LIAISON COMMUNICATION ROLES IN A FORMAL ORGANIZATION Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY DONALD F. SCHWARTZ 1968 ,. ~.._...~ «fig... . , sum “.94 . IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIZIIIIIZIIIIII 3 1293 103818 This is to certify that the thesis entitled Liaison Communication Roles A0: in a Formal Organization presented by ‘ Donald Franklin Schwartz has been accepted towards fulfillment .1 I fat the requirements for M—degrée inmnication 0-169 L181? 4 R Y Michigan Scam University ABSTRACT LIAISON COMMUNICATION ROLES IN A FORMAL ORGANIZATION by Donald P. Schwartz Most previous communication research in formal organ- izations has examined aspects of the formal structure with particular emphasis on superior-subordinate relations. b///’ The focus of this study was to map the extant functional communication structure of an organization (including both formal and informal aspects), then to differentiate this sociometric map into two structural types based on topolo- gical concepts from graph theory, and, finally, to describe differences on specified variables between the two struc- tural types . The primary structural type was the liaison communi- ‘Qgtigg‘;glg. Individuals who function in a liaison role have interlinking communication contacts with two or more sociometrically-defined clique groups in the organization. Essentially, when liaison role persons are removed.from the sociogram.of communication contacts, groups to which they are connected separate. Thus the liaison role, a concep- tual analogue to the articulation point in graph theory, is a critical location in a communication network. All other individuals who were not liaison persons or isolates were termed non-liaison persons. Eleven hypotheses pre- dicted differences between liaison and non-liaison persons in terms of certain communication behavior and personal Donald P. Schwartz attributes perceived by their respective reciprocated non- liaison contacts. The first step in topological analysis was to map the organization's communication structure based on sociometric reports from all members of their usual weekly or more of- ten functional communication contacts. A communication linkage was assumed between two persons 9512 if each re- ported the other as a contact. From a study population of ~142 members composed of all the faculty and professional Astaff of a College within a large university, 225 recipro— cated dyads emerged. This matrix of contacts was analyzed to yield identification of 22 liaison role persons, 18 iso- lates and 102 non-liaison persons who had sociometric mem- bership in 29 separate groups of varying size. Respondents completed questionnaires regarding their communication behavior with, and perceptions of, persons with whom they reported daily contact. Thirty question- naires completed on 17 liaison persons by their reciprocated non-liaison contacts were the sources of data for the liai- son sample. A random sample of 17 non-liaison persons yielded 21 questionnaires completed on them by their non- liaison contacts and these were the sources of data for the non-liaison sample. The individuals evaluated in each samp’; ple differed in the number of monthly committee meetings and span of reciprocated contacts, with liaisons having the higher average in both instances. The samples were compar- atively similar on several other demographic variables Donald F. Schwartz including age, sex, academic rank, time allotted to various duties, publication rate and tenure at the university. Seven of the hypothesis-testing variables were oper- ationalized as multi-item scales, with the remaining two direct estimates of message transaction initiation fre- quencies. Hypotheses were evaluated by the “t“ test for independent sample means. Contrasting liaison persons and non-liaison persons as perceived by their reciprocated noneliaison contacts, support was found for perceptual characterization of lia- isons as having a larger number, more structurally diverse and more important contacts in the organization. Liaisons were perceived to be more influential within the organiza- tional ”power structure.“ In the organization's informa- tion relay network, liaisons were reported to more.frequent- tion for their contacts. Support was not obtained for hypotheses concerning the frequency and directional ratio of deliberate message trans- action initiation, qualification and safety dimensions of source credibility, general persuasiveness with secondary contacts and specific (dyadic) opinion leadership. The study demonstrates a unique conceptual and metho- dological framework (communimetrics) for study of the exp tant communication structure of formalized social systems and provides some empirical evidence on perceived attributes and functions of liaison communication role incumbents. LIAISON COMMUNICATION ROLES IN A FORMAL ORGANIZATION Byve V. (6’0 Donald P. Schwartz A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1968 COpyright by Donald Franklin Schwartz 1969 Ill I I Ill ‘ I l. l [I I l [I III I II II Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the requirements for the Doctor of Philosophy degree. Direcirr o Thesis Guidance Committee: Adda. .~¢.Aur,- , Chairman ' .I' ,' u " 1.11”' 4/ ll, / 14/.1 The list of individuals who contributed directly and indirectly to the present work is of such length to defy enumeration. However, the author wishes to express his sincere gratitude to the following individuals and groups: (1) the respondents to the pretest and main study question- naires for their cooperation; (2) the members of my commit- tee for their patience and counsel: Dr. Hideya Kumata, chairman and director, Dr. Eugene Jacobson, who provided the original impetus and continuing encouragement for this work, Dr. Gerald Miller and Dr. Everett M. Rogers; (3) Dr. Donald J. Leu for his interest and special assistance; (4) the faculty of the Department of Communication, Michigan State University, for a challenging and meaningful learning experience, and especially Dr. David K. Berlo for leading a year of provocative exploration of organizational communi- cation: (5) that unique set of individuals with‘whom I shared graduate student status in the Department of Communi- cation, especially Dr. Jay Weston and Dr. Nan Lin, for stim- ulating intellectual interaction: (6) the staff of North Dakota State University Communications Office for shoulder- ing an extra work load during my absence: and (7) Dr. Robert L. Crom, Dr. H. R..Albrecht and Dr. Seth Russell for arrang- ing sabbatic and other leaves and giving added invaluable incentives toward pursuit of graduate study. The emotional and clerical talents of my famdly pro- vided the major support for accomplishment of the present study and graduate program. My*wife, Lois, played addi- tional roles as typist, data tabulator, coder, keypunch operator, illustrator and grammarian. My two-year old daughter, Daria, unknowingly provided comic relief at appro- priate, and occasionally inappropriate, times. This work is dedicated to the memory of my parents, Frank W. and Mabel E. Schwartz, both of whom died during the author's tenure as a graduate student at.Michigan State University. My father, who was forced to terminate his formal education at the fourth grade, and my mother, who held a bachelors degree in mathematics, provided the en- couragement and considerable economic sacrifice for an academic career which, without their support, the author would have then enthusiastically abandoned at the eighth grade. DPS 11 'llll N I.‘ N'N‘tllllllll ‘I I Ill 1 l I INN" IN" IN! l I ll lllIII I I TABLE OF CONTENTS Page ACKNOWLEDGEMENTS . . . . . . . . . . . . 11 LIST OF TABLES . . . . . . . . . . . . . 1v LIST OF FIGURES . . . . . . . . . . . . . vi LIST OF APPENDICES . . . . . . . . . . . . vii Chapter I. INTRODUCTION . . . . . . . . . . . . 1 Critique of Previous Research I . . . . . 3 The Research Contexts Liaison Communication Roles . . . . . . . . 16 Hypotheses and Rationale . . . . . . . 31 II. RESEARCH DESIGN . . . . . . . . . . . 42 The sample 0 O O O O O O O O O O O 43 Operationalization of Variable . . . . . 45 Data callection O O O O O O O O O 0 60 Data Analysis . . . . . . . . . . . 61 Pretest Procedures . . . . . . . . . 63 III. PRETEST RESULTS . . . . . . . . . . . 67 Pretest Analysis . . . . . . . . . . 68 IV. FINDINGS O O O O O O O O O O O O O 81 Data Collection . . . . . . . . . . 81 Scale Analysis . . . . . . . . . . 85 Sociometric Analysis . . . . . . . . 91 Characteristics of the Respondents . . . . 102 Tests of Hypotheses . . . . . . . . . 121 v 0 CONCLUSION O O O O O O O O O O O O 131 Summary . . . . . . . . . . . . . 131 Discussion . . . . . . . . . . . . 142 Contributions of the Study . . . . . . 159 Suggestions for Future Research . . . . . 163 BIBLIWRAPHY O O O O O O O O O O O O O O 168 APPENDICES . . . . . . . . . . . . . . 173 111 LIST OF TABLES Table Page 1. Frequency of contact and number of contacts reported by twenty-five pretest respondents . . . . . . . . . . . . 71 2. Inter-item correlation matrix and average correlations for the perceived struc- tural diversity of contacts pretest scale 0 O C O O O O O O O O O O O 73 3. Inter-item correlation matrix for the perceived number of contacts pretest scale 0 e e e e e' e e e e e 74 4. Inter-item correlation matrix and average correlations for the first source of information pretest scale . . . . . . . 74 5. Inter-item correlation matrix and average correlations for the importance of secondary contacts pretest scale . . . . . 75 6. Inter-item correlation matrix and average correlations (within type) for the diffuse opinion leadership pretest scale I O O O O O O O O O O O 0 O 77 7. Inter-item correlation matrix and average correlations (within Type 1 items) for the specific opinion leadership pretest scale . . . . . . . . . . . 79 8. Daily cumulative total of questionnaire packets delivered and returned . . . . . 83 9. Inter-item correlation matrix for the diffuse opinion leadership main study scale . . . . . . . . . . . . 87 lo. Inter-item correlation matrix for the diffuse opinion leadership scale based on "several times daily” frequency of contact questionnaires . . . . 89 ll. Inter-item correlation matrix for the specific opinion leadership main BtUdY Beale O O O O O I O O O O O O 89 iv Table Page 12. Inter-item.correlation matrix for the source credibility scales . . . . . . . 9O 13. Distributions of reported contacts by frequency of contact category . . . . . . 92 14. Summary of the number of reported communication contacts by liaisons, non-liaisons and isolates . . . . . . . 105 15. Characteristics of members of the study population by type . . . . . . . . . 109 16. Comparison of selected characteristics of the liaison and non-liaison samples . . . . 118 17. Summary of results from tests of hmtheses e e e e e e e e e e e e 143 LIST OF FIGURES Figures Page 1. Graph theory representation of articulation point and bridge . . . . . . . . . . 24 2. Reproduction of the matrix of reported communication contacts . . . . . . . . 97 3. Sociogram.of the communication structure of the organization . . . . . . . . . 103 4. Frequency polygon for the choices received by liaisons, non-liaisons and isolates . . . 107 5. Frequency histogram of the reciprocated communication contacts of liaison and non-liaison individuals . . . . . . . . 108 V1 LIST OF.APPENDICES Appendix Page A. Entry Letter . . . . . . . . . . . . 174 B. Questionnaire Packet a. Cover Letter . . . . . . . . . . . 176 b. Part I - Personal Data Questionnaire . . . 177 c. Part II - Personal Contact Checklist . . . 180 D. Part III - Personal Contact Questionnaire . .' . . . . . 181 C. Variable Index for Part III - Personal Contact Questionnaire . . . . . . . . 189 vii CHAPTER I INTRODUCTION Examination of the literature on formal organizations reveals little explicit discussion or empirical research on the structure and process of communication. There is, how- ever, general consensus that communication is an important process in organizations. Barnard (1938, p. 9) was one of the early theorists to posit communication as a central concern in organization theory: ...in any exhaustive theory of organization, commu- nication would occupy a central place, because the structure, extensiveness, and scope of organization are almost entirely determined by communication techniques. Many contemporary social scientists place similar emphasis upon organizational communication, but from differing view- points. Leavitt (1958, p. 300), for example, sees decision- making as a central theoretical concern but underlines the importance of communication: Prom management's perspective, we can think of ...(an organization) as an elaborate set of interconnected communication channels designed to collect and col- late, analyze, and sort out information: also as a system for making deciSions, acting them out, getting feedback information and correcting itself. Viewing organizations from the standpoint of management theory, Dorsey (1957) explicitly identifies administration as a communication process. More recently, Katz and Kahn (1966) have defined or- ganizations from a systems viewpoint, characterizing an organization as both an energic and an informational system 1 2 with the function of the information system being manage- ment of the energic system. With this conceptual back- ground, their emphasis on the importance of communication is expected: The closer one gets to the organizational center of control and decision-making, the more pronounced is the emphasis on information exchange. In this sense, communication -- the exchange of information and the transmission of meaning -- is the very essence of a social system or an organization. The input of physical energy is dependent upon infor- mation about it, and the input of human energy is made possible through communicative acts. ...Communication is thus a social process of the broadest relevance in the functioning of any group, organization or society. It is possible to subsume under it such forms of social interaction as the exertion of influence, cOOperation, social contagion or imitation, and leadership. (p. 223, 224) As a final example, Thayer (1967, p. 85) clearly ele- vates communication above other organization processes when he states, "...the essential and preeminent organismic, interpersonal, and organizational function is communication -- i.e., the programmatic inputting, processing, and out- putting of 'information.'“ Given these and other commonalities on the theoretic assumption that communication is, at least, a central pro- cess in an organizational context, it is surprising that little empirical research has been accomplished within the specific context of formal organizations. Walton's (1962, p. 4) description of the state of knowledge about one aspect of organizational communication is still apt: Although the organization's communication system may be the most important factor accounting for its beha- vior, we have insufficient knowledge of the dynamics of the system.writ large that clearly explains and 3 accounts for the movement of messages within the organizational environment. Thayer (1967, pp. 80-1) states the case even more strongly: Perhaps more has been 'communicated' about 'communi- cations problems' in organizations than any other single topic in the field. Yet this plethora of commentary has not been conducive either to theory- building or to theory-validation. This bulk, most of which represents individual curiosities rather than scientific endeavor, defies cataloging on two points: first, the studies themselves have not been carried out in the context of any superordinate theory and hence offer little possibility for conceptual systemization: second, most have been based upon nor- mative and not empirical assumptions. Consequently, their conceptual usefulness is severely diminished. ...no single integrated body of literature has been accumulated. Since there is general consensus on the importance of communication as an organizational process but a poverty of empirical literature on organizational communication, the development of a conceptual and empirical approach necessar- ily rests on an analysis of why so little evidence appears to have been accumulated to date. Critique 9; Previous Researgh Perusal of the few investigations directed at the study of communication within the context of "real" organi- zations suggests several reasons why the field has not ad- vanced substantially beyond a problem identification stage: r—A‘(l) the bulk of the research has maintained a “machine theory“ orientation which encourages a clinical, therapeutic />-result: (2) little attention has been paid to the develop- ment of methodology for analyzing the extant communication structure of an organization sans the formal, imposed 4 hierarchial structure: (3) too often the research which has been attempted was not conceptualized within the framework of a superordinate theory of either communication or organ- ization. These major reasons are not mutually exclusive. Theoretig Limitations The concept of organization is predicated on the g priori prescription of a structured set of interrelation- ships between members and task-groups. Katz and Kahn (1966, pp. 36-7) describe organizations as contrived social systems and observe that "the essential difference between social organizations and less structured social systems is the greater reliance upon formal prescriptions of acceptable as against unacceptable behavior in the organization.” In other words an organization is a “social machine" designed to organize relationships between “human parts“ in order to efficiently accomplish collective goals. Within this con- text it is not surprising that the usual approach to organ- izational communication research has been an attempt to ascertain how communication operates within the formal “design" of the organization. Communication has typically been studied in terms of “vertical“ or “horizontal" rela- tionships between or among hierarchial levels or formally structured task units. Often the underlying goal of this approach has been to devise improvements for practicing organizational members: i.e., a therapeutic approach. In the same vein Etzioni (1961, p. 137) observes that most organizational communication research has been carried 5 out primarily in a human relations tradition. Emphasis is usually upon superior-subordinate relations. Again, the focus of research is upon the formal structure of the organ- ization with little attention paid to the impact of the informal communication structure1 on formally prescribed relationships. Thayer (1967, p. 97) suggests: The traditional basic problems of 'downward' and 'upward' communication have been variously solved by traditional methods. The solutions are equivocal, however, suggesting that our traditional approaches and methods have not brought us far theoretically -- or practically....Perhaps we are getting the right answers to the wrong research questions. A.basic conceptual limitation prevents research con- ducted in the machine theory tradition from contributing substantially to inclusive theory-building: by encompassing only a few levels or units of the organization within the research problem, this approach is never able to conceptual- ize or adequately account for the total interactive dynamic of the organization's communication system. The result is Ionly a partial analysis which has restricted theoretic utility. The conceptual limitation is based on “a much too limited, sometimes unworkable, sometimes confused, concept of what communication is in organizations“ (Thayer, 1967, p. 83). When the conceptual formulation underlying research is restricted to a view of communication as a form.of con- trol or a tool with motivational implications between I 1With a machine theory orientation, the usual definition of I the formal structure has been in terms of the imposed task- specialization, authority hierarchy. The informal struc- ture has been negatively defined as a residual category. 6 superior and subordinate the fundamental role of communica- tion as a functional information transmdssion process is obscd§ked in terms of overall organizational dynamics. Significant examples of organizational communication research conducted within the machine theory tradition in- clude Burns' (1954) study of communication patterns of de- partmental executives, Berkowitz and Bennis' (1961) and Simpson's (1959) descriptions of communication patterns be- tween hierarchéhl levels, Wager's (1962) description of message diffusion in an organization, and Read's (1962) study of the relation between the accuracy of upward com! munication and a subordinate's upward mobility aspirations and trust of his superior. Other studies have attempted to describe the frequency, direction or effectiveness of formal channel usage; e.g., Goetzinger and Valentine (1962), Hin- richs (1964), Walton (1959) and Dahle (1954). Mischler and Tropp (1956) explored the relation between occupational status and frequency of formal and informal channel usage. Triandis (1959a,!fl studied the relation between cognitive similarity and effectiveness of communication within superior-subordinate dyads. ‘A conceptual framework related to the machine theory tradition is the approach underlying the studies of communi- cation networks in small groups: e.g., Bavelas (1950) and Leavitt (1951). The relevance of this research to the more complex level of formal organizations is still not clear (Thayer, 1967, p. 75). Although he justifies generalizing 7 small group findings to the organizational level, Guetzkow (1965, p. 535) admits a speculative basis: The richness of materials at the individual and group levels has induced extrapolation of findings perhaps inappropriate for rigorous analysis of communications in organizations. Yet, with the dearth of studies about organizations, either from the field or labora- tory, one can but Join with others in speculation. These observations on the generalizability of small group research to the organization level underline an inability, at this point in time, to translate such research into the framework of a superordinate theory of organization. Methodologigal Limitations Concomitant with the limitations on communication-in- organization research imposed by the machine theory orien- tation is the lack of attention to methodology compatible with a more inclusive conceptual framework. The problem is a chicken-egg paradox. The lack of readily available methods may partially explain the lack of attention to a broader framework. On the other hand, it may be the lack of a broader framework in the first place which has hindered development of ”new" methodology. In either event, there exist limitations in both data-collection and data-analysis methods. Organizational communication research is by definition focused on the structure and/or process of message trans- actions. The heterogeneity of content or function and the temporal nature of these transactions pose unique problems in data collection when the researcher goes beyond the transactional structure imposed by formal prescription. 8 If the research problem involves content-functional delinea- tions, there is necessity for a viable taxonomy to cope with the heterogeneity of transactions. No consensual taxonomy for this purpose exists.1 This problem becomes manifest in attempts to operationalize, for example, sociometric ques- tions. The temporal nature of message transactions also creates data-collection problems. Such non-continuity poses the question of whether message transactions should be sampled or censused and over what time period. Message sampling, as well as selective observations of message transactions based on content, may result in only partial analysis. Likewise, a message census may raise serious questions of reliability if organizational members are asked to record all their message transactions during a given time period because involvement in record-keeping may bias communica- tion processes. Pour general data-collection methods appear in the literature: (1) sociometric techniques (e.g., Jacobson and 1Taxonomies have been attempted. Katz and Kahn (1966) pro- pose five organizational subsystems: production, suppor- tive, maintenance, adaptive, and managerial. It might be assumed that if such subsystems can be identified there must be an extractable communication network associated by content of message transaction with each, although with overlapping membership. Thayer (1967, pp. 94-6) proposes three basic information systems: operational, regulatory and maintenance-development. Guetzkow (1965, pp. 542-50) classifies research on communication networks into five categories based on content of transactions in the net: authority, information, task-expertise, friendship and status. The deductive efficacy of these taxonomies has not been demonstrated. 9 Seashore, 1951), (2) tracing of a given message after it diffuses through the organization (e.g., Davis, 1953a,ln, (3) the communication log or audit wherein a census of messages is recorded by members over a given time period (e.g., Burns, 1954) and (4) timed random sampling of mes- sages (e.g., Hinrichs, 1964). A fifth method, use of trained observers to record all or a sampling of message transactions (e.g., Bales, 1951) has not been attempted in field research. Problems in Operationalization due to content heterogeneity and noncontinuity of transaction are inherent in each data-collection method. Another methodological problem related to both data collection and data analysis is that imposed by the inde- pendent random sampling assumption necessary to most statis- tical models. In the machine theory tradition, individuals or dyads easily can be randomly selected from one or sever- al organizations. But this approach does not permit rela- tional analysis of a total social system which in essence requires a census of the members of the system (saturation sampling). Coleman (1964, p. 442) observes: The individual remained the unit of analysis. No matter how complex the analysis, how numerous the correlations, the studies focused on individuals as separate and independent units. The very techniques mirror this well: samples were random, never inclu- ding (except by accident) two persons who were friends: interviews were with one individual, as an atomistic entity, and responses were coded on separate IBM cards, one for each person. .As a result...(the problems were) never problems concerned with relations between people. One of the more obvious problems of saturation sampling as a conceptual-methodological approach is that of generaliza- 10 bility of the findings to other similar, or dissimilar, organizations. The problem is not unique to organization research and is described by Katz (1960, p. 439) as "how to take account of interpersonal relations and still preserve the representativeness of a sample." Implications £2; Future Reseaggh Analysis of previous communication-in-organization research reveals several implications for future research, both in terms of conceptual and methodological orientation. What appears called for is a shift from the machine theory approach to a more inclusive theory of organizational com- munication as the structure and process of a network of . interpersonal relations within which the formally prescribed network is but one aspect. This shift also requires a reorientation in methodology. Coleman (1964, p. 442 ) points out that such a shift is ”quite a difficult one to make, both conceptually and technically.“ A more inclusive theory of organizational communication) would accept the communication network(s) of an organizationI as basic structural characteristics without initial regard I for other structural characteristics such as authority, status, prescribed function, or spatial relation. Succeed- ing steps in this approach would superimpose these other structural characteristics upon the communication network(s)f in an attempt to explain the variance found in their struc- I tural or operational functions. Within this context, before superimposing other organizational characteristics upon 11 communication networks, directionality of message trans- actions has no meaning in the usual machine theory sense: i.e., there can be no concepts of vertical or horizontal or "criss-cross“ communication, but only communication be- tween or among members and/or groups. Also within this context, the extraction of communication networks might be related to specific types of content or function yielding analyses based on specific types of networks, or the sea- grig ggmmunication structure could be extracted on a con- 1 tent-free basis. The latter approach would attempt to extract the most general extant communication network. For either approach, minimum criteria2 for frequency of channel3 usage within any network would have to be esta- blished to define exact limits for the existence of a channel (reflecting the temporal nature of message trans- actions between dyadic linkages) or dyadic linkages within a network could be assigned weights to reflect frequency of channel usage. Unfortunately, there is no existing empiric foundation upon which to base minimum criteria of 1Utilizing, for example, a sociometric question like, “Who do you communicate with in this organization?" as opposed to a question worded “Who do you communicate with about (subject matter)?" 2For example, up to daily contact, up to once weekly con- tact, etc. 3The terms channel and dyadic linkage are used interchange- ably. Either term means simply the occasion of a previous message relation of some frequency between members of the dyad. 12 channel usage by relating frequency of usage to maximum ex- planation of variance in operating characteristics and com- munication effects. The generic communication structure1 would represent a “pure type“ within the framework suggested here. The pure type would provide a record of all message transactions between all members of the organization, regardless of con- tent-type or historic frequency, in order to describe all the dyadic linkages making up the fabric of the organiza- tion's total communication structure. This over-all struc- ture then might be redefined in terms of several content- frequency substructures with each being a separable network of its own, but most likely with overlapping membership across substructures. However, a complete “pure type" such as this may remain an ideal type as a result of practical limitations on data-collection procedures in real organiza- tions. Nonetheless, procedures yielding a best-approxima- tion of extant communication structures represent an im- provement over previous research strictured by the machine theory conceptualization. To correct for previous inadequacies, Thayer (1967, p. 97) opts for an operational-functional study of organi- zational communication and offers this suggestion: ...if the individual is viewed as a complex informa- tion-processing system, research on human behavior 1The terms communication structure and communication net- work are used interchangeably. 13 in organizations could be based upon a view of the individual as the focal point of a set of information vectors which define that individual's functional role in the organization. Combining Thayer's suggestion with the generic communication structure concept points to a two-phase or togglogical- analygic procedure. The first step would involve mapping, as an abstracted isomorphic system, the extant communication structure (general or specific depending on the research question) of the organization. This topological map would permit locating any given member of the organization as the focal point of a unique set of information vectors: i.e., his set of previous message transaction linkages with other members of the organization. The second step in this pro- cedure would be to determine the “meaning" of an individual's location within his unique set of vectors. For maximum utility, the topological-analytic procedure would neces- sarily rely on the development of descriptive structural concepts such as Jacobson and Seashore's (1951) "liaison role", Walton's (1962) “magnetic centers", or Davis' (1953a, b) "single strand“, “gossip chain", “probability chain", and "cluster chain.