ SUMMARY........ APPENDIX A - ITEM ANALYSIS.... APPENDIX B - TABLE 17. .00....-0.00.0000...00.000.000.00. Descriptive Statistics: All Respondents APPENDIX C - PROCEDURES FOR DISCRIMINANT CLUSTER ANALYSIS....... BIBLIOGRAPHY..... iv 0.0...OOOOICOOOOOOOOOOOOOO... 13 14 23 2n 25 26 31 35 52 54 56 56 10. ll. l2. l3. 1”. 15. l6. 17. 18. 19. 20. LIST OF TABLES Page Network Roles for Employees by Category...................... 32 ICLS Variables............................................... 33 Descriptive Statistics: Group and Non-Group Respondents..... 36 t-Tests Between Group and Non-Group Populations.............. 38 Variables with Significant t Differences Between Group and Non-Group POPUlationSOOOOOOOOOOO0.00.00.00.00...000...... no One-Way Anova: Variable by Group............................ 43 Network and Predicted Groupings.............................. uu Variables by Connectedness Rank.............................. “7 Chi-Square Goodness of Fit Tests - Variables by Connectedness NB Chi Square Goodness of Fit Tests - Variables by IntegrativeneSSOOOOCOIOCOOOOOOIOOOOOOOOOOOOOO00......00...... SO Variables by Connectedness................................... Sl Inter-Item Zero Order Correlations........................... Sl Partial Correlations Controlling for Connectedness........... 52 ICLS Scales Inter-Item Correlations.......................... 61 Median and Range of Inter-Item Correlations, Item-Cluster, and Item Communalities of Scales in ICLS..................... 62 Alphas Of ICLS scaleSOOOOOOOOOIOOIOOOOCCOOOOOOOOOOOOOOOOOOOOO 62 Descriptive Statistics: All Respondents..................... 63 Discriminant Analysis--Step Wise Variable to Enter Sequence WiJ-ke's Lanlbda ProcedureOOOOOIOOOOIOOOOOCOOOOCOOOO0.0.0.0.... 61+ Discriminant AnaIYSiS--GI’OUP 83-28 for AnaIYSiS. o o o o o o o o o o o o o o 65 Wilkes A for Entered Variables............................... 65 V LIST OF FIGURES Figure Page 1. Alternative Levels of Integration Within a Network.......... 21 2. Various Levels of Connectedness ...... ..... ..... ............. 22 3. Comparison of Mean Variables for Group and Non-Group PopulationOOOOOOOOOOOOOOOOCOOOOOOOOOOOOOIOOOOOOCOOOOO..00... 37 u. Average Standard Deviations for all Variables by connectedness RankOOIOOOOOOOUOOOOOOOOOOOOOOOOOOOO...00...... 1+6 5. Mean Standard Deviations for Variables by Integrativeness RarIkOOOOOOOOOOOOOOO0.000000COCICOOOOOOOOOOOOOCOOOOOOCOOOOOOO “9 vi INTRODUCTION It has been an accepted canon of organizational theory that the in- formal system of relations prevailing among participants in an organiza- tion has significant impact upon the nature of the organization (Selz— nick, 19u7). Selznick advanced three propositions: (1) every organiza- tion creates an informal structure; (2) in every organization, the goals of the organization are modified by processes within it; and (3) the process of modification is effected through the informal structure. Further, Selznick argues that informal groups are norm-providing systems of relations which arise spontaneously within rational systems as a con- sequence of the idiosyncratic intrapersonal needs of system members. Following the studies at the Hawthorne plant of Western Electric Company, Rothlesberger and Dickson (1939) commented that: The study of the bank wiremen showed that their behavior at work could not be understood without considering the informal organization of the group and the relation of this information organization to the total social organ- ization of the company. (p. 551) The importance of the role played by the group in industrial organi- zations has been emphasized by Lewis (1952), Nelson (19u9), and Blake and Mouton (1968). Empirically, however, these studies have focused on groups as formally structured entities rather than as informal liaisons among mutually attracted individuals. Following Selznick, this thesis presents the view that people who interact regularly andrepetitively among themselves create a conjoint infbrmation space wherein values, attitudes, and beliefs are exchanged l through the process of information transmission among the members in the space. This view is held irrespective of purposiveness; rather, purpo— siveness is held to be "not something inherent in the situation, but rath- er something in the way a situation is evaluated" (Rappoport, 1963). By developing a theoretical model which demonstrates that individuals form groups which have the effect of restricting both information inputs and outputs (and assuming for the moment that no member of the system "in- tended" to form this group), we can derive a model which will account for the emergence of what has been called "the organizational man." It is often noted in both the literature on organizational relations and in common parlance that people who join organizations are changed out of all proportion to the significance of the work they perform (White, 196“). Individuals from diverse backgrounds become similar in their beha- vioral patterns and in the way they structure their symbolic environment. Laski (19u2) argues that this process of behavioral differentiation arises from the conflict between organizational units competing for con- trol. He states: (1) Cooperative effort, under the conditions of increas— ing number and complexity of fUnctions, requires the dele- gation of functions. Thus action which seeks more than limited, individual results becomes action through agents. It is the activity of officials acting as agents with which the discussion of bureaucracy is concerned. (2) The use of intermediaries tends to create bifurcation of interest between the initiator of the action and the agent employed. This is due to the creation of two sets of problems: for the initiator, the achievement of the goal which spurred him to action, and for the intermedi- ary, problems which are concerned chiefly with his social position as agent. The character of the agent's new val- ues are such as to generate actions whose objective con- sequences undermine the professed aims of the organization. This conflict need not be between the employer as a person or a group and the agent, for the latter may be able to manipulate the ideas of the former, but between the actual course of the organization and those aims formally asserted, whether the employer recognizes the conflict or not. (3) This bifurcation of interest makes dominant, for initiator and agent alike, the issue of control. What is at stake for each is the control of the conditions (the organizational mechanism) which each group will want to manipulate (not necessarily consciously) toward solving its special problems. In this struggle fbr control, an informal structure is created, based large- ly on relationships involving personal influences rath- er than formal rules. As individuals enter an organizational system, they become exposed to a new set of behavioral expectations which are communicated from many sources. While one's tenure increases, the individual continually lim- its the number of successive input stimuli by choosing those with whom one wishes to interact, from the variety of personalities within the or— ganization. Attitude change is induced by usual and frequent interaction patterns with this chosen body of personalities. It will be argued that attitude change stems from the messages an individual receives about the kinds of activities and behaviors which are considered appropriate and expected by the group of which the source is a member. Implicit in the source's messages is an attitude set which condi- tions the type of information and the bias attached to the information being transferred. This attitude set will be adopted over time by the new member, although not holistically. The new member will also contribute some subset of his attitude set to the perceptions of the group. Thus, attitudes related to the specific environmental context change continu- ously within groups. This thesis will examine the role of the group as the psychological field within which organizational norms are developed and maintained. While many observers (Schein, l967; Huse, 1975; Lawrence and Lorch, 1966) hatns indicated that informal groups establish these norms, a methodology by tailich informal groups can be detected within large systems has only recently become available. This thesis is organized into three sections. In the first section we discuss some of the classic research performed within the group con- text. In particular, our interest focuses on the nature of group commun- ication processes and the manner in which the group influences the atti- tudes and beliefs of participants. The second section advances a theoretical explanation which attempts to parsimoniously account for the phenomenon of interest. We will dis- cuss the concept of an interactive human system as a network and describe a method for heuristically characterizing relations within the network. In the third section, we will present data in order to test several aspects of the theory. We will then attempt to draw conclusions and point to further research which needs to be done. Chapter 1 STATEMENT OF THE PROBLEM In reviewing the literature on group process, Merton (1952) points to 26 different properties which have been articulated by various research- ers from a wide variety of paradigms and disciplines. Principal among these properties have been studies into the relationship between size of membership and structure (Simmel, 1950M stratification, or hierarchic- al levels (Likert, 1967; Landecker, 1951); group openness versus closed- ness; duration of group; and clarity of social definitions of membership in a group (Merton, 1953; Goldberg and Larson, 1974). Each of the 26 properties listed by Merton has been used as the de- pendent variable in a set of experimental situations. Of particular in- terest here, however, are the variables related to the clarity of the social definition of group membership. Certainly, clarity of membership varies across many social situa— tions. The individual, for example, may be a member of the Elks Club, a member of a Friday night Poker Club, an employee of an organization, or all three simultaneously. In each of these cases the social unit has some formal manner in which membership in the group is acknowledged. The Elks Club has membership cards and initiation rites. The Poker Club, likewise, has an approval process, and a chair at the table. The corporation has hiring processes, formal work routines, and status conferral systems. In each case the membership is a formal, acknowledged process. The member of the group acknowledges both his membership in the group, and the exist- ence of the group. Such an acknowledged collective framework fits the conceptualization of many scholars studying the social psychology of the group. According to Smith (19H5), the group as a psychological entity may be defined as, ". . . a unit consisting of a plural number of organisms who have a col- lective perception of unity and who have the ability to act, or are act— ing in a unitary manner toward their environment." For Kretch and Crutchfield (1953), the group is defined in terms of the psychological processes of self-recognition. They wrote, "The criter- ia for establishing whether or not a given set of individuals constitute a psychological group are mainly two: (1) all the members must exist as a group within the psychological field of each individual, i.e., be per- ceived and reacted to as a group, and (2) the various individuals must be in dynamic interaction with one another" (p. 62). Both definitions are fairly representative of the thinking of social psychologists on the distinction between an aggregate or collectivity and a group. However, several objections must be raised to these view- points. First of all, each definition contains an inherent tautology. In a sense, each says, in order for a group to exist, a group must exist. Second, it is difficult to operationalize the perception of the relation- ship between the members and the common distinction each group member makes between his membership and the environment. Third, each requires that the set of relations inherent in "groupness" comprise some objective- ly ascertainable unity of mind or purpose —- a group "consciousness." This implies that aggregated behavior is imposed by the goals of the group rather than emerging from needs of the individuals. An alternate approach which avoids this particular notion of collec- tive consciousness emphasizes instead the role of independent expectations as they interact within the forum of the group. Cattell (1953) defined a group as a "collection of organisms