University of Southern California Dissertations and Theses

A structural equivalence and contingency theory perspective on media usage and communication performance: The case of voice messaging

(USC Thesis Other)

A structural equivalence and contingency theory perspective on media usage and communication performance: The case of voice messaging

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A STRUCTURAL EQUIVALENCE AND CONTINGENCY THEORY PERSPECTIVE ON MEDIA USAGE AND COMMUNICATION PERFORMANCE: THE CASE OF VOICE MESSAGING by Douglas Edward Shook A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA in Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY {Communication Theory and Research) January 1988 Copyright 19 88 Douglas Edward Shook UMi Number: DP22434 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Dissertation Publishing UMI DP22434 Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SC H O O L UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90089 PbD. CM This dissertation, written by Douglas Edward Shook under the direction of h.i.f. Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies D ate iL U i.ll’ . . . . 1 .? . ? . 8 . . DISSERTATION COMMITTEE Chairperson Dedicated to my Mother and my Father ACKNOWLEDGMENTS I would like to acknowledge Dr. Bonnie Johnson, Greg Hoffman, Jan Muller, Dan Wiford, and Jeri Pantalone from the organizational site for their cooperation in a very complex and time-consuming data-collection process. Without very close cooperation from the organizational site, this research and this dissertation would not have been possible. I also would like to thank and acknowledge the Annenberg School of Commu nications for their generous support over the past three years. I consider it a privilege to have conducted my doctoral studies at this rigorous institution. I owe a great debt of gratitude to Dr.Omar el-Sawy and Dr. Everett M. Rogers who provided critical insight and exceptional guidance from the inception of this dissertation. They have been invaluable sources of knowledge and en couragement and I thank them for their time and willingness to work with me on such an arduous task. I especially thank them, however, for being generous, humane scholars, and good friends over the past few years. I cannot adequately acknowledge or thank my mentor Dr. Ron Rice. Profes- ; sor Rice has given far more of his time, knowledge, and energy than I ever could have expected. He is a brilliant scholar, tireless worker, and compassionate teacher. He represents an epitome of the ideal professor, dissertation commit tee chairman, mentor, and loyal friend. Working with him has been the highlight of my studies at the Annenberg School. I thank you Ron. Ultimately, I thank my mother and father who have stood beside me when i only truly loving parents could have done so. Although the successful defense of | one’s dissertation is a rare privilege, its importance pales in comparison to the (honor I derive from being my mother and father’s son. I ( 1 TABLE OF CONTENTS Page LIST OF TABLES ................ ............................................................................ viii LIST OF FIGURES................................................................................................ x ABSTRACT........................................................................................................... xii CHAPTER ONE: MOTIVATION AND DESCRIPTION OF THE MODEL 1.1 M otivation................................................ ........................... ................ 1 1.2 Description of M o d e l..................................................................... .. 3 1.2.1 Structural Equivalence and Contingency T h e o ry ................... 3 1.2.2 Voice Messaging and Communication Perform ance 4 1.2.3 Strengths of the Present Research ...................................... 5 CHAPTER TWO: THEORY AND LITERATURE REVIEW 2.1 Organizational Contingency Theory............................................... 9 2.1.1 Organizational Technology .................................... 10 2.1.2 Mechanistic and Organic Systems ........................ 12 I 2.1.3 Differentiation and Integration.................................................... 14 2.1.4 Interdependence in Workflows ................................................. 15 i 2,1.5 Task Certainty and Anaiyzability ....................................... 17 2.1.6 Information P rocessing.............................................................. 19 2.1.7 Information Richness.............................. ............................... 19 2.1.8 Summary ..................................................................................... 20 2.2 Formal and Informal Organizational Structure . .................... 22 2.3 Structural Equivalence . ................. 25 Page , i 2.4 Media Usage Considerations........................................................ 29 | 2.4.1 Social P resence.................................................................... 29 ! 2.4.2 Information Richness..................................................................... 32 2.4.3 Social Information Processing.................................................. 37 i 2.5 Com puter-M ediated M e d ia ............................................................. 39 2.5.1 Performance and Benefits ........................................................ 39 2.5.2 Voice Messaging ........................................................................... 42 2.6 The Proposed M o d e l........................................................... 44 2.6.1 Structural Equivalence................................................................. 46 2.6.2 Task Characteristics.................................................................. 48 2.6.3 Communication P erform ance.................................................. 50 2.6.4 Control Variables ....................................................................... 51 2.6.5 Summary ................... 53 CHAPTER THREE: METHODS 3.1 Data C o lle c tio n .................................................................................... 54 i 3.1.1 Sam ple............................................................. 54 j 3.1.2 Data S ources................................................................................ 55 j 3.1.3 Measures of Predetermined Variables ............ 56 I 3.1.4 Measures of Media Usage ......................................................... 57 3.1.5 Measures of Communication Perform ance........................... 58 3.2 Data T ransform ations........................................................................ 59 i 3.2.1 Normalizations............................... 59 3.2.2 Structural Equivalence Transformations ............................... 59 j .3.2.3 Scales and Dichotomizations ................................................... 61 i t v Page 3.3 A n a lyse s............................................................................................... 63 CHAPTER FOUR: RESULTS 4.1 Descriptive Statistics ..................................................................... 64 4.2 Tests of Hypotheses................................. 66 I 4.2.1 Hypotheses 2.1 and 2.2 ......................................., ................. 66 4.2.2 Hypotheses 1.1 and 1,2 ............... 70 4.2.3 Hypotheses 2.3 and 2.4 ....................................................... 73 4.2.4 Hypotheses 3.1, 3.2, and 3 . 3 ................ ................................ 76 4.3 Structural Equation M o d e ls ............................................................. 80 4.3.1 Hypotheses 4.1 and 4.2 .......................................................... 84 4.3.2 Analyses of Hypotheses 1.1 and 1 .2 ........................................ 84 4.3.3 Communication Performance ................................................... 85 4.3.4 Control Variables ........................................................................ 86 4.3.5 Task Environment and Messaging Behaviors . , .................. 87 4.3.6 Full Matrix M odel.......................................................................... 87 4.3.7 Low Analyzable Environments ..................... . 91 4.3.8 High Analyzable Environments................................................ 94 4.3.9 Comparisons of Task Environments............... 94 4.3.9.1 Messaging and Performance ....................................... . 98 4.3.9.2 Low Analyzable Environments.......................................... 100 4.3.9.3 High Analyzable Environments......................................... 105 , 4.3.10 Summary...................................................................................... 105 CHAPTER FIVE: DISCUSSION AND CONCLUSIONS 5.1 Purpose of the D issertation............................................................. 107 j 5.2 Implications of Hypotheses Tests ............................ 108 ! I 5.2.1 Position Effect .................. 108 | 5.2.2 Positional Adjacencies ....................................... 110 I vi i Page 5.2.3 Task Environment and Active Voice M essaging.................. 111 5.2.4 Task Environment and Position Effect ................................... 113 5.2.5 Communication Performance . . . ......................................... 114 5.2.6 Additional Hypotheses and Discussion of the M o d e l 117 5.3 Empirical Comparison of Test Model With a Conventional, Individual-level M o d e l................ 120 5.4 Specific Strengths and Weaknesses of the Proposed Model ....................................... 125 5.5 Implications of the Present Research for Communications Applications........................................................................................... 129 5.6 Implications of the Present Research for Communications Theory and R esearch........................................................................ 133 Notes .................................................................................................................. 135 .References........................................................................................................... 136 Appendices ......................................................................................................... 145 Appendix 1 Algorithms Used to Create Structural Equivalence Variables ............................ 146 ! Appendix 2 Time 1 Questionnaire . . ............................................................ 149 .Appendix 3 Time 2 Questionnaire........................................ 159 Appendix 4 Example of Computer-Monitored D a ta .....................: ........... 170 l f vii LIST OF TABLES Table Page 1 Contingency-Based Concepts Central to Present W o rk................................... 21 2A Continuum of Information Richness ................................................ 33 2B Extended Continuum of Information Richness................................ 36 3 Scale Creation S ta tistics........................................... 62 4 Summary Statistics ............................................................................ 65 5 Zero-Order Correlation Coefficients ............. 67 6 T-Tests of Mean Differences Based Upon Task Environment ............................................................................. 68 7A Impacts of Structural Equivalence on Active Messaging Behaviors........................................................................ 71 7B Impacts of Structural Equivalenceon Passive Messaging Behaviors........................................................................ 72 7C Z' Test of Regression Coefficients - Comparison of Position Effect on Comunication Performance ..................... 75 8A Impacts of System Usage on Communication Performance (Active System Usage) ....................... 77 8B Impacts of System Usage on Communication Performance (Passive System Usage) ........................................ 78 9 Full Effects and Phi Identification for Figure 3 . , ...................... 83 10 Full Effects and Phi Identification for Figure 5 ...................... 90 viii LIST OF TABLES Table Page 11 Full Effects and Phi Identification for Figure 6 ............................ 93 12 Full Effects and Phi Identification for Figure 7 ............................ 96 13 Horizontal Differentiation and Messaging ...................................... 97 14 Full Effects and Phi Identification for Figure 9 ........... 102 15 Full Effects and Phi Identification for Figure 10 .......................... 104 16 Test of the Effect of Non-Independent Observations On Measures of Position ........................................ 127 6 45 82 88 89 92 95 99 101 103 106 x LIST OF FIGURES Basic Conceptual Model ................ .................. Model of Voice Messaging Use, Structural Equivalence, and Communication Performance . . . Structural Equivalence, Voice Messaging Behaviors, and Communication Performance (Full Matrix) .................................................................... Theoretical Model of Structural Equivalence and Voice Messaging Behaviors .............................. Structural Equivalence and Voice Messaging Use (Full Matrix) .......................................... .............. Structural Equivalence and Voice Messaging Use (Low Analyzable Environments) ....................... Structural Equivalence and Voice Messaging Use (High Analyzable Environments) ....................... Theoretical Model of Voice Messaging Use and Communication Perform ance..................... Voice Messaging Use and Communication Performance (Full Matrix) .......................................... Voice Messaging Use and Communication Performance (Low Analyzable Environments) .... Voice Messaging Use and Communication Performance (High Analyzable Environments) .... LIST OF FIGURES Figure Page 12 Individual-Level Analysis of Task Environment, Voice Messaging Use, and Communication Performance .............. 122 13 Individual-Level Analysis of Task Environment, Voice Messaging Use, and Communication Performance (Full Matrix Model) ................................................... 