“ From a methodological viewpoint, the topological- analytic approach has special data-collection and data- analysis implications. Of the four data-collection methods used to date, sociometric and communication audit methods appear the most amenable procedures for mapping the extant communication structure. However, the requirement for having all members of the organization record all of their 14 daily message transactions may render the communication audit impractical in field research. Use of timed random sampling of message transactions may prove practicable after it can be established that a given time period for 1 of the data collection yields a reliable approximation extant communication structure. In a similar vein, Davis' (1953a, b) ECCO technique (tracing a given message after it diffuses through the organization) would require the trac- ing of some minimum number of messages for each content 2 to yield a reliable approximation. At this point, category content-free, non-directional3 sociometric methods appear the most parsimonious and practical approach. The methodological problem imposed by the independent random sampling assumption may be partially met in topolog- ical-analytic research because the many dyadic linkages 1For example, given the heterogeneity of message content and the temporal nature of dyadic linkages, collection of a random sample of message transactions over a five-day period may be inadequate to description of the many dyadic linkages making up the extant communication structure. 2Davis' technique might be most useful in combination with another topological method where, first, the extant commun- ication structure is mapped and then ECCO analysis is used to study, for example, sharpening, leveling and assimila- tion effects on given messages or to study uncertainty absorption (March and Simon, 1958, p. 165). 3A directional sociometric question on the order of "Who do you seek for information" yields a partial mapping of the communication structure: in this case, only the informa- tion seeking network. Informational message transactions can take place when neither member deliberately seeks the other, or when one member seeks another to give and not receive information. 15 within the communication structure or structures of large formal organizations may lend themselves to random sam- pling, depending upon the nature of the structural concepts utilized in a particular study. But the fact that most organizational communication research is usually done within one organization still poses a question of generalizability to other organizations. What appears necessary are compar- ative studies between several organizations or organization- types (Cf. Etzioni, 1961). I A major technical problem related to the first step in topological-analytic research is the need for descriptive structural concepts which have mathematic properties or which can be translated, for purposes of analysis, into more abstract mathematical models. As Weiss and Jacobson (1955, p. 661) observe: The use of sociometry to determine the overall struc- ture of a complex organization probably owes its rarity to an absence of both structural concepts and of efficient methods for the manipulation of large masses of sociometric data. Although contemporary concepts and procedures for analyses of communication structures are, in Coleman's words (1964, p. 442), “only halting steps toward a full-fledged metho- dology,“ some of the work in matrix algebra (e.g., Festing- er, 1949) and graph theory (e.g., Harary and Norman, 1953) offer promise. For example, the liaison role concept is an analogue to the articulation point in graph theory (Weiss, 1956, p. 88: Ross and Harary, 1955, pp. 251-8). Unfortunately, little of the work accomplished in these 16 two related1 fields has been programmed for computers in order to deal with the large sociometric matrices which typify organization research. But even though such a com- puter program would greatly increase methodological effi- ciency, its unavailability need not totally impede topo- logical-analytic research. The focus of the present study, in fact, is directed at the examination of the liaison com~ munication role concept. The Research Context: Liaison Communication 3215; Of the organizational communication research which has been conductedponly the study of liaison communication roles (Jacobson and Seashore, 1951: Weiss and Jacobson, 1955; Weiss, 1956) bears conceptual relatedness to what has been described here as the generic communication struc- ture. Not only is that study unique in its conceptual foundation, but it also presents a method for analyzing sociometric data to yield a differentiated topology of a communication structure based on analogues in graph theory. These analogues are generalized structural concepts which permit second-phase description and analysis of variables related to topological features. 1;; Thgggetig ggntexg The present research is based upon the concept of the liaison communication role developed by Jacobson, Seashore 1Connected graphs may be expressed in matrix form as sets of binary relations (1 or 0) representing communication linkages between components. 17 and Weiss. Undergirding the liaison role concept is a broader theoretical base which accepts the extant communi- cation structure of the organization as the most general structural concepts 'Organization structure' can be conceptualized in terms of communication events which connect pairs of individuals, and thus establish patterns of contact among individuals and among groups (Jacobson and Sea- shore, 1951, p. 33). Implicit in this definition of structure and in the pre- ceding discussion in this paper is a distinction, and yet an inextricable relation, between the concepts of structure and process. Before examining the liaison role concept in more detail, clarification of the more basic relation be- tween structural and processual concepts is necessary be- cause that relation has both theoretic and methodological implications. Structure and Process J. G. Miller (1965a, pp. 209-11) defines the structure of a system as “the arrangement of its subsystems and components in three-dimensional space at a given moment of time.“ Process is defined as ”all change over time of matter-energy or information in a sys- tem." ”Structure,” he maintains, "is the static arrange- ment of a system's parts at a moment in three-dimensional space....Process includes the on-going function of a system, reversible actions succeeding each other from moment to moment.” He further defines agtion as one form of process: “any change of state of matter-energy or its movement over space, from one point to another.” 18 Miller's definition of structure appears to be a limiting conceptualization of a behavioral system. Katz and Kahn (1966, p. 31) suggest a reason why: There has been no more pervasive, persistent, and futile fallacy handicapping the social sciences than the use of the physical model for the understanding of social structures....This figurative type of thinking ignores the essential difference between the socially contrived nature of social systems and the physical structure of the machine or the human organism. To counteract application of the ”biological metaphor” to social systems, Katz and Kahn observe: Social systems...1ack the fixed physical structure of biological and other physical systems. Social systems have structure, but it is a structure of events rather than physical parts, a structure there- fore inseparable from the functioning of the system (p. 69). Miller's definition of structure as an arrangement of come ponents in three dimensional space ggygygiggg moment Qg gigs ignores the temporal nature of message transactions in the functioning of a social system. Miller further sug- gests, “A high-speed photograph...would reveal the three- dimensional spatial arrangement of the system's components as of that instant" (p. 209). However, a ”high-speed photo- graph" of an organization would not reveal the interconnec- ted linkages between members (i.e., the communication structure),gg'§hgy endgre 92;; gigs, but would only capture a small sample of the message transactions in process at the moment of photograph. In fact a “time exposure” would be necessary to describe the communication structure of a social system because message transactions between individual 19 members of the system do not exist in continuous physical relation, but are temporal ”parts” or processes of the sys- tem. Thus a “high-speed photograph“ may be very misleading if one attempts to deduce the communication structure of the system from that single moment in time, or, for that matter, from the tracing of a single message over time, or from a random series of “photographs“ at various moments in time: i.e., a timed random sampling of message transactions. In terms of organizational communication research, Turner's distinction between structure and process may be most useful: The concept 'structure' predicates that a phenomenon consists of identifiable parts organized in functional relation....The concept 'process' predicates that a structure under the play of external forces and through its own energy undergoes action or acts so that change affects it....Structure, however enduring, exists in terms of process, and process, no matter how slowly or rapidly it operates, always moves through structure. (Quoted by Miller, 1965a, p. 210-11) Turner's discussion implies that it is most useful to conceive of structure as the history of process; i.e., the communication structure is mapped as the message transac- tion linkages between system components (members) within a given duration of time. These linkages can be described as ”channels“ defined, in Miller's terms, as “a route in a spatial region by which markers bearing information can be sent1 from a transmitter to a receiver" (1965b, p. 349). 1This definition might read "has been sent” with “can be sent" reserved to describe pgtential channel in order to be consistent with the suggested definition of structure as the history of process. 20 In further refutation of Miller's definition, communi- cation structure initially can be conceived as existing in swordimensional space, rather than three-dimensional space. Thinking of the members of an organization as points on a two-dimensional plane, the communication structure can be represented by drawing lines between certain of those points based on information about the history of message transac- tions between them during a given time period. In a pro- cessual sense, one can visualize lines representing messages in transit moving out from one point to another, leaving a “trace“ on the base plane to represent the history of the transaction: or, one of the points may physically move across this conceptual two-dimensional plane to connect with another point in a face-to-face message transaction, return- ing again to its original position and leaving a “trace" to record the process. The more lines existing between all points, the greater the saturation of communication contacts within the organization. This basic conceptualization of the communication structure of an organization moves to a three-dimensional representation only when another characteristic, measured as the attributes of members, is introduced and as members are characterized as having more or less of the attribute(s) under consideration: e.g., authority, power, sociability, expertise. Therefore, for purposes of simple description, the communication structure based on message transactions as the unit of analysis is a two-dimensional, static 21 construct. For purposes of analysis, the superimposition of one or more other structural characteristics of the or- ganization based on membe; attributes as the unit of analy- sis moves the abstracted system.from static structure in two-dimensional space to static structure in three-dimen- sional space. 2 It is useful, then, to maintain a conceptual distinc- tion between the communication structure as a descriptive two-dimensional concept and analygig structures as three- dimensional conepts and to conceive of the mapping of a behavioral structure as the history of process. One caution is in order, however, for utilization of this conceptual framework. Miller notes that "it is notor- iously hard to deduce process from structure, and the re- verse is by no means easy" (1965a, p. 219). In utilizing the concept of structure there may be a tendency to reify the structural map. If the concept of structure is accep- ted as the history of process it must be remembered that the concept of process involves action and resultant change. Regardless of the historical regularity of the transac- tional process which forms the basis for a structural comp munication map, that structure is not a concrete system. As a result, prediction of future communication processes in a given system.may not be totally accurate because the structure may have changed from time of abstraction to time of prediction. The Liaison Role Concept The conceptual framework of 22 the Jacobson, Seashore and Weiss study contains the essen- tial assumptions relating structure and process outlined above. Thayer (1967, p. 81) asserts: The Jacobson and Seashore study is significant by virtue of the authors' implicit emphasis on the in- terdependence of the structure and functions of come munication events, and the manner in which they con- ceive of organization structures as dynamic patterns of communication events. Equally significant in this study is the presentation of structural concepts which can be applied to a sociome- tric matrix in order to classify topological data of the matrix for analyses of functional processes. The formal constitutive definition of these concepts is presented by Weiss and Jacobson (1955): a. Work group was defined as a set of individuals whose relationships were with each other and not with members of other work groups (except for contacts with liaison persons or between groups). b. Liaison pprson was defined as an individual who worked with at least two individuals who were members of work groups other than his own. c. Contagt between groups was defined as a single working relationship between members of sets of indi- viduals who would otherwise be classified as separate work groups. The definition of work groups was not based upon the organ- ization's formal prescription of task units although some congruence was found between formal task units and socio— metric work groups. Because the structural classification of the sociometric data was based on the extant patterning of communication contacts, the term work group has relevance only within that context. For example, it would be entirely possible for the extant communication structure of a formal 23 task unit to break into several work groups and it would also be possible for boundaries of work groups to exceed the boundaries of formal task units in instances where information or coordination across task unit lines was necessary. Another clarification of the work group defi- nition is given by Jacobson and Seashore (1951, p. 37): ...some individuals have frequent, reciprocated and important contacts with a limited number of other individuals who in turn are closely interrelated, and have few, non-reciprocated and unimportant contacts outside of this group. The separation of work groups into independent entities is basically accomplished by removing liaisop‘pgppgpg and gpntaptp between groups from the sociometric matrix. The liaison person is a communicative link between multiple work groups and they “characteristically have many, frequent, reciprocated and important contacts which cut across the contact group structure" (Jacobson and Seashore, 1951, p. 37). The special criterion for the liaison person is that he , must have contacts with pp lpgpp two persons in work groups. other than his own. A single contact between two members g of separate work groups is not defined as a liaison, but as a contact between groups. In graph theory the liaison person is an analogue to the‘pppigplgpigp‘ppipp and the contact between groups is an analogue to the pgiggp as 11- lustrated in Figure 1. A liaison person may or may not have membership in one of the separate work groups. In Figure 1, the liaison (ar- ticulation point) is characterized without group membership. 24 Articulation ppint Adapted from Weiss, 1956, pp. 88-9 Figure 1. Graph theory representation of articulation point and bridge In the Jacobson, Seashore and Weiss study, one-third of the identified liaison persons could not be characterized as members of any separate work group because they "had many contacts with each other and few with any single work group” (Weiss and Jacobson, 1955). This group of liaisons were termed members of a liaison app. The other liaisons who could be identified as having group membership were each termed a liaison gpppp member, and the remaining were ter- med liaison individuals. It appears possible that the individuals who emerge as liaisons in this topological description might be exe- cutives who by location in the formal hierarchy have con- tacts within their task units and contacts at their own ex- ecutive level for coordination of work at that level. But Jacobson and Seashore report that "this was not the case. They...(liaisons) are found at all status levels in the structure" (p. 37). 25 The special functional significance of the liaison person's position in an organization's communication struc- ture is underlined by Jacobson and Seashore (p. 37); These liaison persons appear to be of critical im- portance to the conceptualization of organization in communication terms as they are in a position to in- fluence significantly or to control the communications to and from certain groups. Through them, it is ex- pected, it will be possible to trace differential in- fluences throughout the...(organization) as they are reflected in differences in attitudes among the sever- al sub-groups. In other words, the liaison person functions at least in a 1 for the various sub-groups to which he “gatekeeper role“ is connected. He may selectively relay information to sub- groups or he may function as a structurally important “un- certainty absorber." This is not to say, however, that a given liaison person's control over information to any given sub-group may be absolute. Information may be transmitted to a sub-group through contacts sub-group members have with other liaisons or through bridge contacts. The complex web of the total extant communication structure of a formal or- ganization is not as simple as the structure implied in Figure 1. Nonetheless, an analysis based upon this topo- logical differentiation of communication structure clearly emphasizes the unique locational role of the liaison person in that structure. What remains to be learned is the phen- omenological, behavioral and influential meaning of this 1Katz and Lazarsfeld (1955, p. 119) define the gatekeeping function as “controlling a strategic portion of a channel ...so as to have the power of decision over whether what- ever is flowing through the channel will enter the group or not.“ 26 unique locational communication role in organizations. gpprethodological Context Because topological-analytic research methods are in a developmental state and have not been programmed for computers in order to cope with large sociometric matrices, this study will replicate the manual method of sociometric differentiation developed by Jacobson, Seashore and Weiss. This study will also closely replicate the sociometric data-collection method of that study. . Sogiometrig 233g Qollegtion The basic sociometric instrument in the previous research was termed a Pepgonal Contagt Chegklis . Instructions were: Now go over the past two or three months and think of the people (in the organization) with whom you have worked most closely. We would like to get the names of the people with whom you work most closely. Write their names in here. Following the instructions were 18 blanks in which respon- dents could list the names of their contacts. In addition, respondents were asked to check for each individual prede- termined categories for frequency of contact, subject mat- ter, reason for contact, and importance of contact. This Personal Contact Checklist provided the basic data for topo- logical analysis. Although the instructions in the checklist yield non- directional sociometric contacts, they do not yield com- pletely content-free contacts and in this sense the result- ing structure is not the generip ppmmunigation structure in entirety. Specifically, the structural description is 27 of the work-related, or functional, communication network. Functional contacts were selected as the sociometric cri- terion because the result would ”presumably include all or nearly all of the important communication relationships be- tween pairs of persons” (Jacobson and Seashore, 1951, Eh 35). In other words it was assumed that the most frequent and most relevant information exchanges between individuals would take place in the course of contacts brought about by individual and group task requirements.. Partial support for this assumption is reported by Blau and Scott (1962, p. 134): “Work pressures increased the frequency with which colleagues consulted each other about their professional problems.” It was also assumed that functional relation- ships, because of the demands of task roles, would remain relatively stable over time (Weiss and Jacobson, 1955, p. 661). Sogiometrig Analysis Preliminary to topological anal- ysis of sociometric data in the Jacobson, Seashore and Weiss study was the need to establish a criterion for oper- ationalizing the existence of a contact between two persons. Approximately 2400 work relationships were reported by 196 respondents in the study or an average of 12 contacts per respondent. Of this number 44 percent were reciprocated. Weiss and Jacobson report: The technique used in isolating work groups required that the matrix be symmetric across the main diagonal ....The original unreciprocated matrix could be made into a symmetric form either by adding entries in the proper cell when a report was not reciprocated, or by deleting unreciprocated entries. The second 28 method was chosen because the close relationship between reciprocation and importance and frequency of contact suggested that the simplification would not be at the cost of essential information. The basic sociometric data, then, was graphically repre- sented in a 196x196 matrix containing approximately 1000 non-zero entries designating reciprocated contacts between pairs of individuals. However, in the final stages of dif- ferentiating the matrix into sub-groups it was found neces- sary to use unreciprocated contacts of some liaison per- sons in order to determine whether they would be classed as ligipon individuals or liaison gpppp members. After the basic sociometric data was arranged in a 2 symmetric matrix, the next step involved applying a cri- terion of group membership in order to differentiate se- parate work groups: The technique...consisted of applying, initially, a stringent criterion of group membership in order to locate the most clearly differentiated groups and to identify their core membership: this initial cri- terion was followed by the application of a series of less stringent criteria of group membership until all, or nearly all, of the members were identified as belonging to one or more of the groups. (Jacobson and Seashore, 1951, p. 36) 1In the same article, Weiss and Jacobson report that 80 per- cent of the contacts listed as “several times daily“ and of “utmost“ or “great” importance were reciprocated while only 19 percent of the low frequency and low importance contacts were reciprocated. 2In order to accomplish the succeeding steps in matrix ana- lysis, it is necessary that contacts be clustered as much as possible, around the diagonal. Listing members accord- ing to the formal organizational chart resulted in clusters around the diagonal since most work-related contacts were generally within formal task units. ’29 Briefly, the first step in the procedure1 was to arbi- trarily break the original matrix into smaller segments of not more than 50 persons. Segments were divided so that members in the segment had a minimum number of con- tacts outside the segment. Each segment could contain several clusters of contacts around the diagonal. Separ- ation of sub-groups within a segment was accomplished by removing tentatively identified liaison persons from the segment matrix, reordering the matrix to maintain clusters around the diagonal, removing tentative liaisons, reorder- ing the matrix, etc. The end goal of this procedure was to identify separate work groups within each segment by in- spection. Specified criteria were maintained throughout the procedure and a number of checks applied to guard a- gainst improper classification. Careful attention was paid to the final identification of separate groups to be sure that the original arbitrary boundary of a segment did not in fact divide one or more work groups into different segments. At the end of this procedure, the yield was a list of the membership of each separate group, a list of liaison persons, and a list of isolates. It was necessary at this point to utilize high-frequency unreciprocated con- tacts of liaison persons and isolates in order to ascertain 1The procedure is described in detail by Weiss, 1956, pp. 88-108. Weiss and Jacobson (1955) report that, after the basic reciprocated matrix is prepared, one person can pro- bably complete the procedure for a 200x200 matrix in ap- proximately 40 hours. 30 whether or not they could be characterized as having mem- bership in one of the separate groups. Weiss and Jacobson (1955) describe their end product: a. About 82 percent of the respondents could be classed as members of the 22 separate work groups that formed the basic framework of the organization. b. The 22 work groups were held together by a network of liaison persons who were the remaining 18 percent of the respondents. The result of this sociometric analysis is the identi- fication of sub-groups within an organization based entire- ly upon the egpant communication structure and the identi- fication of individuals within that structure who play an important locational communication role by linking sub- groups together. In the organization studied, the procedure allowed classification of the 196 members into two discrete groups: 35 liaison persons and 161 non-liaipon pgrsons (the members of the 22 sub-groups). Because the analysis is based on the extant communication structure, it is not pos- sible to predict g priori the number of liaison persons who will emerge from a given organization. Jacobson and Sea- shore (1951, p. 36) suggest, "It can be expected that organ- izations will differ greatly as to their uniformity and clarity of sub-group differentiation.“ Concomitantly, it probably can be expected that organizations will differ greatly in the number of liaison persons in a given organ- ization based on such factors as, for example, task coordin- ation demands or the degree of rigidity in adhering to for- mal hierarchial lines. 31 Given differentiation of an organization's communica- tion structure into the liaison and non-liaison categories, the second step in the topological-analytic approach is descriptive and/or functional analysis of these roles. Hypptheses gpg Rationale The focus of this study will be upon the comparison of two sets of individuals, liaison persons vs. non-liaisog rsons, as to certain aspects of the phenomenology of 5 their communication role in the organization, their commun- ication behavior and their interpersonal and organizational influence potential. The only currently available evidence that liaison persons are different from non-liaison persons is the fact of the unique locational position of the liaison role in the organization. Justification for study of the liaison communication role can be based on two factors: 1. differentials between liaisons and non-liaisons in regard to actual communication behavior and influence; and/or 2. differentials between liaisons and non-liaisons in regard to pgrceived communication role and influence potential in the organization.1 Within this broad context, a large number of research lIn a similar vein, Yadav and Rogers (1966, p. 4) opt for study of opinion leadership in terms of behavioral attri- butes, personality attributes and attributes perceived by group members. ' 32 questions might be posed. The questions selected for this study are directed primarily toward specifying certain dif- ferentials based on perceptions reported by non-liaison persons in direct contact with selected liaisons and non- liaisons. Thus the sampling unit will be liaisons and certain non-liaisons, but the source of data will be non- liaison persons directly linked to them in order to attempt definition of the phenomenological meaning of the liaison communication role. In Thayer's terms, the focal point of the information vectors will be defined in terms of percep-‘ tions by persons immediately adjacent to that point. Theseg “immediate others“ will be referred to as dyadig gpntagts. E. Selection of this analytical perspective is conditioned by the necessity for collecting sociometric and descriptive data at the same time and the fact it is not possible to identify the liaison persons until topological analysis of the sociometric data is completed. The specific variables selected for this study are assumed directly related to communication functions. Whe- ther differentials are antecedents to, or consequents of, the communication functions of liaison persons will not be detected. The differentials, if any, will only be descrip- tive of the discrete categories -- liaison and non-liaison. General classification of the variables selected for study include (1) awareness of the existence of liaison roles, (2) interpersonal communication behavior of liaison role persons, (3) information relay functions of liaison 33 role persons, (4) perceived personal attributes of liaison role persons, and (5) diffuse (organizational level) and specific (interpersonal level) influence functions of liai- son role persons. While in-depth study of a range of vari- ables within each of these categories would be possible and desirable, this pilot study'will focus on only certain var- iables within each in order to provide a beginning assess- ment of the efficacy of the liaison role concept in organi- zational communication research. . A major limitation in attempting to derive descriptive hypotheses regarding the liaison communication role is that there is no exact analogue in previous research to this structural concept. Awareness 9; Liaison gplpg One factor which may affect the communicative function of the liaison person is the extent to which his dyadic contacts are aware of his strategic location in the organi- zation's communication structure. Findings from small group communication network research indicate that group members who are not initially aware of the predetermined structure of the group are, with some experience within the group, able to detect elements of the actual structure (Bavelas, 1960, p. 6761). The actual pattern of interpersonal contacts of liaison persons is by definition more structurally diverse 1Specifically, group members are able to identify the per- son occupying the central location in the network as the group leader. 34 than that of non-liaison persons. H l: Liaison persons are perceived to have more structural- ly diverse communication contacts in the organization than are non-liaison persons. By inspecting the sociogram drawn from the Jacobson, Seashore and Weiss study (reproduced in Weiss, 1956, pp. 60-1) to identify liaison persons, a random sample of liai- sons and non-liaisons revealed a broader span of recipro- cated contacts for liaisons. The median span for liaisons was six with a range from four to twelve (N=9): for non- liaisons the median span was four with a range from two to six (N=9) . H 2: Liaison persons are perceived to have more communica- tion contacts in the organization than are non-liaison persons. Interpersonal Communication Behavior The fact that the communication contacts of liaison persons are more structurally diverse, at least within the extant communication structure, suggests that there may be physical or psychological propinquity differentials between liaisons and their dyadic contacts versus non-liaisons and their dyadic contacts. To the extent such differentials exist, it might be assumed that initiation of message transactions requires more effort in liaison -- non-liaison dyads than in non-liaison -- non-liaison dyads: i.e., mes- sage transaction initiation within liaison -- non-liaison dyads may be more deliberate or purposive. Non-liaison -- non-liaison dyads, the members of which are tied together in a closely related work group, appear more likely to 35 happen to meet in the course of their regular work schedule. H 3: Liaison -- non-liaison dyads more frequently partici- pate in deliberately initiated message transactions than do non-liaison -- non-liaison dyads. Any set of deliberately initiated message transactions between two persons may be further differentiated by the directionality of the initiation: i.e., the proportion of times A seeks B to the times B seeks A. One person may seek another either to get or give information or opinions. There is no evidence to suggest why individuals assume liaison roles. Perhaps either (or both to a degree) of two conditions obtain: (1) the liaison may be a gregarious per- son who seeks others in diverse locations to get or give information or opinions1 , or (2) the liaison may be a “mag- netic center" (Walton, 1962) who possesses unique attributesz which attract others to him for information or opinions. If either condition uniquely applies to the liaison role, the implication is that directionality of message transac- tion initiation will be proportionately different between liaison-non-liaison dyads and non-liaison -- non-liaison dyads: i.e., directionality of initiation will be more equal in the latter dyad than the former. 