in which the existence of all (in their given relationships) is necessary to the satisfaction of certain individual needs in each." The underlying assumption of this definition was expressed by Rappoport (1963). Rappo- port, in discussing the characteristics of operational philosophy, com- mented that one of the fundamental invariants in the human condition is the need to belong. Groups fulfill this need by providing a field where- in behavior can take place and needs can be satisfied. According to Cattell, individuals belong to groups only because they achieve certain satisfactions, and in the absence of satisfaction, the group will dissolve. Cattell comments that implicit in his formulation is acceptance of the view that psychodynamic events can be operational— ized through the use of any one of three sets of concepts: (1) the stim- ulus situation-cue—response, emphasizing the detailed environmental con— ditions; (2) the satisfaction of "ergic" or "metanergic" motives which concentrates on the drizg_phenomena; (3) the energy expression formula- tion which directs attention to the work done in the motivational situa— tion. However, it is difficult to clearly define the character of the motives of force within these concepts. Cattell, echoing Durkeim's (1933) sociological conceptualization of group consciousness, asserts that groups have a pseudo-personality (syn- tality) which creates and is created by the expectations and perceptions of the members. The components of this personality provide the drive re- duction mechanisms which direct the group behavior and the conforming re- sponse patterns of individual members. The behaviors of members, then, would seem to be directed by an unobservable whole, while the reinforcers of that behavior are measurable at the individual level. Sherif (1967) has assembled a large body of literature on groups which further suggests that the "self-determinancy" concept is not con- tinuously present among researchers into group processes (c.f., Hiller, 19u7; Monge and Lindsey, 197a; Richards, 1974; Whyte, l9u3; Shaw, 1930; Roethlisberger, 1937). Sherif claims that, in general, researchers have applied three criteria before a collectivity may be said to be a group. First, members must exhibit some sort of structural interdependency. Second, status levels and role differentiation must be present at some level. Third, a set of norms or values common to the group must be iden- tifiable. In addition, Sherif requires that interaction episodes among the participants be more than transitory. The question of groupness, then, directly corresponds with the de- gree to which these properties are present. It is not enough to identify dichotomously the nominal presence or absence of any of these variables. Groupness, in this sense, is a continuous emergent property which springs from the needs and gratifications of the participants and is given pre- dictive power by the patterning of interaction among the participants. The precise formulation has been stated by Borgatta and Cattell (1955): Even if the assumption is made that some collectivities are groups, and others are not, and that there is a difference in kinds, when the effort is made, we find ourselves considering variables on which all collectiv- ities can be ordered, and in terms of which they can be variously classified for different purposes. Thus, the question shifts from whether an aggregate is a group or not to one considering the degree to which such an ag- gregate is characterized by a specific complex of vari- ables assumed to be the components of groupness. Such a formulation points to the necessity for identifying the relevant critical component variables in terms of which any collection of persons may be assessed and, at a given point, classified. Some of these variables with continuous properties have already been mentioned. For our purposes, we will be primarily concerned with variables which vary, at least in part, as a function of the integrativeness of the group, the isolation of the group, and the homogeneity of structure with- in groups. By structure (Danowski and Farace, 197%), we mean the degree to which relations are homogeneously distributed within the group. A group itself may be linked relationally to a greater or lesser degree in- to a group of groups, or an organization. The degree to which it is iso- lated affects the pertinence of information pertaining to the organiza- tion which circulates within the group. Finally, groups may be more or less similar on a comparative, aggregated basis. The structure of the linkage paths, the structure of between—group links, and the relative isolation or predominance of individual members also affects the nature of the group. All of these variables may be continuously scaled. The higher the loading for any complex of variables, we would assume, the greater the groupness. Bales (1953) supports this view when he argues that it is necessary to include both the overt behavior of individuals and the sit- uational environment which spawned the behavior (the organization) in an analysis in order to arrive at valid empirical generalizations. In looking at groups, as we have noted, it is often heuristically useful to employ a notion of group syntality or a consensual conscious- ness. Hence, one would expect to find that, within groups, the whole is not the sum of the individual component members, but a uniquely deter- mined set of behaviors which arise from the expectations and objectives caf individual members. It is at the group level that investigators begin tc> observe conformity and socially altruistic behaviors. Talcott Parsons (1&9657) makes a distinction which is useful in understanding the theoreti— cal assumptions being maintained here. He wrote: 10 The latter concept referred to a body of beliefs and sentiments held in common; the collectiveness of it consisted of the "in commonness." Now the collective- ness consists in the nature of reality exterior to the individual to which the individual's "representa— tions" refer. It is not a subjective community of belief which is the source of solidarity, but ration- al orientation to the same set of phenomena in the environment of action, an "objective" source of uni- formities. (p. 141) In other words, the group collectivity provides an arena within which the individual learns to represent reality. It is not a community of belief about sets of social objects, but orientations toward classes of phenomena from which commonalities consisting of beliefs may be said to be independently and objectively derived by any single member of the collectivity. Therefore, it would seem wise to focus on the processes by which groups have come to common orientations. When individuals adopt a common orientation toward an object of perception through socially facilitated learning processes we say they have conformed (Campbell, 1961). While often misrepresented in the pub- lic controversy surrounding the inducement of conforming acts and atti- tudes within corporate and governmental organizations, researchers have taken great pains to point out the normative nature and acceptability of conformity as a phenomenon (Schein, 1967). The language we use, the clothes we wear, the toys we play with, are largely acquired or chosen as the outcome of a "socially facilitated learning process." Social—exchange theorists (Adams, 1965; Blau, 1964; Gouldner, 1960; Homans, 1958, 1962; Jones, 196A; Thibault and Kelly, 1959) view conform- ity as a social process in which certain behaviors (referred to as "ex- pected behaviors") are the reinforcing acts for others who engage in in- teraction. In this sense, a conforming behavior is a kind of advance pay- ment for the positive effects the ego anticipates will emerge from a 11 social interaction. In the organizational context, this might take the form of deference to status others, or persons with greater seniority. It would also take the form of listening or paying attention to others with whom one is interacting. In repetitive interaction episodes with the same individual,a conforming behavioral cycle could be anticipated, which would lead to greater and greater amounts of learning. For example, a person is seeking interaction with another person for the reinforcing value of interaction alone. In the course of that in- teraction, reinforcement (social exchange) takes the form of attentive- ness, or voiced agreement, toward the interacting other. The other per- son is reinforced to engage in interaction and provides cues, implicitly or explicitly reflecting a value orientation. This orientation facili- tates learning in the organizational novitiate or reinforces liking (Moreno, 1934), making subsequent interaction more probable. Related to the social-exchange conceptualization is the idiosyncracy credit model proposed by Hollander (1958, 1964). Hollander looks at con- formity as one input to the accumulation of status. In this case, a per- son is awarded credits by others for displaying approved behaviors. For the novice, these credits can accumulate rapidly, making the installation of long-lasting behavioral sets more likely. Hollander's view provides for a greater latitude of behavior than does a strict interpretation of the social exchange model. Conformity and nonconformity are not invariably defined relative to a fixed norm as in the Sherif's (1966) paradigm. Rather, an accumulation of credit allows actors to be variably perceived (in terms of their stat- us) and they are thus allowed to behave variably with respect to certain norms or conventions. Conformity is thus considered to be to some degree 12 person specific and functionally related to status. Such a view accounts for the finding that individuals with higher status or esteem have a wi- der latitude for social deviation (Berkowitz and Macaulay, 1961; Julian and Steiner, 1961; Wiggins, Dill and Schwartz, 1965). Further, it could be argued that the existence and distribution of this kind of organiza- tional "slack" to higher status individuals would motivate novitiates to conform through time, not simply at the moment a reinforcing episode occurs. Clearly, conformity facilitates socialization processes within an organization. While social—exchange theory may be only a partial answer to the total causal mechanism it does provide a basis for predicting the changing of attitudes when an individual enters an organizational sys- tem. The totality of the requirement for change is defined by the sys- tem's culture, which is the aggregate of rules, folkways, and behavioral patterns which are adopted by members during tenure within the system, and for which Hollander's credit is distributed. As Asch (1959) has put it: Each social order confronts its members with selected portions of physical and social data. The most decisive feature of the selectivity is that it presents condi- tions lacking in perceptable alternatives. There is no alternative to the language for one's group, to the kin- ship relations it practices, to the diet that nourishes it, to the art it supports. The field of the individual is, especially in a closed society, in large measure circumscribed by what is included in the given culture setting. Within the organization an individual will either learn the ropes or will not survive. How then does the socialization process take place? We are arguing here that the individual comes to perceive that the value set spawned by the organization is both necessary and right, even if it is at wide variance to his former beliefs. The change is what Parsons 13 called orientation, and it is fundamental in that it manifests itself on the totality of a person's organizationally relevant value systems. Summary Groups have been the subject of a considerable body of behavioral research. The focus of this research has usually been upon intact groups with known goals. However, "groupness" can be taken to be a continuous property which varies with respect to the amount of conscious acquiescence required of members. Groups, including informal groups, provide an arena wherein the ego's expectations and rewards may be clarified. To the ex- tent that an individual accepts his position in the group and engages in approved behavior, he has conformed. By conforming, he achieves the sa- tiation of needs and expectations. Chapter 2 THEORY AND HYPOTHESES A theoretical rationale for the assumption that shared communication necessarily leads to a convergence of attitude among a set of communi— cants can be derived from a general theory of attitude change provided by Woelfel and Saltiel (1974). They argue that every message has an