123 r ABSTRACT The present research proposes and tests a model of organizational media usage drawn from a confluence of traditional contingency theory and concepts of structural equivalence. Specific foci include structural determinants of voice 'messaging system usage and the effect of system usage upon communication performance. The model is tested using a longitudinal study of a large, distributed service organization during a pilot test of a voice messaging system (a com puter-assisted telephone system capable of storing and forwarding messages in much the manner that electronic mail systems store and forward text-based messages). Data-collection included behavioral data continuously monitored by ; computer, pre- and post-test questionnaires, and extensive use of organizational archival data. Data-analyses consisted of researcher-written programs, LISREL estimation, and conventional descriptive and inferential statistical approaches. Conceptualization and operationalization were conducted at a work group level of analysis as determined by a structural equivalence analysis of the formal organ izational structure. Hypotheses relating structural equivalence measures of task environment to system usage, and system usage to changes over time in com munication performance, received strong support. Structural equivalence was a powerful predictor of voice messaging behaviors. Low analyzable and high uncertainty task environments were more strongly associated with use of mes- ■ saging features hypothesized as possessing the capacity to facilitate collabora- . tive work. Low analyzable and high uncertainty environments also were associ ated with greater increases in relative communication performance as a result of ! active use of voice messaging. Additionally, structural equivalence-based mod- < i j | i xii ;els of messaging behaviors and performance offered much greater explanatory i 'power than conventional models based upon individual-levels of analyses. CHAPTER 1: MOTIVATION AND DESCRIPTION OF THE MODEL 1.1 - Motivation Organizations are operating in increasingly complex and turbulent environ ments, Many tasks facing large organizations demand a variety of individual talents and skills, and thus require the organization to assign portions of tasks to organizational units to be addressed by specific organizational members. Addi tionally, during basic research, new product development, or initiating new pro jects or approaches, etc., it is common for personnel to be organized around projects, or to work collaboratively on various portions of organizational tasks . (Burgleman and Sayles, 1986; Galegher, Kraut, and Egido, 1987; Katz and Tushman, 1979; Keller, 1986). As tasks are “ disassembled” and portions of them are assigned to organizational units, or as organizational members are re quired to operate in collaborative arrangements for the completion of tasks, the ability to transfer information and communicate rapidly and effectively becomes critical for effective group performance. Chapter 2 provides a more comprehen sive discussion of differentiation and integration (Lawrence and Lorsch, 1967) and organizational information processing (Galbraith, 1973). | Concurrent with increasingly complex tasks and turbulent environments, or ganizations are confronted by an increasing number of technologically sophisti- ' cated communication systems (Huber, 1984). Electronic messaging via comput ers (electronic mail or computer-conferencing), teleconferencing, videocon-■ ferencing, facsimile, and voice messaging systems (computer-aided telephone systems capable of storing and forwarding spoken messages in much the man- 1 ner that electronic mail systems store and forward text-based messages) are now used in a variety of organizations in conjunction with more traditional organ izational media. Current theory intended to explain media usage patterns has been unsatisfac tory for two major reasons: 1) models of media usage based only upon media attributes and task demands are generalizable across a wide variety of contexts, but have offered little explanatory power in non-experimental settings, and 2) theoretical approaches attempting to explain organizational media usage through psychological factors offer insight into individual behavior patterns, but are unable to offer a systematic or generalizable organizational-level model of media usage. The present research is an attempt to offer a model of organizational media use and communication performance based upon a realistically complex integra tion of organizational contingency theory with concepts of structural analysis and media attributes. More explicitly, this dissertation rejects the general proposition that organizational media usage is best explained by a rational, individual-level model of matching media attributes with conventional measures of task environ- * merit. The present research argues that an organization is far more than a simple aggregation of its members. From a contingency and structural equivalence , perspective, organizations are best understood and measured as systems of differentiated units. Individual-level conceptualizations and operationalizations re- ■ move the organizing effect inherent in organizations, thus creating a vastly differ- ; ent and weaker model of organizational media usage. Individual-level analyses 1 basically strip from the organization what makes it “ organized” and what differen- i tiates it from a random grouping of persons. [ This dissertation attempts to use structural equivalence approaches to pro- i vide powerful, yet parsimonious model of organizational media use. j 1.2 - Description of the Model i Although the following chapters offer a more detailed description of the theo retical model presented and tested in this dissertation, the present section offers a brief overview of key concepts and relationships. 1.2.1 - Structural Equivalence and Contingency Theory The model presented here views formal groupings of individuals in organiza tions (e.g., work units) as products of the same forces that shape the structure and form of organizations. While contingency factors traditionally are hypothe sized to affect organizational structures such as the degree of hierarchy and degree of differentiation, it is argued here that they also create (1) formal group ings of individuals, and (2) chains of authority and responsibility (see section 2.1 for a description of organizational contingency theory). It is further argued that these groupings and chains of formal interactions may be modeled through a ■network analysis approach called structural equivalence, and can offer additional insight into media usage behaviors at a work unit level of analysis (see section 2.3 for a description of structural equivalence). This is to say that organizational subunits and authority links are, to a large i j degree, manifestations of organizational responses to environmental demands. ' These groupings and linkages represent specific environments produced by for mal organizational structure (Brass, 1981; Downey and Slocum, 1975). This use I of forma! work units is seen, then, as a logical progression or extension of con ventional contingency theory. It is viewed as a missing piece in the overly simpli fied and rational model of organizations generally painted by traditional contin- | gency theory. This approach is an attempt to maintain the logical and j generalizable theoretical appeal of contingency models, while incorporating the I j influence of formally designated, sociological structures (e.g., work units), i The present research uses the organizational subunit level of analysis as de termined through a structural equivalence "mapping” of formal organizational i charts. Although it will be described in greater detail in following sections, hori zontal and vertical organizational differentiation create structurally equivalent roles i :or positions. A central element in this organizational mapping is the link between superiors and subordinates. Organizational work units are defined as groups of persons linked to the same superior as determined by formal organizational hierarchy. A -igroup of persons linked to a single superior is defined as a position (i.e., persons 2 *who are similar with respect to horizontal differentiation), A position is the basic ■ element of organizational differentiation used in this research, Linkage between a given position and the adjacent higher position in the for- ; mal organizational hierarchy is the basis of measuring vertical organizational dif ferentiation, The structure of an organization, then, is seen as many horizontally differenti- I ' ated positions linked through a series of ties to superiors in vertically differenti ated positions, These two kinds of positions are the units of analysis for all meas urement in this research. I 1.2.2 ~ ~ Voice Messaging and Com munication Performance The proposed model posits that horizontal differentiation creates specific task environments which Influence the usage levels of voice messaging (for a discus sion of the relationship between task environment and organizational media use, see section 2.4), Additionally, usage levels will be influenced by communication 1 through the authority/responsibility iinks created by vertical differentiation. It also argues that voice messaging offers capabilities which particularly could enhance ^ communication performance in task environments with less analyzable tasks or where collaboration is necessary. The model also proposes that usage levels will be higher and that perform- ' ance gains will be greater in task environments characterized by collaborative or unanalyzable tasks. Figure 1 offers a basic conceptual depiction of the model. i 1.2.3 - Strengths of the Present Research This research offers several conceptual and empirical contributions to our understanding of organizational media use. Conceptually, the study attempts to unite contingency theory with structural equivalence. This marriage is seen as a : logical extension to the rather sizable body of theory under the heading of contin gency theory as well as the considerable research in network analysis based on structural equivalence. Additionally, the study conceptualizes and operationalizes at the organizational work unit or subunit level of analysis. This research also examines voice messaging (a computer-assisted tele phone system capable of storing and forwarding messages in much the manner that electronic mail systems store and forward text-based messages) as an : example of the state of the art in computer-mediated communication systems i i for organizations. Where there is a considerable body of literature on I text-based, computer-mediated communication systems (Rice, 1987), very lit- | tie published research exists on computer-mediated, voice-based systems. The j present research is timely. Empirically, the study offers several contributions. Though it is not a multi-or ganizational study, the sample is drawn from a wide variety of organizational I 5 FIGURE 1 - BASIC CONCEPTUAL MODEL TIME ONE COMMUNICATION PERFORMANCE SUPERVISOR’S INNOVATIVENESS ORGANIZATIONAL LEVEL INNOVATIVENESS HORIZONTAL DIFFERENTIATION TIME TWO COMMUNICATION PERFORMANCE VOICE MESSAGING BEHAVIORS VERTICAL DIFFERENTIATION EDUCATION TENURE departments in three cities (one large city, and two medium-sized cities). It also is a longitudinal study of media usage. Performance measures were taken prior to the implementation of the system and after the system had been in use for approximately five months. The study also uses continuously monitored behavioral data. A variety of ■ system usage measures were captured by the voice messaging computer and :are the source of the voice messaging behavior data used in this study. : Very detailed measures of messaging system usage were monitored includ ing: number of messages sent, number of messages received, lengths of mes sages sent and received, number of messages stored, use of the system as an “ answering machine” versus use of the machine as an “ active messaging sys tem ,” etc. It is the premise of this research that human communication proc esses are very complex and that it is inappropriate to attempt measurement of these processes with "coarse” instrumentation (e.g., a single, overall measure of usage). The number of messages sent or received, for example, is a behavioral measure of communication frequency or intensity, while length of messages may Indicate the depth or complexity of the communication. As much of the present ! research involves the number of exceptions in a task environment (e.g., reflected , in number of messages) as weli as the relative analyzability of those exceptions ; (e.g., reflected in the length and complexity of communication), the appeal of • these two measures becomes apparent (Perrow, 1972). i Additionally, these multiple measures more accurately monitor variations in i I communication preferences (e.g., persons may routinely 'send a relatively ^ greater number of short messages in comparison to another user who sends relatively fewer, longer messages). Clearly, the use of multiple indicators in I general is preferable to reliance on single indicators (Cook and Campbell, 1983). ; A variety of organizational archival data were collected on organizational hier archy and superior-subordinate links. Researcher-written programs were used to convert these archival data into structural equivalence “ maps” of organiza tional vertical and horizontal differentiation. This approach allows the study to use a non-arbitrary, organizational subunit level of analysis throughout all analyses. Finally, data analysis techniques include LISREL. The LISREL approach en ables the creation and evaluation of realistically complex, causally-oriented co- i ' variance models involving systems of interdependent, simultaneous equations. This analytical technique is highly beneficial when examining models with multiple, interdependent endogenous variables. , ■ The intent of this research is not to test formally theories of organizational contingency or structural equivalence. This dissertation rather is an attempt to offer a generalizable, a priori model of organizational media use which applies conceptual approaches drawn from both theoretical bodies to answer four re search questions: 1 What is the effect of task environment, as defined by horizontal ! structural equivalence, on voice messaging use? I i ! 2 What is the effect of vertical differentiation, as defined by struc tural equivalence, on voice messaging use? 3 What is the effect of voice messaging upon communication per formance? 4 How do the previous associations vary when measured across task | environments of different levels of analyzability? 8 CHAPTER 2: THEORY AND LITERATURE REVIEW The first five sections, of this chapter examine the theoretical approaches and literature relevant to the present research. A summary of central conceptual thinking as well as the particular strengths and weaknesses of each approach for the present research are discussed. The final section presents hypotheses de rived from the theories and literature discussed, and presents an empirical model for estimation. 