1Katz and Lazarsfeld (1955, pp. 287-9), for example, found gregariousness an important variable in opinion leadership, especially in regard to public affairs content. 2Walton's research on magnetic centers was based on attri- butes related to authority, power, expertise and sociabi- lity. To this could be added: information related to task coordination, charismatic personality, awareness of the liaison's locational role and others. 36 H 4: The directionality of deliberate message transaction initiation is more disproportionate in liaison -- non-liaison dyads than in non-liaison -- non-liaison dyads. Information Rplpy Function In their discussion of the gatekeeping function, Katz and Lazarsfeld (1955, pp. 119, 113) distinguish between the information transmission role and the influential role of a gatekeeper. While this distinction may be difficult to maintain operationally unless limited to source-oriented analysis, it is possible to differentiate gippp sources of information from later, redundant sources without regard for source purpose or the effect of the information on the receiver. In addition, first sources may be studied re- gardless of whether the content of information relayed was evaluative or non-evaluative or the original source was the relayer or some other person. Wager's (1962) study of rumor transmission in a for- mal organization suggests that individuals who have access to rumor information play an important communication role by serving as first sources of information for their con- tacts. The relay of information across formal organiza- tional lines also appeared to serve a status enhancement function for the relayer. Davis (1953b, p. 46) reports a characteristic of the information relay behavior of "liai- son persons“ as: ...the sooner he knows of an event after it happened, the more likely he is to tell others. If he gets the information late, he does not want to advertise his 37 late receipt of it by telling it to others.1 The strategic location of the liaison person in the communication structure implies that he is in a position to have early access to information available in the organ- ization and that, although his position is peripheral to a given sub—group (especially liaison individuals and mem- bers of the liaison set), he is central among at least two sub-groups. Studies of network centrality in problem-sol- ving small groups (e.g., Leavitt, 1951) indicate that cen- trals assume a position of "information relay leadership“ by virtue of their strategic location. If the liaison role can be assumed analogous to centrality in a small group and if Davis' interpretation is appropriate, it might be ex- pected that liaison persons will be more consistent first sources of information for their dyadic contacts than will non-liaisons. H 5: Liaison persons are more likely to serve as first sources of organization-related information than are non-liaison persons. Perceived Personal Attributes Berlo, Lemert and Mertz (undated) assert: Whether one's judgement is based on the classical ad- monitions of Aristotle, common-sense observations, in- trospection, or the empirical research evidence of the past forty years, it is safe to conclude that an in- dividual's acceptance of information and ideas is based in part on 'who said it'. 1Davis does not report using the Jacobson, Seashore and Weiss sociometric analysis for identifying liaison persons. His identification appears based on simple inspection of the data. Interpretation of his findings must be tempered by this apparent shortcoming in his analysis. 38 The construct used to label this phenomenon is termed source credibility. The generally consistent finding is a positive relation between degrees of source credibility and degrees of acceptance of information contained in messages from specified sources. The factor analytic study con- ducted by Berlo, Lemert and Mertz provides an instrument for Operationally defining credibility as a perceived at- tribute of sources. The construct factored into three di- mensions: qualification, safety and dynamism. The latter factor was found less stable than the former two and its function was suggested as an intensifier for the others (Berlo, Lemert and Mertz, p. 21). A perceived personal attribute, then, which may be predictive of the acceptance of information relayed within an interpersonal communication network is the source credi- bility of the relayer. The maintenance of an individual in a liaison communication role suggests that his dyadic contacts may attribute greater source credibility to him than do the dyadic contacts of non-liaison persons. This assumption is based on the notion that leadership, or in this case information relay leadership, is a role conferred by followers rather than a role usurped by the leader (Katz and Kahn, 1966, p. 301). Furthermore, to the extent that the liaison role can be conceived as a type of leadership, there is evidence that perceived competence (partially ana- logous to the qualification factor) is positively related to attributed leadership (e.g., Katz, 1957, p. 73: Lippitt, 39 pp p_l_, 1958, p. 261). ' H 6: Liaison persons are perceived to have higher source credibility (qualification and safety dimensions) than are non-liaison persons. Influence Potential Sociometric mapping of the extant communication struc- ture of an organization permits separating influence func- tions according to specified pairs of individuals. Fur- ther, from the phenomenological viewpoint of any individual within that communication structure, influentiality can be differentiated on at least three levels: (1) my dyadic contact's influence over me (specific opinion leadership): (2) my dyadic contact's influence over his other contacts (diffuse opinion leadership): and (3) the "importance“ of the other (secondary) contacts which my dyadic contact hasl. As Katz (1957, pp. 74-5) observed, ”An individual may be influential not only because people within his group look to him for advice but also because of whom he knows out- side his group.“ In addition, influence over a dyadic contact may be related to how much influence is perceived to be exergised over those other secondary contacts. Katz and Lazarsfeld (1955, p. 123) observe that "the gatekeeping role is often coupled with influentiality within the group -- although the two functions need not necessarily go together." Leavitt (1964, p. 238) asserts that "preferen- tial access to information...is a major source of power in 1Importance is defined here as accessibility to the organi- zational “power structure“ and to individuals knowledge- able of activities in the organization. 40 any organization." The liaison person's more diverse con- tacts within the organization suggest he should have greater access to information than non-liaisons and thus may be attributed greater power. Lippitt, _e_p _a_l, (1958, p. 259) report that “the recipient of attributed power will make . more frequent attempts to influence the behavior of others and will be more successful in these attempts." Again, to the extent that the liaison role can be conceived as a type of leadership, Beer, pp p._]_., (1959, pp. 49-56) report that elected leaders of social groups are perceived to be more forceful and persuasive than non-leaders. The implication of these statements for the liaison communication role is that this strategic social position may be related to in- ,fluentiality both in terms of power over information relay and the concomitant opportunity to exercise opinion leader- ship over dyadic contacts. H 7: Liaison persons are perceived to have more important I secondary contacts in the organization than are non- liaison persons. H 8: Liaison persons are perceived to have more influence over their other contacts in the organization than are non-liaison persons (diffuse opinion leadership). H 9: Liaison persons are more likely to be perceived as personal opinion leaders for their dyadic contacts than are non-liaison persons (specific opinion leader, ship). ' V n I Tests of the hypotheses presented in this study are anticipated to provide a beginning assessment of the re- search efficacy of the sociometric description of the extant communication structure of an organization proposed by Jacob- son, Seashore and Weiss. Evidence regarding these hypotheses 41 will serve as a basis for further exploration of the specific functions of unique locational roles within the communication structure of an organization. The analytic framework is more inclusive than the fragmentary study of communication within only the formally prescribed struc- ture. CHAPTER II RESEARCH DESIGN The study population from which the samples will be drawn consists of all the professional faculty and staff of one College who are officed in a single, multifloor building of a large Midwestern university. Professional faculty and staff are defined as those individuals holding academic rank of instructor or higher and/or administrative titles denoting professional functions in a separate task unit within the formal structure of the College. The Col- lege selected is not known to be unique in any significant manner from other subject matter Colleges at the same Uni- versity, with the exception of size and subject matter orientation. Sociometric data will be collected only on functional communication relationships among individuals meeting the professional faculty or staff criterion and whose office is in the main building housing the College. One hundred and forty-two individuals meet these two criteria. The sources of data from each respondent will be a Personal Contact Checklist, a questionnaire concerning primarily demographic data, and a Personal Contact question- naire to be completed for each individual named in the "daily" or “several times daily" categories of the Personal Contact Checklist. All instruments will be self-completed. Instruments and data-collection procedures will be pretested 42 43 to determine ambiguities in procedures, instrument instruc- tions, and item.wording: to determine potential instrument- completion success, and to analyze the scales developed for this study. Egg Sample Although sociometric and questionnaire data will be collected from all members of the study population, only a portion of the members of this population will be included in the final sample. The first step in determining the sample will be to complete matrix analysis of the socio- metric data in order to classify the population into lépp- ppp person and non-liaison pprson categories (if any gpp- 135;; are found, they will not be included in either cate- gory). The primary sample of liaison persons will include all of the individuals identified in that category. A ran- dom selection of non-liaison persons equal to the number of liaison persons will constitute the primary non-liaison person sample. From the remaining population, non-liaison persons will be identified who are dyadic contacts of individuals in the liaison and non-liaison primary samples on a “daily” or “several times daily“ basis. These non-liaison dyadic contacts, the secondary samples, will be the source of data; i.e., data used to test the hypotheses will be taken from the Personal Contact questionnaires these individuals com- pleted on their reciprocated contacts who are in either the 44 liaison sample or the non-liaison sample. In the event that one individual might be the source of data for one person in the liaison sample and another person in the non-liaison sample, a random procedure will be utilized to select one or the other in order to maintain independence of source data between the two sample categories. Additionally, in the event that two individuals in the non-liaison secondary sample are a reciprocated dyad, a random procedure will be utilized to select only one as a source of data. Both of these procedures should insure independence of source data between and within the two secondary samples. The rationale for selecting a primary sample of non- liaison persons equal in number to the primary liaison per- son sample is to equalize a major source of variance be- tween the two sample categories. When dyadic contacts are the source of data, for example, two sources of variance are: (1) within-dyadig pg; differences and (2) between- dyadic pp; differences. Differences among perceptions of a given person by his set of dyadic contacts (within-dyadic set differences) are likely to be somewhat similar because the same person is the perceived object. This within- dyadic set variance is expected to be less than between- dyadic set variance. Thus, selecting equal numbers into the liaison and non-liaison primary samples should tend to minimize the standard error of the difference between sam- 1 ple means more than any other sampling procedure applicable 1Cf. Edwards, 1967, pp. 214. 45 to sociometric cluster sampling such as utilized here. If an organization much larger than the one in this study were investigated, it might be possible to order the dyadic contacts of all liaison persons as a sampling popu- lation, order the dyadic contacts of all non-liaison per- sons as a sampling population, and then randomly select individuals into two sample categories from these two popu- lations. However, it is not expected in this study that a sufficiently large number of individuals will be available to permit random sampling from within two such sampling populations. Operationalization pf Variables The key variables to be used in this study are: l. Perceived structural diversity of contacts 2. Perceived number of contacts 3. Deliberately initiated message transactions 4. Directional ratio of deliberate message transaction initiation 5. First source of information 6. Perceived source credibility (qualification and safety dimensions) 7. Importance of secondary contacts 8. Diffuse opinion leadership 9. Specific opinion leadership Six of the nine variables are operationalized as Likert-type scales (Oppenheim, 1966, pp. 133-42), two are 46 direct frequency estimates, and one utilizes Semantic Dif- ferential scales (Oppenheim, 1966, pp. 204-8: Kerlinger, 1964, pp. 564-80: Berlo, Lemert and Mertz, undated). Infor- mation from appropriate reported research was utilized in selection and construction of scale items. The Operationalization of variables is for the pretest instrument (Personal Contact questionnaire). It is antici- pated that analysis of pretest data will result in deleting certain items from the final instrument in order to shorten, as much as possible, the completion time required for each Personal Contact questionnaire. Each of the items in the following sections is com- pleted by a respondent in terms of a specific individual he named as a dyadic contact on the Personal Contact Check- list. Perceived Structural Diversity 9; Contacts Cues for itemslin this scale and the following Per- ceived Number of Contacts scale are derived from respon- dent's open-end descriptions of reasons for going to a cer- tain individual (a “magnetic center”) in an organization for information as summarized and reported by Walton (1962, 1 The number preceding each item in this and following scales refers to the item number in the pretest instrument. Items in the pretest questionnaire were ordered on a random basis. 4. 16. 25. 26. 47 would you say that this person works with people in more or fewer different departments or administrative offices than most other members of the faculty in the College of (---)? . a great many more quite a few more just a few more about the same number just a few less quite a few less . a great many less \JO‘Ul-bUNl-J C All of this person's contacts are concentrated within one somewhat separate group of faculty in the College Of (---)e ' This statement is: 1. definitely false 2. probably false 3. might be false 4. neither true nor false 5. might be true 6. probably true 7. definitely true Because of his work this person really ”gets around“ among faculty from many different areas in the College Of (---) e . agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement . disagree slightly with this statement . disagree somewhat with this statement . disagree strongly with this statement “mmlfiwNH e In any organization like this College, clique groups develop. Would you say that this person has contact with people in more or fewer clique groups in the Col- lege than do most other faculty members? 1. a great many more clique groups 2. quite a few more clique groups 3. just a few more clique groups 4. about the same number of clique groups 5. just a few less clique groups 6. quite a few less clique groups 7. a great many less clique groups 48 35. This person works with people whose offices are located in several buildings in the College of (---). 1. agree strongly with this statement Nam-5U” s . agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement . disagree strongly with this statement These five items will be reflected and summed to a to- tal scale score for each respondent. The larger the total score, the more structurally diverse are the dyadic con- tact's relationships. Perpeived Number 9; Contagts Items in the pretest scale include: 7. Other than myself, very few faculty or administrators in the College communicate with this person. agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement person probably works with more other faculty in the College than do most other members of the faculty. 1. I 2. I 3. I 4. I 5. I 6. I 7. I 11. This 1. I 2. I 3. I 4. I 5. I 6. I 7. I agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement 49 20. About how many academic or administrative people in the College of (---) would you say this person has contact with in an "average" week compared to the number with whom most other members of the faculty have contact? 1. a great many more quite a few more just a few more about the same just a few less quite a few less a great many less fid‘U‘IbWN s 29. This person has access to more members of the faculty of the College of (---) than do most others in the College. ' l. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement 32. This person communicates with very few administrators and members of the faculty in the College of (---). 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement Numeric values of response alternatives for items 11, 20 and 29 will be reflected and the five items summed to a total score. The higher the total score, the more contacts the dyadic contact is perceived to have. Deliberately Initiated.Message Transactions This variable will be operationalized as a direct estimate of the frequency of deliberate message transaction initiation with dyadic contacts. The item takes the form: 50 34. Out of 100 times you might have contact with this per- son, about how many times would: a. you seek him or initiate the contact (you go to see him, place the phone call, write) . . . . . . . . . b. he seek you or initiate the contact (he comes to you, places the phone call, writes). . . . . . . . . c. none of the above, we just happen to meet (neither of us goes to the Other ) e e e e e e e e e e 0 TOTAL = 100 contacts The frequency of contact reported in response (c.) will be taken as the reflex of estimated deliberate initiation: i.e., 100 minus (c.) equals the estimated frequency of de- liberate message transaction initiation based on 100 trans- actions. This variable will also be a direct estimate of inter- action frequency. Estimated frequency of respondent seek- ing behavior to contact's seeking behavior (response alter- natives a. and b. in item 34 above) will be taken as a mea- sure of the directional initiation ratio for deliberate message transactions within the dyad. The specific measure will be the fraction (a.) over (b.) pp (b.) over (a.) with the larger number always in the denominator. The more this fraction deviates from one, the more disproportionate is message transaction initiation in the dyad. ggppp Source pg Information The reported research of Walton (1962, pp. 89, 91-2) 51 and Wager (1962) provide information for construction of items in this scale. 2. 19. 28. 36. When you learn about some change or new idea being pro- posed or discussed in the College of (---) or any of its departments or institutes, how likely are you to hear about it first from this person? 1. extremely likely 2. very likely 3. somewhat likely 5. somewhat unlikely 6. very unlikely 7. extremely unlikely A great deal of the planning for the College and its sub-units is accomplished by faculty committees. How frequently do you learn about the activities and dis- cussions of certain committees first from this person? 1. extremely infrequently . very infrequently somewhat infrequently about half the time somewhat frequently . very frequently . extremely frequently \lO‘U‘IfiWN s As new developments occur in the College, I usually "get the word" from someone other than this person. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement ‘ 6. I disagree somewhat with this statement 7. I disagree strongly with this statement Thinking back over the contacts you've had with this person in the past month, about how often have you learned something new from.him about programs, acti- vities, or people in the College? 1. almost every time 2. very often 3. slightly more than half the time . about half the time slightly less than half the time seldom almost never \lO‘U’IIF s 52 Items 2 and 36 will be reflected and the four items summed to a total scale score. The higher the total score, the more likely the dyadic contact is to serve as a first source of information. Perceived Source Credibility The five most representative scales for each of the two major dimensions of source credibility, qualification and safety, identified in the factor analytic study by Berlo, Lemert and Mertz (undated, p. 18) will be used to assess the perceived credibility of dyadic contacts. Each dimension will be tested as a separate variable. References to both dimensions were manifest in respondent evaluations of "magnetic centers“ as reported by Walton (1962: qualifi- cation: pp. 80, 84, 88, 96, 100-1: safety: pp. 81, 105). The form of each scale will follow semantic differen- 1 The word-anchors tial procedures with word-anchors added. used with response alternatives for each adjective pair will be: extremely, quite, somewhat, so-so, somewhat, quite, extremely. The recommended adjective pairs for the safety dimension are: safe-unsafe, just-unjust, kind-cruel, friendly- unfriendly, honest-dishonest: for the qualification dimen- sion: trained-untrained, experienced-inexperienced, skilled- unskilled, qualified-unqualified, informed-uninformed. 1The addition of word-anchors to semantic differential scales was recommended by V. C. Troldahl, Department of Communication, Michigan State University, in a personal conversation. 53 Response alternatives will be scored from one to seven with seven representing the most positive evaluation for each adjective pair. The scales for each dimension will be summed to a total score with a range from 5 to 35. Impprtance 9; Secondary Contacts Secondary contacts are defined as other persons in the organization with whom the respondent perceives his dyadic contact to have a regular communication relationship. Pre- test scale items developed to measure perceptions of the importance of these secondary contacts were derived again from the reported research of Walton (1962, pp. 83, 89, 92, ‘98). Importance of the secondary contact is defined in terms of access to individuals in the “power structure“ or knowledgeable of organizational activities. 5. If there is anything noteworthy going on in the College of (---), this person works with people who usually know about it. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement 8. How important are the members of the College of (---) faculty with whom this person works most closely? 1. very important 2. moderately important 3. somewhat important 4. so, so . somewhat unimportant . moderately unimportant . very unimportant \lO‘U‘ S4 13. Very few College of (---) faculty members with whom this person works would be very "useful” contacts if one wanted support for an idea he was trying to get adopted in the College. 1. qmmbww O thhiHldhiH agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement 23. Of the College faculty with whom this person works, hardly any of them are in a position to officially or unofficially influence important decisions made about activities of the College or the faculty. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement 27. This person has access to individuals who are rela- tively high in the “power structure" of the College of --- . l. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement 37. This person works with faculty in the College of (---) who are “in the know“ about what's going on in the College. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement Items 5, 8, 27, and 37 will be reflected and six items summed to a total score. The higher the total score, the 55 more important secondary contacts are perceived to be. Diffuse Opinion Leadership Diffuse opinion leadership is defined as the extent to which a respondent's dyadic contact is perceived to in- fluence the opinions of others in the organization (i.e., the latter's secondary contacts). Walton's (1962, pp. 83, 87, 91) respondents suggest this perceived influence over secondary contacts as one reason for initiating communica- tion with a "magnetic center.“ These reports suggest items for a Diffuse Opinion Leadership scale: 9. Those relatively high in the "power structure“ of the College of (---) have been very receptive to this person's suggestions and opinions. 10. Many HHHHHHH agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement of the “official" and "unofficial" influential people in the College look to this person for opinions and advice. agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement this person communicates his views to others in the College, they usually listen and heed his suggestions. 1. I 2. I 3. I 4. I 5. I 6. I 7. I 12. When 1. I 2. I 3. I 4. I 5. I 6. I 7. I agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement 15. 17. 21. 24. 30. 56 I see this person as a key figure who can support or block proposals made in the College of (---). 1. agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement qmubww O O HHHHHHH Those individuals who have a lot to say about what goes on in the College respect the suggestions this person makes. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement This person doesn't have much influence with the other people he works with in the College of (---). . I agree strongly with this statement . I agree somewhat with this statement I agree slightly with this statement I neither agree nor disagree with this statement I disagree slightly with this statement . I disagree somewhat with this statement . I disagree strongly with this statement N¢WI§WNH a When interacting with other faculty members in the Col- lege, this person isn't a very good convincer. 1. I agree strongly with this statement 2. I agree somewhat with this statement 3. I agree slightly with this statement 4. I neither agree nor disagree with this statement 5. I disagree slightly with this statement 6. I disagree somewhat with this statement 7. I disagree strongly with this statement This person has a way of getting what he wants from the other people in the College of (---) with whom he has contact. agree strongly with this statement agree somewhat with this statement agree slightly with this statement neither agree nor disagree with this statement disagree slightly with this statement disagree somewhat with this statement disagree strongly with this statement \lO‘U‘athI-J e HHHHHHH 57 Items 9, 10, 12, 15, 17, and 30 will be reflected and the eight items summed to a total score. The higher the total score, the more diffuse opinion leadership the dyadic con- tact is perceived to exercise. Specific Opinion Leadership Specific opinion leadership is defined as the per- ceived influence a dyadic contact has over the respondent as reflected in information and opinion giving-seeking be- havior and attempts to convince. . Items in this scale are adapted from self-designating opinion leadership scales developed by Rogers and Cartano (1962, PP. 435-41) and Troldahl and Van Dam (1965). Rogers (1962, p. 231) reports a split-half reliability of .703 for his six-item scale and a coefficient of reproducibility of 91.4 percent. Both previous scales require respondents to estimate their potential Opinion leadership behavior across 311 their communication contacts. Adaptation of the scales for this study is based on asking respondents to estimate potential opinion leadership behavior with a specifip indi- vidual. Opinion leadership is thus estimated by one member of a dyad on a zero-sum basis. Reported perceptions of opinion leadership in reference to a specific individual offer the potential of greater accuracy in estimating in- terpersonal behavior. Items in the pretest scale include both content-re- lated and content-free communication behavior. 58 1. Which of the following has usually occurred during 14. 18. discussions you've had with this person in the past week or so about activities, programs, or people in the College of (---)? I asked him questions: 1. much more often than he asked me somewhat more often than he asked me slightly more often than he asked me about as often as he asked me slightly less often than he asked me somewhat less often than he asked me . much less often than he asked me \lO‘Ul-fiwN e Whenever you communicate with this person which one of the following is usually the case? 1. almost always he talks and I listen 2. very often he talks and I listen 3. often he talks and I listen 4. we usually talk and listen about equally 5. often I talk and he listens 6. very often I talk and he listens 7. almost always I talk and he listens Considering the relationship you have with this person, who do you think depends on the other more for advice on matters related to teaching, research or consulting? I depend on him: 1. much more than he depends on me . somewhat more than he depends on me . slightly more than he depends on me about as much as he depends on me slightly less than he depends on me . somewhat less than he depends on me . much less than he depends on me NO‘U‘IfiWN a When you and this person discuss activities of the Col- lege or any of its sub-units, which of the following happens more often across a number of these discussions? He tells me about: 1. a great many more things than I tell him 2. quite a few more things than I tell him 3. just a few more things than I tell him 4. the same number of things I tell him 5. just a few less things than I tell him 6. quite a few less things than I tell him 7. a great many less things than I tell him .l I'll III II I I. Jill 59 22. During the last discussion you had with this person, which one of you attempted to do the most convincing? He tried to convince me: . much more than I tried to convince him somewhat more than I tried to convince him slightly more than I tried to convince him as much as I tried to convince him slightly less than I tried to convince him . somewhat less than I tried to convince him . much less than I tried to convince him samcneuuaaw O 31. Generally speaking, does this person depend on you more than you depend on him for information about activities and personalities in the College of (---)? 1. much more than I depend on him 2. somewhat more than I depend on him 3. slightly more than I depend on him 4. about as much as I depend on him 5. slightly less than I depend on him 6. somewhat less than I depend on him 7. much less than I depend on him 33. Across a number of contacts you may have with this per- son, which one of the following usually happens? I am: 1. much more likely to ask his opinions than he is to ask mine . somewhat more likely to ask his opinions than he is to ask mine slightly more likely to ask his opinions than he is to ask mine about as likely to ask his opinions as he is to ask mine slightly less likely to ask his opinions than he is to ask mine somewhat less likely to ask his opinions than he is to ask mine much less likely to ask his opinions than he is to ask mine \I 0‘ U‘ 05 on N 0 Items 1, 3, 14, 18, 22 and 33 will be reflected and the seven items summed to a total score. The higher the total score, the more specific opinion leadership the dyadic con- tact is perceived to exercise. 