effect which changes the way in which concepts related to the message are ar- rayed with respect to one another within the mind of the receiver. The amount of movement or change can be described by a vector between the ad- vocated position and the receiver's previous opinion. Total (net) change is a function of the amount of change advocated and some value for the importance of the source and the salience of the information at the time of input. Under this formulation we would expect that a communication group would increase the redundancy with which any specific group member would advocate a position for another member. Since Woelfel-Saltiel argue that an initial position is the weighted lin- ear sum of all previous inputs with respect to a set of attitude objects, increased redundancy will, over time, continuously shift a receiver's balance point towards that of source. This is not necessarily a recursive process. As the receiver inter- acts with and changes in the direction of the source, the source is sim- ultaneously responding to the receiver's feedback. Hence, the discrepancy between the two is being reduced. Given a number of such give-and-take relationships within a communication, one would predict that, over time, 14 15 the opinions, rules, drive and motivations of organizational colleagues will shape the attitudes of the novice as he advances toward the organi- zational equivalent of tenure. In turn he passes at least a part of what he inherits onto later members. It should be noted that each individual, to the extent that he is communicative, adds messages to the total pool and influences the direction of the change processes. The Woelfel-Saltiel theory differs from traditional theories of attitude change or formation (Heider, 1946; Newcomb, 1956; Festinger, 1957; Osgood, Suci and Tannenbaum, 1957) by suggesting that attitudes are embedded in a continuously changing, multivariate "balance point." Atti— tudes, themselves, are conceptualized as a predisposition to respond; however, the W-S theory emphasizes that the individual processes infor— mation related to the attitude through a multidimensional cognitive space. This space consists of an array of attitude-objects wherein the distances between the objects are the scalar representations of the un- derlying predispositions. Attitude—objects are taken to be those phenom- ena within the environment to which an individual assigns a valence, either positive or negative, for evaluation purposes. The space, which is essentially an unstandardized orthogonal factor rotation, specifies the dimensions which underlie the relative locations of the attitude- objects. Unlike Heider (1956), the Woelfel-Saltiel theory specifies the rela- tionship between message volume, the significance of the source, and attitude mass. Message volume is the quantity of input to the receiver across all sources. Attitude mass is that characteristic of an attitude which reflects its resistance to change; mass is a function of the number of messages a person has previously received about the objects of the l6 attitude. In an organizational context, this means that an individual with low tenure is likely to have low-mass attitudes toward informal work rules. Hence, his attitudes will be highly amenable to change. As he passes through the organization, the message history behind his balance point grows large, making change less and less likely. Heider attributed attitude change to search processes initiated by the individual as a result of some internal state of attraction. The individual attempts to remain consonant with both his state of attraction to another person and an incongruent attitude between ego and the other. While this scenario is not excluded by the Woelfel-Saltiel formulation, it also includes other circumstances in which the individual encounters valenced information towards the objects of an attitude. In other words, according to Woelfel-Saltiel, all_information, from all_media are seen as contributing to the magnitude, valence, and mass of an attitude. Woelfel-Saltiel also deviate significantly from Festinger's (1950) conceptualization of the role of dissonance and cognitive consistency as the impetus to attitude change. Again, internal states may initiate an information search which changes the locus of a balance point. However, other motives are not excluded by Woelfel—Saltiel. Similarly, Newcomb's A-B-X model may be seen as a special case of the Woelfel-Saltiel theory. In dyadic interactions, particularly those taking place in a laboratory setting, the issues of discussion are likely to not have had a large mes- sage history (e.g., low mass). A great many messages may be exchanged rapidly which will induce considerable movement in the locus of the bal- ance point in a relatively short period of time. Woelfel-Saltiel cover the individual attitude change case (c.f., Woelfel and Haller, 1972) and the generalized case of attitude change across an entire culture (Taylor, Barnett and Serota, 1975). 17 At its simplest level the theory suggests that an attitude is the joint effect of a set of messages: xl,x2,x3,...,xn. The consequent attitude a_is the linear sum of the messages divided by the number n_of messages. Attitude a can be represented as: X l n ' R" ix + E-x + l-x l— + i 3-- n l n 2 n 3 ° ' ' nXn i u v15 This equation assumes that each incoming message stimulus has a unique effect equal to the effect of all other incoming stimuli. Further, it assumes that no other variables have a substantial effect. Each message, Xi’ is postulated to be a force which pulls the atti— tude in one direction or another. The mean (xi) of all fbrces consti— tutes the balance point at which all forces sum to zero since: This conceptualization can be expanded to explain complex empirical phenomena at either the individual level (Woelfel and Haller, 1972) or cultural level (Barnett and Wigand, 1975). The theory allows the inves- tigator to specify the impact of messages given control over certain con- tingent variables. Messages can be weighted either for the significance of the source (Woelfel and Haller, 1972), or the salience of the informa- tion for the receiver. In fact, the precise effect of an additional num- ber of messages required to change an attitude where the message history, or the mass, is known, can be specified. In field studies, however, lack of experimental controls prevents adequate empirical examination of these equations. Attitude change, then, is actually treated as a simple quan- titative function of the number of messages an individual has received about a given attitude-object. Thus, the greater the information history 18 about the objects of an attitude, the more difficult it becomes to foster attitude change. Pour factors are, therefore, causally related to attitude change according to this theory: (1) the number of new messages (2) the number of messages comprising the initial balance point (3) the amount of discrepancy between the old attitude and the mean position advocated by the new messages. Woelfel and Saltiel state: . . . the amount of attitude change is directly related to the product of the average discrepancy between in- coming information and the old attitude (average change advocated) and the number of such messages, and inverse- ly related to the sum of the number of messages out of which the change message and the original message is composed. (pp. 4-5) Our interest here is particularly concerned with attitude change processes and attitude formation processes as a longitudinal activity. For the organizational novice, attitude formation processes about the na- ture of work and the organizational environment begins with socializa- tion processes during childhood and adolescence. Organizations like the Boy Scouts, Girl Scouts, educational institutions, and the family stress the development of leadership qualities and confbrming behavior with an eye to "learning to live in the adult world." At least some component of these childhood exercises influence the formation of attitudes toward the working environment. It is argued here that these attitudes are of low mass; that is, experiences are likely to have been infrequent, and inconsistent. In effect, the organizational novice is an attitudinal tabula rasa. Within organizational groups, since inputs are by definition limit- ed, the opportunity to receive variably valenced infOrmation is also 19 limited. Were any single individual free to experience the entire range of opinion on a given attitude-object, the theory would predict that his self-position would converge on the mean for the whole system. However, the group has to a certain extent self-selected itself (Moreno, 1934), according to affiliation needs, task-similarity, proximity, and other rea- sons. Prior to an individual's entrance into a group, attitude formation processes have already been at work within the group; hence, depending on the integrativeness of the group, attitudes may already be said to be homogeneous on a set of culturally determined objects to some degree. Thus, we would predict that on admission to a communication group (or more precisely, when an individual becomes integrated such that he is a member of a communication group), the associated information stimuli are a_priori restricted in variability. The degree of restriction will be a function of the group's cohesiveness, connectedness, and integrativeness. Cohesiveness (Jacobson and Seashore, 1951) is the sum of cognitive and environmental forces which act to keep a group together. These forces include intrapersonal needs for social clarification and social definition, particularly in the work context. Hence, the greater inte- grativeness, or the degree to which people who are linked to others are also linked to each other, is a measure of network redundancy. The more a group is integrated, the greater the probability of encountering re- dundant information, and hence, the greater the attitude similarity and the lower the variance. A low level of integration indicates that mem- bers of the group have a large number of non-group contacts (since groups are, by definition, connected at a minimal level). Thus integrativeness is a function of the ratio between-group and within-group interaction. Connectedness, or the degree to which individuals are connected by one- step links, would limit distortion, increase the opportunity for face—to— 2O face interpersonal exposure, and hence, high connectedness reduces vari- ability also. Cohesiveness, then, is a consequence of integrativeness and connect- edness. Festinger, Schacter, and Back (1951) point out that chance prop- erties of a neighborhood communication network, particularly propinquity, determine the affiliation behaviors of members of the network. As affil- iation increases, liking increases, and liking reinforces the pattern. Balance theorists would argue that as individuals interact, their atti- tudes will grow more similar over time, and the perceived agreement will constitute another force for holding the group together. Further, habit- ual use of a link, like any other contiguous behavior pattern, is pre- dictable from the pattern of reinforcement developed early in the rela- tionship. Thus, cohesiveness is a complex variable which will predict whether the group will be sustained. Integrativeness is an individual measure of the degree of intracon— nectedness among all_of an individual's organizational contacts. To the extent that a group member has a wide variety of non—group connections, the probability grows small that these links will be connected with each other (see Figure 1). Therefore, these links will provide the group mem- ber with unique and unpredictable stimuli. An individual may also have a large number of links outside the group, and still have a high integra- tion score. This would be the case wherein the individual is connected to several members of another group, and they are connected to each other. It is not, however, argued that the maximum integrativeness score is a desirable goal for a group. Moderate integration gives a group what Mitchell (1969) identifies as "reachability." That is, the extent to which a person can be reached or can reach other groups or individuals within the system. However, if a group is not reachable, that is to say, A l 21 FIGURE 1. Alternative Levels of Integration Within a Network. 16 ¥ ' '20 2. )0 C b . 