2.1 - Organizational Contingency Theory Contingency theory is presented first as it establishes a broad conceptual foundation upon which to build the theoretical framework for this dissertation. The .theory offers an attractive, process-oriented mode! of organization relating : structure to performance. Although some disagreement exists as to a definitive description of contin gency theory, Khandwalla (1977) describes contingency theory as similar to open systems theory in that it gives a great deal of weight to the interface of the : organization with its environment, The difference between the two sets of theo- [ ries lies in the tendency for open systems theory to dwell on the dynamics of the ' processes through which an organization adapts itself to the environment, while j contingency theory is more concerned with the end result— organizational per- j formance as mediated by an organization’s structural responses to environmental | demands. Contingency theory has been compared to a Skinnerian approach— a ( stimulus/response model which ignores the process in between. i 9 | Khandwalla (1977) includes the theories of Woodward (1958, 1965), Burns land Stalker (1961), Thompson (1967, 1973), Lawrence and Lorsch (1967), : and Perrow (1972) as contingency theories. Miner (1982), on the other hand, .classifies only the theory of Lawrence and Lorsch as contingency theory. He classifies the work of Woodward, Perrow, and Burns and Stalker as theories of technological imperative; the work of James Thompson as sociological open systems theory; and the work of Katz and Kahn as psychological open systems theory. Although Miner's distinctions are valid and helpful in many applications, Khandwalla's less detailed categorizations will suffice for the present overview. The most basic premise of contingency theory is that the structure of an , organization is an product of the interaction of its size, legal incorporation, char acter of its markets, constraints, nature of dominant task, problems, and tech nology, Technology as used in this literature describes the basic processes through which an organization executes its tasks— technological, social, etc. The concept of contingency lies in the belief that there is no single “ best” struc ture for an organization— structure is, and should be, contingent upon the overall conditions in which the organization operates. Organizational effectiveness, then, is dependent upon how well the structure of the organization “ fits” these overall 1 conditions, I '2.1,1 Organizational Technology l j Joan Woodward's research in the late 1950’s generally is regarded as the seminal work in this area (Miner, 1982; Khandwalla, 1977). Her theory is an j inductive one as it was born out of a study of 100 English firms where she was I attempting to test the principles of management process (Woodward, 1958). ; Woodward found that generally, organizations ignored management principles 1 and that the differences she found in organizational structures were closely asso ciated with organizational technology. She developed a scale for measuring what she called technological complexity which used three major classifications: 1) unit and small batch, 2) large batch and mass production, and 3) process. The scale measures technical complexity increasing as production processes be come more controllable and yield more predictable results (Woodward, 1958). The basic hypothesis of her work is that organizations most clearly approxi mating the “ typical” structure for their technology should be the most successful. 'Success is a function of an appropriate fit between technology and structure. As an example, organizations with relatively stable and certain technology (i.e., large batch and manufacturing) should be characterized by reliance on hierarchy and formally delegated rules and responsibilities— the classical principles of manage ment (e.g., Fayol, 1937; or Taylor, 1911). Organizations on either end of the scale (i.e., unit and small batch or process), however, should be characterized ' by either line domination or functional form with specialists making key decisions (Miner, 1982), The clean iogic and appeal of Woodward's original theoretical statements were moderated, however, when she found, ironically enough, that manufacturing ' firms were not as closely related to the technology scale as the organizations on either end. With this finding she incorporated concepts of control. I | The genera! impression...that there was greater variation in the way pro duction operations were planned and controlled in firms in the middle ranges of the scale was confirmed by the follow-up studies (Woodward, 1965). With these findings, Woodward incorporated the additional conceptual com- i ponents of control and uncertainty. Technology and control determine uncer- jtainty ieveis, and uncertainty, in turn, influences structure. Technology, control, uncertainty, and structure are, to greater or lesser degrees, key components in most contingency-based theory which followed Woodward’s work. They also i ;have become central elements in organizational media usage theories focusing Ion associations between task and media characteristics (Daft and Lengel, 1984, ; 1986). j Some scholars have judged Woodward’s work to be methodologically sloppy i ] (Khandwaila, 1977), and since the theory was drawn inductively from the data, !her findings do not qualify as tests of theory. Her theory appeared to be in transition at the time of her death, with an increasing emphasis on uncertainty, but it is not entirely clear where the theory was heading (Miner, 1982). 2.1.2 Mechanistic and Organic Systems A somewhat different approach to examining organizational structure and per form ance is found in Burns and Stalker’s theory of mechanistic and organic sys tems (Burns and Stalker, 1961). In their theory, organizational structure is de pendent upon change rather than technology as the contingency variable. Burns and Stalker conducted qualitative analyses of 20 British electronics organizations I using unstructured interviews. Their general findings were that there are two op- ! posite management styles: the mechanistic and the organic. The mechanistic style is characterized by the elements of bureaucracy: for- j mal hierarchy, strict adherence to rules, and structured communication, rules, and responsibilities, It also incorporates some elements of the classical manage ment school including the importance of status cues and team loyalty. The organic style, conversely, is characterized by informality, flexibility, free . and open communication both vertically and horizontally, authority resting in the situational need rather than the formal hierarchy, goals are universally held, etc. I Clearly, these two styles are "ideal types" in the Weberian sense of the term : (Weber, 1946). Burns and Stalker did, however, observe that firms character ized as mechanistic had difficulty in adapting to change or innovation. In mecha- ■ nistic organizations, organizational "exceptions" are continually passed vertically . up into the hierarchy for management action. Given enough change or innova tion, management and the communication channels leading to them are over loaded. While the mechanistic approach may be very efficient for a stable and certain task environment, change is not well accommodated due to the formal structure. The organic style, however, is very effective at accommodating change as it, in a sense, can "restructure” itself around the innovation for most effective ac tion. The increased open communication and situational authority allows for ap propriate behavior in the face of change. Burns and Stalker (1961) note several potentially negative situations arising from an overloaded mechanistic system including: 1) competent individuals may I "fail” because they are placed in departments whose policies prohibit needed change; 2) chains of command are bypassed due to overloaded links, thus fur- | ther overloading key persons; 3) a "jungle" situation where new branches are i I I simply grafted on to the existing bureaucratic tree in an attempt to handle over-' load rather than making needed change, The theory is relatively vague as far as predicting when these negative situ ations would arise other than in times of change. Burns and Stalker never put j their theory to an empirical test, and, although it contains the well known and ! I 13 1 I often cited concepts of organic and mechanistic systems, it has received little j ; direct empirical attention. I 2.1.3 Differentiation and Integration Lawrence and Lorsch. (1967) borrowed heavily form the works of Woodward ; and Burns and Stalker to create a more compiex theory of organizational struc- , ture centered upon the concepts of differentiation and integration. The concept of differentiation is central to the present research: [An organization is] a system of interrelated behaviors of people who are performing a task that has been differentiated into several distinct subsys tems, each subsystem performing a portion of the task, and the efforts of each being integrated to achieve the performance of the system (Lawrence and Lorsch, 1967). Lawrence and Lorsch argue that differences in task environment cause the organization to become differentiated. For example, assuming a word processing pool's task environment is more certain than that of the research department, the word processing pool should become more structured than the research department. Differentiation could, exist in a great variety of task characteristics (e.g., goals, planning, time demands, etc.), all of which would contribute to the creation of an organization which is functionally a composite of dissimilar parts. The more differentiated the organization becomes, the greater the need for integrative forces to arise to coordinate the organizational activities. Integrative forces listed by Lawrence and Lorsch (1967) include the formal organizational, hierarchy, designing chains of command, use of integrative committees or liai sons, establishing standard operating procedures, and training managers in hu man relations. Explicit analyses of communication, however, received minimal attention. | Aside from creating a more comprehensive and complex model of organiza tional structure, Lawrence and Lorsch's major contribution for this research was the recognition that uncertainty is associated with systematic variation of struc ture within organizations as well as between organizations. By dealing with organ izational subunits and their environments separately, Lawrence and Lorsch avoided the trap of organizational averaging and abstractions (Miner, 1982). An organizational subunit or work unit level-of-analysis is at the very core of the , present research. It is posited that this level of analysis is an appropriate concep tual and operational approach for contingency-oriented research, Additionally, the present research is rooted in the belief that low explained variance typical of most contingency oriented research is in part attributable to a failure to opera tionalize at this level. 2,1.4 Interdependence in Workflows James Thompson’s work (1967) contributed the concept that organizations have a “ technical core.” This technical core houses the organization’s crucial throughput mechanisms. In order to optimize effectiveness, the core needs to be buffered from the environment and input and output operations of the organi zation (e.g., for a factory, the assembly line would represent the technical core ( and the stockpiling of raw materials and the wharehousing of assembled units i I would be a way to buffer the core). This buffering allows the core to be struc- i , tured and operated in as efficient manner as possible as it need not adjust for i environmental or intraorganizationai vicissitudes. | Thompson also conceptualized environments as having two primary dimen- I sions: 1) the degree to which it is stable/unstable, and 2) the degree to which it is heterogeneous/homogeneous, which affect organizational structure. The more [stabie and homogeneous the environment, the better able the organization is to | 'manage the environment by using rules and categories for applying those rules. i The greater the instability and heterogeneity, the greater the need for organiza tional subunits, contingency planning, decentralized decision making, and moni- jtoring. One of Thompson’s most important contributions to the present research, ; however, is the concept of intraorganizational subunit dependence in workflows. ,This concept of dependence is significant as it specifies the actual mechanism through which organizations are structured or change their structure (a factor curiously absent in most contingency literature). Although Thompson specifies various levels of dependence and the characteristics associated with each, the basic notion is that when subunits are dependent upon one another, their activi ties must be coordinated. Organizations minimize these coordination costs by grouping the dependent units “ close together. ’’ The closeness here may be .hierarchical as in putting several subunits under the control of one manager, or physical as in locating all faculty in a particular discipline in close geographic 'proximity (e.g., the same office or building). When it is impractical for an organi zation to physically or hierarchically move dependent units closer together, an ( organization tries to minimize coordination costs by instituting rules for the interac- : i ' tions of interdependent personnel. Thompson has taken the hypothesis of uncertainty as a force which shapes j ■ structure to a logical end by specifying the actual mechanism through which this 1 ' process occurs. His model depicts dependency shaping rules, hierarchies, physi- \ cai location, communication linkages, etc. ^2.1.5 Task Certainty and Analyzability The final contributor to contingency theory to be mentioned is Charles Per- row. Perrow's work appears to be influenced quite heavily by Thompson’s earlier work (Miner, 1982), and his contributions to contingency theory, though some what limited, have special salience for the present communication research. Per- row conceptualized technology as having two major dimensions: number of ex ceptions to routine procedures, and given an exception to routine procedures, the analvzabilitv of the search necessary to determine the appropriate response (Perrow, 1972). Clearly, this 2 by 2 approach results in four quadrants: Perrow’s Task Categories 1 “craftsman" - few exceptions/unanalyzable search • 2 “nonroutine” - many exceptions/unanalyzable search 3 “engineering” - many exceptions/analyzable search 4 “routine" - few exceptions/analyzable search ENGINEER NONROUTINE ROUTINE CRAFTSMAN DECREASING ANALYZABILITY -----------------------------► 17 1 Perrow, for example, relates "nonroutine” tasks to Burns and Stalker’s or ganic system, and “ routine” tasks to their mechanistic system. Perrow attempted to extrapolate these dimensions, which are appropriate for ! analysis at the level of task environments or work groups, to categorizations of • entire organizations. In view of the work of Lawrence and Lorsch and Thompson on subunits, this effort was, perhaps, simplistic. Lawrence and Lorsch described organizations as assemblages of many dif ferentiated subunits. An attempt to use Perrow's dimensions at an organizational level demands that the varied task