60 Data Collegtion Adaptations suggested by the pretest will be incor- porated into the main study data-collection procedures, plus additional steps assumed necessary to insure as close as possible to 100 percent return of the Personal Contact Checklist. One day prior to the start of data collection, a let- ter from a high prestige faculty member in the College will be distributed to all members of the study population. The entry letter will briefly introduce the study and the in- vestigator, and will ask the cooperation of each individual. After the letter is distributed, personal telephone calls will be made to each member of the population to enlist their cooperation and, if obtained, to make an appointment for delivery and pick-up of the questionnaire packet. Inso- far as possible, appointments will be made for delivery and pick-up on the same day. Each questionnaire packet will include: (1) an intro- ductory letter from the investigator, (2) Part I - Personal Data questionnaire, (3) Part II - Personal Contact Check- list and (4) Part III - ten Personal Contact questionnaires. The instructions for Part III will ask respondents to com- plete a separate Personal Contact questionnaire for each person they indicate contact with in Part II on a “daily” or "several times daily“ basis, writing in the name of the individual at the top of the questionnaire. When delivering the questionnaire packet, the 61 investigator will talk briefly with each respondent to re- quest that they not reveal the nature of the questionnaires to other members of the study population. It is anticipated that the entire population of 142 persons, excepting those who might be temporarily out-of- town, can be contacted and questionnaires delivered in a three-day period. Follow-up telephone calls and personal visits will be made to respondents who miss the pick-up deadline in an attempt to keep the time from delivery to pick-up at a minimum. In the event that a respondent re- fuses to complete the questionnaire packet after receiving it, special effort will be made to get, at minimum, the Personal Contact Checklist because of the critical need for complete sociometric data on the entire study population. Data Analysis The first step will be sociometric analysis to deter- mine membership of the liaison and non-liaison categories from which the primary samples will be drawn. Reciproca- 1 tion of contact will be operationalized as mutual listing on the Personal Contact Checklist by each member of a dyad regardless of the frequency of contact, or discrepancy in the reported frequency between members of the dyad. When non-respondents in the study population are mentioned as a contact by a respondent, reciprocation will be assumed for those non-respondents named only in the daily or more often contact-frequency category. The close relation between 62 reciprocation and high-frequency contact reported by Weiss and Jacobson (1955) suggests this is a realistic assumption for operationalizing reciprocation with non-respondents. Personal Contact questionnaires with less than 50 per- cent completed items will be discarded. The treatment of missing data within a partially completed questionnaire will be as follows: (1) For item 34 (estimate of deliberate and directional message transaction initiation frequency) if two of three response alternatives are completed, the third can be accurately predicted because one degree of freedom exists for the response alternatives: if only one or none of the response alternatives are completed, questionnaires completed by other members of the population for the per- son named will be identified and the missing data coded as the mean response alternative of the distribution of other responses directed toward the person named. This "profile mean“ transformation is utilized instead of transforming to the mean of all responses to an item because of the expec- tation that within-dyadic set variance will be less than between-dyadic set variance. (2) For all other items in the questionnaire, if the respondent completed other items in a scale, the no-response item will be coded as the mean of the responses to other scale items: if no items in a scale are completed, each item will be coded to the “profile mean“ by the method described above. McOuitty cluster analysis will be applied to the inter- item correlation matrix of each main study scale to assess 63 the dimensionality of items in comparison to the pretest results. The nature of the hypotheses and the characteristics of the data indicate that each of the nine hypotheses can be evaluated with the “t“ test for independent sample means (McNemar, 1962, p. 103). Pretest Procedures The purposes of the pretest are: (1) To determine if data-collection procedures result in sufficiently high co- operation from respondents: (2) To evaluate ambiguities in questionnaire instructions and item wording: (3) To deter- mine potential instrument completion success, especially the Personal Contact Checklist: (4) To determine whether 16 blank spaces on the Personal Contact Checklist are suffi- cient for all respondents: (5) To estimate the maximum num- ber of persons named in the “daily" and ”several times daily" categories on the Personal Contact Checklist in or- der to determine if ten Personal Contact questionnaires are sufficient for each questionnaire packet in the main study: and (6) To analyze the scales developed for the pretest in order to select items for scales used in the main study. The pretest sample will consist of 30 randomly selec- ted professional faculty and staff from another College within the University. Each pretest respondent will be asked to complete two Personal Contact questionnaires, one for each of two randomly selected persons named on the 64 Personal Contact Checklist. A potential total of 60 Per- sonal Contact questionnaires will be available for pretest scale analysis. The 30 randomly selected members of the pretest sample will be contacted by telephone to enlist their cooperation, and, if obtained, to make an appointment for delivery and pick-up of the questionnaire packet. The contents of each questionnaire packet will include: (1) a letter briefly introducing the study, (2) Part I - Personal Data questionnaire, (3) Part II - Personal Contact Checklist adapted from the instrument reproduced in Jacobson and Seashore (1951, p. 34), (4) Part III - two Personal Con- tact questionnaires, and (5) Part IV - a questionnaire eval- uation form. For the pretest, respondents will not be required to return the Personal Contact Checklist since sociometric analysis is not necessary to purposes of the pretest. The Checklist is included because respondents must complete it in order to complete the two Personal Contact questionnaires. Instructions on each Personal Contact questionnaire will read: "Please complete this questionnaire in terms of per- son #(---) you named on the Personal Contact Checklist.“ A randomly selected number between one and ten will be in- serted in each blank by the investigator. In instances where respondents named fewer than ten persons on the Per- sonal Contact Checklist, it is possible that one or both of the Personal Contact questionnaires will not be completed. 65 However, if the number of persons named approximates the findings of the Jacobson, Seashore and Weiss study (i.e., an average of 12 named), these instances should be rare. The first step in pretest scale analysis will be to determine the internal consistency of each scale based on inter-item Pearson product moment correlations. Items in a scale which correlate zero or negatively with two or more other items will be deleted from the scale. If there is a negative correlation between any two individual items, the item which has the lowest average correlation with all other items in the scale will be deleted. The requirement that main study respondents complete a Personal Contact questionnaire for each person named in the "daily“ or “several times daily" contact-frequency categories suggests that each questionnaire should be as short as possible to assure the highest number of completed questionnaires. It is conceivable that some respondents may be required to complete as many as ten Personal Contact questionnaires. One purpose of the pretest will be to re- duce the 37 pretest scale items to an approximate maximum of 25. This will require that any given scale be limited to an approximate maximum of five or six items, with a min- imum of two. The reduced scales will be determined by the following method. After deleting inconsistent items from the scale inter-item correlation matrix by the method des- cribed above, McOuitty cluster analysis (McOuitty, 1957, pp. 207-29) will be applied to the remaining items to evaluate 66 scale dimensionality. If more than one dimension emerges, item selection will be based on inspection to determine the dimension most suited to the concept. If more than one dimension seems consistent with the general conceptual definition of the variable, those dimensions will be re- tained but will be treated as sub-scales and tested as se- parate variables for the appropriate hypothesis. If, after completion of these procedures, further reduction in the number of items is necessary for a given scale or sub-scale, the average correlation1 of each scale or sub-scale item will be calculated and those items with the lowest average correlation will be deleted. 1Since the magnitude of difference between correlations is not linear, the average correlation is calculated by the following formula: {r2 where r equals the square of each correlation andllequaIs the number of correlations. CHAPTER III PRETEST RESULTS Members of the pretest population were identified by department and academic rank or administrative title from the university telephone directory. The 69 individuals so identified were listed alphabetically by department and name and each was assigned a unique identification number ranging from 01 to 69. Thirty two-digit numbers were drawn from a table of random numbers to select population members into the pretest sample. The day prior to distribution of the pretest question- naire packet each member of the sample was contacted by telephone and asked if he would be willing to participate in the pretest. All 30 agreed to do so. Each agreed to having the pretest packet delivered to his office the next morning with completion time agreed upon for later that afternoon or early the next morning. The 30 questionnaire packets were delivered by 10 a.m. the next morning. Thirteen packets were returned by late that afternoon, eight more by noon the second day, five more by 9 a.m. the third day, and two more by the end of the fourth day. Two members of the sample (6.67%) refused to complete the questionnaires after receiving them. Of the last two packets picked up, one was not completed and the reason given by the respondent was that the question- naire was “too personal“ (even though the Personal Contact 67 68 questionnaires were anonymous). One non-respondent, al- though he had agreed to complete the questionnaires in the telephone contact, gave as his reason that he ”hates research.“ The total non-response rate was ten percent of the pretest sample, indicating that additional steps in securing respondent cooperation would be necessary in the main study. Each of the 27 respondents were asked in the instruc- tions to complete a maximum of two Personal Contact ques- tionnaires. Two respondents did not complete either ques- tionnaire because the random numbers inserted on both their questionnaires were larger than the total number of contacts they listed on the Personal Contact Checklist. Six res- pondents completed only one questionnaire because the se- cond random number was too high, and one respondent refused or neglected to complete the second. Eighteen respondents completed the maximum of two questionnaires. Thus a total of 43 Personal Contact questionnaires were available for scale analysis. Pretest Analysis Part IV of the pretest questionnaire packet was a brief questionnaire asking respondents to indicate problems they may have encountered in instrument instructions or item wording, and the approximate length of time required to complete each of the three different types of questionnaires. No major instruction or item wording ambiguities were 69 mentioned by respondents. Some minor changes were suggested and incorporated into the main study instruments: e.g., adding "and ad hoc" to the types of committee assignments listed in item (10) of Part 1: adding “none“ to the list of administrative titles in item (3) of Part I. The reported completion times for the Part I question- naire by 25 respondents ranged from two to ten minutes, with a mean of 4.72 minutes and a mode of 5: for Part II, the Personal Contact Checklist, the range was one to ten minutes, with a mean of 3.70 minutes and a mode of 3 (N323): for Part III, the Personal Contact questionnaire, the range was five to fifteen minutes, with a mean of 10.10 minutes and a mode of 10 (N=22). These estimates of completion times reinforced the need to reduce the over-all length of the main study questionnaires, especially the Personal Contact questionnaire. An “average“ respondent who would be asked to complete as many as seven Personal Contact ques- tionnaires in the main study would have an approximate com- pletion time of 40 minutes for the entire set. Since the questionnaires were unsupervised and self-completing, it was deemed desirable to keep total completion time within a range of 20 to 30 minutes. The Personal Contact question- naire would have to be reduced to an approximate total of 20 items, plus the ten Semantic Differential scales. In order to keep the pretest questionnaires anonymous, respondents were not asked to return the Personal Contact Checklist but were asked in the Part IV questionnaire to 70 report the number of names listed in each frequency of con- tact category and whether or not they would have listed the same, fewer, or no names if the Checklist would have had to be returned. Twenty-one of 24 respondents indicated they would have listed the same persons, and three respondents indicated they would have refused to list any names. Com- bining this information with the previously reported non- response rate to the entire questionnaire packet, indicates that 20 percent of the pretest population would not have completed the Personal Contact Checklist. Because of the critical need for data from the Checklist for the main study sociometric analysis, it was obvious that special effort would be needed to insure greater coOperation from main study respondents in completing the Personal Contact Check- list. The total number of contacts listed on the Personal Contact Checklist by 25 respondents was 213. Of this total, 13 percent were in the ”several times daily“ category, 22 percent in the “once daily“ category, 34 percent in the "2-3 times per week” category, and 31 percent in the “once weekly" category. Combining the first two categories into a "once or more daily“ category yields 35 percent of the total in this category. The raw data are presented in Table 1. Only two of the 25 pretest respondents used all 16 blanks on the Personal Contact Checklist (i.e., listed 16 persons as contacts), but indicated they would not have 71 Table 1. Frequency of contact and number of contacts repor- ted by twenty-five pretest respondents Number of contacts listed by each respondent 012345678910121516 Total Frequency contacts of contact repprted A. Several times daily 8 104 2 l (N=25) 28 3. Once . daily 3 9 7 2 3 l (N=25) 47 C. (Once or more daily) 6 7 5 1 2 2 2 CN=25) (75) D. 2-3 times per week 3 4 7 4 3 1 1 1 1 (N=25) 73 E. Once weekly 8 6 5 1 4 1 (N225) 65 F. (Once or more weekly) 1 4 2 2 3 l 3 3 2 2 2 (N225)(213) listed more persons if additional blanks had been provided. The highest number of contacts listed in the combined “daily or more" category was seven. Based on this infor- mation it was decided that nine Personal Contact question- naires would be included in each main study questionnaire 1 packet. Instructions for Part III in the main study pac- kets were written to indicate that more questionnaires 1A total of 1278 Personal Contact questionnaires were needed to make up 142 questionnaire packets. 72 would be available from the investigator if needed. Pretest §pplg Analysis The first step in scale analysis was to determine the internal consistency of items in each scale by inspecting the inter-item correlation matrix for negatively correlated items. None of the items in any of the scales correlated zero or negatively with any other item in its respective scale. The second step was to assess scale dimensionality by applying McQuitty cluster analysis to the inter-item corre- lation matrix of each scale. By applying cluster analysis and the method of average correlations, each scale was re- duced to a minimum of two and a maximum of five items, yield- ing a total of 21 questionnaire items in the main study Per- sonal Contact questionnaire. Perceived Structural Diversity 9; Contacts The inter- item correlation matrix and the average correlation of each scale item in this scale is presented in Table 2. Cluster analysis of the inter-item matrix yielded the following single typal structure:1 26 > 4 “$35 25 \16 Items (4) and (35) emerged as the core of the matrix with items (25) and (26) most closely associated with item (4). lMcQuitty (1957, p. 3) defines a typal structure as "one in which every member of a type is more like some other mem- ber of that type (with respect to the data analyzed) than it is like any member of any other type.” 73 Table 2. Inter-item correlation matrix and average corre- lations for the perceived structural diversity of contacts pretest scale Item Average number* 4 16 25 26 35 correlation 4 x .443 .693 .754 .786 .745 16 x .596 .404 .427 .473 25 x .562 .608 .623 26 x .659 .663 35 x .688 * Item numbers in this and following tables may be compared to the item numbers in Chapter II under Operationaliza- tion pg Variables to identify the item. Based on the average correlation, item (26) was selected over item (25) to comprise the three-item scale used in the main study. Perceived Number 9; Contagts The inter-item correla- tion matrix is presented in Table 3. Cluster analysis of the inter-item matrix yielded two typal structures: Type I: 32:29 Type II 11<._"‘> 20 Items (32) and (29), the highest single type in the matrix, were selected for the two-item scale used in the main study. gippp Source 9; Information Table 4 presents the inter- item correlation matrix and average correlations for items in this scale. 74 Table 3. Inter-item correlation matrix for the perceived number of contacts pretest scale Item number 7 11 20 29 32 7 x .358 .537 .346 .532 11 x .773 .694 .625 20 x .677 .676 29 x .808 32 x Table 4. Inter-item correlation matrix and average corre- lations for the first source of information pre- test scale Item Average number 2 19 28 36 correlation 2 x .607 .740 .634 .663 19 x .537 .545 .564 28 x .567 .621 36 x .583 Cluster analysis of the matrix yielded the following typal structure: 36 \‘2—>28 197 Items (2) and (28) formed the core of the single type which emerged and were selected to comprise the two-item main study scale. Impprtance pg Secondapy Contacts See Table 5 for the 75 Table 5. Inter-item correlation matrix and average corre- lations for the importance of secondary contacts pretest scale Item Average number 5 8 13 23 27 37 correlation 5 x .582 .428 .526 .598 .731 .582 8 x .437 .527 .603 .570 .547 13 x .478 .462 .493 .460 23 x .414 .644 .523 27 x .717 .569 37 x .637 inter-item correlation matrix and average correlations for this scale. The following single typal structure emerged: 27<=——8 5:37423 13 Based on the emergent typal structure, the average corre- lations, and a three-item limitation on the scale, pro- cedure would dictate selecting items (5), (37) and (27). However, in order to retain item (8) for comparative pur- poses with the Jacobson, Seashore and Weiss study1 and still remain within the three-item limitation, item (5), which is highly correlated with item (37), was deleted and 1Degree of "importance of contact" was one of the response categories in the original Personal Contact Checklist. 76 the three items retained for the main study questionnaire were (37), (27) and (8). Support for retaining item (8) is found in its relative centrality within the sub-matrix when items (5) and (37) are removed. The following typal structure then results: 13\\Q /}7 23 In addition, item (8) emerges with the highest average cor- 8:;f227 relation from among these four items: Item Average correlation 8 .527 27 .499 23 .475 13 .459 Diffuse Opinion Leadership The inter-item correlations and average correlations within types for the Diffuse Opin- ion Leadership scale are reported in Table 6. Two typal structures emerged from the matrix with the following con- figurations: 15 K Type I: 9:;i317:::: 30 Persuasiveness within 10 organizational "power structure” Type II: 24:;5121 General persuasiveness ‘12 with contacts Inspection of the items in each emergent type indicates that Type I might be labeled “persuasiveness within the or- ganizational power structure“ and Type II might be called "general persuasiveness with contacts." With the exception of item (30), which is weakly associated within its cluster, the items in Type I refer to the receptivity members of the 77 Table 6. Inter-item.correlation matrix and average corre- lations (within type) for the diffuse opinion leadership pretest scale Average Item correlation number 9 10 12 15 17 21 24 30 within type 9 x .667 .657 .645 .784 .555 .546 .360 .633 10 x .491 .567 .691 .504 .528 .449 .601 12 x .617 .649 .658 .562 .575 .612 15 x .734 .638 .603 .682 .660 17 x .669 .600 .566 .698 21 x .728 .590 .694 24 x .575 .650 30 x .528 "power structure“ have to suggestions, opinions and advice from the individual. Items in Type II refer to the indi- vidual's general ability to influence or convince others. Items in Type II are most appropriate to the concep- tual definition of diffuse opinion leadership, while items in Type I are more specific in defining a particular group with which the individual wields influence. Further, items in Type II are worded to indicate diregt contact between the individual and others in the organization (e.g., item 24 states, "When interacting with other faculty members in the College, this person isn't a very good convincer.“), while items in Type I may be interpreted to mean either indirect or direct relationships (e.g., item 17: “Those 78 who have a lot to say about what goes on in the College respect the suggestions this person makes.“). However, because both types relate to general conceptual dimensions of opinion leadership, the most representative items from each type were retained for the main study but as separate variables. Based on the configuration and average corre- lations, items (9), (l7) and (15) were selected from Type I and items (21) and (24) from Type II. Sppgifig Opinion Leadprship The inter-item and average correlations for the Specific Opinion Leadership scale are presented in Table 7. Again, two typal structures emerged with the following configurations: 3 /1a< Type I: 14::i3l 22 Information-advice dependency Type II: l:§333 Asking behavior Inspection of the items in the Type I configuration suggest this set might be labeled "information-advice dependency“ and the items in Type II can be called “asking behavior.“ Conceptually, the second type might be assumed a sub-set of the Type I items, since both are consistent with the gener- al conceptual definition of opinion leadership. The Type II items are more specific to the direction of message transaction initiation, but still reflect an information- advice dependency. Based on the emergent typal structures and the aver- age correlations, items (3) and (22) were deleted from the 79 Table 7. Inter-item correlation matrix and average corre- lations (within Type I items) for the specific opinion leadership pretest scale Average cor- relation with- 33 in Type I Item number 1 3 14 18 22 31 1 x .269 .600 .321 .102 .562 .689 xxxx 3 x .367 .601 .287 .593 .182 .482 14 x .504 .095 .826 .670 .520 18 x .423 .736 .187 .578 22 x .283 .083 .296 31 x .530 .643 33 x xxxx scale, with all other items retained. Type I and Type II item sets will be treated as separate variables in addition to treating all five items as a summary measure of specific opinion leadership. pp;p,§ppgy Instrument Design The word anchors used for response alternatives of certain items in the main study Personal Contact question- naire were altered slightly from those used in the pretest instrument. Two pretest respondents pointed out that in the word-anchor set -- “agree strongly", “agree somewhat", and “agree slightly“ -- the last two response alternatives are near scale equivalents. While this problem may have contributed some measurement error in pretest data, it was assumed that the ordering of response alternatives and the 80 number-anchors preceding each response alternative partial- ly corrected this potential source of error. Based on a combined synthesis of the word-scale interval equivalence research reported by Dodd and Gerbrick (1960) and Jones and Thurstone (1955), the following word-anchors were selected for use with appropriate items in the main study instrument. Degree of feeling: Amount 8 Number: Degrees in general: Wprd-ancho; very strongly quite a bit moderately neither-nor a lot more a good deal more a little more about as ( ) very many more many more a few more the same number of extremely very fairly so-so (50-50) 0.80 s s H N N G Ital h'ltdl ::-: h'l O O O o c: <3 0 c: <3 0 O O lr-l h-Ir-l O O O O O O Interval weight (by inter- polation) Items in the main study Personal Contact questionnaire were first grouped for similarity of response alternative wording, and then randomly ordered within these groupings. CHAPTER IV FINDINGS Three weeks were required to complete pretest data collection, scale analysis and preparation of the 142 main study questionnaire packets. Data collection began during the second week of the last full month of the aca- demic year. The present chapter contains a summary of the data-collection procedures and response rates, the main study scale analysis, findings of the sociometric analy- sis, a description of the respondents and tests of the hypotheses. 2353 Collection The data-collection procedures described in Chapter II were followed with minor exceptions and consisted of (1) distribution of an entry letter1 from a faculty member in the College, (2) telephone contact with each person to enlist cooperation and make appointments for delivery and pickup of the questionnaire packet, (3) personal delivery of the packet to each member and (4) pickupof the packet at the agreed upon time. Only one of the 142 individuals refused to participate in the study when contacted by telephone and only 12 addi- tional individuals refused, or were unavailable, to complete 18cc Entry Letter, AppendiXAA. 81 82 any part of the questionnaire packet after receiving it. Unfortunately, three other studies had been conducted with faculty respondents in the College during the month pre- ceding this study. It was originally hoped that the entire study pOpula- tion could be contacted and questionnaires delivered within a three-day period. By the end of the third day, 88 per- cent of the population had been contacted by telephone, with 96 percent contacted by the end of the fourth day. The remaining respondents, all of whom were out-of-town during the initial days of data collection, were contacted by the end of the sixth working day. The daily cumulative total of packets delivered and returned is reported in Table 8. Every effort was made to obtain 100 percent completion of, at least, the Personal Contact Checklist and to keep respondents' lag time from delivery to pickup at a mini- mum. Whenever possible personal visits were made to res- pondents who missed the agreed upon pickup date to set a second pickup time. Of the tardy respondents who were not in fact refusals, most were extremely busy or out-of-town shortly after delivery of the questionnaire packet. Sever- al of these individuals turned out to be liaisons whose Personal Contact questionnaires were not used as a source of data in the study. Most of the 12 refusals came at the end of the tenth and fifteenth days of the data-collection period. Additionally, the investigator, who was officed 83 Table 8. Daily cumulative total of questionnaire packets delivered and returned Percent working Number Percent Number of number days delivered delivered returned* delivered 1 55 38.7 21 .38.1 2 92 64.7 50 54.3 3 112 78.8 70 62.5 4 124 87.3 86 69.3 5 137 96.4 165 76.6 6 139 97.8 109 78.4 10 140 98.5 120 85.7 15 141** 99.2 137 97.1 17 l39*** 98.5 * Including refusals ** One individual refused to accept a packet. *** Two individuals had not completed packets at the close of data collection. in the building, and two other informants heard little com- ment on the study among faculty during the data-collection period. The combination of these factors led the investi- gator to believe that although the data-collection period was extended several days beyond the time originally anti- cipated, it did not appear that questionnaires from tardy respondents would be invalid. From a total study population of 142, 129 individuals (90.8 percent) completed a Part I Personal Data questionnaire. 84 Part II Personal Contact Checklists were completed by 127 individuals (89.4 percent). Of the 15 non-respondents who refused or were unavailable to complete a Personal Contact Checklist, six were determined to be isolates in the socio- metric analysis, seven refused because they were “too busy“, or the questionnaires were “too personal“, “unethical“l, or “impossible to complete“, and two individuals simply did not complete their packets by the close of the data-collec- tion period. . Of the 127 respondents who completed Personal Contact Checklists, 96 (75.6 percent) listed at least one daily or more frequent contact. Respondents were asked to complete a Personal Contact questionnaire (PCQ) for each individual they named in this frequency of contact category. From ' among these 96 respondents who could have completed one or more PCQs, 86 (89.5 percent) completed at least one ques- tionnaire and ten did not complete any questionnaires on their daily contacts. Of the 86 respondents who completed at least one questionnaire, 78 (90.6 percent) completed questionnaires for all their daily or more frequent con- tacts. A potential of 33 PCQs were lost from the ten indi- viduals who did not complete any questionnaires on their daily contacts and an additional 13 PCQs were lost from the eight individuals who completed fewer questionnaires than 1Scales for the safety dimension of source credibility were especially singled out in this category by res- pondents. 85 the number of daily contacts they listed. The 96 respondents who each listed at least one daily contact reported a total of 270 contacts. A total of 224 Personal Contact questionnaires were completed, or a res- ponse rate of 82.9 percent. Sgale Analysis Only 32 of the 224 completed Personal Contact ques- tionnaires lacked one or more responses to questions within the instrument. The longest set of no responses occurred on one questionnaire in which the respondent skipped the ten source credibility scales plus nine other questions. Three questionnaires were not complete for the item esti- mating deliberate and directional message transaction ini- tiation frequency: all three were transformed to the “pro- file mean" of the person evaluated as described in Chapter II. Of the remaining 29 questionnaires, 14 required missing data transformations for only one question each, with most of the remainder requiring only two to five question trans- formations. The procedures specified in Chapter II were followed in all cases. The raw data were coded and punched into IBM cards for- ming a raw data deck. Next, negatively worded or coded items were reflected and repunched into a transformed data deck to facilitate scale construction procedures and so that all inter-item correlations would printout in the “true“ direction. 