1. u 01 O A N A 13'", ll 12 Person 3 is highly integrated. All individuals with whom 3 is connected are connected to each other by short (one-step) paths. Person 2 is moderately integrated. Person 2 is connected to Person 6 and Person 15, who are only linked by the path: 15-—¥ 17—*S-—*8 or 7—96. Person 5 is poorly integrated. With the exception of within- group links (7,8), no node with whom 5 is connected is connected to another. Node 5 has the most unpredictable, most highly variable, information sources. 22 if a group has a low level of integrativeness, it will be relatively un- constrained by disturbing, countervailing information. Connectedness, as conceptually defined here, is a measure of the density of interaction within the group (Danowski, 1974), and density (Mitchell, 1969). Connectedness is the indicator of the affective strength of within-group relationships. As Danowski (1974) has noted, "The more friends exchange information, the greater the probability of homogeneous information being exchanged." Connectedness is calculated by determining the number of actual links within the group, and dividing that number by the total number possible, or: Actual links C = n(n-1) ’ 2 where n equals the number of possible links. To the extent that a group is completely connected we would expect it to exhibit a greater degree of attitudinal similarity. Completely con— nected groups will guarantee that at least some amount of meaningful in- formation travels undistorted between all possible dyads, and at some continuous rate. Conversely, in the poorly connected group, information will move more slowly, and with greater distortion and lower effect. Figure 2 Various Levels of Connectedness 5 4 5 4 5 4 Completely Connected Partially Connected Poorly Connected C=l.0 C=0.5 C=0.4 23 The Woelfel—Saltiel theory suggests the following hypotheses: H Communication groups will be more homogeneous with respect to a set of work-related attitudes than will non-group organizational members. The groups are designated as a function of frequent and repetitive interactions which pass between members. As a result, attitudes should converge as the number of within-group messages mount. As the attitudes converge, variance will be decreased, irrespective of the mean value for any given group. Hence, greater redundancy of message stimuli should de- crease variability among communicants. H The grand mean for group members with respect to any work- related attitude should not vary from the grand mean for the non—group population. We are assessing the validity of the null hypothesis in this case. Since each group member has a restricted ratio of input stimuli, each group should converge on a unique mean. These means, assuming that groups themselves are not highly interconnected, should vary about the mean for the non-group organizational population randomly. H Groups will differ from one another with respect to a set of work—related attitudes. This hypothesis addresses the issue of within-group cultural varia- bility. That is, the organization has an overall set of norms and values which constrain the behavior of any individual member. The extent to which any norm acts to constrain is a function of the degree to which that norm is held to be important by a substantial proportion of the or- ganizational or subsystem population. Since the groups have lowered var- iability due to the homogeneity of input, each group should converge on a unique locus or mean position for each normative item. Thus, not only will groups be more uniform with respect to these norms, they will also be different from each other. 24 H The greater the within-group connectedness, the lower the within-group variability. Not only may groups have greater or lesser communication with the environment, they may also be variably linked within the group. The more that individuals are linked together, the greater the probability of re— dundant information. The more that communication is restricted to indi— viduals who are tightly linked together, or the more that the group can be said to be connected, the more uniform that group will be in their attitudes toward organizational policies and practices. In statistical terms, the consequences of repetitive contact will be the reduction of variance around a mean response givenby members of a group on some atti- tude item. H The greater the integrativeness, the lower the within-group variability. Since integrativeness is a ratio that is dependent upon both within- group and outside—of-group contacts, we would predict that a high score reflects a reduction of the variability of both environmental and within— group stimuli. When integrativeness is low, variable and unpredictable stimuli from the environment are likely to disturb the normal group proc— esses. H5 The greater the connectedness, the greater the cohesiveness. Cohesiveness has been defined as the sum of all forces acting to keep a group together (Jacobson and Seashore, 1951). Cohesiveness, in this sense, can be taken to be the consequence of repetitive interaction. Festinger (1953) has demonstrated that familiarity breeds liking. Farace and Danowski (1974) have discussed cohesiveness as a "function of the in- terpersonal attraction of a group for each other, and will be a function of the perceived potential rewards to be obtained relative to the amount of energy required to participate in the work group." Farace and Danowski 25 note that cohesiveness is mutually causally related to greater within- group message exchange. Instead of arguing that similarity builds fam- iliarity (Sherif, 1964), we are arguing the information contiguity be- tween individuals both causes them to become more similar through time, and causes them to like one another. Summary The observation of attitude can be treated as a longitudinal activ— ity rather than a discrete event and change can be treated as motion in a multidimensional space. As Woelfel and Saltiel have shown in their dis- cussion of cognitive processes, the concept of attitude may be treated as a set of interrelationships which define any cognitive element's proxim- ity to all other cognitive elements. Definition and evaluation of ele- ments must be regarded as a function of the information an individual receives. Within organizations groups may be detected which are based on the communication behavior of system members. These groups limit the variability of infOrmation available to members causing attitude means for the group to converge on a value different from that of the parent organization and with lower variability. These processes were identified as functions of the connectedness, the integrativeness, and the cohesive— ness of the group. Hypotheses relating these variables to the distribu- tion of work related norms are presented. Chapter 3 METHODS AND RESULTS Methods of group detection have been discussed from a graph-theoret- ic perspective by Harary, Norman and Cartwright (1965); Flament (1963); and Roby (1968). The application of network analysis to the study of hu- man communication systems has been extensively reviewed by Farace and Danowski (1973); Guetzkow (1971); Festinger, Schacter, and Back (1951); Jacobson and Seashore (1951); Collins and Ravin (1965); and Porter and Lawler (1972). The technique employed in this study —- network analysis (Farace and Russell, 1972; Danowski, 1974; Richards, 1974b) —- is a method developed out of sociometric techniques which takes a set of paired data specifying the interactions between individuals within a social environment and dis- plays those interactions into an N by N person matrix. Each cell enter- ing is an interaction frequency between two individuals. The matrix is rearranged in such a way as to place individuals who communicate fre- quently with each other proximate to one another in the column and row entries. Cluster analysis techniques (Richards, 1974b) exist which de— compose the dense clusters of the matrix into "groups." These "groups," arising from the data matrix, are ordinarily defined by a set of parameters. Richards (1974) has developed an algorithm which enables these groups to be formed from large organizational data sets. The algorithm allows the investigator considerable flexibility in the de- tection of such groups, but in general, groups usually meet the following 26 27 specified parameters: 1. The amount of communication by members within the group, must be greater than the amount members engage in outside the group. 2. No group member may be more than a pre-set number of links (that is, communication "steps") from any other member of the group. 3. There must be a path of communication such that any two members of the group are linked together. 4. There must be no subset of nodes or links making up more than a small pre-specified percentage of the total links or nodes, which, when removed, would cause the groups to become disconnect- ed. 5. No group may be smaller than four members. Notice -- the criteria do not require that the group recognize its own existence. Festinger, Schacter and Back's (1952) classic study of friendship patterns within a housing community found sociometric clique detection procedures useful for analyzing group properties. Schacter demonstrated that contact, principally a function of proximity, determined the devel- opment of friendship relationships. The more individuals interacted, the more they grew to like one another. Festinger e£_gl: found that co- hesiveness of a psychological group was an important determinant of the number of deviants and how effectively a group norm was maintained. In other words, the greater the cohesiveness, the greater the number of de- viants (in absolute terms), and the easier it was for a group to maintain obedience to a norm. We would argue that this is supportive of our con- tention that group membership leads to attitudinal uniformity. Note that Festinger's definition of a psychological group is conceptually similar to Smith et al.; however, his clique detection routine is conceptually all .I. .I u n 28 equivalent to that employed by Richards. Sociometric groups have, then, been hypothesized to be referential, norm providing, and socially rewarding. It is argued that communication groups within organizations perform functions similar to those of the social group. Similarly, Bales (1953) found it useful to use communication behavior to describe the functions of social systems. Brown (1974) used interac— tion patterns to detect patterns of racial and male chauvinism within a college community. The preceding studies differ from the present effort in an important respect. At the operational level, each requires some- thing of a respondent beyond a report of his communication behavior. The Festinger studies ask the respondent to name his friends. Bales (1951) was more interested in measuring the individual components of a group than differentiating a group from its environment. In the present case, our interests lie in extracting a group from an environment accord- ing to a rigid set of criteria and then demonstrating that the fact of patterned interaction has consequences for the individual and the organi- zation. At its simplest level, network analysis requires individuals to in- dicate with whom they communicate and how often. The naming can be done either by asking a person to list individuals with whom he/she has fre- quent contact, or by providing the person with a list of the names of individuals believed to be members of a common social system. Specifying the names to be used on such a list is not a trivial problem. Often one of the most difficult operational problems is specifying who should and should not be included in the system census. This is generally known as a boundary problem, and in the present case has been unambiguously re- solved by including only those individuals who are on the payroll of the 29 organization under study. Often, network analysts will specify the content which is relevant to a particular study. Hence, the investigator might be interested in change processes, would attempt to have respondents indicate with whom they discuss "new ideas," or innovations. In the present investigation content was not restricted. Subjects were asked to choose those people with whom they interacted on the job. Channel -— that is, face—to-face, telephone, memo or other mediated transmissions —- can also be used to discriminate different networks within a common organizational boundary (Farace and Johnson, 1974). In the present study, channel was restricted to face-to-face contacts. After data are gathered, they are placed in an N by N matrix where N_is equal to the population of the network. The matrix is rearranged by moving individuals into proximity along the diagonal of the matrix with those with whom they interact. This is accomplished by shifting the position