environments across many work units are measured as a mean. Because organizations are differentiated, generally, they would include a variety of task environments. This level of analysis tends to aggregate potentially disparate units and does not reflect the structure of organ- ■ izational differentiation. Perrow's dimensions of analyzability and numbers of exceptions are the bases , of the conceptual and empirical classifications of task environments in the pre- ' sent research. The dimensions, however, are used at the work unit level of analy sis. Though other have built upon and modified these dimensions (Galbraith, ' 1973, 1977; Daft and Lengel, 1984, 1986), Perrow's original schema is highly , attractive for this research. i Although formal research in contingency related areas has declined, the infor- i ' mation processing model (Galbraith 1973, 1977), and information richness i 'm odel (Daft and Wigington, 1979; Daft and Lengel, 1986) are deeply rooted in ■ the basic premises outlined previously. Additionally, the media appropriateness I considerations of information richness are very similar to social presence con- t structs developed by Short, Williams, and Christie (1976). Information richness | will be discussed in two parts: 1) its relationship to contingency theory and infor- 18 ;mation processing (section 2.1.7), and 2) its relationship to social presence as a (theory of organizational media use (section 2.4). .2.1.6 - Information Processing A sizable body of work in itself, information processing primarily builds upon Thompson’s concept of intraorganizational subunit dependencies (Thompson was Galbraith’s Ph.D. seminar professor) and Perrow’s concept of numbers of exceptions to routine operations. In the information processing model, subunit : dependencies manifest themselves as information flows for the purpose of reduc in g organizational uncertainty. Uncertainty is defined as the difference between the amount of information required to perform the task and the amount of infor mation already possessed by the organization (Galbraith, 1977). As it is used here, uncertainty is nearly indistinguishable from Perrow's dimension of numbers of exceptions to routine operations. The greater the gap between available information and needed information, the greater the uncertainty. Accordingly, the greater the level of uncertainty, the greater the need to process additional information. Galbraith described the vari- , ations in organizational form as being dependent upon the amount of information necessary to reduce uncertainty and thereby attain an acceptable level of per- i iformance (Daft and Lengel, 1986). Much of Galbraith's writings concern design , or structural mechanisms through which an organization can control information I flows (i.e., either facilitate or limit). t I . 2.1.7 - Information Richness : While Galbraith’s work centers on uncertainty and the quantity of information i j flow between organizational subunits, Daft and Lengel (1986), and Daft and 19 |Wigington (1979) also incorporate concepts of equivocality and information rich ness. Equivocality is nearly the same as Perrow’s dimension of degree of task analyzability (i.e., the analyzability of the search required to obtain information necessary to resolve an exception to routine operations), Equivocal or un- analyzable tasks are not resolvable simply by acquiring additional information: Equivocality seems similar to uncertainty, but with a twist. Equivocality i presumes a messy, unclear field. An information cue may have several ; interpretations. New information may be confusing and even increase un- | certainty. New data may not resolve anything when equivocality is high. Managers will talk things over, and ultimately enact a solution. Managers reduce equivocality by defining or creating an answer rather than by learn ing the answer from the collection of additional data (Daft and Lengel, 1986). This model breaks somewhat from Galbraith’s approach in that it creates an ; information dimension that is less restricted or less limited by rationality. It be comes directly relevant for communication research as it proposes that media possess varying capacities for conveying rich information— organizational media are not equally appropriate for resolving equivocality. Information richness will be defined formally and discussed in greater detail in section 2.4 on media usage considerations. i i i I j 2.1.8 - Summary : Miner (1982) states that the literature related to organizational technology is I ' about as confused as any in the area of organizational theory, and Donaldson i ' (1976) and Reimann and inzerilii (1979) state that major reviews of research come to near polar opposite conclusions. Additionally, much of the research has j failed to uncover significant associations, and many of the significant associations , uncovered have not been in the hypothesized directions (Miner, 1982). i 20 TABLE 1 Contingency Based Concepts Central to the Present Model Theorists Woodward (1958, 1965) Burns & Stalker (1961) Lawrence & Lorsch (1967) Theoretical/Conceptual Contributions organizational structure varies systematically with technology organic and mechanistic structural responses: appropriateness contingent upon environmental stability differentiation and integration: organizations are complex systems of interrelated subunits - differentiation enables unit specialization while integration allows coordination of subunits Thompson (1967) Perrow (1972) Galbraith (1973, 1977) Daft & Wiglngton (1979) Daft & Lengel (1986, 1987) intraorganizationai subunit dependencies in workflows: costs of coordinating dependent subunits are minimized by reducing hierarchical or physical distances among interdependent subunits organizational task environments possess two independent dimensions: number of exceptions to routine and the analyzability of the search for an appropriate response to the exception information processing: subunit dependencies manifested as information flows to reduce organizational uncertainty information richness: addition of equivocality and information richness of organizational media to the information processing model 21 | Although contingency theory will be discussed in relationship to structural equivalence and media attributes in the following sections, particular contin- . gency-based implications for the structural equivalence model proposed in the 'present research build upon the following observations: 1) Structural features of successful organizations are not random, but are rational efforts adapting the organization to its environ ment for the completion of its tasks. 2) Organizational differentiation results in subunits with particular re sponsibilities in a chain of task completion. 3) Subunits develop specific characteristics in response to task de mands. 4) Subunits are staffed by persons sharing similar environments, de mands, and capabilities. 2.2 - Formal and Informal Organizational Structure Structure may be formal or informal, but it refers generally to the arrange ment of components and subsystems within an organization (Rogers and Rogers, 1976). Blau and Meyer (1971) describe structure as referring to the properties ; of the organization, not its members. Generally, structure is what makes an or- ' ganization “ organized” or discernible from a simple inadvertent grouping of peo- The organization maximizes its effectiveness in achieving its goals by re quiring its members to work with certain individuals and not with others, to take orders from some persons and not from others....So the organiza tion’s structure acts as a constraint on the individual...it is structure that makes an individual’s behavior distinctive in an organization (Rogers and Rogers, 1976). 22 ! The organizational chart reflects the formally prescribed channels of commu- jnication and chain of command, or in other words, the formal structure of that ! organization. The organizational chart can reflect many attributes of the organi zation's formal structure including the number of vertical layers in the hierarchy , (how "tall" the bureaucracy is), span of control, line to staff ratios, work units, superior-subordinate positions, etc. ! As Weber (1947) described, there is efficiency inherent in hierarchy as it 'funnels communication vertically only after passing through formal gatekeepers. This gatekeeping function helps to prevent upper management from becoming overloaded by information that should have been handled by a lower level posi tions. The system restricts access and filters information vertically. With this efficiency, however, comes much slower communication from the lower levels of the organization to upper levels where decisions eventually are made (time here is a function both of distance and mediating steps). Also, because formal structure stresses vertical communication, if the formal structure is strongly adhered to, most horizontal communication would be eliminated. Simi larly, if members are restricted to vertical communication, not only is the direc tion of communication limited, but also the content (i.e., persons may tell co-workers information they would not tel! their superiors) (Rogers and Rogers, 1976). Additionally, there is informal communication structure or emergent structure in organizations: Both formal and informal systems are necessary for group activity, just as two blades are essential to make a pair of scissors workable. Both for mal and informal organizations comprise the social system of the work group (Davis, 1967). 23 , Informal structure arises whenever communication exists between organiza tional members which is not formally designed into the system (i.e., not included in the organizational chart or in job descriptions). The degree to which the informal structure, usually characterized by emergent communication networks, does not correspond to the formal structure is an indication of the inadequacy of i the formal structure to provide the necessary communication channels. I In general, formal structure is what channels and constrains communication in ; an organization. Informal structure is manifested through communication net w o rks which arise to complete the day-to-day tasks and to fulfill the personal ■ and social needs of organizational members. % The present research attempts to relate measures of formal organizational .i structure to organizational media usage. This approach offers the following : strengths: • 1 Some structural traits of organizations are quantifiable and can be i generalized across a variety of settings. I 2 Structural traits are sociological rather than psychological. Socio logical measurement can operationalize concepts at an organiza tional role or work unit level of analysis. If an interest is in organ izational media use (not individual media use), psychological ap- | proaches operate at an inappropriate level of measurement. j j 3 Though media usage clearly can be influenced by psychological ; factors, even these forces operate within a structural context. 24 2.3 - Structural Equivalence They say you are not you except in terms of your relation to other people. If there weren’t any other people there wouldn’t be any you because what you do, which is what you are, only has meaning in relation to other peo ple. -Robert Penn Warren All the Kings Men There exist many approaches to conceptualize and measure organizational structures. Structural equivalence is a network-based approach which uses measures of linkages between organizational members as its primary source of data. A system of such links entails a network of social interactions. Analysis of social networks is a movement away form a atomistic, or individ ual-level conceptualization of social structure. It is an acknowledgment that the social systems wherein a person exists can affect individual beliefs, perceptions and behaviors. ....network analysis incorporates two significant assumptions about social behavior. Its first essential insight is that any actor typically participates in a social system involving many other actors who are significant reference points in one another's decisions. The nature of relationships a given ac tor has with other system members thus may affect that focal actor's perceptions, beliefs, and actions...Its second essential insight lies in the importance of elucidating the various levels of structure in a social sys tem, where structure consists of “regularities in the patterns of relations among concrete entities” (White, Boorman, and Brieger, 1976). Network analysis offers a researcher a “ two edged sword" in that it can illuminate entire social structures, and focus on individual elements within the structure (Knoke and Kuklinski, 1983). It allows researchers to examine social ! interaction as a unit of analysis, instead of the more conventional singular reliance i upon individual attributes. Two major schools of thought exist in network analysis: 1) relational ap proaches, and 2) positional approaches (Burt, 1980). Although a positional approach is used in the present research, an understanding of both approaches sheds some light on the rationale for this strategy. Reviews of such approaches are provided by Burt, (1980), Rice and Richards (1985), Rogers and Kincaid (1981), and Knoke and Kuklinski (1982). The relational approaches are somewhat more intuitively obvious as they or ganize systems of network measures around observed links between individuals (actors) (i.e., iinkages may entail talking, commerce, familial relations, organiza tional chains of command, etc.). Actors who are more frequently interlinked are categorized as a group. In network analysis terminology, linked actors are adja cent. Moreover, a network of interactions can be modeled through the analysis of direct and indirect linkages among all members of a given system. The positional approaches, on the other hand, describe network structures based upon similar patterns of the presence or absence of linkages. The pres ence and absence of links is no longer the basic unit of measurement, but rather, ; the network position assumes this function. A position or role is a sociological ■ concept of related status or behavior. It is measured by similar patterns of ' relations of a given actor with those of all other actors in the system. An actor’s position is a function of the “ distances” between that actor and all other actors ■ in the system. 