86 Following the scale analysis procedures used with the pretest instrument, scale inter-item Pearson product moment zero-order correlation coefficients were inspected for zero or negatively correlated items and McOuitty Cluster Analy- sis applied to the three five-item scales to assess scale dimensionality. All of the obtained correlations within each scale were significantly different from zero at the 0.001 level, two-alternative test, N=224 (critical value of r = 0.227). . The obtained inter-item correlations for the three- item Perceived Structural Diversity of Contacts scale were as follows: items 14, 191, r = 0.557: items 14, 21, r = 0.703: items 19, 21, r = 0.709. The correlation between the two items of the Perceived Number of Contacts scale (items 12, 17) was 0.699. The two-item First Source of Information scale (items 8, 20) correlated 0.683. Inter- item correlations among the three items of the Importance of Secondary Contacts scale were: items 10, 15, r = 0.762: items 10, 22, r = 0.729: and items 15, 22, r = 0.748. The inter-item correlation matrix for the Diffuse Opinion Leadership scale is presented in Table 9. Appli- cation of McQuitty Cluster Analysis to the matrix of cor- relations yields the following single type: 9-——\13 411.2318 <—l6 1Item numbers may be compared to item numbers in the Per- sonal Contact questionnaire, Appendix B, to identify the item. 87 Table 9. Inter-item correlation matrix for the diffuse opinion leadership main study scale Item number 9 ll 13 16 18 9 x .636 .683 .460 .618 11 x .736 .663 .780 13 x .560 .673 16 x .663 18 x‘ This typal structure differs from the pretest results where- in two types emerged with items ll, 16 and 18 forming Type I (persuasiveness within the “power structure“) and items 9 and 13 forming Type II (general persuasiveness with con- tacts). The questionnaires upon which these correlations are based were completed on respondent's contacts in two frequency of contact categories: ”once per day" and "sever- al times daily.“ Partitioning the data on these two fre- quency categories, on the assumption that so doing might provide an indication of the stability of the scale, yields the same single type from the "once per day“ frequency cat- egory (N=136). But the inter-item correlations from the "several times daily“ category (N=88) yields the following two types each of which contain the same items as those found in the pretest: Type I: 11:;E:18-:—-16 Persuasiveness within the "power structure" Type II: 9:55:13 General persuasiveness with con- tacts 88 The correlation matrix is presented in Table 10. Based on the results of this analysis and those of the pretest, the Diffuse Opinion Leadership scale items were treated as two variables in accord with the two emergent types from.the data of Table 10. The inter-item correlation matrix for the Specific Opinion Leadership scale is presented in Table 11. Clus- ter analysis of the matrix yields the following single type: 4—>2fise 3e 6 Although two types emerged from this scale in the pretest it was argued in Chapter III that the face validity of the items suggested one type to be a subset of the other, indi- cating a single general type. This contention is supported by the findings from the main study data. Even when the data are partitioned into the two frequency of contact cate- gories, a single type emerges from each separate matrix al- though with a slightly different configuration for the "several times daily” category. Based on these results, the five items were treated as a single variable, but, for purposes of cross-checking, the two types which emerged from the pretest analysis were tested separately. The inter-item correlation matrix for the ten scales of the safety and qualification dimensions of the source credibility concept are reported in Table 12. Cluster anal- ysis produces two typal structures, each of which contains exactly those scales originally input for each dimension: 89 Table 10. Inter-item correlation matrix for the diffuse opinion leadership scale based on “several times daily“ frequency of contact questionnaires Item number 9 11 13 16 18 9 x .597 .648 .453 .617 11 x .644 .623 .714 13 x .481 .562 16 x .642 18 x Table 11. Inter-item correlation matrix for the specific opinion leadership main study scale Item number 2 3 4 5 6 x .660 .647 .727 .533 x .588 .676 .556 x .557 .530 x .548 x 01$wa 9O x on see. x as new. «on. .x as ~Hs. new. sHN. x on son. ems. use. sad. x as one. How. Hem. «mm. ode. x we Nov. can. Hem. sew. eve. ems. x on man. ems. Nme. Hes. adv. one. see. x oh men. «as. Has. mes. Hem. sum. v~m. ace. x an ace. ems. Now. use. «on. new. Hem. com. «mm. x as as as as on as as on as as .4s senses sauH season huaaanaoeuu sounds on» How kahuna sodusaouwoo swaslwousH .~H manna 91 Type I : 71 : 7H w:— 78 <— 7F <—-7D Qualification dimen- sion Type II : 7J : 7E v—7C v— 7A Safety dimension 7G Each type was treated as a separate variable as originally anticipated. Spgiometrig Ana is The source of data for sociometric mapping of the com- munication structure of the organization was the Personal Contact Checklist (FCC), Part II of the questionnaire packet. A total of 895 contacts were listed by 127 members of the study population, or an approximate average of seven con- tacts per person (i = 7.047, s.d. = 3.441, range 0 - 16, median1 = 6.625). The total number of contacts reported within each frequency of contact category was: ”several times daily" - 102: "about once per day“ - 168: "2-3 times per week" - 271: and “about once per week“ - 354. The fre- quency distributions of reported contacts by frequency category are summarized in Table 13. In the study reported by Weiss and Jacobson (1955, p. 663) the average span of reported contacts per respondent was approximately 12, while the average in this study is approximately seven. The findings are not comparable, how- ever, because the Personal Contact Checklist utilized in 1Medians for contacts are calculated without midpoints since communication choices are discrete. 92 Rsuz 4.34.133 mmmmmvvsmmomfifimammm x33 Hem shoe HOOUGO Bets. «mm oooooooosmmmsflzmmag 3689. 00:0 e3.~ 2h ooooeoooaevmosmmmmmmsm smashed mass—mum emiwxosd ooooHHoHONm m3 2 R 8 Hm 336 0H9: no coco mmmémfl oooooooHoH~memsmmmmev s33 00:0 :6de oooooooooOemmoimmos aims Hohm>mm sum: muosusou as me as me NH as 0H m m s e m e m m H o unsusoo no deuce unaccommmu some an penned muomucoo no Hmaeoz anamoooum huommumo uomucoo mo hucmsomum an mucousoo omuuommu mo occausnfihumfin .MH manna 93 in the previous study included a more extended range of con- tact-frequency categories: their categories extended from “2-3 times per week" to "several times monthly“ and "sever- al times yearly“, whereas the upper limit in the present study was “about once per week.“ The 895 contacts reported by 127 respondents were dis- tributed among the 142 members of the study population. The approximate average number of gpppgpp geceived was six (32 s 6.301 , s.d. = 5.561, range 0 - 29, N = 142, median = 5.56). The grouped frequency distribution is: Number of choices received: 0-4:5-9:10-14:15-19:20-24:25-29 Number of members (N = 142): 62 56 16 1 2 5 Completion of the sociometric analysis to identify the necessary topological properties of the communication struc- ture required determining reciprocation of contact from the contacts reported in the Checklist. Since this topological analysis was accomplished in matrix form, the most efficient method of determining reciprocation was to begin with con- struction of the 142x142 matrix. During the period of data collection a preliminary matrix had been constructed in which respondents were originally listed sequentially ac- cording to the department or administrative unit in which they were officed. Groupings of individuals in the final matrix was accomplished by inspecting the locations of 1The mean number of choices received is definitionally less than the mean number of choices given because there were 15 non-respondents to the Checklist. If 100 percent of the study population had completed a checklist the two means would be equal. 94 tentatively identified reciprocated choices in the prelim- inary matrix and relisting individuals within the second matrix in order to bring each into adjacency with a major- ity of the other individuals with whom he had contact. The purpose of the reordering was to create clusters of reci- procated contacts around the diagonal as required for ma- trix analysis. The final matrix was plotted on a 34-inch square section of h-inch-square graph paper. Respondents' identification numbers were listed down the left hand edge of the sheet (for rows) and across the top of the sheet (for columns) in identical sequence beginning in the upper left hand cor- ner. The data contained in the Personal Contact Checklists were then transferred to the matrix by placing a number re- presenting a code for the frequency of contact in the ma- trix intersect cell of the respondent and his reported con- tact. This first transfer was made using the identifica- tion numbers along the side of the matrix (rows) for the respondent (the person making choices), and the identifica- tion numbers along the top of the matrix (columns) for the contacts listed (the persons chosen). At the completion of this step the matrix contained entries in 895 cells, re- presenting the total contacts listed on the Personal Contact Checklists. Next, to make entries in the matrix symmetric and to determine reciprocation in such a way as to have a record 95 of the frequency of contact listed by both members of a reciprocated dyad, the matrix was rotated counterclockwise one-quarter turn and the data from the Checklists again transferred to the matrix. Now, however, the identifica- tion numbers for the contacts listed (persons chosen) were along the bottom of the matrix: i.e., rows in Step 1 be- came columns in Step 2. At the completion of these two steps, those cells of the matrix which contained two entries represented reci- procated contacts. At this point there were a total of 400 cells with double entries indicating 400 reciprocated one- way communication linkages, or 200 two-way pairs of indivi- duals. The last step in construction of the matrix of reci- procated contacts was to assume reciprocation of contact for non-respondents based on their selection in the ”several times daily" or “about once per day" contact-frequency categories.1 The 15 non-respondents were indicated as daily contacts by 25 respondents (12 on a "several times daily" basis and 13 on a “once daily" basis). These 25 choices were assumed to be reciprocated, adding 50 reciprocated one-way contacts to the total, or 25 two-way pairs. The 1As reported earlier, Weiss and Jacobson (1955) found a close relation between reciprocation of contact and repor- ted importance of the listed contact. They indicate that 80 percent of the contacts listed as "several times daily" and of “utmost“ or “great“ importance were reciprocated while only 19 percent of the low frequency and low impor- tance contacts were reciprocated. In this study only 59 of the 270 daily or more often contacts (22%) were not reci- procated. 96 final total of reciprocated contacts used for topological analysis was 450 or 225 reciprocated dyads. The "best mea- sure"1 of the percentage of reciprocation was 50.63 percent. The approximate average number of reciprocated contacts per person was three (2 = 3.169, s.d. = 2.571, range 0 - 13, N = 142, median = 3.241, mode = 3). The grouped frequency distribution is: Number of members (N = 142) : 64 56 16 ‘4 2 A reproduction of the final working matrix2 is presented in Figure 2. The shaded cells represent reciprocated con- tacts, the small numbers are unreciprocated contacts (1 = several times daily, 4 = once weekly). Using only the reciprocated contacts in the final ma- trix, the topological analysis was accomplished following the procedures specified by Weiss (1956, pp. 88-108). The first step was to partition the large matrix into smaller segments by inspection. The segments were selected so mem- bers within the segment had a minimum of contacts outside the segment. By this procedure the large matrix was divided 1The "best measure“ is based on subtracting the 105 times non-respondents were chosen (all frequency categories) from the total number of choices made (895 - 105 = 790) and dividing this number into the total number of reciprocated contacts not including those assumed for non-respondents: i.e., 400/790 = 50.63 percent. 2The final working matrix reproduced in Figure II is actual- ly a 136x136 matrix, rather than 142x142. Six persons who were positively identified as isolates before construction of the matrix were deleted because isolates are deleted from the working matrix in the first steps of segment ana- lysis. 97 Figure 2. Reproduction of the matrix of reported com- munication contacts The shaded cells represent reciprocated contacts, the small numbers unreciprocated contacts. Boundaries of the four segments are outlined. llll'llllllllllllll' In." 98 into four smaller square matrices. Segment A contained 33 individuals, Segment B contained 28, Segment C had 31 and Segment D contained 44. The segments were copied on separate sheets of graph paper and each analyzed individually in order to isolate separate work groups (cliques). The basic procedure of segment analysis is to remove tentatively identified liaison persons from the segments until separate groups can be identified by inspection as a cluster of reciprocated contacts around the diagonal. The only permissible contact outside a group cluster (i.e., with individuals in other clusters) are single contacts between two individuals in two separate groups. These are bridge contacts in graph theory terminology. In Segment A the separation of groups in approximately three-quarters of the segment was not clear. It could not be determined by inspection if there were several small se- parate groups or one large, loosely connected group (a large number of bridge contacts and/or liaison contacts be- tween separate smaller groups would make the group struc- ture ambiguous for separation by inspection). At this point it was necessary to use the method of two-step chain analysis (matrix squaring) described by Weiss (1956, pp. 88- 98). This was the only time the two-step chain method was utilized because the group structures in the other three segments could be readily identified by inspection, probably due to the relatively small number of reciprocated contacts in the matrix. 99 The result of this procedure was a list of the member- ship of 29 separate work groups (cliques) and a tentative list of 27 liaison persons (those who had been removed from the segments or who were originally listed as tentative liaisons because they had two or more contacts outside of their segment, not counting contacts with individuals al- ready tentatively identified as liaisons). The problem was then to construct a sociogram of the total population and attempt to replace the tentatively identified liaisons to determine if they met all the criteria of the liaison role definition. Based on the information presented by Weiss and the conceptual meaning of an articulation point in graph theory, the following criteria were utilized for the ggppl identification of liaison persons: 1. To be considered a member of a separate group, a liai- son must have a majority of his contacts within the group, not counting contacts he has with other liaison persons} These are liaison gpppp members. 2. Not counting contacts he has with already identified liaison persons, a liaison who has membership in a se- parate group must have a minimum of two contacts outside his group. An exception to this rule is the individual who has contact outside of his group with only one other person in a separate group, but has one or more contacts 1Weiss suggests using high frequency (daily or more often) unreciprocated contacts of liaisons to clarify potential group membership. This procedure resulted in the assign- ment of only one liaison person to group membership. 100 with members of the liaison set. 3. A liaison,ipg;y;gggl is one who does not have member- ship in a separate group, but has contacts with persons in a minimum of two separate groups. These contacts may be with other liaisons pply if these liaisons have mem- bership within their respective groups (in this case, the liaison set may be treated as a separate group). 4. There will probably be a group of liaisons who cannot be characterized as having membership in any separate group, but who have all or nearly all their contacts with other liaisons, at least two of whom must be mem- bers of two separate groups. These are members 9; ppg liaison ppp. Liaisons peripheral to the liaison set may be characterized as members of the liaison set if they have a majority of their contacts within the set, or if they have contacts with a majority of the members of the set. 5. There may remain a group of individuals who cannot be characterized as members of a separate group and who have a majority of their contacts with liaison persons. If they cannot be characterized as liaison individuals (based on the matrix segment analysis) consider them non-liaison individuals. (Weiss' instructions on this point are not clear. To be consistent with the graph theory definition of an articulation point, these per- sons should not be treated as liaisons because they are not associated with separate groups.) Two non-liaison 101 individuals were found, each linked to a separate liaison. 6. A non-liaisonlgpppp member may have no more than one contact outside his own group (except with liaisons) and must have a majority of his contacts within the group. These criteria which were especially appropriate to the ana- lysis of this study population represent only an essential summary of the procedures described by Weiss. The complete instructions contain other criteria for treatment of special cases not encountered in the present study, and a more com- plete description of the operations necessary in each step of the analysis. The final sociogram of the communication structure of the organization should be such that if all of the liaisons are removed from the sociogram, the work groups will separate into individual clusters except for bridge contacts between groups. The final test of a liaison person's identifica- tion should be as follows: Ignoring bridge contacts between groups to which a liaison is connected and ignoring other liaison contacts among those groups and treating the liaison set as a separate group, if the liaison is removed from the sociogram the groups in question should separate. If the liaison has membership in a separate group, when he and only he is removed his group should separate from the other groups to which he is connected leaving no connections among the groups in question (exceptions noted above). If either one of two connected persons could be removed to separate 1‘ I'll 1'. ll... ‘1 1.!- 5‘ ‘l 102 any two groups, then this is an improper solution: this is a bridge contact. The sociogram of the communication structure of the study population is reproduced in Figure 3. The end result of the topological analysis was the identification of 22 liaison persons (15.49 percent of the study population), 18 isolates (12.68 percent) and 102 non-liaison persons (71.83 percent), 100 of whom were members of 29 separate groups of varying size. There were two non-liaison individuals. The addition of the liaison set as a separate group brings the total number of sociometric groups to 30. Of the 22 iden- tified liaison individuals, five were members of the liaison set, six were liaison individuals and the remaining 11 were liaison group members. The range in group size (including liaison group mem- bers in their respective groups) was from two to eight mem- bers: Group size : 2 :_§_: 3 : §,: § : 1 : § Number of groups: 7 JJ. 3 1 4 1 2 N = 29 Sociometric and demographic characteristics of indi- viduals in the tOpological categories are reported in the following section. Charagteristigs pg the Resppndents Sociometric data on work-related communication contacts from the 127 Personal Contact Checklists and demographic information from 129 members of the study population 103 .nuomucoo caduceucofifioo :euuo duos no thGuS ucenmwmow masope>soce mseuooccoo mused .oopoaoofi no: one anMHOna ma och scausuecsmwo on» no esopuosum GOaDMOHGSEEoo esu no Emumo«oom .m choose «a / \ \\_/ r . \\x// 4“ \ t ’1“, E MKIVQ Am. J b ‘ IV inky/«\W‘flflfi \ 4’, .ll.fl\|I.// \xlul/ ’ AMWMWWW\~ (Mn, / /!la //“\ .33. an: dew \1 h... 3m I {III II In lrllllrr l {I I 104 on the Personal Data questionnaire can be utilized to char- acterize the liaisons, non-liaisons and isolates. These data are presented as descriptive parameters of the study population because the universe of professional organiza- tional members, non-respondents excepted, are represented. The value of the data is in describing certain character- istics of the population and for later use as comparative statistics for other empirical case studies1 conducted in similar, or dissimilar, organizations; Sociometric Description p§_§ppgy Population Partitioning the reported contacts according to the three topological categories reveals that liaison persons listed nearly twice as many ppppl contacts as non-liaisons and slightly more than twice as many as the isolates. A summary of the reported contacts by classification and fre- quency category is presented in Table 14.2 Regarding the number of phoices received, the mean choices directed toward liaison individuals was 12.14, for non-liaisons 5.83 and for isolates 1.83. The liaisons 1Evans and Leppmam (1967, p. xviii) term studies like the present one where all members of a “real life" organiza- tion serve as the sources of data (saturation sampling) as “research case histories.“ The "history“ terminology is more appropriate to their study of the introduction of an innovation in a university over a period of two years. 2Data presented in Table 14 are based on the 127 Personal Contact Checklists: data in Table 15 are based on the 129 Personal Data questionnaires. Data on choices received and reciprocated contacts (Figures 4 and 5) were tabulated on all 142 members of the study population. 105 h.~ H.H v.n Now mm mm mm ha muomucoo # Ameuz. neusHOnH ¢.m 5.H H.H O.H umvum N.m h.m m.H moN mod ood 5.0 m. Avmuzv meanamdalcoz men n.m mvm v.m 66H ~.m e a oew m.H mes m.~ om m.m on m.H Amwmmwsoo a .e.m hoad How m.v mom ®.m mvN 05H mm mu mm mm Om gowns xamuzc accessed once so waxed: coco 0kg .HO sexes: mum once so Sauce 8 1. see: Mom coco xee3 Hem mum hot mom 00:0 sense Hmue>om Nwomousm aocmsomwm mmusHOms use accessed loos .n:0mesaa he muomucoo scausoecseeoo ounuoaou mo Henson use no hwseeom .ea edema 106 received an average of slightly more than twice as many choices as did the non-liaisons. The total choices re- ceived by 22 liaisons was 267 and the total received by 102 non-liaisons was 595. The 18 isolates received a total of 33 choices. The raw score frequency polygon for groups appears in Figure 4. It should be noted that the heights of the curves are not comparable due to the differ- ences in the number of individuals in each category. Excluding three non-liaison individuals who received extreme choices, the general range of choices for non-lia- isons is 0 to 12 (the non-liaison who received 29 choices was a non-respondent who was chosen only twice in “once per day" frequency category, thus had only two assumed reciprocated contacts). No liaison individual received less than five choices, while 45.1 percent of the non- liaisons received less than five choices. The number of reciprocated gontacts of liaisons and non-liaisons is charted in Figure 5. The mean number of reciprocated contacts for liaisons was 7.14 (range 2 to 13). The mean for non-liaisons was 2.87 (range 1 to 7). Iso- lates are not included in Figure 5 but do appear in other sociometric and demographic tables because the operational definition of an isolate is the absence of reciprocated contacts based on work-related content of a minimum fre- quency of once per week. These persons are "isolated“ only within the limits of this operational definition. 107 meusHOna one moanasualcoc .mcomasea an om>smumu meadono use How soo>H0d unconvenm .v muomsm om>amumu nmoHOSU Mo uwnEoz mm . mm mm . Hm ON b,mH vH hrNH HA OH m m h m m w m N H o 171.. i «mauzv mousaomH ..... , \\ Amofinzv ncOmHMHHIcoz.II:II. .4 Ammuzv accessed O‘CDI‘VOLGV'MNHO meoeono mcabaeomw muonEmE mo Menace 108 Liaisons (N=22) 29 - [:3 Non-liaisons (N=102) Number 25 r_ of respon- 20 F dents 15 ' 10 5 o E Number of reciprocated contacts Figure 5. Frequency histogram of the reciprocated communi- cation contacts of liaison and non-liaison indi- viduals Demographig Characteristics pg ppplgppgy Population A summary of selected personal characteristics of in- dividuals in the three topological categories is presented in Tables 15 and 15-A. This information is based on the 129 Personal Data questionnaires.1 Comparison of the average ages of individuals in the three categories revealed slight differences, with liaisons averaging approximately four years older than non-liaisons and isolates eight years older than liaisons. Differences 1No-response transformations in the Personal Data ques- tionnaires were required for only two respondents, one question each. One respondent neglected to list his age: the missing data was coded as the mean year of the re- maining responses. The other respondent did not report the estimated percentage of time devoted to teaching, re- search, etc. Based on this person's administrative title, it was assumed he spent 100 percent of his time in admin- istrative work. 109 Table 15. Characteristics of members of the study population by type M W 15.9.1923 a. Mean a. Mean a. Mean b. St. dev. b. St. dev. b. St. dev. Variable c. Range c. Range c. Range (N=21) (N=96) (N=12) Age a. 46.24 a. 42.33 a. 54.42 b. 10.26 b. 10.04 b. 16.76 c. 29-66 c. 26-68 c. 24-77 Number of years a. 9.90 a. 7.34 a. 12.67 at university b. 6.43 b. 7.29 b. 9.25 c. 2-24 c. 1-31 c. 2-25 Percentage of time allotted: 1. Teaching a. 26.52 a. 47.13 a. 39.00 b. 33.02 b. 43.60 b. 37.25 c. 0-100 c. 0-100 c. 0-100 2. Research a. 15.00 a. 22.17 a. 26.00 b. 29.30 b. 25.97 b. 31.11 3. Consulting a. 6.95 a. 7.32 a. 9.58 b. 7.34 b. 13.61 b. 14.69 4. Administrative a. 41.14 a. 16.80 a. 23.33 duty b. 37.31 b. 29.60 b. 37.19 5. Committee a. 10.52 a. 5.54 a. 2.08 work b. 22.38 b. 8.37 b. 4.50 c. 0-100 c. 0-50 c. 0-15 Number of committee memberships: 1. Departmental a. 1.95 a. 1.61 a. 0.75 level b. 1.66 b. 1.59 b. 0.87 c. 0-5 c. 0-7 c. 0-2 2. College a. 1.05 a. 0.61 a. 0.25 level b. 1.53 b. 0.81 b. 0.62 Ce 0-7 c. 0-3 es 0-2 3. University a. 0.95 a. 0.32 a. 0.33 level b. 1.20 b. 0.24 b. 0.26 c. 0-4 c. 0-5 c. 0-2 11 0 Table 15. Characteristics of members of the study popula- tion by type (continued) Liaisons Non-liaisons Isolates a. Mean a. Mean a. Mean b. St. dev. b. St. dev. b. St. dev. Variable c. Range c. Range c. Range Total number of committee memberships a. 3.95 a. 2.55 a. 1.33 b. 2.65 b. 2.34 b. 1.83 c. 0-11 c. 0-10 c. 0-5 Number of committee meetings in a typical month a. 7.38 a. 4.41 a. 2.50 b. 4.95 b. 5.09 b. 2.84 c. 0-15 c. 0-30 c. 0-10 Number of articles published or read in past two years a. 3.62 a. 4.00 a. 3.25 b. 2.06 b. 4.66 b. 4.43 c. 0-8 c. 0-25 c. 0-15 Number of books published in past five years a. 0.81 a. 0.97 a. 0.58 b. 1.53 b. 1.85 b. 0.79 c. 0-5 as 0-9 Ce 0-2 Percent of time allotted in College a. 99.76 a. 96.73 a.100.00 b. 1.91 b. 11.74 b. ----- (2.95-1.00 Ce33"100 Ce ----- 111 Table 15-A. Characteristics of members of the study popu- lation by type Liaisons Non-liaisons Isolates Variable N % N x N % Academic rank 1. Instructor, lecturer, specialist l 4.76 17 17.71 3 25.00 2. Assistant professor 4 19.05 25 26.04 0 ---- 3. Associate - professor 2 9.52 25 26.04 2 16.67 4. Professor 14 66.67 29 30.21 6 50.00 5. None 0 ---- 0 ---- 1 8,33 100.00 100.00 100.00 Administrative title 1. Head or assis- tant head of academic or research unit 10 47.62 11 11.46 2 16.67 2. Head or assis- tant head of special units or projects 4 19.05 22 22.92 3 25.00 3. None 7 33,33 63 65.63 7 58.33 100.00 100.01* 100.00 Highest earned degree 1. Doctorate 19 90.48 81 84.38 7 58.33 2. Masters 2 9.52 13 13.54 4 33.33 3. Bachelors 0 ---- 2 2.08 1 8,33 100.00 100.00 99.99* Joint appointments 1. None 17 80.95 82 85.42 10 83.33 2. Within College 1 4.76 3 3.13 1 8.33 3. With other College 3 14.29 11 11,46 1 8,33 100.00 100.01* 99.99* Appointment basis 1. Twelve months 16 76.19 63 65.63 7 58.33 2. Nine months 5 23.81 33 34.38 4 33.33 3. Less than nine 0 ---- 0 ---- 1 8.33 100.00 100.01* 99.99* 112 Table 15-A. Characteristics of members of the study popu- lation by type (continued) Liaisons Non-liaisons Isolates Variable N ‘% N % N % Sex 1. Male 18 85.71 82 85.42 12 100.00 2. Female 3 14.29 14 14,58 0 ---- 100.00 100.00 100.00 * Rounding error in the number of years employed by the university were also minor and ordered the same as age: liaisons had an average of approximately 28 more years of service at the institution than non-liaisons, and isolates had approxi- mately the same interval more tenure than liaisons. Exactly two-thirds of the liaison persons held aca- demic rank as full professor, while approximately one-third of the non-liaisons were full professors. Considerably larger percentages of non-liaisons held rank as instructors and associate professors than did liaisons. Differences between the liaisons and non-liaisons in highest earned degree were slight with a minor tendency for more doctor- ates among liaisons and more masters degrees among non- liaisons. Isolates, however, showed higher percentages in the masters degree category in comparison to liaisons and non-liaisons. Approximately three-fourths of the liaisons were 113 employed on a twelve month basis (vs. nine months) with two-thirds of the non-liaisons and three-fifths of the isolates on twelve month appointment. On one common professional evaluative measure -- publication -- there were only slight differences among the three types, with non-liaisons having the highest book and article publication average and isolates the smallest average. In the estimates of the percentage of time devoted to several major activities typical of faculty, non-liaisons reported nearly double the average of liaisons in time devoted to teaching, and a slightly higher average in time devoted to research activities. There were minor differ- ences among the types in time devoted to consulting. In estimated time devoted to administrative duties, liaisons reported an average nearly 2.5 times the non- liaison average. The average time devoted to administrative duties by liaisons (41 percent) approached equaling the average time non-liaisons devoted to teaching (47 percent). The stronger representation of administrators in the liai- son group is also revealed by the percentages holding ad- ministrative titles -- two-thirds of the liaisons held for- mal administrative titles (including major titles, assis- tant unit heads, administrative assistants, and directors of special projects), while only one-third of the non-lia- isons held similar titles. Surprisingly, a little over 40 percent of the isolates reported holding an administrative 114 title and devoting an average of nearly one-fourth of their time to administrative duties. Membership in committees which may cut across clique group and formal unit lines would appear to put an indi- vidual in a structural position from.which liaison-type communication contacts could develop. As anticipated lia- isons reported devoting an average of nearly twice as much time to committee work as did non-liaisons (11 percent vs. 6 percent) and approximately five times as much as isolates. In terms of the average number of committees in which they had membership, liaisons reported about one and one-half as many £2531 committees as non-liaisons, and three times as many as isolates. Comparing only liaisons and non-lia- isons, the tendency was for liaisons to belong to an in- creasing number of committees as the administrative level of the committees increased: i.e., liaisons reported roughly the same average number of memberships on depart- mental level committees, slightly less than twice the aver- age of non-liaisons on college level committees, and on the average of three times as many university level committees. The greater involvement of liaisons in committee work is also verified by the estimated number of committee meet- ings attended in a “typical" month -- nearly double the average number reported by non-liaisons and three times that reported by isolates. The remaining variables reported in Table 15, namely sex, joint appointments and the related percentage of time 115 allotted in the college, did not reveal any noteworthy differences among the three topological categories. Selection pf ppp Samples Following the procedures specified in Chapter II, all of the liaison individuals on whom Personal Contact ques- tionnaires had been completed by their reciprocated non- liaison contacts constituted the liaison person sample. Of the 22 identified liaisons, 17 met this criterion (three of the five liaisons who dropped out of the sample were members of the liaison set and were the tOp level ad- ministrators of the College). A total of 30 Personal Con- tact questionnaires1 were completed on the 17 liaison per- sons by 26 of their reciprocated non-liaison contacts. These 30 questionnaires served as the source of data for the tests of the hypotheses. Of the 17 liaisons, Personal Data questionnaires were available on 16 of them and served as the source of data for the description of the sample to follow. 