location of a person's column and row entries such that they lie next to or nearly adjacent to others with whom they interact. The clus- ters of individuals along the diagonal are then analyzed according to the criteria we have discussed and groups are formed. In the Richards' algo- rithm, the diagonal is treated as the base of a histogram and the inter- action frequency between individuals forms the length of the perpendicu- lars of the histogram. In addition, the distances between the frequency lines are manipulated according to the strength of the link between in- dividuals. As strong links are identified, this distance is continuously shrunk such that dense clusters of lines fbrm. Cluster analysis is per- formed on these dense areas and groups are discriminated from the histo- gram. 3O Individuals are assigned different roles as a result of their inter- action patterns. Dense clusters become groups. Single individuals link- ing groups are called liaisons since they provide a means by which infor- mation may be exchanged between two groups. Certain group members are linked to other group members in other groups, and are referred to as bridges. Other persons have no links and are referred to as isolates. TWO types of isolates are specified: Type 1 isolates are linked to one other person within the network, either a group member or a dyad member; Type 2 isolates are not linked to anyone in the network. While this may be an artifact of measurement error, it is also possible that certain in- dividuals are remote from the entire organization communicatively. Also, Type 2 isolates are likely to be found in topic specific networks. Fin- ally, certain individuals meet none of the other criteria and are classi- fied as Others. Within the algorithm itself, the investigator has a number of para- meters at his control. In the present study only reciprocated links, links which are acknowledged by both members of any given dyad, have been included. The links are assumed to be bidirectional, or initiation of contact is presumed to be equal for both members of each dyad. This simplifying assumption is subject to the criticism that it does not rec- ognize the normative force of status rules within hierarchically con- strained relationships. Links may also be characterized as having vary- ing strengths. In the present data set the link strength was inadequate- ly operationalized as frequency of contact only, disregarding the length for an average contact. Link strength was operationalized as: How frequently do you talk to ? 5 Very frequently 4 Frequently 3 Occasionally 2 Infrequently 1 Very infrequently 31 Responses in the "Very infrequently" category were eliminated from the data set. Thus, while strength, and strength discrepancies, are not em- ployed as independent variables, link strength values 2-5 are considered to constitute a "communication link" for purposes of this research. Description of Data Source The data employed in this study were gathered by survey instruments from a Midwest manufacturing concern of moderate size (approximately 450 employees).1 The data were originally gathered during 1970 as a part of an effort to monitor an ongoing organizational development program. This program originated with the introduction of a Scanlon Plan2 in 1952. The Scanlon Plan is a theory and method of management which rests on the assumption that employee motivation can be facilitated through the distribution of decision—making authority to committees. According to Heinen (1972) the Scanlon Plan stresses the need for: l. a system of sharing with the employees infbrmation about the organization's objectives and problems, 2. a system for increasing organizational effectiveness through the participation of employees in decisions that affect them- selves and their jobs, and 3. a system for establishing a reward structure and leadership practices that foster cooperation. Subsequent behavioral science inputs have modified the original Scanlon Plan design. Management grid training programs were introduced through Likert (1967) and the Institute for Social Research at the Univ- ersity of Michigan. These programs transformed the committee structures into a series of work teams. A work team was defined as a group of people with common responsibility for the organization's work. Each The data were gathered by Professors Heinen (1972) and E. Jacobson, Dept. of Psychology, M.S.U., and made available for secondary analysis. 2 Thorough discussions of the Scanlon Plan can be found in McGregor (1952), Lesieur (1958), and Doyle (1970). 32 foreman or supervisor was responsible for from one to three work teams of approximately 13 individuals per team. The work team functions to plan the work, identify and solve problems, and to evaluate the activities of the team. It should be stressed that this organization is unusually sensitive to the need for cooperative or participative management systems. The re- sultant communication policy may be judged to have enhanced the amount of communication. Survey data were obtained from the Inter-Company Longitudinal Survey (ICLS) of the Institute of Social Research. Attitude items were measured on five-point Likert scales (see Table 2). The characteristics of the items used in measuring each variable from the ICLS questionaire were examined through a cluster analysis (Heinen, 1972). As a result of this analysis (Appendix A), 23 items were chosen from a possible 212 on the basis of their ability to discriminate between subjects. Means and stan— dard deviation for each variable are presented in Appendix B. In addi- tion, variables related to age, education and tenure with the firm were added to the set. The breakdown of individuals into network roles is given in Table 1. TABLE 1. Network Roles for Employees by Category. ROLE SUBJECTS Group members (includes bridges) 180 Liaisons 5 Dyads 2 Isolates T1 16 Isolates T2 63 Others 148 TOTAL ulu 10. ll. 12. 13. 14. 15. 33 TABLE 2. ICLS Variables. To what extent do persons in your work group keep each other informed about important events and situations? How friendly and easy to approach are the persons in your work group? When you talk with persons in your work group, to what extent do they pay attention to what you're saying? To what extent are persons in your work group willing to listen to your problems? How much do persons in your work group encourage each other to give their best effort? To what extent do persons in your work group maintain high standards of performance? To what extent do persons in your work group help you find ways to do a better job? To what extent do persons in your work group provide the help you need so that you can plan, organize, and schedule work ahead of time? To what extent do members of your work group plan together and coor- dinate their efforts? To what extent do you feel that you and the other persons in your work group belong to a team that works together? To what extent do you have confidence and trust in the persons in your work group? When did you first come to work here? Into what age bracket do you fall? How much schooling have you had? 0n the basis of your experience and information, how would you rate your work group on effectiveness? How well does it do in fulfilling 16. l7. 18. 19. 20. 21. 22. 23. 34 its missing or achieving its goals in comparison with other work groups in the company? How much do the members of your work team want to stay in your work team rather than join any other work team? In discussing a problem in your work team, to what extent do team members feel free to suggest ideas that are different from the maj- ority opinion? In arriving at a decision to a work team problem, how often does your work team outline a detailed action plan? To what extent does your work team feel free to change a previous team decision? To what extent are your interests and needs taken into account when work activities are assigned? To what extent are you important to the smooth operation of your work team? To what extent do the members of your work team see eye to eye on most matters? To what extent do the members of your work team agree with you on what are the basic purposes of your work team? 35 Nineteen groups (in Table l) were generated with a mean size of 9.5 and a range of three to 26 members. Two groups with less than five mem- bers were eliminated from the analysis of groups for statistical consid- erations. They were, however, retained as a part of the group population when analysis calls for the aggregation of all group member data. The communication groups cut across work-team boundaries. All respondents were, first of all, members of a work team. 0f the 180 group members, 68, or approximately 38%, were assigned to a communication group which differed from their work team. Results H Communication groups will be more homogeneous with respect to a set of work-related attitudes than will non-group organizational members. Figure 3 shows the trajectories of the mean variances (Table 3) for both the non-group and the group populations. Of the 23 items, the var- iance for the grouped population is exceeded by the variance for the non-group respondents on 12 items. Since this represents just barely more than 50% of the variables, Hypothesis 1 is not supported. HlA The grand mean for group members with respect to any work- related attitude should not vary from that for the work group population. The results of t-tests for all variables broken down by group and non-group populations are given in Table 4. For 18 of the 23 variables the t-value is non-significant (p<.05). Given the nature of the excep- tions noted below, Hypothesis 1A is regarded as supported. For those variables in which significant differences (p<.05, Table 5) are obtained, it is possible to observe a pattern. TABLE 3. Variables Provided Information Peer Friendly Peer Attention Peers Listen Work Group Encouraged Work Group Has High Standards Work Group Innovative Work Group Plans Ahead Work Group Coordinates Team Work Trust Tenure Age Education Rate My Group Stay With My Group Suggest Ideas Detailed Plan Change Decision Your Needs Self Importance Team Agrees Goal Agreement 36 Descriptive Statistics: n E', s 52 178 3.40 1.01 1.02 160 3.02 1.21 1.45 156 4.40 .66 .44 154 4.27 .69 .48 155 4.18 .86 .73 157 4.26 .70 .49 155 4.49 .64 .41 153 4.18 .77 .59 157 4.36 .66 .44 169 3.43 .91 .83 170 3.37 .93 .87 165 1.29 .46 .21 167 2.47 1.06 1.13 167 3.18 1.96 3.85 161 2.61 1.17 1.37 144 4.02 1.07 1.14 123 3.51 1.36 1.84 140 3.27 1.07 1.13 134 2.94 1.04 1.08 133 3.15 .93 .86 137 3.21 .93 .86 135 '3.47 .94 .88 132 3.26 .85 .73 n = sample size s §'= sample mean 2 s n Group and Non-Group Respondents Non-Group X S 225 209 199 197 195 199 196 190 196 212 212 209 214 207 213 209 195 207 198 190 204 201 201 3.47 3.19 4.46 4.37 4.31 4.39 4.58 4.35 4.34 3.58 3.60 1.31 2.64 3.76 2.97 4.02 3.49 3.38 2.90 3.38 3.40 3.54 3.50 variance 1.09 1.36 .67 .67 .73 .60 .53 .70 .63 .95 1.05 .46 1.17 1.94 .97 1.91 1.33 1.22 .98 1.02 1.02 1.03 .80 standard deviation 1.18 1.85 .45 .45 .54 .36 .28 1.42 1.76 1.48 .96 1.03 1.03 1.06 .64 Comparison of Mean Variances for Group and Non-Group Population. FIGURE 3. 