26 , The relational approach focuses essentially on the pathways in networks; ! the positional approach on patterns of similarity in relational configura- ' tions. The former lends itself to analyses in which the social processes ; occurring in the network are conceptualized in terms of "communica tion, ” using that term in the broadest sense, and the later to a delineation of hierarchical structure. Neither approach excludes the other, and nei ther is reducible to the other. Indeed, they ultimately complement each other (Blau and Alba, 1982). A basic concept in relational approaches is the clique, or set of actors in a ’ network who are mutually finked by strong associations. In a positional approach, a corresponding concept is the jointly occupied network position (Burt, 1980). .Although it varies somewhat depending upon the positional approach taken, this concept expresses the notion that two actors occupy a similar position in relation to other actors in the system: The extent to which specific relational patterns occur repeatedly across multiple actors is captured by the concept of structural equivalence. Bor rowing the political concept of equivalence (e.g., Kelley, 1955:9-10; Lorrain & White, 1971:63) discuss two actors as structurally equivalent if they have identical relations with all actors in a system. This is a strong criterion of equivalence. Two actors I and J are structurally equivalent under this strong criterion when the Euclidean distance between their re- * spective network positions is zero (Burt, 1976, 1977) (Burt, 1980). ; Though relatively straightforward to demonstrate via a few simple matrix op erations, structural equivalence centers around an interest in identifying groups of , actors who are structurally similar in their relational patterns with other actors. In r ! structural equivalence, it is just as important if an actor does not interact with I J other actors as it is if he or she does interact with other actors. The approach , operates at a structural level of analysis and is concerned with social roies and ; stratification. r The ability of structural equivalence approaches to detect social roles in or ganizational structure makes it particularly appropriate for the present research. It allows conceptualization and operationalization at the level of organizational position. Position indicates shared tasks, perceptions, and behaviors as dictated by organizational differentiation. Thus, structural equivalence is more valid theo retically and empirically for the present research interests than conventional case-oriented, aggregate statistical approaches. It is argued here that structural equivalence analysis techniques offer uniquely 1 suitable, conceptual meaningful, and empirically valid opportunities for the study of organizational structure and use of media, Although structural equivalence will • be discussed in relationship to contingency theory and media attributes in the ■ following sections, particular contingency-based implications for this research lie , in the following points: 1 Structurally equivalent positions are products of social differentia tion within organizations. 2 Groups of actors occupying similar positions within a system’s structure can be identified using a structural equivalence network analysis approach. I [ 3 Structurally equivalent actors are characterized by similar commu nication patterns as measured by network analysis methods. i 2.4 - Media Usage Considerations Attributes of media - Various media possess different characteristics which may: a) influence the types of communication for which a medium can or should be used, and b) restrict the type or amounts of information conveyed, : An awareness of the importance of the use of media extends back at least as far as the initial confrontations between the Church of the Middle Ages and peas- : ants over who should have access to the Word, Penny papers, the telegraph, radio, and television all sparked interest in what potential effect media have upon their users and messages, Harold Innis (1951), however, was one of the first theorists to formally shift interest away from the content of the media to that of the form of the media itself, Innis examined the relationships across history between changing commu nication technologies and concurrent changes in the host’s socio-cultural char acteristics. He envisioned technological innovation as more than a portent of change for society; he believed it was a causal force— a shaper of society, 2.4.1 - Social Presence Other theorizing on the importance of media characteristics includes the work McLuhan (1960, 1962), Pool (1977, 1983), Bell (1973), and others too numer- i ous to list. A conceptual springboard to most current interest in the use and i , impacts of media in organizations, however, is the work of the Communications ! Studies Group in London, Short, Williams, and Christie (1976) summarized much j of this research in The Social Psychology of Telecommunications. Much like Woodward’s seminal work in contingency theory, Short et al.’s research and I j conceptual contributions have been questioned. Their work did, however, sensi- tize organizational communication researchers to the fact that the form of media used for various communication tasks may make a difference in the outcomes of performing those tasks. Over the past decade, many technological changes in organizational media usage have occurred. Electronic mail, computer conferencing, voice mail, the use of group decision support systems, and computerized databases to name a few. Running concurrent with organizational adoption of these newer forms of communication and information has been a renewed interest on the part of com munication researchers concerning the potential benefits, costs, advantages, dis advantages, and positive and negative interpersonal and organizational impacts i (Rice and Associates, 1984), Short et al. posited two initial attributes of media: 1) media are able to tran scend time or distance, and 2) media transmit only a portion of a communica tor’s presence (Short et al., 1976; Reid, 1977). Clearly, the first attribute can be a advantage for communication while the second attribute could serve to constrain or limit the effectiveness or appropriateness of the media for various communication needs. Short et al. described the ability of a medium to convey a communicator’s ; presence as, logically enough, the social presence of the medium. Social pres- i : ence depends not only upon the words conveyed, but also upon a range of | non-verbal cues including facial expression, direction of gaze, posture, attire, ■ and physical distance and many verbal cues including, timing, pauses, accentua tions, tonal inflections, etc. Clearly, high levels of social presence are available in direct, proximate fa ce -to -fa ce communication, but various media tend to | “ strip away” many of these cues. Short et al. rated media in the following order 30 of decreasing social presence: 1) face-to-face, 2) video, 3) handset tele phone, 4) monaural audio, 5) business letter. Short et al. used a variety of experiments to determine the relative appropri ateness of media of varying social presence for a variety of organizational com munication tasks. These experiments usually involved testing the time required to accomplish the task via fa ce -to -fa ce compared to various media. The most important implications were that, given the general acceptance of the importance of non-verbal communication, surprisingly, most task-related communication could be conducted rather handily through media. In fact, some tasks which simply required the exchange of information were actually accom plished more quickly via mediated means than through fa ce -to -fa ce communi cation (e.g., one task was the assembly of a small trash can cart— it was as sembled more quickly via telephone mediated instructions than through face to face instruction). In these cases, the researchers contend that the increased ’ social presence, when unnecessary for the task, becomes a distraction and a detriment to the completion of the task. Generally, fa ce-to -fa ce communication, however, was found to be superior for tasks involving conflict resolution, negotiations, getting to know someone, j resolving problems, etc. These types of tasks, it was hypothesized, require high j levels of social presence. As mentioned previously, much of this work could be challenged on both , conceptual and empirical grounds, but, it has, nonetheless, motivated thinking ^ and research into the area of appropriate use of media in organizations. A model of organizational media usage evolved from the work of the Communica tions Study Group which posits that both tasks and media have reasonably under- i standable characteristics— tasks require varying amounts of social presence, ; and media transmit various amounts of social presence. Maximum effectiveness I is achieved when there is a match between the social presence requirements of | a task and the amount of social presence inherent to the mode of communica- : tion used. The concept of a "fit" between task demands and media characteris tics is central to the social presence piodel. Rice and Case (1983) extended • the concept of social presence to include such factors as organizational level, i tenure, job classification, and prior media habits into a construct called "media style. ” 2.4.2 - Information Richness Daft and Wigington (1979), and Daft and Lengel (1984, 1986), proposed the ! t concept of information richness, which is very similar to Short et al.’s concept of social presence. Daft and Lengel define information richness as the potential i information carrying capacity of medium and the ability of information to change ■ understanding within a time interval. Daft and Lengel proposed a ranking of media with potential for information l richness which aligns rather well with Short et al.’s ordering for social presence: face-to-face, telephone, personal documents such as letters or memos, imper- I i sonal written documents, and numeric documents (i.e., listed according to de- ' creasing capacity for conveying rich information). Table 2A contains Daft and Lengel’s information richness continuum. i Daft and Lengel also suggested criteria which determine the "richness" of : media: 1) capacity for immediate feedback, 2) number of cues and channels' I utilized, 3) personalization, and 4) language variety (Daft and Wigington, 1979), TABLE 2A Continuum of information Richness Richness Medium Feedback Channel Source Language HIGH \ Face-to-Face Immediate Visual, Audio Personal Body Natural Telephone Fast Audio Personal Natural Written (personal) Slow Limited Visual Personal Natural Written (formal) Very Slow Limited Visual Impersonal Natural u )W Numeric (formal) Very Slow Limited Visual Impersonal Numeric w m a m sm m 1 (From Daft & Lengel, 1984) Face-to-face is the richest medium because it provides immediate feed back so that interpretation can be checked. Face-to-face also pro vides multiple cues via body language and tone of voice, and message content is expressed in natural language (Daft and Lengel, 1984). As an example, telephone provides immediate feedback and natural lan guage, but it does not provided proxemic or kinesic cues. A written letter pro vides natural language but does not offer immediate feedback or nonverbal cues. Finally, a numeric document provides neither natural language, nor other sources of information the richer media provide. Though developed from different theoretical backgrounds (i.e., Daft et al. do not cite the much earlier work of Short et al.), the concepts of social presence and information richness are very similar. Daft et al. combined contingency concepts of analyzability (equivocality) and numbers of exceptions (uncertainty) with social presence (information richness) within an information processing framework. The information richness model spans traditional contingency theory approaches and organizational media usage considerations. The model also improves upon conventional information processing models by incorporating no tions of equivocality which help free the model from an overly rational perspective of organizational behavior, Though all the theoretical “ pieces” had existed for a decade or more, Daft et al.’s unique contribution is synergism. The information richness model does, however, take a rather limited perspec tive both on the range of media considered and on the types of attributes con sidered. Most of the information richness literature does not consider a variety of newer, technologically advanced organizational media (e.g., electronic messag ing, voice messaging, videoconferencing, audioconferencing, computer con- 34 jferencing, etc.). These exclusions limit the applicability of the model to less technologically advanced organizations. Their continuum is rooted in an information processing perspective, which, ultimately, is a model of organizational structure based upon the earlier work of contingency theorists. It is not, however, at its origins, a theory or model of human communication. It is, perhaps, this lack of communication orientation that results in a rather limited conceptualization of media attributes. Basic media concepts such as synchroneity or number of communication participants are either given cursory treatment or are skirted entirely. Steinfield and Fulk (1986) extended Daft and Lengel’s information richness , continuum to include both videoconferencing and electronic messaging. Stein field and Fulk ranked media in the following order of decreasing capacity for richness: face-to-face, videoconferencing, telephone, electronic messaging, personal written, formal written, and formal numeric. Though an improvement over the original Daft and Lengel continuum, Steinfield and Fulk’s model excludes a several important media and is based upon the same rather limited set of attrib utes as the original continuum. Clearly, the number of active participants in a communication process may influence the outcome. A business meeting, for example, is the conventional, approach for enabling debate and reaching consensus. The ability to have syn- . chronous communication among all pertinent individuals is considered mandatory . i i for many organizational decisions. Based upon Daft and Lengel’s original contin- ' uum, Table 2B proposes an extended model incorporating a more extensive I range of media, synchroneity, and number of active communicators. i 35 TABLE 2B Extended Continuum of Information Richness | (Media in Order of Decreasing Capacity for Social Presence) i Medium Synchroneity/ Feedback Active Participants Mode HIGH Face-to-Face A Meetings synchronous/ immediate more than two visual, audio,nonverbal kinesic, proxemic, etc. Face-to-Face Dialogue synchronous/ immediate two visual, audio, nonverbal kinesic, proxemic, etc. Video- Conferencing synchronous/ immediate two or more limited visual and audio Audio- Conferencing synchronous/ immediate two or more limited audio Telephone Dialogue synchronous/ immediate two limited audio Voice Messaging asynchronous/ variable one limited audio Computer Conferencing usually asynchronous/ variable one or more text Electronic Messaging asynchronous/ variable one text Personal Written asynchronous/ slow one text Formal Written asynchronous/ slow one text Online Databases asynchronous/ none one text Numeric Databases asynchronous/ none or slow one numbers 36 It would be naive, however, to restrict the classification of media to the singular dimension of information richness. As an example, based upon the infor- ; mation richness continuum, voice messaging has less capacity for richness than does conventional telephone. While technically true, the asynchroneity and computer-mediation (covered in section 2,5) provide voice messaging with ca pabilities beyond that of normal telephone for many applications. Though the richness continuum may suffice reasonably well for rudimentary classifications of conventional media, computer-mediated media do not classify ; as "neatly” along this single dimension. Computer-mediated media incorporate features unrelated to bandwidth or information richness which affect their appro- - priateness for various communication tasks (e.g., enhanced abilities to connect groups of persons, abilities to adjust for temporal or geographic differences, etc.). Moreover, “ computer-enhancements” possessing potential to facilitate coordination of organizational information flows may be more important for col- laborative-oriented communication than the relative information richness of a given medium. Additionally, as described in the next section, social factors not directly re lated to objective technical characteristics of media may be instrumental deter minants of media usage. 