1Of the total of 224 Personal Contact questionnaires com- pleted by members of the study population, 166 were com- pleted by at least one member of a reciprocated dyad (72 of the 166 were completed by reflex pairs: i.e., both mem- bers of the dyad). Of the 166, 30 were completed by non- liaisons on liaisons, 45 were completed by liaisons on non-liaisons, 18 were completed by liaisons on liaisons, and 73 were completed by non-liaisons on non-liaisons. Of the latter group, 28 questionnaires were completed by non-liaisons who were sources of data in the liaison sam- ple and, to maintain independence of data sources between the two samples, these questionnaires were deleted from possible selection into the non-liaison sample. 116 Procedure required selecting 17 non-liaison persons on a random basis and utilizing Personal Contact question- naires completed on them by their reciprocated non-liaison contacts as the source of data for the tests of the hypo- theses. The first step in selection of the sample consisted of identifying all the non-liaison persons in the study pop- ulation on whom Personal Contact questionnaires had been completed by other non-liaisons who were their reciprocated contacts. Then these reciprocated dyads were listed, de- leting questionnaires completed by reciprocated non-liaison contacts who had served as sources of data for individuals in the liaison sample. This resulted in the final identifi- cation of 38 non-liaisons on whom a total of 45 Personal Contact questionnaires had been completed by reciprocated non-liaison contacts. Each of the 38 individuals was as- signed an identification number and a table of random num- bers used to select the 17-member non-liaison sample. In order to maintain independence of source data within the sample (as described in Chapter II), whenever both members of a reflex pair1 were selected into the sample, a random procedure was used to delete one source respondent of the reflex pair. If this procedure deleted a non-liaison on 1A reflex pair is defined as the instance when both mem- bers of a reciprocated dyad complete a Personal Contact questionnaire on the other member. For example, given the_dyad A, B: if A completes a questionnaire on B and B completes a questionnaire on A, the dyad is termed a reflex pair. 117 whom only one questionnaire had been completed, an addition- al random number was selected to replace this person in order to reach the predetermined total of 17 non-liaisons in the sample. The result of these procedures yielded a total of 21 Personal Contact questionnaires completed on the 17 ran- domly selected non-liaisons by 16 of their reciprocated non-liaison contacts. These 21 questionnaires served as the source of data for the tests of hypotheses. Of the 17 individuals in the non-liaison sample, Personal Data ques- tionnaires were completed by 14 of them and these question- naires became the source of data for the description of the sample to follow. Description 9; ppp Samples A summary of selected characteristics of individuals in the liaison and non-liai- son samples is presented in Tables 16 and 16-A. Of 23 var- iables reported in the tables, only three indicate statis- tically significant differences1 between the two samples. Liaison individuals reported a higher mean number of com- mittee meetings attended in a typical month than did non- liaisons (.05 level of significance, two alternative “t“ 1Statistical tests of differences in this section and in the hypothesis testing section are applied in a heuristic sense to attempt to provide an approximation of potential para- meters. Two problems plague the application of statistical tests to the data of this study: (1) the study population from which the samples were taken is not randomly drawn and the generalizability of the findings must depend on future comparative studies to determine the representativeness of the pOpulation: (2) the "sample“ of liaison persons is not a random sample hence violates a basic assumption of the statistical tests. 118 Table 16. Comparison of selected characteristics of the liaison and non-liaison samples Liaison Non-liaison Mean sample sample difference a. Mean a. Mean tests: “t“ Variable b. St. dev. b. St. dev. value Age a. 45.75 a. 43.36 0.5752 b. 11.05 b. 11.72 Number of years at a. 10.25 a. 9.57 0.2450 university b. 7.02 b. 8.16 Percentage of time allotted: 1. Teaching a. 27.94 a. 42.86 1.2786 b. 30.58 b. 33.32 2. Research a. 18.63 a. 27.14 0.7686 b. 32.86 b. 27.01 3. Consulting a. 6.37 a. 13.21 1.0107 b. 7.50 b. 25.91 4. Administrative a. 33.69 a. 13.93 1.6659 duty b. 35.22 b. 28.83 5. Committee work a. 13.38 a. 2.85 1.5307 b. 25.08 b. 5.71 Number of committee memberships: 1. Departmental a. 2.31 a. 2.00 0.4832 level b. 1.62 b. 1.92 2. College level a. 1.31 a. 0.79 1.0385 b. 1.66 b. 0.97 3. University a. 0.81 a. 0.64 0.3630 level b. 1.22 b. 1.34 4. Total a. 4.44 a. 3.43 0.8856 committees b. 2.85 b. 3.39 Number of committee meetings in a a. 7.69 a. 4.07 2.1816* typical month b. 4.71 b. 4.29 119 Table 16. Comparison of selected characteristics of the liaison and non-liaison samples (continued) Liaison Non-liaison Mean sample sample difference a. Mean a. Mean tests: “t" Variable b. St. dev. b. St. dev. value Number of articles published or read a. 3.56 a. 6.07 1.3437 in past two years b. 2.19 b. 7.11 Number of books published in past a. 1.00 a. 0.93 0.0947 five years b. 1.71 b. 2.40 Percentage of time allotted in the a. 99.69 a. 98.07 0.9537 College b. 1.25 b. 6.66 Total number of contacts reported on Personal Contact a. 11.69 a. 7.28 3.1971** Checklist b. 4.01 b. 3.45 Number of recipro- a. 7.29 a. 3.53 4.3927*** cated contacts b. 3.18 b. 1.55 Liaison sample N Non-liaison sample N 14 * Significantly different means, 0.05 level, native test, df=28 ** Significantly different means, 0.01 level, native test, df=28 two alter- two alter- *** Significantly different means, 0.001 level, two alter- native test, critical value of "t" computed by Cochran and Cox formula = 4.0143 was significant at the 0.02 level, two alternative test: obtained F = 4.2246: critical F Liaison N=l7, Non-liaison N=17. (the F ratio of variances (16,16) 002:3.37) 3 120 Table 16-A. Comparison of selected characteristics of the liaison and non-liaison samples Liaison Non-liaison Differenc sample sample tests: X Variable N % N % value Academic rank 1. Instructor 1 6.25 3 21.43 2. Assistant professor 2 12.50 1 7.143 3. Associate professor 2 12.50 4 28.57 4. Profgssor 11 68.75 6 42.86 1. Less than I professor 5 31.25 8 57.14 0.9843 2. Professor 11 68.75 6 42.86 n.s.* Administrative title 1. Have title 10 62.50 5 35.71 1.0588 2. No title 6 37.50 9 64.29 n.s.* Highest earned degree 1. Doctorate 15 93.75 11 78.57 n.s.** 2. Masters 1 6.25 3 21.43 Joint appointment 1. None 13 81.25 12 85.71 n.s.** 2. Within College 1 6.25 0 3. With other College 2 12.50 2 14.29 Appointment basis 1. Twelve months 12 75.00 12 85.71 n.s.** 2. Nine months 4 25.00 2 14.29 Sex 1. Male 14 87.50 11 78.57 n.s.** 2. Female 2 12.50 3 21.43 * x2 values calculated with Yates correction, df=1. ** X2 values were not calculated due to expected frequencies The obtained dif- ferences between types are less extreme than the two of less than 5 in one or more cells. tested and would not be significant. The different percentage in this cell compared to Table 15 appears to be an aberration of random selection made more likely by the larger number of categories under this heading. 121 test, df=28). Comparing the mean number of contacts repor- ted by members of each type on the original Personal Con- tact Checklists, liaisons had listed a significantly higher number of contacts than had the non-liaisons (. 01 level of significance). Comparing the number of reciprocated contacts, liaisons had a significantly higher mean number than non-liaisons (.001 level, two alternative test, df=l6 and 16 for the Cochran and Cox corrected "t“ value). Al- though the differences on a number of other variables appear substantially large, the relatively small sample size and, in many cases, large standard deviations reduced the possibility of finding statistically significant dif- ferences between the samples. With the exception of the three variables cited above (and at appropriate confidence levels), differences are within the expectations of random error and the samples can be assumed relatively homogen- eous on the reported variables. Tests pg Hypptheses Findings dealing with the tests of hypotheses are presented under each major variable category. In each case the method of evaluating the hypothesis was to compute the “t“ test for independent sample means with unequal numbers in each sample (McNemar, 1962, pp. 102-3). The assumption of homogeneity of variance between populations was tested for each variable by computation of the F ratio (Walker and Lev, 1953. p. 186: Downie and Heath, 1965, p. 141). 122 Awareness p; Liaison gplpg Two hypotheses were postulated regarding awareness by non-liaison persons of the actual structural diversity and span of contacts of their respective reciprocated liaison and non-liaison contacts. H1: Liaison persons are perceived to have more struc- turally diverse communication contacts in the organ- ization than are non-liaison persons. The variable was measured by a three item scale with a po- tential summed range from 3 to 21. The findings are as follows: Liaison Non-liaison sample sample “t" value F ratio Mean scale score 15.67 12.81 2.4401 Sample standard deviation 3.59 4.77 1.7618 Obtained range of scores 8-21 5-20 The obtained difference between the sample means is signi- ficant beyond the 0.05 level (two alternative test, df=49, critical value of "t" = 2.0116): the hypothesis is con-' firmed. The obtained value of F does not reach signifi- cance at the 0.05 level (critical value of F(20 29)=2.24, I two alternative test): the assumption of homogeneity of variance is not rejected. H2: Liaison persons are perceived to have more communi- cation contacts in the organization than are non- liaison persons. This variable was measured with a two item scale with a potential summed range from 2 to 14. The findings: 123 Liaison Non-liaison sample sample “t" value F ratio Mean scale score 10.80 8.76 2.3114 Standard deviation 2.61 3.70 2.0174 Obtained range of scores 5-14 2-14 The obtained difference between the sample means is signi- ficant beyond the 0.05 level (two alternative test, de49): the hypothesis is cpnfirmed. The obtained value of F(20'29) does not reach significance at the 0.05 level. Interpersonal Communigation Behavior Two hypotheses posited differences between liaison -- non-liaison dyads and non-liaison -- non-liaison dyads in terms of the frequency and directional ratio of deliberate message transaction initiation. H3: Liaison -- non-liaison dyads more frequently parti- cipate in deliberately initiated message transac- tions than do non-liaison -- non-liaison dyads. The variable was measured by respondents' direct estimates of the frequency of deliberate initiation of contact out of a hypothetical 100 contact occasions. The estimates could vary from 0 to 100. The findings: Liaison-- Non-liaison-- Non-liai- non-liaison son dyad dyad sample sample "t" value F ratio Mean frequency estimate 87.17 90.71 0.7870 Standard deviation 17.60 12.87 1.8695 Obtained range 40-100 60-100 124 The obtained mean frequency estimates are in the opposite direction from predicted, but the difference does not reach significance at the 0.05 level (two alternative test, df= 49): the hypothesis is not supported. The F value does not reach significance at the 0.05 level (critical value of 2.358, two alternative test). F(29.20)= H4: The directionality of deliberate message transaction initiation is more disproportionate in liaison -- non-liaison dyads than in non-liaison -- non-liaison dyads. ' The measure of this variable was respondents' estimates of the frequency of respondent deliberate initiation of con- tact compared to his contact's initiation of contact. The ratio of the larger frequency to the smaller frequency provides a measure of the proportionality of contact ini- tiation in the dyad. The more this ratio deviates from one, the more disproportionate is the directionality of message transaction initiation. Theoretically the ratios could vary from 0.00 to 1.00. The findings are: Liaison-- Non-liaison—- Non-liaison non-liaison dyad dyad sample sample "t" value F ratip Mean directional ratio 0.7529 0.7035 0.6161 Standard deviation 0.2851 0.2772 1.0581 Obtained range 0.111- 0.111- 1.000 1.000 The obtained mean directional ratios are in the opposite direction from predicted, but the difference does not reach 125 significance at the 0.05 level (two alternative test, df= 49): the hypothesis is not supported. The obtained value of F(29.20)is not significant at the 0.05 level. Information Rplpy Fungtion A single hypothesis in this category predicted dif- ferences between liaison persons and non-liaison persons in information relay functions, specifically in terms of serving as a first source of information. H5: Liaison persons are more likely to serve as first sources of organization-related information than are non-liaison persons. This variable was measured as a two item scale with a po- tential summed range from 2 to 14. The findings: Liaison Non-liaison sample sample “t“ value F ratio Mean scale score 9.33 6.62 3.2835 Standard deviation 2.90 2.91 1.0013 Obtained range 2-14 2-13 The obtained difference between the sample means is signi- ficant beyond the 0.01 level (two alternative test, df=49, critical value of "t“: 2.684): the hypothesis is confirmed. The obtained value of F is not significant at the (20,29) 0.05 level. Perceived Personal Attributes The variable of interest in this category was source credibility, more precisely, the qualification and safety dimensions of that concept. 126 H6: Liaison persons are perceived to have higher source credibility (qualification and safety dimensions) than are non-liaison persons. Each dimension was measured with a set of five semantic differential scales with the possible summed range for each being 5 to 35. The findings for the qualification dimen- sion are: Liaison Non-liaison - sample sample "t" value F ratio Mean scale ' score 30.97 31.62 0.6102 Standard deviation 4.54 2.16 4.4440 Obtained range 14-35 28-35 The obtained sample means are in the Opposite direction from predicted, but the difference does not reach signifi- cance at the 0.05 level (two alternative test, df=49): the hypothesis is not supported. The obtained value of F is significant at the 0.02 level (two alternative test, df= 29,20) indicating that the population variances were not homogeneous. However, it was unnecessary to compute a cor- rected critical value of "t" by the Cochran and Cox for- mula (Downie and Heath, 1965, pp. 143-4) since the obtained value of “t" was not significant at the tabled critical value and the adjusted critical value is larger than the tabled value. The findings for the safety dimension are as fol- lows: 127 Liaison Non-liaison sample sample "t“ value F ratio Mean scale score 31.87 30.76 1.1555 Standard deviation 3.3604 3.3601 1.0001 Obtained range 20-35 21-35 The obtained difference between sample means does not reach significance at the 0.05 level (two alternative test, df= 49): the hypothesis is not supported. The obtained value of F is not significant. (29,20) Influence Potential Three original hypotheses predicted differences between liaisons and non-liaisons in terms of the perceived impor- tance of their secondary communication contacts, diffuse opinion leadership, and specific Opinion leadership. How- ever, findings from the scale analysis of both the pretest and main study indicated the Diffuse Opinion Leadership scale had two dimensions, leading to two corollary hypo- theses in place Of the original. H7: Liaison persons are perceived to have more important secondary contacts in the organization than are non- liaison persons. This variable was measured by a three item scale having a potential summed range from 3 to 21. The findings: Liaison Non-liaison sample sample "t" value F ratio Mean scale score 17.63 14.29 3.5336 Standard deviation 3.31 3.36 1.0352 Obtained range 9-21 10-21 128 The obtained difference between sample means is signifi- cant beyond the 0.001 level (two alternative test, df=49, critical value of "t" = 3.510): the hypothesis is confirmed. The F(20'29)value does not reach significance at the 0.05 level. Corollary H8A: Liaison persons are perceived to have more influence over members of the "power structure” than are non-liaison persons. This variable is one Of two dimensions of the original Diffuse Opinion Leadership scale. Its measure consisted of a three item sub-scale with a potential summed range from 3 to 21. The findings are: Liaison Non-liaison sample sample "t" value F ratio Mean scale score 16.03 13.33 2.4728 Standard deviation 3.81 3.88 1.0361 Obtained ' range 7-21 6-20 The obtained difference between sample means is significant beyond the 0.05 level (two alternative test, df=49): the corollary hypothesis is confirmed. The obtained F(20 29) I is not significant. Corollary H88: Liaison persons are perceived to be more generally persuasive across all their communication contacts than are non- liaison persons. This variable is the second of the two dimensions of the original Diffuse Opinion Leadership scale and was measured with a two item sub-scale having a potential range from 2 to 14. The findings are: 129 Liaison Non-liaison sample sample "t“ value F ratio Mean scale score 11.67 10.67 1.4721 Standard deviation 2.43 2.33 1.0831 Obtained range 4-14 6-14 The obtained difference between sample means does not reach significance at the 0.05 level (two alternative test, df= 49): the corollary hypothesis is not supported. The ob- tained F(29’20) ratio is not significant. The original five item Diffuse Opinion Leadership scale was also submitted to evaluation. The obtained mean scale score for liaisons was 27.70 and for non-liaisons the mean was 24.00. The obtained "t" value was 2.3134 which is significant at the 0.05 level, but untenable in support of the original hypothesis since the scale actually contained two dimensions. H9: Liaison persons are more likely to be perceived as personal opinion leaders for their dyadic contacts than are non-liaison persons (specific Opinion leadership). The specific Opinion leadership variable was measured by a five item scale with a potential summed range from 5 to 35. The findings are as follows: Liaison Non-liaison sample sample ”t" value F ratio Mean scale score 23.30 21.19 1.4408 Standard deviation 4.62 5.82 1.5831 Obtained range 12-30 13-35 130 The obtained difference between sample means does not reach significance at the 0.05 level (two alternative test, df= 49): the hypothesis is not supported. The obtained value of F(20'29) does not reach significance at the 0.05 level. In the pretest scale analysis the Specific Opinion Leadership scale emerged as having two dimensions, although these results were not obtained in the main study scale analysis. To cross-check, however, each sub-scale was tested. On the "information-advice dependency“ dimension, the obtained mean for the liaison sample was 14.13 and the mean for non-liaisons was 12.76. The “t" value was 1.3853 which does not reach significance at the 0.05 level. For the "asking behavior" dimension of the scale, the Obtained mean for liaisons was 9.17 and for non-liaisons 8.43 with a "t" value of 1.2376 which also does not reach significance at the 0.05 level. CHAPTER V CONCLUSION The basic objectives of the present work were (1) to develop a conceptual and methodological framework for study of the extant functional communication structure of a formalized social system and (2) to apply this frame- work to an empirical examination of phenomenological at- tributes associated with certain topological features of the communication structure. The present chapter contains a summary and discussion Of the findings of the study, concluding with suggestions for future research. Summary Most previous communication research in formal organ- izations has focused on aspects of the formal structure with particular emphasis on superior-subordinate relations. The intent of this study was to map the extant functional communication structure of an organization including both formal and informal aspects, then to differentiate this sociometric map into two structural types based on topolo- gical concepts from graph theory, and, finally, to describe differences on specified variables between the two struc- tural types. An assumption undergirding this approach is that the most definitive understanding of organizational processes may be arrived at by study of generic communica- 131 132 tion patterns and events in organizations as opposed to partial analyses based on elements of only the formal structure. The primary structural type examined was the liaison communication role. Individuals who function in liaison roles have interlinking communication contacts with two or more separate sociometric work groups (cliques) in the organization. Essentially, when liaison role persons are removed from the sociogram of communication contacts, work groups to which they are connected separate into isolated entities with the exception of single communication con- tacts (bridges) between two members of different groups. Thus, the liaison role, which is a conceptual analogue to the articulation point in graph theory, is a critical lo- (“f cation in the communication structure of an organization. The organization selected for this study was a Col- lege within the formal boundary Of a large university. All of the 142 professional staff of the College (faculty and administrators) who were officed in a single building served as the study population. Each staff member was asked to complete a short biographical questionnaire, a Personal Contact Checklist on which they listed all of their regular, work-related communication contacts of a frequency of once per week or more often, and a Personal Contact questionnaire for each contact named on a daily or more often basis. The 127 respondents to the Personal Contact Checklist listed a total of 270 daily or more 133 Often communication contacts and completed a total of 224 Personal Contact questionnaires. A total of 895 regular communication contacts of a minimum frequency of once per week were reported by the 127 respondents. In preparation for differentiating the communication structure of the organization into topological types, in- formation from the Personal Contact Checklists was utilized to determine reciprocation of contact among members of the organization. The topological and empirical analysis was based upon reciprocated contacts only. From a population of 142 members with a potential total of 10,011 reciprocated pairs, 225 pairs were found. Approximately 50 percent of the reported contacts were reciprocated. These reciprocated contacts were cast into a sociomatrix and, using procedures described by Weiss (1956, pp. 88-108), analyzed to yield identification of 22 liaison role persons, 18 isolates, and 102 non-liaison persons who had sociometric membership in 29 separate work groups of varying size. For the tests of hypotheses regarding differences be- tween liaison and non-liaison role persons perceived by their respective non-liaison communication contacts, all of the Personal Contact questionnaires completed by non- liaison persons on liaisons were the sources of data for the liaison sample. A total of 30 questionnaires were com- pleted on 17 of the 22 liaison role persons. A random sam- ple of 17 non-liaison persons yielded a total of 21 ques- tionnaires completed on them by their non-liaison contacts. 134 These questionnaires were the sources of data for the non- liaison sample. The liaison and non-liaison persons evaluated in each sample were comparatively similar in terms of age, sex, academic rank, degrees held, administrative positions, years employed by the university, percentages of time al- lotted to teaching, research, consulting, administration and committee work, number of committee memberships at var- ious levels, number of books or articles published recent- ly, appointment time basis, and number with joint appoint- ments. The two samples did differ in the average number of reported committee meetings per month (0.05 level, two alternative "t” test, df=28) with liaison persons reporting a higher mean number of meetings, and in the average num- ber of reciprocated contacts (0.001 level, two alternative "t“ test, df=l6 and 16 for Cochran and Cox formula) with liaisons reporting the larger mean number Of contacts. Nine primary hypotheses predicted differences between liaison and non-liaison persons in terms of certain com- munication behavior and personal attributes perceived by their reciprocated non-liaison contacts. Each hypothesis was evaluated by the "t” test for independent samples with degrees of freedom equal to 49 in each case (two alterna- tive test). Awareness g; Liaison Rplpp Two hypotheses dealing with awareness of actual struc- tural attributes of liaisons and non-liaisons by their 135 respective reciprocated non-liaison contacts were postu- lated and tested. Hypothesis 1 stated that liaison role persons would be perceived to have greater structural diversity of com- munication contacts in the organization than would non- liaison persons. By definition, the liaison role person does have greater structural diversity of contacts among sociometrically defined groups. Since these groups are determined only through analysis of sociometric data and may not be concretely visible to members of the organiza- tion, the question was whether or not individuals who have contact with liaisons and non-liaisons are aware of their actual pattern of contacts. The obtained means were in the hypothesized direction and the difference between the means was significant at the five percent level. The hypothesis was supported. Hypothesis 2 was also a reality testing hypothesis and stated that liaison role persons would be perceived to have a larger number of communication contacts than non- liaisons. The obtained means were in the direction hypo- thesized and the difference between means was significant at the five percent level. The hypothesis was confirmed. Interpersonal Communication Behavior Two hypotheses posited differences between liaison -- non-liaison dyads and non-liaison -- non-liaison dyads in terms of the frequency and directional ratio of deliberate message transaction initiation. In this case, dyadic 136 communication behavior as perceived by one member Of the dyad was the unit of analysis. The expectation stated in Hypothesis 3 was that out of 100 hypothetical contact situations liaison -- non- liaison dyad members have more frequently sought one another deliberately (as Opposed to chance meeting) than has been the case between members of non-liaison -- non- liaison dyads. The obtained means were opposite from the direction predicted, but the difference was not signifi- cant at the five percent level. The hypothesis was not supported. Both obtained means (87 and 91) indicated a general tendency among members Of the study pOpulation for approximately only one in ten message transaction occasions to be on a chance basis. Hypothesis 4 suggested that, of the deliberately ini- tiated message transaction occasions, the directional ra- tio Of initiation in liaison -- non-liaison dyads would be more disprOportionate than in non-liaison -- non-liaison dyads: i.e., the frequency of each member of the dyad seeking the other would be more nearly 50-50 in non-liai- son -- non-liaison dyads. The obtained mean directional ratios were Opposite from the direction predicted, but the difference between means was not significant at the five percent level. The hypothesis was not supported. Information gplpy Function Hypothesis 5 stated that liaison persons would be more likely to serve as the initial source of organization- 137 related information for their contacts than would non- liaisons for their contacts. The obtained means were in the direction predicted and the difference between means was significant at the one percent level. The hypothesis was confirmed. Perceived Personal Attributes Source credibility was the concept examined in this category, specifically the qualification and safety dimen- sions of that concept. Hypotheses 6A and 68 stated that as information sources liaisons would be attributed higher qualification and safety by their contacts than would non- liaisons by their contacts. The Obtained means for the qualification dimension were Opposite to the predicted di- rection, but in the hypothesized direction for the safety dimension. However, the difference between means was not significant at the five percent level for either dimension. The hypotheses were not supported. Influence Potential Originally three hypotheses were presented in this category, but pretest and main study analysis of the scale developed to measure the diffuse Opinion leadership con- cept indicated the scale contained two dimensions requir- ing treatment as two sub-scales and recasting of the ori- ginal hypothesis into two corollary hypotheses. Hypothesis 7 concerned the perceived importance1 of 1Operationally defined in terms of access to the ”power structure" and individuals knowledgeable of activities in the organization. .1— 138 individuals other than the respondent with whom liaisons and non-liaisons had contact (secondary contacts). The expectation was that liaisons would be perceived by their non-liaison contacts to have more important secondary con- tacts than would non-liaisons by their contacts. The ob- tained means were in the predicted direction and the dif- ference significant at the 0.001 level. The hypothesis was confirmed. The original statement of Hypothesis 8 concerning 1 was that liaisons would be diffuse Opinion leadership perceived by their non-liaison contacts to have more in- fluence over their secondary contacts than would non-liai- sons as perceived by their non-liaison contacts. The ob- tained means were in the predicted direction and signifi- cantly different at the five percent level, but the find- ing could not be accepted in support of the hypothesis because the scale contained two dimensions requiring treatment as separate variables. One sub-scale was termed ”perceived persuasiveness within the 'power structure'" and the second "general persuasiveness with other contacts." Restatement of Hypothesis 8 in terms of the first sub- scale was to the effect that liaisons would be perceived to have more influence over members of the power structure than would non-liaisons. The obtained means were in the 1 Defined as the extent to which an individual is perceived to influence the Opinions Of others in the organization. 