37 3.5 3.0 2.5 04 :3 0 $4 040 8 3: $4 0 (D Z ) / I (I I I I I \ ,1. I ’l I I / 4. \ \\ 7 I c—“J I” If ’I 0’ I ‘x 5 I I J \ \ ‘x I )5 I I I < \ \ \ \ I I I I r: \ \ \ \ LO 0 Ln PI H 0.0 Goal Agreement Team Agrees Self Importance Your Needs Change Decision Detailed Plan Suggest Ideas Stay With My Group Rate My Group Education Age Tenure Trust Team Work Work Group Coordinates Work Group Plans Ahead Work Group Innovative Work Group Has High Standards Work Group Encouraged Peers Listen Peer Attention Peer Friendly Provided Information TABLE 4. Provided Information Non—Group Group Peer Friendly‘ Non-Group Group Peer Attention Non-Group Group Peers Listen Non-Group Group Work Group Encouraged Non-Group Group Work Group Has High Standards Non-Group Group Work Group Innovative Non-Group Group Work Group Plans Ahead Non—Group Group Work Group Coordinates Non-Group Group Team Work Non-Group Group Trust Non-Group Group Tenure Non-Group Group Age Non-Group Group 38 t-Tests Between Non-Group and Group Populations. Means t—Value df 3.47 .59 402 3.40 3.19 1.19 367 3.02 4.45 .84 353 4.39 4.38 1.32 349 4.28 4.31 1.56 348 4.18 4.38 1.83 354 4.26 4.58 1.39 349 4.49 4.54 2.15 341 4.17 4.43 1.03 351 4.36 3.58 1.11 379 3.47 3.59 2.21 380 3.37 1.31 .42 372 1.29 2.64 1.44 379 2.47 Sig. .56 .24 .40 .19 .12 .07 .17 .03 .30 .27 .03 .152 Education Non-Group Group Rate My Group Non—Group Group Stay With My Team Non-Group Group Suggest Ideas Non—Group Group Detailed Plan Non—Group Group Change Decision Non-Group Group Your Needs Non—Group Group Self Importance Non—Group Group Team Agrees Non-Group Group Goal Agreement Non-Group Group 39 t-Value .40 3.28 .02 —.1O .87 -.37 2.11 1.49 .66 2.66 372 351 316 345 330 321 338 334 331 Sig. .69 .98 .39 .71 .04 .13 .51 .01 4O TABLE 5. Variables with Significant t Differences Between Group and Non-Group Populations. Mean t-Value Work Group Has High Standards Non-Group 4.38 1.83 Group 4.26 Work Group Plans Ahead Non-Group 4.54 2.15 Group 4.17 Trust Non-Group 3.59 2.21 Group 3.17 Rate My Group Non—Group 1.16 3.28 Group .97 Your Needs Non-Group 3.38 2.11 Group 3.50 Goal Agreement Non-Group 3.50 2.66 Group 3.25 Note that the mean values for these (Table 5) variables are gener- ally lower for the group members than for the non-group individuals. Lower values represent less support for work team processes. Having knowledge not available to the rest of the work team about the functioning of the rest of the organization may enable the group member to assess the ability of his work team to plan ahead, relative to other work teams. Significantly, use of outside sources might reflect greater trust in the outside sources and a consequent (or as a consequence of) lowering of trust within the work team. Following this line of reasoning, one would 41 then predict the differences found for rating of the work team, relative desire to stay with the work team, and the perceived importance of self to the functioning of the work team. Since communication groups do not strictly reflect the work team structure, members of communication groups are more likely than non—group members to receive information from a wider variety of sources. Hence, an individual would be less likely to rely upon the work team for social and work-related support. Thus we would expect that group members would vary significantly from the non-group population on variables like goal agreement, intrapersonal needs, rating fer the work group, work group standards, work group planning, and work group trust. H Groups will differ from one another with respect to a set of work-related attitudes. This hypothesis indicates that groups will converge on a unique value for the various independent variables. The one-way analysis of the variance table (Table 6) shows that the groups significantly varied on nine of the 23 items (p<.05). While this is supportive of the hypothesis, additional analysis is required to eliminate the possibility of spurious findings. Additionally, it is hypothesized that if the groups are creating a unique informational environment for members, the complex of all atti- tudes should discriminate the groups. That is, since each group is unique- 1y achieving a mean value for each work-related attitude, it was predicted that rotating the group members through an N-dimensional discriminant space where N is equal to the number of variables should recreate the net— work groupings from clusters of the items alone. A step-wise discriminant analysis was performed in the data with hierarchical clustering. Appendix C gives the entering sequence and the 42 F probability of each entered variable. The procedure employed allowed the investigator to indicate the initial group structure. Following the cluster analysis, clusters are compared to the original group and Table 7 gives the comparison between actual and predicted results on a group by group basis. To read the table, look across the row to see how individ- uals from the network groups are distributed into discriminant clusters. Reading down the columns gives the membership of the discriminant clus- ter. For example, six individuals from network group 1 were placed in discriminant cluster 1, seven in cluster 6, and one in cluster 17. Note that due to list—wise deletion of cases only 121 subjects are included in this analysis. (By list-wise deleting, we mean that when missing data were encountered for a respondent on a variable, all values for all variables for that respondent were deleted from the analysis. Therefore, the number of subjects was reduced.) Further, because 60 of the respondents did not complete one section of the questionnaire, eight variables (stay with my team; suggest ideas; detailed plan; your needs; self importance; team agrees; goal agreement) were eliminated from the analysis. The results of the discriminant analysis provide strong support for this hypothesis, and for the theoretical value of the network clique- detection routine. In 59.6% (p<.0001) of the cases, individuals were correctly placed in their predicted groups. Given 17 groups of known size, random procedures would correctly place only 6%. In other words, simply knowing the complex of attitudes and the age, tenure, and the edu- cation of the communication groups enabled the groups to be reproduced with nearly 60% accuracy. Hypothesis 2 is, therefore, supported by the evidence. 43 TABLE 6. One—Way Anova: Variable By Group. YEEEEEEE- 532a::s .215; F Ratio F Prob. Provided Information 212.16 175 1.77 .04 Peer Friendly 227.97 157 1.49 .11 Peer Attention 66.62 153 2.29 .01 Peers Listen 72.39 151 1.02 .44 Work Group Encouraged 110.88 152 1.93 .02 Work Group Has High Standards 75.68 154 1.93 .02 Work Group Innovative 62.21 152 1.45 .13 Work Group Plans Ahead 88.17 150 .74 .75 Work Group Coordinates 67.48 154 1.26 .23 Team Work 139.63 166 .36 .99 Trust. 145.62 167 1.23 .22 Tenure 33.87 162 3.90 .00 Age 187.07 162 3.21 .00 Education 625.90 164 3.47 .00 Rate My Group 216.34 158 15.23 .00 Stay With My Team 162.94 141 1.33 .21 Suggest Ideas 222.25 .20 .54 .88 Detailed Plan 154.12 137 2.69 .00 Change Decision 138.39 131 1.47 .15 Your Needs 110.95 130 1.63 .08 Self Importance 116.99 135 .77 .68 Team Agrees 115.02 132 .26 .99 Goal Agreement 89.57 129 1.00 .45 44 TABLE 7. Network and Predicted Groupings. Network Groups Size Group 1 16 6 7 2 1 Group 2 9 4 l l 2 Gnmp3 8 6 l 1 Group 4 8 4 l l 1 Group 5 7 5 2 Group 6 21 2 16 l 1 Group 7 4 l 3 Group 8 4 l l l 1 Group 9 5 4 1 Group 10 7 l 2 l 2 1 Group 11 4 4 l 2 1 Group 12 4 1 3 Group 13 4 3 Group 14 2 2 Group 15 6 2 4 Group 16 2 2 Group 17 3 1 l Digigiginant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 TOTAL SIZE 8 5 8 6 6 2 5 2 7 4 4 4 3 4 9 4 6 (Discriminant) Classified: n = 121 59.6% Cases Correctly Classified p < .000 45 H The greater the within-group connectedness, the lower the within-group variability. Hypothesis 3 is not directly supported by the evidence. Figure 4 gives the average standard deviations for all variables across all groups. The groups are ranked for mean connectedness along the Xsaxis, and clear- ly there is no increasing trend from left to right. It was felt that perhaps size was suppressing the relationship. The Pearson correlation between size and connectedness was a -.76. However, the partial correla- tions between connectedness and the independent variables (see Table 8) show little change. While most correlations became slightly positive, the absolute value of the correlations reflect little change. However, chi—square results indicate that connectedness does enable us to discriminate between the groups (Table 9). Significant differences are found for ten variables (p<.05) and nearly significant variations are found for two other variables (p<.10). Therefore, it can be concluded that connectedness does manifest itself in the attitudes of the group members. H The greater the integrativeness, the lower the within~group variability. The prediction of a linearly decreasing variance with respect to in- tegrativeness is not supported. Again, the pattern of the average stan- dard deviations of the variable by group (Figure 5) is linear, but the overall trend is negligible. Controlling for size (with integrativeness, Pearson r = -.76) only marginally affects the correlations (Table 10). The relationship between integrativeness and connectedness was examined to remove the possibility of non-independence between the variables. The Spearman~Rank order correlation coefficient between the two variables is equal to .25 (p<.01). While this is a significant correlation, it is still relatively low, the shared variance being equal to 6.3%. 46 FIGURE 4. Average Standard Deviations for all Variables by Connectedness Rank.* Average Standard Deviations 2.0 1.75 1.5 1.25 1.0 .75 .25 l 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Connectedness Rank * Variables limited to those used in discriminant analysis to maintain sample sizes. 47 TABLE 8. Variables by Connectedness Rank. .Variable Zero Order Correlation Controlling for Size Provided Information —.08 -.01 Peer Friendly -.08 .04 Peer Attention -.08 .05 Peers Listen -.003 .09 Work Group Encouraged -.09 .06 Work Group Has High Standards -.11 .10 Work Group Innovative .03 ‘ .13 Work Group Plans Ahead .12 .16X Work Group Coordinates -.04 .11 Team Work -.03 -.003 Trust .02 .07 Tenure .19x .11 Age —.19x -.23x Education -.13 -.11 Rate My Group -.36 -.05 Stay With My Team -.004 .16x Suggest Ideas -.05 -.04 Detailed Plan -.21 —.09 Change Decision -.09X -.16 Your Needs -.19x -.11 Self Importance -.05 .06 Team Agrees -.02 -.05 Goal Agreement -.09 .02 (p<.05) TABLE 9. Chi Square Goodness of Provided Information Peer Friendly Peer Attention Peers Listen Work Group Encouraged Work Group Has High Standards Work Group Innovative Work Group Plans Ahead Work Group Coordinates Team Work Trust Tenure Age Education Rate My Group Stay With My Team Suggest Ideas Detailed Plan Change Decision Your Needs Self Importance Team Agrees Goal Agreement 48 Fit Tests - Variables by Connectedness. 12.95 12.18 15.67 7.72 11.73 16.68 9.23 20.37 9.16 11.71 11.14 18.93 17.85 14.99 33.97 23.52 11.61 28.05 24.42 35.59 15.58 22.75 13.34 dif. 16 16 12 12 16 12 8 16 8 16 16 u 16 16 8 16 16 16 16 16 16 16 16 sig. .70 .73 .20 .80 .76 .16 .32 .20 .33 .76 .80 .00 .33 .52 .00 .10 .77 .03 .08 .00 .48 .12 .64 49 Chi—square goodness of fit tests indicate that integrativeness does account (p<.05) for differences in the distribution for seven independent variables (Table 10). While it is difficult to determine an overall pat- tern for the chi-square results, they demonstrate that the groups can be affected by different levels of integrativeness. H5 The greater the connectedness, the greater the cohesiveness. The correlations between the measures of cohesiveness are presented below. The code variables were chosen on the basis of a cluster analysis of all 23 variables (see Appendix B). Two items which otherwise were eliminated from the analysis are included, despite the fact that the items do not effectively discriminate between subjects. 71 FIGURE 5. Mean Standard Deviations for Variables by Integrativeness Rank. 7— Average Standard Deviation 1.5 1.25 1.0 .75 .50 .25 1 2 3 4 5' 6 7 8 9 10 11 12 13 14 15 16 17 Integrativeness Rank TABLE 10. Chi Square Goodness of Fit Tests -- Variable by Integrativeness. Provided Information Peer Friendly Peer Attention Peers Listen Work Group Encouraged Work Group Has High Standards Work Group Innovative Work Group Plans Ahead Work Group Coordinates Team Work Trust Tenure Age Education Rate My Group Stay With My Team Suggest Ideas Detailed Plan Change Decision Your Needs Self Importance Team Agrees Goal Agreement 50 2 x 16.31 27.83 16.67 23.25 24.43 5.97 14.94 9.54 10.38 13.47 3.80 20.83 10.43 34.15 16.10 9.33 21.88 22.44 20.57 11.33 11.82 19.19 dif. 16 16 12 12 16 12 16 16 16 16 16 16 16 16 16 16 16 16 16 sig. .87 .43 .01 .16 .10 .Ol .65 .53 .29 .85 .64 .43 .18 .84 .00 .45 .89 .15 .13 .19 .78 .75 .25 51 TABLE 11. Variableslanonnectedness. . Pearson XEEiEElEE- Correlations Peers Friendly -.05 Stay with my team .004 . xx Get along w1th my team -.24 Team sticks together .03 Help each other on the job -.03 (xx = p<.01) The results clearly do not support the hypothesis. Arguably, the items relate to peers in the work team, and forces which act to keep the work team together. Because one's psychological peers are only partial- ly within the field provided by the work team, these correlations would necessarily be less. However, if this argument holds, then the zero or- der correlations between the cohesiveness items ought to be substantial- ly increased when the effects of connectedness are partialled out of the relationship (see Table 12). TABLE 12. Inter-item Zero Order Correlations. (1) Peers Friendly (2) Stay with my team .18X (3) Get along with my team .18X .06 (4) Team sticks together .12X .06 .24X (5) Help each other on the job .10 -.01 .25x .71x (1) (2) (3) (4) (x = p<.05) 52 TABLE 13. Partial Correlations Controlling for Connectedness. (1) Peers Friendly (2) Stay with my team .08 (3) Get along with my team .04 .06 (4) Team