2.4.3 - Social Information Processing Previous applications of attribute-oriented models of media usage (e.g., re lating individual perceptions of task environment to selection of media according to attributes), have been criticized as overly rational and have offered low ex planatory power (Fulk, Power, Schmitz, and Steinfield, 1986; Steinfield, Jin, and Ku, 1987; Rice, Grant, Schmitz and Torobin, 1988). Steinfield et al. (1987), and Fulk et al.'s (1986) recent work has combined attribute models of media usage with social information processing theory in at tempts to create a model of media usage less restricted by rational assumptions. These efforts offer certain appeal in comparison to more restricted models of media usage based solely upon task environment-media attribute matches. This approach offers a model of media usage where an individual’s media perceptions are proposed to be a function of: 1) social information about supervisor and co-worker perceptions, 2) objective media attributes, and 3) media experience (Steinfield et al., 1987). A central concept to this approach lies in the belief that while media and tasks have discernible traits which could affect usage (i.e., a notion of a “ fit” similar to .■Daft and Lengel's, 1986 approach), social information processing intervenes • and moderates organizational media usage behaviors. This approach creates somewhat of a theoretical "hybrid.” The theory operates at both the structural i level (i.e., includes concepts of rational task environments), and the psychologi c a l level (i.e., user perceptions of co-w orker and supervisor perceptions). Though the model is attractive, the inclusion of social information processing as an intervening variable creates a far less parsimonious model of media usage. I The present research proposes a model which is able to incorporate many of the ; features of social information processing while maintaining a structural perspec- I j tive without a complex mix of levels of analyses. The next section discusses media attributes with particular relevance to com- ■ puter-mediated communication systems in organizations. i i i ( I 38 2.5 - Computer-Mediated Media i 2.5,1 - Performance and Benefits Several characteristics of newer communication media (computer-mediated) distinguish them from more traditional forms of media. These systems generally require the user to enter and receive information from some form of a computer terminal. Characteristics of these newer systems which have implications for or ganizational communication included the following: (Rice and Shook, 1988) 1) Asynchroneity - users do not have to be on the system simultane ously in order to send or receive messages. This removes the constraint of temporality inherent in fa ce -to -fa ce or telephone communication. 2) Feedback - users may interact as quickly as they wish, to clarify points or request further information. This removes the time lags inherent in memos and letters. 3) Electronic transmission and storage of information - messages ; can be accessed wherever a user has access to a terminal. This re- ; moves the constraint of geographic proximity inherent in fa ce -to -fa ce communication, and the constraint of point-to-point communication in herent in letters and memos. , 4) Structuring of communication - users may use the capabilities of a j computer to structure their communication. For example, using pre-es- i tablished distribution lists removes the constraint of having to send sepa rate letters or make separate phone calls to transmit the same message to many people. Users also have greater abilities to index and retrieve messages. I j 5) Connectivity - users typically can contact other users on the system i without having to know them in advance, or, by using keywords for inter- I est areas or distribution lists, without even knowing the person exists. I This also increases the potential value of the interaction because users become sources of information value themselves, and the value of the communication component rises exponentially with the number of users, I 6) Integration - users have access to multiple applications and technolo- ! gies without learning multiple interfaces, and can transform information ! from one media to another. This removes the constraint that multiple or ganizational media impose on the sequential transmission or formatting of communication. Because of these and other characteristics, com puter-m ediated communica tion systems have been shown to reduce delays in information exchange, im prove maintenance of records and information received, increase coordination of geographically dispersed groups, and improve users’ abilities to process large amount of information (Johansen, de Grasse, and Wilson, 1978; Rice and Bair, . 1984; Rice, 1987). These newer systems generally offer flexibility and power which hitherto were unavailable, Because of their general asynchroneity and text , or audio only modes, however, these systems tend to be characterized by rela- i tively lower capacity for information richness, Based upon these attributes of computer-mediated communication systems, i-the introduction of such a system potentially could affect users’ communication behaviors in two main areas: 1) augmentation of the extant communication ac tivities and media by a) increasing the relative number of communication con- ] ! tacts one has within the organization and b) increasing a user's ability to initiate ; vertical communication within the organization; and 2) substitution of comput- er-mediated media for traditional media of similar capacity for social presence. A potential consequence of a computer-mediated communication system is an increase in the number, diversity, and direction of communication contacts as : a result of the system's ability to augment or complement existing communica tion activity. A variety of studies have reported increases in overall communica tion and in interpersonal linkages as a result of implementing such systems (Free- j man, 1980; Biltz, 1983; Hiltz and Turoff, 1978; Johansen, de Grasse, and Wil- i ; 40 ^son, 1978; Johansen, Vaile, and Spangler, 1979; Kerr and Hiltz, 1982; Leduc, i ! 1979; Palme, 1981; Rice, 1980; Rice and Bair, 1984; and Rice and Case, . 1983). Computer-mediated communication can facilitate communication i through the development of communities of interest and by assisting the ex change of information across locations, time, and hierarchies (Hiltz and Turoff, 1 1978; Kerr and Hiltz, 1982; and Uhlig, 1977). ; The ability to contact any one in the system asynchronously from multiple > • locations without busy signals or fear of inopportune timing could increase not only the number of communication contacts, but also could increase a user’s ability to initiate communication vertically within the organization. The lowered social presence afforded by the computer-mediated format may in fact encour age task-related communication across hierarchies which otherwise may have been discouraged by more imposing fa ce -to -fa ce or telephone conversations. Based upon the previous discussions of social presence and information rich ness, computer-mediated communication can be seen as a potential substitute for traditional limited bandwidth media for communication tasks that do not de mand high levels of personal involvement. It has similar capacity for social pres ence as, but with the many advantages of computer processing for enhanced communication functionality: In general, computer conferencing is at least as good and sometimes better than fa ce -to -fa ce contact (in some experiments, video contact) for the following tasks: 1) exchanging information (especially technical in formation and particularly in short bursts), 2) asking questions, 3) ex changing opinions or orders, 4) staying in touch, and 5) generating ideas. These interactions do not demand high personal involvement and are co operative ventures (Rice, 1980). ! Research on media substitution generally has found that the implementation i of a computer-mediated communication system reduces somewhat the use of memos, phone calls, and letters (Byrne, 1984; Conrath, 1978; Miller and Nichols, 1981; and Rice and Case, 1983). A study by Picot, Klingensberg, and Kranzle (1982) suggested that electronic mail could substitute for 4 percent of organizational travel, 9 percent of face -to -fa ce communication, 19 percent of telephone calls, and 63 percent of mail traffic. Similar figures were suggested by Dormois, Fioux, and Gensollen (1978). Most research on computer-mediated communication systems does find or ganizational benefits in the areas mentioned. The reported benefits are, how ever, moderate. Typical reductions in the usage of other media and typical increases in communication contacts range between 10 to 20 percent (Rice and Shook, 1988). Rice (1987) provides a comprehensive review of these attributes, applications, and outcomes of computer-mediated communication systems. 2.5.2 - Voice Messaging Voice messaging is a form of computer-mediated communication, but it is accessed through the nearly universal telephone. Voice messaging systems ba- i sically combine the ability of computer storage and processing with the conven- j tional abilities of a private branch exchange (PBX). Voice messaging users use I the keypad of a touch-tone telephone to issue commands to the system. Each i ' user has a voice mailbox and ID number. He or she may record and send mes- i sages directiy and asynchronously to another user’s mailbox. Users also have the ability to store incoming messages for reference, forward messages to other users, record and store a message for a timed, future deliv- ; ery, "broadcast” a single message to a number of users, or trigger a "rollover” .function which diverts conventional incoming calls from one’s extension into the voice mailbox (answering machine function). Generally, the user’s own recorded voice answers the phone on incoming calls. Similar to electronic mail, a typical system has built-in prompts and helps, passwords are required to access a mailbox, messages can be labeled as ur- igent or private, and messages can be reviewed, deleted, etc. While electronic mail systems require a user to have access to a computer terminal or computer network, voice messaging requires the user only to have ' access a touch tone telephone, Additionally, while electronic mail requires the user to type messages, the user of a voice messaging system simply "talks.” Because of its more natural interface, generally universal physical access ■ potential, less potential for the system to be rejected by persons higher in the organizational hierarchy due to stereotypes concerning keyboards (keyboards ■ may be seen as clerical or low-level work stations), and voice’s inherent ability ■ to carry additional information through tone, pauses, inflection etc., voice mes saging could have wider appeal than the current crop of text-based messaging : systems for many applications, For a more comprehensive treatment of organ izational media’s capacity for carrying information, and their relative appropriate ness for tasks of varying characteristics and demands, see Rice (1987), and Rice and Shook (1987). I 1 The present research conceptualizes voice messaging as possessing two | dimensions: 1) answering, and 2) messaging. A voice messaging system can be used "passively" as an answering machine, or " actively” as an asynchro nous, voice messaging system. ! The answering or passive use simply is enabled by triggering the rollover fun c-, ; tion mentioned above. A person allows the phone to ring a predetermined num- : 43 ber of times, and his or her voice messaging mailbox answers the call. For a j caller, little exists which would distinguish the voice messaging system from a I typical answering machine. Clearly, passive use of voice messaging could en- ^ able a person to intercept and store calls. This function potentially allows one to | manage more incoming calls, and or manage one’s time more effectively by j avoiding interruptions. j A user also has the ability to access active or messaging features which ; particularly could enhance coordination and collaboration (e.g., time shifting of messages to compensate for varying schedules of collaborators or persons con tributing to a project, use of asynchroneity to avoid interrupting unnecessarily others on a project for the simple delivery of non-timely information, storing messages as referent material or for future delivery, “ broadcasting” a single ! message to a number of persons involved with a project, forwarding important or insightful messages to others, etc.). Distinctions between these two modes of operation will be discussed further in section 2.6 describing the proposed model. 2.6 - The Proposed Mode! i i ! The model proposed in this research posits relationships in three areas: 1) ! effects of task environment upon voice messaging behaviors, 2) effects of hori- ; zontal and vertical differentiation upon voice messaging behaviors, and 3) effects : of voice messaging behaviors upon communication performance, The model is an effort to a structural model of organizational media usage. Figure 2 depicts the theoretical model of voice messaging use, structural equivalence, and c o m -1 munication performance. 