139 predicted direction and the difference significant at the five percent level. The first corollary hypothesis was confirmed. Restatement of the original hypothesis for the second sub-scale contained the expectation that liaisons would be perceived to be more persuasive across all their secondary communication contacts than would non-liaisons. The obtained means were in the predicted direction but the difference was not significant at the five percent level. The second corollary hypothesis was rOt supported. Hypothesis 9 concerned the concept Of specific Opin- ion leadership1 wherein, again, the dyad as perceived by one member was the unit of analysis. The expectation was that liaison persons would be perceived to exercise more Opinion leadership over their non-liaison dyadic contacts than would non-liaisons over their non-liaison dyadic con- tacts. In other words, a non-liaison reporting on the personal relationship he has with a liaison would be more likely to indicate the liaison as exercising more opinion leadership for him than would another non-liaison reporting on the personal relationship he has with another non-liai- son. The obtained means were in the hypothesized direction, but the difference was not significant at the five percent level. The hypothesis was not supported. Other Findings Consideration of some of the characteristics of the 1Defined as the extent to which one member of a dyad influ- ences the Opinions of the other member of a dyad. 140 responding population of liaison (N=21) and non-liaison (N=96) persons suggests heuristic descriptive parameters for individuals in the two tOpological classes. On the Personal Contact Checklist, liaisons listed nearly twice the average number of total contacts as non- liaisons (11.7 vs. 6.1) and received more than twice the mean number of choices as non-liaisons (12.1 vs. 5.8). NO liaison person received fewer than five choices, while 45.1 percent of the non-liaisons received less than five choices. Considering only reciprocated contacts, the overall average (mean) number of reciprocated contacts for the 142 members of the study population was 3.17: however, breaking the number of reciprocated contacts down by classification reveals that liaisons had nearly two and one-half times the average number as non-liaisons (7.14 vs. 2.87). Comparison of certain demographic characteristics in-; dicated liaisons to be slightly older on the average than 1 non-liaisons (46 years vs. 42) and to have slightly longerj tenure at the university (9.9 years vs. 7.3). Considering) the four usual categories of academic rank, exactly two- thirds of the liaisons were full professors while the non-; liaisons tended to be fairly evenly distributed throughout; the four categories of rank. Administrators were more L strongly represented in the liaison group as indicated by the larger number holding some administrative title (67 percent of the liaisons vs. 34 percent of the non-liaisons) 141 and reporting a greater percentage of their time devoted to administrative work (an average Of 41 percent among liaisons vs. an average of 17 percent among the non-liai- sons). As would be expected from the latter findings, non- liaisons reported a considerably larger portion of time de- voted to teaching than did liaisons (an average of 47 per- cent for non-liaisons vs. 27 percent for liaisons). In committee work, liaisons reported an average proportion of time devoted to committees nearly double that reported by non-liaisons (11 percent vs. 6 percent), a larger average number of total committee memberships (4 vs. 2.5), and nearly twice as many committee meetings in a typical month (7 vs. 4). There was an apparent tendency for liaisons to have membership in an increasing number of committees as the administrative level of the committee increased from departmental level to university level. There were only minor differences between liaisons and non-liaisons in rates of publication, with non-liaisons holding a slight edge over liaisons. Differences were also minor in terms of the numbers of individuals holding joint appointments, average percentage of time allotted within the College, appointment basis, and highest earned degree. The male-female distribution was almost exactly equal be- tween the two types (approximately 85 percent males and 15 percent females). 142 Discussion A total of nine main hypotheses and two corollary hypotheses were postulated and tested. Of the 11 hypothe- ses, five were supported by the data and six were not. A summary of the findings is presented in Table 17. Before discussing the findings it is appropriate to review major limitations which must be placed upon the generalizability of conclusions drawn from the present study. 1. The study pOpulation from which data was collected and samples drawn was selected for availability and access and is not a randomly selected population. The study de-. sign is essentially that of an empirical case study utili- zing saturation sampling. In addition, one sample upon which tests of hypotheses were based was a non-random sam- ple, although the second was randomly drawn from the study population. Whether or not the findings are generalizable-T beyond the study population is unknown and can only be de- monstrated by later comparative studies. One qualitative ,J’ contrast with the previous study reported by Jacobson and Weiss (1955) is apparent in the nature of the work rela- tionships: although the pOpulations of both studies were academic personnel, the nature of the work performed was different. The Jacobson and Weiss study was of a govern- ment agency concerned with the processing and granting of research project applications: the work-flow likely required more inter-personal and inter-unit coordinating communica- tion than would be the case in the present study of 143 Table 17. Summary of results from tests of hypotheses Hypo- Signi- thesis ficance Variable number Result level* Perceived structural diver- sity of contacts 1 Supported 0.05 Perceived number of contacts 2 Supported 0.05 Frequency of deliberately initiated message trans- actions 3 Not supported Directional ratio of -de1iberate message trans- action initiation 4 Not supported First source of information 5 Supported 0.01 Source credibility Safety dimension 6A Not supported Qualification dimension 63 Not supported Importance of secondary contacts 7 Supported 0.001 Diffuse Opinion leadership Persuasiveness within power structure 8A Supported 0.05 General persuasiveness with secondary contacts 83 Not supported Specific Opinion leadership 9 Not supported * ”t" test of difference between two independent means, two alternative test, df=49 144 '1, university faculty members whose main work -- teaching and research -- places emphasis on independence and individual autonomy. This unique work-pattern may partially account for the relatively small number of reciprocated contacts which emerged from the sociometric analysis in the present m ‘ study. It should also be noted that the study population in the present work is the membership Of a sub-unit of a lar- ger formal organization and does not include the parent organization boundary or contacts outside the sub-unit‘.‘n boundary. The study population is limited to the formal I prescription of the sub—unit boundary and, in addition, includes only individuals officed in a single buildingL’”J This latter geographic boundary criterion was presumed to reduce the probability of a large number of high frequency contacts outside the boundary making the study population similar in most aspects to a single, autonomous formal organization. It should be further noted that the study population includes only professional members of the staff of the or- ganization, thus is not a total census of the communication contacts among all individuals who could be defined within the boundary of the organization. Excluded are clerical staff, maintenance personnel and students. It was assumed, however, that the nature of the sociometric criterion uti- lized for mapping the communication structure was not es- pecially relevant to the information functions which might 145 be performed by the excluded personnel and their inclusion would likely result in the mapping of loosely connected sub-structures within which information content would dif- fer qualitatively from that among the professional staff. 2. A second limitation stems from the phenomenological nature of the data, especially in regard to estimates of dyadic behavior. The assumed congruence of actual dyadicfi‘T behavior with the behavior perceived and reported by one 1 member of the dyad must be tempered by recognition of the possibility of perceptual distortion on the part of the .rJ perceiver. 3. A third limitation of generalizability of the find- ings is that conclusions drawn from the study must be lim:%1 ited to topological concepts identified on the basis of reciprocated contacts of a minimum contact-frequency of once per week, and hypothesis testing data based on reci- procated contacts of a minimum frequency of once per day.‘_ The latter contact-frequency criterion was assumed to pro- vide more reliable data than would be obtained from less frequent contact categories. I 4. A final limitation relates to the nature of the sampling model used to define sources of data for the tests of hypotheses. The sampling unit was liaison and non-lia- ison role persons, but the source of data was non-liaisons who had direct reciprocated contact with members in the samples. Perceptions of liaisons and non-liaisons by others in the organization who are not in direct contact 146 with them are not included in the analysis. The sampling model also did not provide for random selection of the liaison sample. This limitation was a practical necessity based on the small number of liaisons who emerge from topological analysis of a single organiza- tion of this size. The result is that the application of statistical tests to the data is heuristic and can only be assumed to provide an approximation of potential parameters. Unsuppprted Hypptheses ' Of the six unsupported hypotheses, three were in the expected direction and three were opposite to the direction predicted. Following are some of the methodological and theoretic factors which may be related to these outcomes. Methodological Factors One of the major methodologi- cal reasons which may have contributed to Type II error could be that the sample sizes were not adequate. There were 30 source questionnaires in the liaison sample and 21 in the non-liaison sample. Two of the six unsupported hypotheses approached significance: Corollary Hypothesis 88 (general persuasiveness with secondary contacts) and Hypothesis 9 (specific Opinion leadership) were in the predicted direction and significant at the 0.20 level (two alternative test) indicating there were four chances in five that the Obtained differences were "real“ and that Type II error had been committed. A third hypothesis re- garding the safety dimension of source credibility reached significance at the 0.30 level. 147 In regard to Hypothesis 9, another test of the speci- fic Opinion leadership data is to treat the measure as a discrete variable (scale scores of 21 or higher)1 and apply a one-sample X2 test to the frequency distribution of dyadic opinion leaders in the liaison sample, using the non-liaison sample distribution as a check on expected. The expectation stated in the hypothesis implies that spe- cific Opinion leadership among non-liaison -- non-liaison dyads will be equally distributed: i.e., about half of the dyads will have the respondent as the opinion leader and half will have,the respondent's contact as the opinion leader. The data support this expectation: 11 of the 21 dyads (52 percent) had the respondent's contact as the dyadic Opinion leader and ten (48 percent) had the respon- dent as the dyadic opinion leader Or indicated equal opin- ion leadership in the dyad. In contrast the 30 liaison -- non-liaison dyads included 23 dyads (77 percent) in which the liaison was the Opinion leader and seven (23 percent) with either the non-liaison as the opinion leader or equal dyadic opinion leadership. The setup for the one-sample x2 is as follows: 1Each item in the scale contained seven response alterna- tives. The item midpoint (4) would indicate equal Opinion leadership in the dyad, higher than 4 would indicate Opin- ion leadership on the part of the respondent's contact, and less than 4 would indicate dyadic Opinion leadership by the respondent. Summing across the five items, scores above 20 would indicate a tendency for the respondent's contact to be the dyadic opinion leader. 148 Observed Expected (SO-50) Liaison Opinion leader dyads 23 15 not liaison Opinion leader dyads _Z 12 N = 30 3O 2 The Obtained X equals 8.54 which is significant at the 0.01 level (critical value Of x2 = 6.64, df=l, two alter- native test). There were more liaison -- non-liaison dyads in which the liaison was the Opinion leader than would be expected if Opinion leadership were equally distributed among the dyads. This secondary analysis, however, cannot be accepted as confirmation of the hypothesis because more evidence is necessary to support the contention that speci- fic Opinion leadership is always distributed equally among non-liaison -- non-liaison dyads. Without this evidence, the better estimate of expected distribution is Obtained by use Of a two-sample (2x2) x2 test which, with this data, does not reach significance at the 0.05 level (Obtained x2 = 2.277). The one-sample result does, nonetheless, re- inforce the investigator's feeling that, treating speci- fic Opinion leadership as a continuous variable, a larger sample size may yield support for the hypothesis. A second methodological problem was that the vari- ances Of estimates Of deliberate and directional message transaction initiation frequencies, Hypotheses 3 and 4, were relatively large. It may be that measurement error occurred here as a result Of asking respondents to provide a single estimate of initiation frequency based on 100 149 hypothetical contact situations. Such estimates may be extremely difficult to make and the resulting unreliabi- lity may have contributed tO inflated variances. A third methodological problem, especially regarding source credibility and specific Opinion leadership, may stem from the nature Of the sampling model utilized. Eval- uations Of liaisons and non-liaisons were made by indepen- dent samples Of non-liaisons. Each dyad within each sample category was closely related from a cOmmunication viewpoint (contact-frequency Of at least once per day), probably in- dicating, in addition to a close working relationship, a personal friendship. Homans (1950) asserted that the more frequently individuals interact with one another the more nearly alike they become in the norms and expectations they hold. In this light, perhaps a better sampling model for testing these hypotheses would be one which utilizes the same non-liaison respondent as the source of data for comparison of a liaison and another non-liaison with whom the respondent has contact. The contrast in perceptions~a- Obtained using this model may be more consistent with the different structural positions Of the two individuals evaluated since these are the specific linkages which serve to maintain an individual in his structural role. The theoretic position for evaluating the findings, how- ever, would need to change to account for the fact that now three-person connected chains of individuals are in- volved from which a directional hierarchy Of information E s ’0 as O I t g: S I, g 150 and influence flow may be inferred. Theoretic Factors As pointed out in Chapter I, a limiting factor in the derivation of hypotheses about the liaison role was the complete lack Of previous evidence re- garding the concept. Utilization of other frameworks and findings required making a number of heuristic assumptions. Hypotheses 3 and 4 (deliberate -- directional initia- tion) were undergirded by the exploratory assumption that there might be physical or psychological propinquity dif- ferentials between liaison dyads and non-liaison dyads. Evidence from this study does not reinforce the assumption, or deny it. Further study Of homophily1 within dyads may provide more concrete evidence upon which to formulate later hypotheses. Another problem may be in the nature Of the sociome- tric criterion utilized in the present study. Rather than a general work-related criterion, data based on more speci- fic types Of information content may reveal transaction initiation differences which were masked in the present study. Support for the liaison's function as afirst source of organization-related information (H5) suggests that his information relay function may be more specific than gener- al. Initiation frequencies related only tO this ”grapevine" information may be in the direction hypothesized. 1Defined as the degree to which individuals with a certain attribute have interpersonal communication contacts With other individuals with a similar attribute (Yadav, 1967, p. 167). 151 .lua '_~ i It also appears possible that non-liaison individuals who have contact with liaisons may not perceive the delib- 5 erateness Of some liaison-initiated message transactions. E Perhaps liaison persons purposely initiate some trans- i actions in socially-oriented “information markets" (like g the coffee room) where individuals whom they contact are ; less likely to interpret the meeting as planned. Given j//: the stronger representation Of administrators in the liai- son sample, this misperception may partially account for the higher mean frequency Of reported happenstance commun- ication with liaisons. To avoid the formality Of calling a subordinate into the Office, an administrator may plan to "bump into" that subordinate at coffee or some other non-Office location. Certain subordinates may use the same procedure, sublimating the deliberate nature Of the meeting. Hypotheses 6A and 68 results (qualification and safe- ty dimensions Of source credibility) may also have been confounded by the nature Of the sociometric criterion used. Reciprocal source credibility in closely associated dyads as formed the basis for these data might be relatively high given the likelyhood Of a friendship bond. This as- sumption is reinforced by the skewness Of the Obtained dis- tributions. Given high mutual trust and professional res- pect between closely associated individuals, source cred- ibility differentials may be very specific tO certain types Of information content, or even to specific message 152 transactions within given situational fields. Non-support for the general persuasiveness with se- condary contacts (H88) and specific Opinion leadership (H9) hypotheses is an interesting paradox considering the found support for Hypotheses 5 (first source) and Corollary Hy- pothesis BA (persuasiveness within the power structure). A recent study by Hickey (1968, pp. 49-54) reports a posi- tive relationship between perceived information control and status, and points out the already well established positive link between status and influence. Since non-lia- isons are aware Of the structural diversity and broad span Of contacts Of liaisons to whom they are connected (H1 and H2) and recognize the liaison as a first source Of informa- tion (H5) it might be inferred that they are also aware Of their potential for information control. However, even assuming this inference to be correct, the liaison status which may arise from these cognitions may be Of a special nature and not appropriate to the more general finding. Again, more definitive results might be Obtained by inves— tigating Opinion leadership within the context Of specific information contents and situational fields. The findings in regard to specific Opinion leadership may have tended to err on the conservative side (Type II error) because some non-liaison individuals who were bridge contacts or who had inter-group contacts with liaison per- sons were the Objects Of evaluation in the non-liaison sam- ple. These individuals have, to a lesser degree, some of 153 the structural characteristics associated with liaison role persons. Ten of the 17 evaluated non-liaisons had bridge or inter-group liaison contacts. TO the extent that structural diversity is related to the occurence Of specific Opinion leadership, this group in the non-liaison sample would tend to attenuate the differences between the liaison and non-liaison samples. Another factor which might have reduced differences between the samples is the structural similarity between data-source non-liaisons in the liaison sample and their evaluated liaison contact. Treating the midpoint Of the summed scale score for specific Opinion leadership as a cutting point for determining which member of the dyad was reported as an Opinion leader for the otherl, 23 Of the 30 dyads had the liaison member Of the dyad as the Opinion leader, one indicated equal Opinion leadership in the dyad and six had the non-liaison member as the Opinion leader. Comparing the 23 liaison Opinion leader dyads to the six non-liaison Opinion leader dyads reveals that in the latter all six non-liaisons were either bridge contacts or had other inter-group liaison contacts, while 15 of the 23 non- liaison respondents (65 percent) in the liaison Opinion leader dyads had only in—group non-liaison contacts. In other words there was greater similarity of structural characteristics between members in the non-liaison Opinion 1Refer to footnote on page 147. 154 leader dyads of the liaison sample than in the liaison Opinion leader dyads. This similarity may have contributed to lower scale scores for the six liaisons in the non-liai- son Opinion leader dyads also causing decision-making error on the conservative side. Suppgrted Hypotheses M Two objectively defined structural characteristics of at...” _ . “WWW-d.“ N(.'-~....‘~_ liaison persons are their structural diversity and broader span of communication contacts as compared to non-liaisoni.g Evidence from the present study indicates that individuals connected to liaisons are generally aware of these two structural characteristics, even though the diversity of contacts characteristic is sociometrically defined and not objectively visible to members of the organization in the course of their day-to-day activities. This latter finding, however, may be partially artifactual due to the greater representation of administrators in the liaison sample. Their contacts may have attributed greater structural di- versity to them simply based on the assumption that, as an administrator, they should have more diverse contacts. But whether or not this assumption did in fact affect responses is a moot question. The point is that individuals who have daily contact with liaisons and non—liaisons are differen- tially aware of these two structural aspects of the role. Findings from the Jacobson and Seashore study indi- cate that liaisons and their reciprocated contacts gener- ally report each other as important contacts. The present 155 study provides evidence that, looking beyond the imme- diate dyad, non-liaison contacts of liaisons attribute greater importance to the other contacts of the liaison than do the contacts of non-liaisons for the non-liaison's other contacts. Thus the present study extends the evi- dence reported by Jacobson and Seashore and may suggest one reason why the liaison himself is considered an impor- tant contact by non-liaisons. The finding supplements the Katz assertion (1957, pp. 74-5) that an individual may be valued by a group not only for what he knows, but also for whom he knows outside of the group. Related to the above finding is support for the hypo- thesis that liaisons are perceived by their contacts to hex? more influential within the "power structure" of the organ-E ization than are non-liaisons. The Importance of Second1/,i ary Contacts scale included reference to having assess to members of the power structure. Linking these two findings1 supports the contention stated in Chapter I that an indi- vidual, in addition to being valued by a group for whom he knows outside the group, may be of increased value for how much influence he has, or is perceived to have, with part- icular outside contacts. This finding may imply another reason why the liaison is considered an important contact. 1Without the related items in the two scales, the influence in the power structure finding could be artifactual. If an individual was not perceived to be linked to the power structure he likely would not be perceived as influential within that structure. 156 The exact meaning respondents had for the term "power structure" is, of course, unknown, but the findings pro- vide some general definition Of phenomenologically impor- tant substructures within a formal organization and of the perceived linkages with those substructures within the ex- tant communication network of the organization. Clearly, the liaison role has meaning to organizational members within this context. The findings reported above give some descriptive perceived characteristics of liaison role persons, but do not provide concrete evidence of their function within the organization's extant communication network. One such function for which evidence is provided in this study is the liaison person's greater likelihood of being the fi;;t\] source of organization-related information1 for his non- \m liaison contacts than is the non-liaison for his non-liai- son contacts. The finding leads to the interpretation that\? liaisons are "early knowers" who are also "early dissemin- f Lfl ators“ within the organization's information relay net- work. Whether Or not this function is related to the ap- parent motivational bases Davis (l953b, p. 46) suggested, or simply the result of the unique structural location of the role cannot be detected from the present data. What is supported is the fact that not only is the liaison role 1Operationally defined in the questionnaire items as in- formation relating to changes or new ideas being proposed or discussed, or new develOpments which have occurred in the organization. 157 person in a potentially important location in the extant communication network, he functions to provide early dis- semination of certain information through the network. In summary, contrasting the liaison role person and the non-liaison role person as perceived by their recipro- cated non-liaison contacts, evidence was found in the pre- sent study that liaisons are characterized as having a larger number, more structurally diverse, and more impor- tant contacts in the organization and are perceived as being more influential within the organizational power structure. In the organization's information relay network liaisons were reported to more frequently serve as first sources of'organization-related information for their re- ciprocated daily non-liaison contacts. 9522; Findings Although the primary focus Of the present study was on perceived attributes of liaison and non-liaison role persons, some demographic information was collected from the study pOpulation and bears comment. The broader span of reciprocated contacts of liaisons I \7 has been adequately discussed. The percentage of reported contacts which were reciprocated was 51 compared to 44 percent in the Jacobson and Seashore study. The higher percentage in the present study is probably due to having limited the lowest frequency of contact reported on the Personal Contact Checklist to a “once per week“ category, '9‘“. ‘—\—.\uLx—h',:a4y&— ,...a~...«,.- . - as opposed to a lower limit of "several times per year" //* 158 in the Jacobson and Seashore study. Jacobson and Weiss re- port a lower percentage Of reciprocation among the lower frequency of contact categories -- 80 percent of the "sev- eral times daily“ and reportedly high importance contacts were reciprocated, while only 19 percent of the "several times monthly" or “several times yearly“ and low importance contacts were reciprocated. Also related to a comparison of the two studies is the fewer number of total contacts reported -- an average of seven in the present study comp pared to an average of 12 in the previous study. This dif- ference may be partially accounted for by the difference in the lower limit frequency of contact categories and, perhaps, partially by qualitative and quantitative differ- ences between the two types of organizations studied. There was an apparent tendency for more individuals with administrative positions to be liaisons although nearly one-third of the non-liaisons were also administra- tive personnel. Being an administrator may be sufficient but not a necessary condition for assuming a liaison com- munication role. There was also an apparent tendency for liaisons to have membership in more committees than non-liaisons, to devote more of their time to committee work, and to attend more committee meetings in a "typical" month. These find- ings do not, however, reveal a causal relationship. It may be that individuals develOp liaison contacts as a re- sult of greater participation in committee work, but it is 159 also possible that having liaison contacts makes one more visible to those who create the committees. More evidence is needed before assuming that manipulation Of committee assignments will create regular liaison-like communication channels for an individual. One comment should be made regarding the 18 individ- uals identified as isolates. It is likely that they are not totally isolated from the organization's communication structure as may be implied by their descriptive label. Their identification as isolates is a consequence of uti- lizing a sociometric criterion limited to work-related contacts, and limited to a minimum frequency of once per week. These individuals are isolated only within these limits, plus the possibility of measurement error in com- pleting the sociometric instrument. Contgibutions _g the Study The primary contribution of the present study has been to provide evidence that the liaison communication role does have meaning to members of a formal organiza- tion: meaning at least in the sense of their awareness of the structural characteristics of the role, the importance of the role incumbent's linkages within the communication structure, and one aspect of the role incumbent's influence potential: and meaning also in terms of one facet of the liaison person's function in the organization's information relay network. The consequence of these preliminary 160 indications is to establish an empirical rationale and highlight a need for additional research on the liaison communication role. Jacobson, Seashore and Weiss originally identified the logical importance of the liaison role based on its loca- tion in the communication structure. Ross and Harary \%p (1955, p. 258) have called this view of the concept its static property: “a liaison is critical because his loss . “ “VA-or“; Q destroys the connected unity Of the organization. Thus anz 1 organization or other structure is most vulnerable at arti-E \ culation points.“ Delineation Of the static prOpertiesfofflg the liaison role is simply a mapping process: a topological I" description Of an organization. Leavitt (1964, p» 230) asserts, "These diagrams are structural. They tell us nothing about the people involved -- just something about the system." The thrust of Leavitt's admonition is for investigators to go beyond the simple mapping of structure and examine what Ross and Harary label, in regard to the liaison role, as dynamic properties: “The critical nature of the liaison person for the dynamic or flow functions of an organization arises from his non-substitutability in paths. For example if a liaison person is a 'bottleneck' the organization suffers badly, while if he is efficient he tends to expedite the flow in the entire organization." The present study touches only lightly on examination of the dynamic properties Of the liaison communication role, but helps set the stage for more definitive work. The ll}ll.flll}l [I'll 161 study was exploratory. A more detailed description of actual and perceived characteristics of liaison role incum- bents, the broader functional meaning of the role, and the in-depth implications for development of organizational communication theory remain to be uncovered. What has been demonstrated is a conceptual and methodological frame- work for tOpological-analytic, or ggmmunimetrigl, study of the extant communication structure of formal organizations. A second noteworthy contribution of this study has been the demonstration of a methodological schema, what Coleman refers to as the method of relational analysis, which has two major advantages when applied to examination of extant communication structures. The first advantage is the provision, based on graph theory concepts, of objec- tively determined, discrete categories for classification of topological properties as a prelude to descriptive or functional analysis. The categories are discrete in the same sense as utilized in the analysis of formally pre- scribed tOpOlogical properties (e.g., superior, subordin- ate) but Offer the advantage Of admitting facets of the informal communication structure into the analytical frame. In addition, the categories are objectively defined as opposed to the arbitrary definition of constructs based on unvalidated criteria: e.g., the definition of an opinion 1Defined as that branch of sociometry which utilizes only a criterion of communication contact for operationalizing social system topologies, followed by analysis of struc- tural and/or process characteristics of the system. 