sticks together .18X .04 .25X (5) Help each other on the job .15 -.05 .25X .71X (1) (2) (3) (4) (x = p<.05) With the exception of the correlations with perceived peer friend- liness, the partial correlations are unaffected by connectedness. Wheth- er this is due to the poor quality of the data or the actual absence of any effect due to coneectedness is undetermined. The hypothesis, in either case, is not supported by the results. Discussion The results of this study reflect the validity of Selznick's obser- vations. Informal structure does appear to be a determinant of the posi- tion an individual will adopt with respect to his work. The results are due to an odd reversal on conventional attitude-behavior formulations. Most theorists in the attitude-behavior milieu argue that (Siebold, 1974) attitudes should lead to a set of behaviors in a logical fashion. That is, attitudes are indicants of a predisposition to behave in some consis— tent manner. The results of this study indicate that attitudes are the consequence of behavioral processes, in particular, communication beha— viors. These communication acts, strung together over the career life- span of an individual, lead to developed patterns of behavior which render the individual's work predictable and his perceived association with the 53 organization mutually satisfactory. However, these data are not so strong that such a position can be advanced unequivocally. It is possible, as an alternative explanation, that individuals join or are integrated into communication groups be- cause, a_priori, a similarity of attitudes and expectations exist. Such a hypothesis is not in any way disconfirmed by the results reported above. It seems clear that research ought to be undertaken which casts the two alternative explanations against one another. It is proposed that future research will concentrate on this anoma- ly. Such research would ideally be carried out longitudinally in a de— veloping organization. If the former hypothesis holds, then a pattern should emerge in which organizational novices would enter the system with widely discrepant attitudes. There would be a correlation between change scores in the attitude configuration (that is, changes in the sum of all attitudes) and level of integrativeness in the organization EE£.§E: Secondly, as the novice becomes a member of a group, we should see in that person's attitudes a shift in the direction of the group mean. We have argued, and demonstrated, that the communication group does provide a unique environment within which referential socialization processes take place. Therefore, the final test would rest upon differ— ences between novices as a function of their involvement with a communi- cation group. The alternative explanation would entail novices entering with an attitude set roughly similar to some specific communication group, or the novice would have to adopt such a set rather rapidly after initiation into the organization. Knowledge of a person's a priori attitude set should enable one to predict which group the individual would associate with, and the probability of a high degree of integration into that group. 54 Note, however, that this explanation assumes that each individual has an equiprobable chance of contacting the appropriate communication group. This possibility seems inherently unsatisfactory, given that organiza~ tional structures exist to constrain such an equiprobable distribution of alternatives. In fact, such a distribution would give an organiza- tion an interpersonal uncertainty coefficient of 1.0. Conclusions The results provide strong support for the contention that communi- cation groups are norm—providing mechanisms. The research demonstrates that unique constellations of attitudes can reliably recreate the infor- mal groups detected by the network algorithms. Further, integrativeness and connectedness were found to be related to the process of the develop— ment of group normative structures. It was predicted, however, that the variance for any attitude would linearly decrease with increases in both connectedness and integrative— ness. This did not prove to be the case. An examination of Figure 4 and Figure 5 shows that, as predicted, the mean variances for each group on the set of all variables was very low. The range of variances should have been sufficient for the emergence of the predicted linear effects. The groups were further collapsed into five groups of groups in order to make certain that size was not distort— ing the results, and again, no linear trend was present. Since these hy- potheses were deduced from the propositions advanced by Woelfel-Saltiel, we are forced to consider the possibility that the theory is not supported. It is possible that connectedness and integrativeness are inadequate operationalizations of frequency. Woelfel-Saltiel is based largely on the notion that attitude change (or states) is a linear function of the 55 frequency of interaction. Therefore, it would seem wise to test the strength of the links for the predicted linearity. This data set, as mentioned earlier, lacked an adequate operationalization of link strength, and, hence, this alternative hypothesis cannot be tested here. It should be noted that given the minimal differences between groups and non-group populations (Table 3), the statistical differences observed in the analysis of variance and the chi-square tests take an additional meaning. This study demonstrates that continued research into the infor- mal communication groups will be fruitful for organizational theory. In addition, a study should be conducted which offers greater opportunity to contrast the efficacy of frequency as a predictor against other vari- ables such as within-group hierarchical differentiation, status, size, and longevity. Such a study is currently in the design stage. If Woelfel and Sal- tiel are correct, an adequately operationalized link-strength variable should lead to the identification of individuals who are highly embedded in an interactant with an organizational network. These individuals should have the greatest impact on change within the system. It is proposed that a longitudinal study be carried out which looks at the correlation between changes in the attitudes of the high embedded group and changes in the attitude of the network at large. Secondly, we would argue that the theory requires contact, or interacting opportuni- ties before an individual's messages may impact upon the cognitions of system members. Therefore, a highly embedded group ought also to predict change better than measurements of a management task force. In other words, a second group consisting of members of a self—nominated manage- ment task force will be measured and their messages will be counted. Perceived status of the task force will be measured to make certain that 56 salience of the infrequent messages from this group does not intervene. To the extent that change in the highly embedded group predicts change in the rest of the organization, Woelfel~Sa1tiel will be supported. Implications for ManagementkTheory A final note should be made about the relationship between the com— munication group and the work-team management structure. Communication groups are norm-providing mechanisms which channel the attitudes of the members along paths which are at variance from the organization. In the case of this organization, we find that the mean for attitudes which would facilitate the development of work teams are hampered by the infor- mal group structure. The results show that members of informal groups feel less inclined to trust their work team members; less willing to advocate staying with the work team; less willing to rate highly the performance of their group; and, most importantly, less likely to see themselves as substantially important to the productivity of the team. This can, however, be looked upon two ways. While these attitudes may affect the perfOrmance of the team negatively, these attitudes also might indicate a healthy skepticism. Were each unit to operate in an in— formation vacuum, the importance of the team's existence and the contri- bution of the team would tend to become artificially inflated. Summary A network analysis was performed at a medium-sized midwestern manu- facturing plant. From the network data, two dependent measures were con— ducted -~ integrativeness and connectedness. The data provided strong support for hypotheses which argued that communication groups do have 57 unique attitudes and perceptions with respect to each other. Hypotheses predicting a linear trend as a function of integrativeness and connected- ness were not confirmed. APPENDICES APPENDIX A Item Analysis The characteristics of the items used in measuring each of the var— iables from the ICLS questionnaire were examined by Heinen (1972) and the results are reproduced here. The cluster analysis was based upon a sample of 306 individuals from all of the company employees, who had re- turned completed questionnaires and not merely limited to the members of the communication groups. The total organizational pool of subjects was selected for purposes of generalizability and stability of measurement. The measures used in the study were intended to reflect characteristics of work teams in the whole organization rather than the twenty teams with complete data. Also, the characteristics of cluster analysis re- quire a sample of several hundred to insure the stability of the mea- sures in the analysis. For both these reasons the item analysis used the total organization subject pool. The items used as measures of the variables for this study that were included in the ICLS questionnaire were: Cohesiveness 1. To what extent do you feel that you and the other persons in your work group belong to a team that works together? 2. How much do the members of your work team want to stay in your work team rather than join any other work team? How does your work team compare with the other work teams at on: 3. the way they get along together? 58 4. 5. 59 the way they stick together? the way they help each other on the job? Problem Solving 11. To what extent do persons in your work group keep each other in- formed about important events and situations? 12. To what extent are persons in your work team willing to listen to your problems? 13. To what extent does your work team plan together and coordinate their efforts? 14. To what extent does your work team make good decisions and solve problems well? 15. In discussing a problem in your work team, to what extent do team members feel free to suggest ideas that are different from the majority opinion? 16. In arriving at a decision to a work team problem, how often does your work team outline a detailed action plan? 17. To what extent does the work team feel free to change a previous team decision? Identification 21. To what extent do persons in your work team help you find ways to do a better job? 22. To what extent do persons in your work team provide the help you need so that you can plan, organize, and schedule work ahead of time? 23. To what extent are your interests and needs taken into account when work activities are assigned? 24. To what extent are you important to the smooth operation of your work team? 25. To what extent do the members of your work team see eye to eye on most matters? 26. To what extent do the members of your work team agree with you on what are the basic purposes of your work team? 60 Perceived Team Growth 31. 32. 33. 34. 35. To what extent do you have confidence and trust in the persons in your work group? On the basis of your experience and information, how would you rate your work group on effectiveness? How well does it do in fulfilling its mission or achieving its goals in comparison with other work groups in the company? How do you feel about the progress your work team has made in the last year? To what extent do you believe you have ideas for improving your work team that have not been expressed and fully discussed? To what extent do you think you and the people you work with need training in how to build a more effective work team? Tabler+gives the inter-item correlation matrix for these items and TableJUSpresents the median and the range of the inter-item correla- tions, item-cluster correlations, and the item communalities for each of the scales contained in the ICLS questionnaire. For all the variables the inter—item correlations, item-cluster correlations, and the item communalities were moderately high to high. 61 00.H 0m. H:. mm. 0m. 5m. mm. 0H. HN.. 5N. mm. 0H. 0m. mm. mm. 0m. 5H. 0m. 0m. :m. :m. 5H. mm. mm 00.H mm. mm. mH. mH. 0H. m0.I 00. mm. 0H. 00. 