44 FIGURE 2: MODEL OF VM USE, STRUCTURAL EQUIVALENCE, & COMM. PERFORMANCE ORGANIZATIONAL LEVEL TENURE HORIZONTAL DIFF. RECEIVE MESS. VERTICAL DIFF. RECEIVE MESS. HORIZONTAL DIFF. SEND MESSAGES VERTICAL DIFF. SEND MESSAGES EDUCATION INNOVATIVENESS SUPERVISOR’S INNOVATIVENESS # OF MESSAGES RECEIVED CHANGE IN ABILITY TO OBTAIN INFORMATION # OF MESSAGES SENT CHANGE IN ABILITY TO DISTRIBUTE INFORMATION hypothesized paths control paths ! 2.6.1 - Structural Equivalence ; As described in the previous section on structural equivalence, structure is, to i a large degree, determined by efforts to minimize costs (economic and social) due to intraorganizational subunit workflow dependencies (Thompson, 1967). These dependencies are a natural product of the forces of differentiation and integration in large and complex organizations (Lawrence and Lorsch, 1967). The formai structures of organizations (i.e., of successful organizations), then, are viewed as a record or chart of differentiation and integration efforts by the I organization. Additionally, differentiation may be conceptualized on two axes: 1) horizontal differentiation and 2) vertical differentiation. The creation of roles (horizontal differentiation) within an organization results in work units or positions. These formally differentiated positions are operationalized through network approaches as groups of structurally equivalent actors. A structural equivalence approach is uniquely appropriate for this application as it measures both formal position in 'Organizational structure, and the behaviors of persons occupying those positions. The argument put forth in the proposed model is that media have attributes which may make them more or less appropriate for communication tasks de pending upon the nature of the task (e.g., see section on media usage consid erations). Concurrently, organizational differentiation creates roles with specific communication needs and task environments. Therefore, members of a position I (structurally equivalent actors) should be characterized by similar media usage j patterns. Support for Hypothesis 1.1 is critical to the validity of the present model. It ' stands at the very core of a model relating media attributes to task characteris- Hypothesis 1.1: Structurally equivalent actors in a system will manifest similar voice messaging behaviors (position effect). i i i ■ tics based upon a structural equivalence approach. It directly tests the position effect at the broadest level. Additionally, vertical differentiation creates “ layers” in organizational struc ture. These layers represent positional links based on authority and responsibility 1 and result in systems of superior-subordinate linkages. Although not structurally equivalent, roles on either side of superior-subordinate links are adjacent. Adja cencies in formal hierarchy link roles across hierarchical levels and are the basis , for information flows both up and down the chain of authority (Fayol, 1949; ■Brass, 1981), As information flows are determined by network positions, and conjoint use of media is necessary for communication to occur, actors occupying adjacent posi- : tions will be characterized by similar patterns of organizational media use. More over, as a superior in a superior-subordinate adjacency possesses discretionary authority, the superior’s media use patterns will influence a subordinate’s media use patterns. ; Hypothesis 1.2: Due to positional adjacencies created by vertical differentiation, superiors' voice messaging behaviors are | | a determinant of subordinate positions’ voice messaging f ; behaviors. i Building upon the previous discussions of media attributes and organizational structure, however, more specific hypotheses which build upon the preceding hypotheses are suggested. 2.6.2 - Task Characteristics i Media usage is but one component in the fabric of organizational life. As a i ' subset of communication behavior in general, it will be affected by a multitude of j ; factors including: norms, expectations, location in the formal or informal organ izational structure, status, task characteristics, personality traits, daily vicissi tudes, ease of access to the system, and the list goes on (Rice, 1987). In conjunction with the preceding discussions of task environment, social pres- ; ence, and information richness, task characteristics can be seen as a determi nant of media usage, The social presence model and the information richness model specify that effective performance is dependent upon an appropriate "fit” between task demands and media attributes. For a given level of performance, ! persons occupying task environments characterized by less analyzable tasks : would need to access media with greater potential for information richness. Con versely, persons facing higher levels of analyzability should depend more heavily ; upon less rich media for effective performance. It is an additional premise of this research that, assuming tasks are differenti ated and or collaborative in nature, as the number of exceptions to routine in- | creases, and as task analyzability decreases, effective organizational perform- j ance dictates that increased coordination of workflows is necessary. This in- ] ) creased coordination of workflow could be reflected generally in increased ; amounts of communication, More interestingly, however, it should be reflected in greater use of specific media with inherent capacity for coordination of workflow. In particular, increased numbers of exceptions and decreased task analyzability should be reflected in increased usage of active voice messaging features. Use of active voice messaging (e.g., asynchronous use of messaging I 48 | I features) offers relative advantage or a greater potential to enhance one’s ca- i i pacity to coordinate workflows. The preceding discussion suggests the following hypotheses: Hypothesis 2.1: To the degree that task environment is less routine, it will be characterized by a greater reliance upon active voice messaging. Hypothesis 2.2: To the degree that task environment is less analyzable, it will be characterized by a greater reliance upon active voice messaging. Moreover, as tasks become less analyzable or less routine, structural equiva lence measures of task environment should become more efficient predictors of media behaviors. This is to say that routine tasks or analyzable tasks do not ; prohibit the use of active messaging features— they simply do not require it as much. But as tasks become less routine or less analyzable, greater communica- ■ tion performance is required and less discretion in media choice is permissible. As tasks become less analyzable or less routine, the position or work unit envi ronment should become a more powerful determinant of behavior. Hypothesis 2.3: To the degree that task environment is less routine, horizontal differentiation (position effect) will be a stronger determinant of using active voice messaging features which facilitate coordination of workflows. Hypothesis 2.4: To the degree that task environment is less analyzable, horizontal differentiation (position effect) will be a stronger determinant of using active voice messaging features which facilitate coordination of workflows. 49 2.6.3 - Communication Performance ■ Again, based upon the discussion of characteristics of computer-mediated media, use of voice messaging is posited to have the potential to affect commu nication performance. Specifically, it is hypothesized that use of active voice messaging features has greater potential to affect the ability to distribute and obtain information than does passive use of the system, Hypothesis 3.1: Passive use of voice messaging will not increase communication performance as much as active use of voice messaging will increase communication performance Moreover, it is hypothesized that use of active voice messaging features in less analyzable or less routine task environments will result in greater perform ance benefits than will use of the same features in highly analyzable or more routine environments. This argument is one rooted in the idea of a “ fit” between task characteristics and media attributes, Hypothesis 3.2:Active use of voice messaging in less analyzable task environments will result in greater increases in communication performance than will use of active voice messaging in more analyzable task environments Hypothesis 3.3:Active use of voice messaging in less analyzable task environments will result in greater increases in communication performance than will use of active voice messaging in more analyzable task environments Several additional hypotheses are suggested by the model proposed in Figure 2. As communication generally involves feedback, sending voice messages 'should be closely associated with receiving messages. Similarly, the ability to . distribute information depends upon the ability to obtain information. These prem ises suggest the following: Hypothesis 4.1: A nonrecursive (reciprocal) relationship exists between sending and receiving voice messages. Hypothesis 4.2: An increased ability to obtain information will result in an increased ability to distribute information. 2.6.4 - Control Variables The model put forth in this section posits that differentiation creates positions with similar communication needs and behaviors and that the identification of these positions is the basis of a parsimonious and effective model of organiza tional media usage. This effort to view organizational members as positions, however, does not take into consideration effects attributable to variations , among individual members. In order to disambiguate the effects of organizational differentiation from those related to traits of particular individuals, this research must control for several important individual-level variables that influence organ izational media use (see note 1, Rice and Shook, 1987). | Voice messaging is an innovation in the present organization. Previous re- j search has shown that individuals vary in their readiness to accept innovation I (Rogers, 1983; Coleman, Katz, and Menzei, 1966). Though not a theoretical | concern in the present research, proper identification of the model proposed t jhere requires that variations in individual innovativeness are estimatable by the j research design. As superiors generally have a degree of influence or control l (over subordinates, the supervisor’s innovativeness also may influence a subordi n a te ’s willingness or opportunity to adopt and use the new system. Additionally, using a new communication system requires a degree of learn ing. An ability or willingness to learn a new system may be influenced by one's .previous education. Persons with relatively more education may: 1) view learning a new system as posing less personal cost, and 2) possess backgrounds with similar learning experiences from which to draw. An individual’s job classification or organizational level may influence his or her willingness or opportunity to adopt and use the new system. Organizational level : may associate strongly with increased access to resources, differing task de- ■ mands, experience, organizational commitment, etc. Lastly, organizational tenure could influence a willingness to learn and use a • new system. Persons with greater tenure possess relatively greater time invest- •ments in the status quo. Additionally, persons with greater tenure may see lower potential return from their time investments learning a new system as they have relatively fewer years before them from which to draw benefits from the new ( system. j | Again, while none of the preceding influences are of immediate theoretical j concern for the present research, proper causal model identification demands , j their inclusion. From the previous discussions on theory testing, it follows that the com mon causes for each pair of cause and effect variables have to be intro- j duced in a complete causal theory. If this is not done, tests of causal ef fects are invalid and estimates of the sizes of the effects will be incorrect (Saris and Stronkhorst, 1984). 52 1 As an example, if education and vertical differentiation are correlated, and , both affect messaging behavior, both must be included in a causal model to correctly estimate the effects of vertical differentiation on messaging. Figure 2 ■ depicts the control paths as dashed lines and the hypothesized paths as solid . lines. i I 2.6.5 Summary As previously mentioned, an analysis of organizational media usage at a struc tural level offers a certain appeal in that results potentially are more systematic and generalizable than findings derived from individual level analyses. If the pur- '.pose is to understand organizational use of media, research beginning at that 1 level of analysis possesses inherent advantages. Efforts attempting this level of measurement, however, have only minimally : advanced scientific understanding of media usage due to the low explained vari- .ances characteristic of most research conducted within this genre. The present research proposes that much of the previous failure is not due so much to ' poor-fitting theory, but rather is due to inadequate operationalization and meas urement. ' It is an additional premise of this research that individual-level measurements j of perceived task environments are inadequate and operate at an inappropriate 1 level of analysis (Downey and Slocum, 1975). The focus on structural equiva- ! lence in this research is an attempt to conceptualize and operationalize organiza- | tional structural characteristics at an appropriate level of analysis. This author ; knows of no prior attempts to test organizational media use and structural equiva lence based upon a mapping of the formal hierarchy, I Chapter 3: Methods 3.1 - Data Collection t A large, multi-city service organization on the verge of implementing a pilot test of a voice messaging system was selected for the present study. The purpose of the large-scale pilot was to determine the advisability of adopting voice messaging as a company-wide practice. The company offers financial and insurance services throughout the United States. The organization is quite ' distributed with sites in various cities each with particular regional and or service ■ responsibilities. 