162 leader as one who receives some minimum number of sociome- tric choices within a given social system. Secondly the communimetric approach is of major ad- vantage because it permits partitioning interpersonal re- lationships and data on a number of different dimensions such as other topologically-defined roles, formal roles, various contact-frequency categories, directional flow of messages and so forth. Perhaps one of the more intriguing possibilities of communimetrics is the potential establishment of a bridge between small group communication network research and organizational research. The tOpological definition of an organization can be utilized as a preliminary step to- ward identification Of various configurations in natural small groups within the organization and for the compara- tive study of group structures within natural social sys- tems, or between natural social systems and contrived experimental small groups. Another broader application could be the study of unique Eggs of inter-group linkages or linkage configurations. Finally, there are at least two advantages of the conceptual and methodological framework demonstrated in this study which have both pragmatic and theoretic inter- est. The first is the finding that liaison persons tend to function as first sources Of information for their con- tacts. The implication is that organizations with higher concentrations of liaison persons should be characterized 163 by swift and efficient diffusion Of information within the system. Thus the liaison role concept should be Of inter- est tO administrators and students of administration and of particular interest to students of innovation diffusion who have used time of awareness as a central criterion var- iable in their research. The second advantage of pragmatic and theoretic in- terest rests in the basic assumption which undergirds both the conceptual and methodological framework. The assump~ tion, as defined earlier, is that the approach should per- mit gaining broader insight into the total dynamic Of an organization because it is based on study of extant commu- nication structures instead of the more limiting study of only formal, prescribed structures. The approach can be utilized to reveal how people behave in formalized social systems independent of, and/or concomitant with, formal prescriptions and as a result may be of considerable value in the design of formalized systems based on principles of “natural“ structural development. Suggestions Egg {Eggheg Reseaggh As anticipated, the present exploratory study raises more questions than it answers. A number of implied sug- gestions for additional research have been adequately covered in the previous text. Following are other general suggestions. ' 1. Further exploration of the characteristics and 3 -‘_ 164 behavioral attributes of liaison role persons. Given the problem of application of statistical tests to samples and!-i populations derived from saturation sampling, additional comparative studies are called for to validate the gener- alizability of the present findings, to extend these find- ings, and to compare findings across different types and sizes of organizations. One development which would great- ly facilitate studies of this type would be a computer pro- gram designed to treat communimetric data for indentifica- tion and classification of topological properties such as the liaison role. This tool would be of great benefit not only for descriptive studies, but also for field ex- periments involving the implantation and tracing of mes- sages, and comparative studies over time in single organ- izations intended to examine changes, and variables related to change, in the communication structure. The greatest theoretic utility of the liaison role concept will come from study of dynamic prOperties of the role. It is necessary to distinguish the functions of the information treated and relayed by liaisons both in terms Of source purpose and receiver effect. The liaison person is a critical path in the control Of information within the communication structure. Hickey (1968, p. 51) points out three elements of information control: One can control —- in a transmit or non-transmit sense -- by handling the messages, or one can control by arranging the channels, or one can control by mani- pulating the content of the messages. 165 All three forms of information handling might be examined in terms of the liaison role. Additionally, the following questions may be fruitfully explored: What kind of infor- mation filter is the liaison? What are the uses and pur- poses the liaison makes of his information handling pos- ition? What are the uses the liaison's contacts make of information received? What are the uses and purposes in- tended by information transmitted 59 the liaison by his contacts? and finally, How do these dynamics change for different information contents or situations? More advanced structural distinctions may also be developed in regard to the liaison role. One would be to go beyond treating the liaison role as a discrete category to the development of a continuous measure of "liaisonness" based either on the number of intergroup links of a given liaison, or on the number of reciprocated contacts of a given liaison, or a weighed measure of both. Another more advanced concept of the liaison role may be a higher-order category labeled a "group-set liaison." Careful examina- tion of the sociogram drawn from the present study reveals eight higher order "groups of groups", or group-setsl. With the exception of only two bridge contacts between group-sets, all of the sets are interconnected entirely by 1The identification Of the group sets is admittedly heuris- tic. The found configurations may be an idiosyncrasy of this particular study population, or may be directly re- lated to the formal structure of the organization. Fur- ther exploration is necessary before formalizing the con- cept. 166 liaison contacts across group-set boundaries. Considering the static connectivity of the total organization, these group-set liaisons are of special significance. Further development of this concept may lead to yet another re- search application of the liaison role. 2. Exploration of other topological properties of the communication structure. Study of the liaison role does not exhaust the possible categories of topological concepts which can be examined. One other concept is the bridge contact which has some of the characteristics of the liai- son role. In addition, certain kinds of group configura- tions, inter-group configurations, and interpersonal chains may lend themselves to Objective definition and analysis. 3. Other partitions of interpersonal relations. The present study examined the liaison role only from the standpoint of perceptions of liaisons and non-liaisons by non-liaisons. Other dyadic partitions which may be examin- ed are liaison --liaison dyads, and unreciprocated liaison -- isolate dyads and non-liaison -- isolate dyads. Reverse direction perceptions in liaison -- non-liaison dyads may also be examined: i.e., the perceptions liaisons have of their non-liaison contacts is the reverse of the pattern of the present study. Of particular methodological interest would be exam- ination of the congruence of perceptions of dyadic behav- ior in reflex pairs. Finally, other partitions of interpersonal relations 167 and data may be based on use of questionnaires from unre- ciprocated contacts, different frequency of contact cate- gories and administrative -- non-administrative classifi- cations. The potential application of the methodological and conceptual framework demonstrated in the present work Open a generally unexplored area of research in organization and communication theory. BIBLIOGRAPHY Bales, R. F., g§_gl. ”Channels of Communication in Small Groups," American Sociological Review, 16, June 1951, pp. 462-68. Barnard, C. I. The Fungtions of the Executive. Cambridge: Harvard University Press,"'1'938. Bavelas,.A., “Communication Patterns in Task-oriented Groups,“ Journal g; 323 Agoustical Society 9; Ameri a, 22, 1950, pp. 725-30: Also in: Cartwright, D. and A. Zander, Group Qynamics: Research 32g Theory. New York: Harper and Row, 1960, pp. 669-82. Beer, M., g; 31. ”Some Perceived PrOperties of the Dif- ference Between Leaders and Non-leaders,” Journal 2; Psyghology,,gl, 1959, pp. 49-56. Berkowitz, N. H., and W. G. Bennis, “Interaction Patterns in a Formal Service-oriented Organization,” Adminis- ggative Sgienge Quarterly, Q, 1961, pp. 25-50. Berlo, D. K., J. B. Lemert and R. J. Mertz, "Dimensions for Evaluating the Acceptability of Message Sources," East Lansing, Michigan: Department of Communication, (undated). Blau, P. M. and Scott, W. G. Formal Organizations. San Francisco: Chandler Publishing Co., 1962. Burns, T., "The Directions of Activity and Communication in a Departmental Executive Group: A Quantitative Study in a British Engineering Factory with a Self- Recording Activity," Human Relations, 1, 1954, pp. 73-97. Coleman, J. S., “Relational Analysis: The Study of Social Organizations with Survey Methods" in Etzioni, A. A. (Ed.) Complex Organizations: A Sociological Reader. New YOrk: Holt, Rinehart and Winston, 1964, pp. 441-53. Dahle, T. L., ”An Objective and Comparative Study of Five Methods of Transmitting Information to Business and Industrial Employees," Speech.Mono ra hs,lgl, 1954, pp. 21-8. Davis, K. A., “A.Method Of Studying Communication Pat- terns in Organizations," Personnel Psychology, Q, 168 169 Davis, K. A., "Management Communication and the Grapevine,“ Harvard Business Review,‘§l, 1953b, pp. 43-9. Dodd, S. C. and T. R. Gerbrick, ”Word Scales for Degrees of Opinion," Langgage and Speech, §:l, 1960, pp. 18-31. Dorsey, J. T., Jr., ”A Communication Model for Adminis- tration,“ Administrative Scienge Quarterly, 2, 1957, Pp. 307-240 Downie, N. M. and R. W. Heath. Basic Statistical Methods. New YOrk: Harper and Row, (2nd edition) 1965. Edwards, A. L. Statistical Methodg. New YOrk: Holt, Rine- hart and Winston, Inc., (2nd edition) 1967. Etzioni, A.:§ meparativg Anglygig g; COmplex Organiza- tigns. New YOrk: Free Press, 1961. Evans, R. I. with P. Leppmann. Rggigtappg pp Innovation ;p Higher Edgation. San Francisco: Jossey-Bass, Inc., 1967. Festinger, L., "The Analysis Of Sociograms Using Matrix Algebra,“ Human Relations, 3, 1949, pp. 153-58. Goetzinger, C. and M. Valentine, ”Communication Channels, Media, Directional Flow and Attitudes in an Academic Community," Journal _o_f_ Communi ation, L2, 1962, pp.23-6. Guetzkow, H., “Communications in Organizations" in March, J. G. Handbook g; Organizations. Chicago: Rand McNally and Co., 1965, pp. 534-73. Harary, F. and R. Z. Norman. Graph Theogy pg g.Mathematical gpdel $5 Social Sgiences. Ann Arbor, Michigan: Insti- tute for Social Research, 1953. Hickey, J. R., ”The Effects of Information Control on Per- ceptions of Centrality," Journalism Quarterly, Spring, 1968. Hinrichs, J. R., “Communication Activity of Industrial Research Personnel," Pegsgnnel P cholo , $1, 1964, PP. 193-2040 Homans, G. C. The ngan Grgup. New YOrk: Harcourt, Brace and Co., 1950. Jacobson, E. and S. E. Seashore, ”Communication Practices in Complex.Organizations," Journal 9; Sogial Issues, 1, 1951, pp. 28-40. 170 Jones, L. V. and L. L. Thurstone, ”The Psychophysics of Semantics: An Experimental Investigation," The Jgurnal g§_Applied Ps cholo , Q2, 1955, pp. 31-9. Katz, D. and R. L. Kahn. The Sogial Psyghology pg Organi- zationg. New York: John Wiley and Sons, Inc., 1966. Katz, E., “Communication Research and the Image of Society: Convergence Of Two Traditions," American Journal 9; Sogiology,‘§§:5, March 1960. ., "The Two—step Flow of Communication: An Up-to- date Report Of an Hypothesis,” Public Opinion uar- terl 21 Spring, 1957, pp. 62-78. ' _' . and P. F. Lazarsfeld. Personal Influence: The Part Played py People pp the Flow 9g Mass Communigations. New YOrk: The Free Press, 1955. Kerlinger, F. N. Foundations pf Behavioral Research. New YOrk: Holt, Rinehart and Winston, Inc., 1964. Leavitt, H. J. Managerial Ps holO . Chicago: University of Chicago Press, 1958, Revised edition) 1964. ., ”Some Effects of Certain Communication Patterns Psyghology, gp, 1951, pp. 38-50. Lewin, K., “Psychological Ecology" in Dorwin Cartwright (Ed.) Field Theory ;p Social Sciences. New York: Harper Brothers, 1951. Lippitt, R.,‘gp‘gl., ”Dynamics of Power: A Field Study Of Social Influence in Groups of Children" in Maccoby, E. E., T. M. Newcomb and E. L. Hartley. Readings pp Social Psyghology. New YOrk: Holt and Co., (3rd edition) 1958. March, J. G. and H. A. Simon. Organizations. New York: JOhn Wiley and Sons, 1958. Miller, J. G., "Living Systems: Basic Concepts," Behavior- 3; Science, 19:3, July 1965a, pp. 193-237. ., "Living Systems: Structure and Process," Behavior- 3; Science, 19:4, October 1965b, pp. 337-79. Mischler, E. G. and A. Tropp, "Status and Interaction in a Psychiatric Hospital," Human Relations, 2, 1956, McNemar, Q. Psyphological Statistics. New YOrk: John Wiley and Sons, Inc., 1962. 171 McQuitty, L. L., "Elementary Linkage Analysis for Isolating Orthogonal and Oblique Types and Typal Relevancies," Educational gpg Psyphologigal Measurement, 11, 1957, PP. 207-29 . Oppenheim, A. N. Questionnaire Design and Attitude Mea- surement, New York: Basic Books, Inc., 1966. Read, W. H., “Upward Communication in Industrial Hierar- chies,“ Human Relations, 1;, 1962, pp. 3-16. Rogers, E. M. Diffusion g; Innovations. New YOrk: Free Press, 1962. . and D. G. Cartano, ”Methods of Measuring Opinion Leadership,“ Publig Opinign Quarterly, gg, Fall, 1962, PP. 435-41 . Ross, I. C. and F. Harary, “Identification of the Liaison Persons of an Organization Using the Structure Matrix,” Mpnagement Sgienge, ;, AprilqMay 1955, pp. 251-58. Simpson, R. L., “Vertical and Horizontal Communication in Formal Organizations," Administrative Science Quarterly, 3, 1959, pp. 188-96. Thayer, L., ”Communication and Organization Theory," in F. E. X. Dance (Ed.) Human Communication Theogy. New YOrk: Holt, Rinehart and Winston, Inc., 1967, pp. 70-115. Triandis, H. C., "Cognitive Similarity and Interpersonal Communication in Industry,” JOurnal‘g§.Applied Psygho- lo , 3;, 1959a, pp. 321-26. ., “Similarity in Thought Processes and Boss-Employee Communication“ in Communicatign in Organizations: New Research Findings. Ann Arbor, Michigan: FOundationfi for Research on Human Behavior, 1959. Troldahl, V. C. and R. VanDam, "A New Scale for Identi- fying Public-affairs Opinion Leaders," Journalism Quarterly, fig:4, Autumn, 1965. Wager, L. W. “Interpersonal and Mass Communication in an Organizational Setting," ggiologica; nguigy,.§g:l, Winter, 1962, pp. 88-107. Walker, H. M. and J. Lev. Statigtigal Inferenge. New York: Holt, 1953. Walton, E. A Magnetic Theo§y_ of Organizational Communi- cation. -China Lake, Ca ifornia: U. S. Naval Ordnance Test Station, 1962. 172 Walton, E., “Communication Down the Line: How They Really Get the Word," Personnel, §§, 1959, pp. 78-82. Weiss, R. S. Progesses 9; Organization. Ann Arbor, Michigan: Institute for Social Research, 1956. . and E. Jacobson, “A.Method for the Analysis of the Structure of Complex.Organizations,“ American Sociolo- gigal Review, 29, 1955, pp. 661-68: Also in Etzioni, A. A. Complex Organizations: A Sogiological Reader. New YOrk: Holt, Rinehart and Winston, 1964, pp. 453-64. fusion gp Egg Indian Villages. Technical Report #2, Diffusion of Innovations in Rural Societies. East Lansing, Michigan: Michigan State University, Depart- ment of Communication, 1967. . and E. M. Rogers, "Interpersonal Communication in Innovation Diffusion," Working Paper #5, Diffusion of Innovations in Rural Societies. East Lansing, Michigan: Michigan State University, Department of Communication, December 15, 1966, (mimeo). APPENDICES 173 APPENDIX A ENTRY LETTER We have often wondered why university faculty are fre- quently required to spend so much time communicating to the wrong peOple about the right thing, or to the right peOple about the wrong thing. This is only one of the many ”communication problems“ which exist in a large uni- versity. I could, and often do, go on and on talking about these kinds of problems. Fortunately, I (and you) have a chance to help get this discussion from the "arm- chair“ stage to something more concrete. Sometime this week you will receive a telephone call from Don Schwartz. He will ask for an appointment with you for delivery and pickup of aquestionnaire entitled UNIVERSITY COMMUNICATION STUDY. (Mr. Schwartz has a unique method for studying faculty communication patterns and effects. His study, which is a Ph.D. dissertation, promises to yield better understanding of communication in a college within a university. This study won't reveal all the problems or solutions (that questionnaire would take us all day to com- pletel), but will make a potentially useful contribution to both theory and practice. It's because I believe this study is important to you and me, and because I'm convinced of the integrity and compe- tance of the investigator, that I've agreed to write this letter asking you to participate in the study by completing the self-administered instrument. The special method of analysis requires Mr. Schwartz to get nearly 100% partici- pation by the faculty officed in this building. With this requirement, he is understandably concerned about the pro- portion of response he can Obtain. I'm sure we'll do both him and ourselves a favor by participating. The time in- volved for each of us is about one coffee break. Rest assured that the completed questionnaire will be seen only by Mr. Schwartz. Perhaps that's not important be- cause the questions are not that "personal“, but I mention it because some may wonder. Mr. Schwartz will provide each participant with a summary of the findings when the study is completed. I encourage you to participate when he telephones for an appointment. Sincerely yours, (Signed) Professor 174 a. b. C. APPENDIX B QUESTIONNAIRE PACKET Cover Letter Part I -- Personal Data Questionnaire Part II -- Personal Contact Checklist Part III -- Personal Contact Questionnaire 175 176 UNIVERSITY COMMUNICATION STUDY May 9, 1967 Thank you for agreeing to complete the enclosed study questionnaire. Don't be alarmed by the bulk of this packet. Pretest experience indicates that most res- pondents can complete all necessary forms in 15 to 30 minutes. Broadly speaking, this study is aimed at investigating an aspect of communication patterns and effects among uni- versity faculty members. The plethora of "communication problems“ in large universities foster a need to develop more efficient and effective means of communication. As Dr. (---) suggested in his letter, your cooperation in this study will contribute to better understanding both of the problem and potential solutions. I will provide you with an abstract of the results of the study after it is completed. All individual questionnaires are strictly confidential. No one but myself will see them. In any report written from this study, it will not be possible to identify any individual nor will the College be identified in any way. Names become numbers and after the data are transferred to IBM cards the original questionnaires will be destroyed. The questionnaire is in three parts. It is important that you complete them in the order they are numbered. After you complete all three parts of the questionnaire, insert the completed forms in the enclosed envelope, seal, and give the envelope to your secretary. I will pick it up from her at the time on which we agreed. I will be available this week at (phone number) if you have any questions. Again, thank you. Sincerely, (Signed) Donald F. Schwartz 177 PART I (PINK SHEETS) Project Number 551A Deck Number _2 Subject Number UNIVERSITY COMMUNICATION STUDY First of all, a few vital statistics..... 1. In what year were you born? :7 2. What is your academic rank? (Check one) 1. Instructor 2. Assistant Professor 3. Associate Professor 4. Professor 5. Other (Please specify) 3. If you have departmental (Institute) or College level administrative duties, what is your formal title? (Check one) tute Director) 2. Department Chairman or Institute Director 3. Assistant or Associate Dean 4. Dean 5. Director (Coordinator) of (Please specify) 6. Other (Please specify) 7. None 4. Of your time devoted to only the following activities, approximately what proportion is devoted to each? (Do not list others: just indicate the proportions only among these.) 1. 2. 3. 4. 5. Classroom teaching (include preparation time and student advising). . . . . . . .- % Research (include time spent preparing papers) X Consulting (for the University and/or on a private 136818) . . . . . . . . . _% University, College, Departmental, or Institute level committee work . . . ..__JK Administrative duties (answer this onl if you have a formal administrative title . . ..__J% TOTAL 100% 5. What is your highest degree? (Check one) 1. Doctorate (or equivalent) 2. Masters (or equivalent) 3. Bachelors (or equivalent) . Other (Please specify) 1. Assistant or Associate Department Chairman (Insti- 178 6. Where did you Obtain your highest degree? (Check one) 1. (This University) 2. Other institution 7. In what year were you first employed at (this Univer- sity)? 7A. Have you been continuously employed (no gaps of less than one year) at (this University) since that year? -(Check one) 1. Yes 2. NO ——9 In what year was ur most recent appointment? £f7 8. On what time basis are you employed at (this University)? (Check one) 1. Academic year (9 months) 2. Calendar year (12 months) 3. Other (Please specify) 9. During the present school year (September 1966 to now) have you been absent from the campus for 3 or more con- secutive months? (Check one) 1. No 2. Yes __9.Which months were you gone? (Circle appropriate months) Se t O t. Nov De an Feb Mar.IA r 10. How many faculty or administrative committees do you belong to including both standing and ad hoc commit- tees? (Indicate number in box) 1::Z7'1. Departmental (or Institute) level committees A: 2. College level committees E 3. University level committees 11. In an “average“ month during this year, how many faculty or administrative committee meetings would you say you attended? 12. Do you currently have a joint appointment? (Check one) 1. NO ___2. JOint appointment with another department, insti- tute, or Office within the College of (---) ‘___3. JOint appointment with another department, insti- tute, or office in angther College. 179 12A. If you check (3) above, what actual proportion of your time do you spend in (this College)? 5% 13. How many professional journal articles have you pub- lished (or had accepted for publication) and how many papers have you presented at professional meetings since September 19_6_§? 5:7 (Combined total) 14. How many books have you published, either as sole author (editor) or co-author (editor), since September 199;? 180 PART II (YELIDW SHEET) PERSONAL CONTACT CHECKLIST Project Number 551A. Deck Number‘g Subject Number Now go back over the past two or three months and think of the professional people in the College of (---) with whom ypp‘have worked most glosely. We would like to have you list below the names of the people in the College with whom you work most closely. By “work with most closely“ we mean the professional peOple with whom you usually havelgp least gpg contact ‘pgp week on matters related to programs or activities of the College, or on teaching, research, or consulting in which you or the other person is engaged. YOu need only list people who are officed in (---) Hall. By “professional peOple” we mean faculty with academic rank of instructor or higher angzor administrators. For each of the individuals you list below, check how frequently in an ”average“ week you have contact with (talk to in person or on the phone, write) each of them. Name as many or as few peOple as accurately describe your usual contacts. h ( erson in the College with heck the appropriate fre- (A) List the name of eac B) For each person listed, hem u work most closel . en column. FREQUENCYfiQ§_CONTACT Several About 2 or 3 About NAME (Please print or times once times once write clearly) daily per day pgr wk, ppr wk, 1. .____ .____ .____ _____ 16. .___. .____ .____ .———— 181 _ PART III (WHITE SHEET) INSTRUCTIONS FOR PART III Completion of this part of the questionnaire will require you to work with the Personal Contact Checklist (Part II, yellow sheet). Please place it alongside this sheet for easy reference. Clipped to this sheet are several separate Personal Contact questionnaires. At the top of each questionnaire are the words, ”Name of Contact", followed by a blank space. Take the first questionnaire and, looking at the names listed on the yellow sheet, write in the blank space the name Of the first person on your list for whom you have checked either “several times daily" or "daily" as the frequency with which you have contact with this person. Then complete the ques- tionnaire keeping only that pgpson in mind. The questions generally relate to your communication with that person. Next, take the second questionnaire and repeat the above procedure: in other words, write in the blank space the name of the second person on your list for whom you have checked either ”several times daily" or "daily“, and then complete the questionnaire keeping only this second person in mind. Repeat this procedure for eagp othe; person on your list for whom you have checked either "several times daily“ or "daily" as the frequency of contact. ngpgg complete a questionnaire for anyone for whom you have checked any other frequency of contact column. There probably are more questionnaires than you need: if so, please return the unused ones along with the completed ones. If you need moge Personal Contact questionnaires, please call me at (---) and I will immediately deliver more to you. AFTER YOU COMPLETE a Personal Contact questionnaire for each person on your list with whom you have contact ”several times daily" or “daily", insert Parts I, II, and all Part III questionnaires in the enclosed envelope, seal, and give to your secretary. I will pick up the packet from her. Your OOOperation and patience are greatly appreciated. Thank you. NOTE: In the questions in each Personal Contact questionnaire references to “people (or faculty) in the College“, or just to "people (or faculty)" should be interpreted to mean fagulty with rank g; instrugtor 2; higher anQ‘o; administrators. 182 PART III (WHITE SHEETS) PERSONAL CONTACT QUESTIONNAIRE Name of Contact (Please print or write clearly) Project Number 551A Deck Number‘g Subject Number INSTRUCTIONS: Please complete this questionnaire in terms of the person whose name you have written in above as one with whom you have indicated contact “several times daily" or "daily" on the Personal Contact Checklist (Part II, yellow sheet). Keep only this person in mind when answering the following questions. Work through these questions without dwelling on any one -- give your first impression. * i t i t 1. Out of 100 times you might have contact with this person, about how many times would: (Fill in each blank) a. you seek him or initiate the contact (you go to see him, place the phone call, write). . . . b. he seek you or initiate the contact (he comes to you, places the phone call, writes). . . . . c. none Of the above, we just happen to meet (Neither of us deliberately goes to the other.).___ TOTAL = 100 contacts For each Of the following five items, circle the number preceding one response for each item.which best completes the sentence. 2. Considering the relationship I have with this person, when it comes to advice on matters related to teaching, research, or consulting, I depend on him: 1. a lot more 2. a good deal more 3. a little more 4. about as much : than (as) he depends on me. 5. a little less 6. a good deal less 7. a lot less 183 3. During the discussions I've had with this person in the past week or so about activities, programs, or people in the College of (---), I asked him.questions: 1. 2. 3. 4. 5. 6. 7. a lot more Often a good deal more often a little more often about as Often : than (as) he asked me. a little less Often a good deal less Often a lot less often 4. When this person and I discuss activities of the College or any of its sub-units, usually he tells me: very many more - many more a few more the same number of ?: things than (as) I tell a few less him. many less very many less .1 5. Across a number of contacts I may have with this person, I am: 6. 1. 2. 3. 4. 5. 6. 7. In a lot more likely ‘ a good deal more likely a little more likely about as likely >: to ask his opinions than a little less likely (as) he is to ask mine. a good deal less likely a lot less likely _ regard to lpggpppplg? about activities and personali- ties in the College of --->. generally speaking this person depends on me: 1. 2. 90‘th a lot more a good deal more a little more about as much >: than (as) I depend on a little less him. a good deal less a lot less _ 184 Please evaluate this person as a source of information in terms Of the following adjective pairs. Check only one of the seven points for each pair. For example: Introverted I I x I Extroverted '5‘ :3 z: 2"? 5 m E 3 E 3 5. § 3 S a 8 '3 :2 m OI m m m <3 m 7A. Friendlyl - Unfglendly 7B. Informed Uninformed 7C. Jggpl Unjust 7D. E rien ed Ineypgrienced 7E. Safe Unsafe 7F. Trained Untrained 7G. Cruel Fgggl 7H. Skilled Unskilled 7I. Qualified Ungualified 7J. Honest Dishonest For the remaining, circle the number preceding one res- ponse for each item which best represents your reaction to the item. 8. As new developments occur in the College Of (---). I usually “get the word“ from someone other than this person. 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 185 9. When interacting with other faculty members in the College, this person isn't a very good convinc.r. 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 10. This person has access to individuals who are rela- tiv?1y high in the “power structure” of the College of --- . 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 11. Those individuals who have a lot to say about what goes on in the College respect the suggestions this person makes. 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 12. This person communicates with.yg£y'ggy administrators and faculty members in the College of (---). 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 186 13. This person doesn't have much influence with the other people he works with in the College of (---). 1. agree very strongly . agree quite a bit agree moderately neither agree nor disagree disagree moderately disagree quite a bit disagree very strongly \IC‘UIwa s 14. This person works with people whose Offices are loca- ted in several different areas in (---) Hall (differ- ent floors, different ends of the building). . agree very strongly . agree quite a bit agree moderately neither agree nor disagree disagree moderately disagree quite a bit . disagree very strongly \IO‘UIPUNH . 15. This person works with faculty in the College of (---) who are "in the know“ about what's going on in the College. 1. agree very strongly 2. agree quite a bit . agree moderately . neither agree nor disagree disagree moderately disagree quite a bit . disagree very strongly 3 4 5 6 7 16. I see this person as a key figure who can support or block proposals made in the College of (---). 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly 17. 18. 19. 20. 187 This person has access to more members of the faculty of the College of (---) than do most others in the College. 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly Those relatively high in the ”power structure" of the College of (---) have been very receptive to this person's suggestions and Opinions. 1. agree very strongly 2. agree quite a bit 3. agree moderately 4. neither agree nor disagree 5. disagree moderately 6. disagree quite a bit 7. disagree very strongly In any organization like this College, clique groups develop. Would you say that this person has contact with peOple in more or fewer clique groups in the College than do most other faculty members? 1. very many more 2. many more 3. a few more 4. about the same number 5. a few less 6. many less 7. very many less When you learn about some ghange or pgy dea being proposed or discussed in the College of ---) or any of its departments or institutes, how likely are you to hear about it first from this person? 1. extremely likely . very likely fairly likely about 50 - 50 fairly unlikely very unlikely extremely unlikely ‘10“th . 188 21. Would you say this person works with peOple in more or fewer different departments and administrative offices than most other members of the faculty in the College of (---)? 1. very many more . many more . a few more about the same number . a few less . many less . very many less \lO‘UlofiwN e 22. How important are the members of the College Of (---) faculty with whom this person works most closely? 1. extremely important 2. very important 3. fairly important 4. so, so 5. fairly unimportant 6. very unimportant 7. extremely unimportant APPENDIX C VARIABLE INDEX FOR PART III - PERSONAL CONTACT QUESTIONNAIRE Item NumberIs) 1a, lb 1c A A‘ w NU UI bah- QQ GUI V 7A - 7J (7A,7C,7E,7G,7J) (7B,7D,7F,7H,7I) 8,20 9,11,13,16,18 (9.13) (11,16,18) 10,15,22 12,17 14,19,21 1. 5. 6. Vapiable Directional ratio of deliberate message transaction initiation Frequency of deliberately ini- tiated message transactions (100 - 1c) Specific Opinion leadership a. Pretest Type I: Information- advice dependency b. Pretest Type II: Asking behavior Source credibility a. Safety b. Qualification First source of information Diffuse Opinion leadership a. Type I: General persuasive- ness b. Type II: Persuasiveness within "power structure“ Importance of secondary contacts Perceived number of contacts Perceived structural diversity of contacts 189