50. NH» mm.. Hm. 0m. 5H. 50. mo. 50. 00. mm. :m 00.H 5:. mm. mm. H:. :m. mm. 50. mm. mm. :m. Hm. :m. 0:. 5m. mm. 0:. 0:. 5m. mm. 5:. mm 00.H mm. mm. mm. 0H. 0m. mm. mm. mm. mm. Hm. 0:. Hm. 5m. mm. H:. H:. m:. Hm. mm. mm 00.H mm. 5m. HH. 5m. m:. H:. 0m. mm. mm. 0m. 0:. m:. H:. 0:. 0:. mm. :m. mm. Hm 00.H mm. mm. H:. 5m. mm. mm. mm. 0:. mm. 0:. 5m. 0:. 0:. 0m. :m. mm. m:. 0N 00.H mm. H:. :m. :m. mm. mm. mm. Hm. mm. 5m. ::. mm. mm. Hm. 0H. H:. mm 00.H 0m. m0. 0H. 0H. 0N. mH. 0H. 0H. 5H. mH. mm. mm. mm. :H. 5m. :N 00.H mm. 5m. mm. mm. mm. H:. 5m. Hm. :m. :m. 0H. 0m. HN. 0:. mm 00.H mm. mm. :m. 5m. 5:. mm. H:. H:. Hm. Hm. :m. mH. mm. mm 00.H mm. 5m. :m. Hm. mm. m:. mm. mm. mm. mm. m0. mm. Hm 00.H mm. mm. mm. mm. Hm. 5m. mm. mm. Hm. mm. mm. 5H 00.H 0:. :m. mm. mm. mm. 5N. mH. 0m. mm. mm. 0H 00.H :m. 0m. mm. mm. 5H. 0m. Hm. mm. mm. mH 00.H m0. 5m. N:. Hm. :m. mm. mm. mm. :H 00.H 0m. 5:. m:. mm. mm. mm. H5. mH 00.H mm. mm. mm. mm. :H. H:. NH 00.H 0:. mm. mm. :m. 5:. HH 00.H :0. :0. mH. m:. m 00.H H5. 5m. m:. : 0.H mm. ::. m 00.H 5N. m 00.H H mm :m mm mm Hm mm mm :0 mm mm Hm 5H 0H mH :H mH NH HH m : m m H mEmpH .WCOMPMHmrHcHOU EWPHIranwCH mmHMUm qu0H .3H 33H. 62 TABLE 15. Median and Range of Inter-Item Correlations, Item-Cluster, and Item Communalities of Scales in ICLS. Item Communalities Variable Inter-Item r's Item—Cluster r's Scales median 133; 2.511. median 1'91. Coh 54 44 71 80 55 83 PS 35 30 68 62 58 68 Id 37 34 68 67 58 68 PTG 41 38 47 66 58 70 hig median low high 64 31 69 39 34 46 45 33 47 44 35 49 Alphas were computed for each of these scales for both the total organization sample and also for the twenty production level work teams. The alphas generally are only slightly lower in the smaller production level work team sample (see Table 16). TABLE 16. Alphas of ICLS Scales. Variable Scales Coh PS Id PTG Total Organization 83 77 79 68 Twenty Production Teams 83 71 83 68 APPENDIX B f TABLE 17. Descriptive Statistics: All Reapondents. Y Provided Information Peer Friendly Peer Attention Peers Listen Work Group Encouraged Work Group Has High Standards Work Group Innovative Work Group Plans Ahead Work Group Coordinates Team Work Trust Tenure Age Education Rate My Group Stay With My Team. Suggest Ideas Detailed Plan Change Decision Your Needs Self Importance Team Agrees Goal Agreement 404 369 355 351 350 356 351 343 353 381 382 374 381 374 374 353 318 347 333 333 341 336 333 3? 3.44 3.12 4.43 4.33 4.25 4.54 4.27 4.40 3.53 3.50 1.30 2.57 3.22 3.22 4.02 3.50 3.34 2.92 3.29 3.33 3.51 3.40 1.09 1.30 .67 .68 .79 .65 .58 .74 .64 1.01 .45 1.13 1.95 1.95 1.14 1.34 1.15 1.00 .99 .98 .99 .83 1.19 1.69 .34 .55 .41 .87 1.02 .21 1.28 1.00 .98 .96 .98 .69 63 APPENDIX C. Procedures fOr Discriminant Cluster Analysis. TABLE 18. Discriminant Analysis-~Step Wise Variable to Enter Sequence Wilke's Lambda Procedure. Step F to Number Variable Entered Enter Wilkes A Eéfi 1 Provided Information 13.45 .31 .000 2 Peer Friendly 4.65 .18 .000 3 Peer Attention 2.87 .12 .000 4 Peer Listen 2.03 .09 .000 5 Work Group Encouraged 1.844 .07 .000 6 Work Group Has High Standards 1.56 .05 .000 7 Work Group Innovative .99 .04 .000 8 Work Group Plans Ahead 1.15 .037 .000 9 Work Group Coordinates 1.12 .037 .000 10 Team Work 1.23 .025 .000 11 Trust 1.34 .020 .000 12 Tenure .80 .018 .000 13 Age .98 .015 .000 14 Education .79 .018 .000 15 Rate My Group .41 .012 .000 The entry criterion (Wilkes A) is that the entered variable produce the greatest increase in the overall F ratio at each successive step. 64 65 TABLE 19. Discriminant Analysis-~Group Size for Analysis. [z 12 {z Group 1 16 Group 7 4 Group 13 4 Group 2 9 Group 8 4 Group 14 2 Group 3 8 Group 9 5 Group 15 6 Group 4 8 Group 10 7 Group 16 2 Group 5 7 Group 11 4 Group 17 3 Group 6 21 Group 12 4 TABLE 20. Wilkes A for Entered Variables. Variable Wilkes A ‘5 Provided Information .81 .140 Peer Friendly .77 1.77 Peer Attention .67 2.97 Peers Listen .74 2.10 Work Group Encouraged .78 1.68 Work Group Has High Standards .75 I 2.02 Work Group Innovative .74 2.12 Work Group Plans Ahead .88 .85 Work Group Coordinates .83 1.27 Team Work .93 .45 Trust .86 3.71 Tenure .79 1.65 Age ’ .73 2.27 Education .31 13.24 For Wilkes A the discriminant criterion is the overall multivariate F ratio for the test of differences among the group centroids. B IB LI OGRAPHY BIBLIOGRAPHY Asch, S. E. 1952 Effects of group pressures upon the modification and distortion of judgements. In G. E. Swanson, T. M. Newcomb, and E. L. Hartley (Eds.), Readings in Social Psychology (2nd ed.). New York: Holt. Bales, R. F. 1950 Interaction process analysis. Cambridge, Massachusetts: Addison—Wesley. Barnard, C. I. 1940 The functions of the executive. Cambridge: Howard Press. Barnett, G. B., and R. Wigand 1975 A multidimensional examination of cultural processes: a cross- cultural approach. Paper presented to the International Commun- ication Association. Chicago. Berkowitz, L., and J. R. Macaulay 1961 Some effects of differences in status level and status stability. Human Relations, 14: 135-148. Blake, R. R., and J. S. Mouton 1968 Corporate excellence through grid organizational development. Houston: Gulf Publishing. Borgatta, E. F., and L. S. Cottrell 1955 On the classification of groups. Sociometry and the Science of Man, 18: 409—422. Brown, J. 1974 The role of communication in the development of a sex role orientation. Paper presented to the annual meeting of the Association for Education in Journalism. San Diego, California. Cattell, R. B. 1953 New concepts fer measuring leadership, in terms of group syn- tality. In D. Cartwright and A. Zander (Eds.), Group dynamics. Evanston, Illinois: Row Peterson. Collins, B. B., and G. H. Raven 1969 Group structure: attraction, coalitions, communication and power. In G. Lindzey and E. Aronson (Eds.), Handbook of social psychology. Reading, Massachusetts: Addison-Wesley. 66 67 Danowski, J. 1974a An uncertainty model: friendship communication networks and media related behaviors. Paper presented to the annual meeting of the International Communication Association. New Orleans. 1974b An information processing model of organizations: a focus on environmental uncertainty and communication network structuring. Paper presented to the annual meeting of the International Communication Association. New Orleans. Doyle, R.IJ. 1970 A new look at the Scanlon Plan. Management Accounting, 52, 3: 48-52. . Durkheim, E. 1933 On the divisions of labor in society. Translated by George Simpson. New York: Macmillan. Farace, R. V., and J. D. Johnson 1974 Comparative analysis of human communication networks in selected formal organizations. Paper presented to the annual meeting of the International Communication Association. New Orleans. Farace, R. V., P. Monge, W. D. Richards, and E. Jacobson 1973 Analysis of human communication networks in large social systems. Unpublished manuscript. Michigan State University. Festinger, L., S. Schacter, and K. Back 1950 Social pressures in informal groups: a study of human factors in housing. New York: Harper. Festinger, L. 1950 Informal social communication. Psychological Review, 57: 271-282. 1957 A theory of cognitive dissonance. Evanston, Illinois: Row— Peterson. Goldberg, G., and L. Larson 1974 Communication in groups. New York: Wiley. Guetzkow, H. 1965 Communication in organizations. In James G. March (Ed.), Handbook of organizations. Chicago: Rand-McNally. Heider, F. 1958 The psychology of interpersonal relations. New York: Wiley. Heinen, J. S. 1972 The development of work teams in a complex organization. Unpub- lished doctoral dissertation. Michigan State University. Miller 1928 The strike. Chicago: University of Chicago Press. 68 Hollander, B. P. 1958 Conformity, status, and idiosyncracy credit. Psychological Review, 65: 117-127. 196W Leaders, groups and influence. New York: Oxford University Press. Homans, G. C. 1950 The human group. New York: Harcourt, Brace. 1958 Social behavior as exchange. American Journal of Sociology, 63: 597-606. 1962 Sentiments and activities. New York: The Free Press. Huse, E. 1975 Organizational development. New York: Wiley. Jacobs, R. C., and D. T. Campbell 1961 The perpetration of an arbitrary tradition through several generations of a laboratory microculture. Journal of Abnormal Social Psychology, 1961: 6u9-658. Jacobson, B., and S. E. Seashore 1951 Communication practices in complex organizations. Journal of Social Issues, 7: 28-HO. Jones, E. E. 196% Ingratiation. New York: Appleton-Century-Crofts. Julian, J. W., and I. D. Stiener 1961 Perceived acceptance as a determinance of conformity behavior. Journal of Social Psychology, 55: 191-198. Kuch, D., and R. S. Crutchfield 19u8 Theory and problems of social psychology. New York: McGraw- Hill. Landecker, W. S. 1951 Types of integration and their measurement. American Journal of Sociology, 56: 332—3uo. Laski, H. J. 19u2 Encyclopedia of the Social Sciences, Volume 3: 70. Lawrence, R. R., and J. W. Lorsch 1967 Organization and environment: managing differentiation and in- tegration. Boston: Harvard Graduate School of Business Admin- istration. Lesieur, F. G. (Ed.) 1958 The Scanlon Plan. Cambridge, Massachusetts: M.I.T. Press. Lewis, K. 1952 In G. E. Swanson, T. M. Newcomb, and B. L. Hartley (Eds.), Readings in social psychology (2nd ed.). New York: Holt. 69' Likert, R. 1967 The human organization. New York: McGraw-Hill. McGregor 1953 Line manager's responsibility for human relations. In, Building up the supervisor's job. New York: American Management Asso— ciation, Manufacturing Series No. 213: 27-35. Merton, R. K. 1958 Social theory and social science. Glencoe, Illinois: The Free Press. Monge, P., and G. Lindzey 197u Network analysis. Paper presented to the annual meeting of the International Communication Association. New Orleans. Moreno, J. L. 193M Who shall survive? Washington: Nervous and Mortal Diseases Publishing Co. Newcomb, T. M. 1953 An approach to the study of communicative acts. Psychological Review, 60: 393-u0u. Osgood, C. B., G. J. Suci, and P. H. Tannenbaum 1957 The measurement of meaning. Urbana, Illinois: University of Illinois Press. Parsons, T. 1937 The structure of social action. Chicago: University of Chicago Press. Rappoport, A. 1963 Operational philosophy. New York: Addison-Wesley. Richard, W. D. l97ua Network analysis in large complex systems. Paper presented to the annual meeting of the International Communication Associa- . tion. New Orleans. 197ub Network analysis in large complex systems: technique and methods--tools. Paper presented to the annual meeting of the International Communication Association. New Orleans. 197uc Network analysis in large complex systems: metrics. Paper pre- sented to the annual meeting of the International Communication Association. New Orleans. J Rose, A. M. 1962 In Rose, A. M. (Ed.), Human behavior and social processes: an interactionist approach. Boston: Houghton, Mifflin. Roethlisberger, F. J., and W. J. Dickson 1939 Management and the worker. Cambridge, Massachusetts: Harvard University Press. '70 Schien, E. H. 1967 Organizational psychology. Bnglewood Cliffs, New Jersey: Prentice-Hall. Selznick, P. 19MB An approach to the theory of bureaucracy. American Sociological Review, 8, l: u2-5u. Shaw, C. R. 1931 The natural history of a delinquent career. Chicago: University of Chicago Press. Sherif, M. 1964 In common predicament: social psychology of inter-group conflict and cooperation. Boston: Houghton, Mifflin. 1967 Social interaction. Chicago: Aldine. Simmel, G. 1950 The sociology of Georg Simmel. K. H. Wolff (Trans.). Glencoe, Illinois: The Free Press. Smith, M. 19MB Social situation, social behavior and social group. Psychologic- al Review, 52: 22u—229. Taylor, J. A., G. A. Barnett, and K. Serota 1975 A multidimensional examination of political attitude change. Paper presented to the annual meeting of the International Communication Association. New Orleans. Thibaut, J. W., and H. H. Kelly 1959 The social psychology of groups. New York: Wiley. White 1964 Organizational Behavior. New York: Dorsey. Whyte, W. F. 1983 Street corner society. Chicago: University of Chicago Press Wiggins, J. A., F. Dill, and R. D. Schwartz 1965 On "status liability." Sociometry, 28: 197-209. Woelfel, J., and A. D. Haller 1971 Significant others, the self—reflexive act, and the attitude formation process. American Sociological Review, 36: 7u-87. Woelfel, J., and J. Saltiel 1975 Cognitive processes as motion in a multidimensional space: a general linear model. Paper presented to the annual meeting of the International Communication Association. Chicago. HICHIGQN STQTE UNIV. LIBRARIES 31293101577553 '