3.1.1 - Sample t The sampling procedures included three stages: 1) creation of a roster of 550 potential users located in sites where access to voice messaging would be made available (a roster of potential system users), 2) logging of individuals as they were trained to use the system (a prerequisite for obtaining a voice messag ing system “ mailbox” ), and 3) distribution and return of questionnaires before implementation of the system and five months after implementation. Respondents/subjects included organizational personnel from all hierarchical levels and a wide variety of expertise areas (e.g., sales, management, clerical, technical, litigation, office administration, systems analysis, etc.). Voice mes- : saging behavioral data were available for everyone on the roster; however 62 j persons did not activate their accounts, and 190 persons never accessed the | system for sending voice messaging. : Of the 190 persons who did not send messages but who did receive mes- j sages, 75 had completed a T1 or T2 questionnaire, A series of ANOVA’s for I 54 : unbalanced designs was conducted to determine if the persons who did not send .messages were significantly different on any important criteria from those who i did send messages. The non-senders were proportionally represented from all sites, and were not significantly different from senders in organizational level, education, or tenure. The non-senders were, however, significantly less innova tive than senders (the innovativeness scale is discussed later in this section). ; 3.1.2 - Data Sources i > Data collection included seif-report/survey information, behavioral data, and archival information. A baseline survey conducted prior to the beginning of the 'pilot test included questionnaires from 201 persons out of a possible 389 (52 . percent response rate), For five months after the baseline survey, individual user data from the voice messaging system was monitored continuously and collected weekly on all per sons included in the roster. These data included computer-monitored measures of the number of voice messages recorded, sent, stored, stored for future deliv ery, number of voice messages received, number of telephone answering mes sages received, and the average lengths of messages sent and received. Over 6,400 weekly observations were collected during this process (see Rice and ! Borgman, 1983, for a discussion and review of computer-monitored data). i : A second survey conducted approximately five months after the voice mes- ! saging system data collection had begun included questionnaires from 254 per- ' sons (65 percent response rate). At this point, users had used the system for varying periods of time as people were added to the study as they received . training on the system. Of the 200 responses at T 1 and the 254 responses at I T2, 125 persons were included in both samples. Attrition was due to non-re- 55 ^ponse, transfers, turnover, and users who were added after the T1 question- . inaire was distributed. The mean number of weeks for using the system was 12.3 I (the minimum number of weeks was 1, the maximum was 21). Extensive archival information on all persons included in the roster was pro vided to the researcher by the organization. The archival information was critical in creation of several structural variables. Table 3 contains scale creation statis- i : : tics and Table 4 contains general descriptive statistics. 3.1.3 - Measures of Predetermined Variables The two questionnaires provided basic demographics such as tenure and 1 education. The questionnaires also included a ten-item scale measuring atti tudes toward innovation and change (Hurt, Joseph and Cook, 1977). The i demographic measurements and the innovativeness scale were used to measure ' . individual or personality characteristics which otherwise could confound the ef- [ I forts to measure structural effects on media usage. 1 A two-part, nine-item scale measuring task rputinism and task analyzability 1 (Withey, Daft and Cooper, 1983), was used to measure individual perception of ■ task environment. i j An 7-point measure was used to measure organizationally defined jobtitle : (1=management, 2=professional, 3=agent, 4=sales, 5=technical, 6=administra- i j tive, and 7=clericai), Tenure simply was measured as number of years at the j organization. Superior-subordinate relationships as well as organizational charts for all de- j partments were obtained from archival information. These relationships were I used as the basis to generate variables measuring organizational differentiation. Specifically, persons were identified into positions if they answered to the same 1 56 superior, and the hierarchical influence of a superior's media usage patterns ; could be linked with the subordinates’ media usage. These are important vari- ■ ables as they allow the introduction of structural equivalence factors as part of a 1 contingency-based, structural analysis. 3.1.4 - Measures of Media Usage In addition to the voice messaging behavioral data obtained from Aspen sys tem, the questionnaires also provided self-report measures of voice messaging : behaviors. The percentage of system usage as answering machine compared to usage for actual voice messaging applications was included in the time two questionnaire. A self-report measure of number of messages sent and received aiso is included. Distinctions between these different measures of voice messaging usage be come important in the various analyses to follow. The percent of system usage for active messaging is used to test hypotheses relating task environment to type of system usage (e.g., less analyzable tasks and collaborative tasks will result in greater usage of active messaging, etc.). Number of messages sent and re ceived is a record of active communication activity and intensity. Length of i messages sent and received extends the simpler numeric count of messages by measuring the amount of information conveyed. Longer messages should be : indicative of greater amounts of information, more complex, less routine, or less i [ analyzable communication topics. i I By measuring both length and number of messages, the analyses also can ' compensate for idiosyncratic communicator traits (e.g., persons who communi- j cate via short, but frequent messages versus those who message less frequently i ■ 57 , but with greater comprehensiveness). I i Monitoring the length and number of telephone answering messages received 1 allows parallel measurement and analyses of the effects and determinants of passive messaging behaviors (i.e., unlike active messaging behaviors, passive messaging is hypothesized to lack the ability to enhance communication perform ance). : Underlying the effort to collect and analyze a range of very detailed communi- ' cation behaviors is a belief that human communication processes are very com plex and varied. Attempts to measure communication in collaborative and un- analyzable organizational environments with simplistic indicators, or with a singular reliance upon conventional self-report operationalizations, are viewed as less appropriate than the methods used in the present research. The ability to com bine these various behavioral measures with realistically complex analyses pro vides a sound and sensitive quantitative foundation for the present research. 3.1.5 - Measures of Communication Performance Likert scale questions (7-point, 1=high performance, 7=low performance) asking respondents about their ability to obtain and distribute information were administered at T1 and T2. The differences between the T1 and T2 scores were used as the measure of “ change in communication performance.” The questions asked about ability to distribute information to “ groups of people” (es sential for collaborative work), and ability to obtain information in a “ timely” manner (again, a key communication factor for collaborative work). 58 3.2 Data Transformations 3.2.1 - Normalizations Behavioral voice messaging data collected from computer records evidenced pronounced negative exponential skew, failed the Shapiro-Wilk W normality sta tistic, and thus were inappropriate for analyses using test statistics based on an assumption of normality. In order to use standard test statistics, the data were rank ordered and converted into normal scores using the Blom formula in the 1 Statistical Analysis System (SAS) Rank procedure (see Appendix 1). The sys tem usage measures then were divided by the number of weeks each person had used the system to generate weekly averages. The resultant transformations preserve the natural ordering of the observa tions, and create a mean zero, unit norma! distribution, Raw scores are reported for descriptive statistics (e.g., means, etc.), but all tests and models used the normalized transformations. Because the system computer monitored all messaging accounts whether or not they ever were accessed by a user, 190 accounts that were not accessed for sending messages during the pilot registered 0’s for all sending measures. These measures were set equal to missing data prior to the normalizations. 3.2.2 - Structural Equivalence Transformations For the 550 observations on the initial roster (all potential users of the sys tem), 172 supervisor names were identified from organizational charts and other organizational archival data (supervisors for 24 system users were not obtained and were treated as missing data for analyses involving structural equivalence). From these supervisor names, subordinates were assigned a supervisor number [equal to the individual identification number of their immediate superior, This as signment created positions (work groups). The size of the positions ranged from 1 to 15 (mean size=4.69, std. dev. = 3,10). Voice messaging variables for subordinates were summarized by supervisor ID numbers. The means of the position groups were taken for all voice messag ing usage variables (i.e., this procedure was performed on all of the com puter-monitored, behavioral data). The position mean was multiplied by the : number of persons in the position, and the value of the individual observation was subtracted from the product. Finally, the difference was divided by n -1 . The : remainder represented the position value for a given measure of media usage with the given observation’s value extracted: respondent's _ sum P0S'^C )n values individual value position value ~ (N-1) (see Appendix 1 for the actual algorithms used) This formula generated values for voice messaging usage measures equal to i ! the mean of all others in the position with the value of each given observation j extracted. It measures the effect of horizontal differentiation upon media usage . at the work unit level by measuring the behaviors of everyone in the formal posi- : I | I tion with the exception of the observation under consideration. Measures of the effect of vertical differentiation, or position adjacencies, are extracted directly from variables measuring each respondent’s superior’s media i usage behaviors (i.e., the superior-subordinate link— see Appendix 1). In effect, these two approaches measure communication media usage behaviors at the i 60 I role or subunit level of measurement according to formal patterns of organiza- i ' tional differentiation. Lastly, all individual observation values were weighted by the inverse of the number of persons in each position. This weighting procedure created behavioral values with an equal weight for each position regardless of the number of persons in the position (see Appendix 1). Data transformations creating a positional level-of-analysis were conducted 'on two levels: 1) the creation of structural equivalence-based measures of hori zontal differentiation (position) and vertical differentiation (adjacency), and 2) the weighting of all cases by the inverse of the position size. These two techniques allow all analyses to be conducted at the positional or work group level of analy sis. 3.2.3 - Scales and Dichotomizations Scale singularity was examined using the reliabilities procedure in the Statisti cal Program for the Social Sciences version 10 (SPSSX), Table 3 contains the individual means, standard deviations, grand means, and Alpha coefficients for all scales used in this research. All scales proved to be singular. Alpha coefficients ( were: .79 for task anaiyzability, .81 for innovativeness, and .86 for numbers of I exceptions. Scales were created using the mean of all non-missing items. : A wide variety of analyses were conducted comparing task environments : : based upon high and low anaiyzability, and upon numbers of exceptions. For i ! these analyses, the anaiyzability and exceptions scales were dichotomized using ' the SAS Rank procedure to produce median splits of high and low anaiyzability i and many and few exceptions. 61 i TABLE 3: SCALE CREATION STATISTICS - Innovativeness - 10 item variable mean std. dev. skeptical 5.38 1.56 cautious of accepting new ideas 4.97 1.62 rarely trust new ideas 5.65 1.31 stimulating to be original 5.66 1.45 on of the last to accept new 5.97 1.22 old ways are the best 6.01 1.13 wait for others 5.79 1.17 must see first 6.03 1.05 challenged by questions 5.53 1.37 challenged by ambiguities 5.45 1.33 N = 251 scale mean = 5.64 alpha = .81 low = 1 high = 7 - Task Anaiyzability - 4 item variable mean std. dev. clear known ways to do work 2.85 1.13 clear guidelines for work 3.02 1.10 understandable sequence of steps 2.76 1.11 rely on established practices 2.84 1.15 N = 251 scale mean = 2.87 alpha = .79 low = 1 high = 5 - Number of Exceptions - 5 item variable mean std. dev. is your work routine 2.33 1 01 do the same work most of the time 2.71 1 09 dept, does repetitive work 2.82 1 05 are your duties repetitive 2.49 1 02 same tasks from day to day 2.31 1 02 N = 251 scale mean = 2.53 alpha = .86 low = high = 1 5 62 : 3.3 Analyses I A range of analyses was conducted including simple descriptive statistics, , pair-wise t-tests, multiple regressions, and a series of LISREL models. The basic decision rule used throughout all analyses required a statistical significance level of £ <.05. Unless otherwise noted, all tests of significance were based upon two-tail distributions. The sample sizes vary according to the analyses. The range includes: 189 to 450 observations for the computer-monitored measures, 318 total questionnaire-based observations, and 125 questionnaire-based ob servations at both T1 and T2. As LISREL requires positive definite correlation matrices, list-wise deletion of ; missing observations was required to create each matrix. The combination of list-wise deletion and analyses based upon median splits of task anaiyzability resulted in some models based upon 30 to 40 observations. Due to this propen sity for shrinkage, submodels were created in specific tests to maintain adequate .sample sizes. Detailed information on the available sample is given for each analysis. Other than the LISREL models, all analysis was conducted with the Statistical Analysis System (SAS). SAS also was used to generate correlation matrices for the LISREL program. All generated matrices were written in double precision (12 'digit) format to ensure accurate estimates from LISREL’s iterative, maximum ! likelihood procedure. i 63 I CHAPTER FOUR: RESULTS i ! This chapter contains results from a variety of statistical analyses. The chap- | ter begins with descriptive statistics and zero-order correlations. The following i I j section contains a range of explicit hypotheses tests using mean-based t-tests ,and a series of multiple-regression models. Tests first are made with all task i 'environment covariances pooled (i.e., called “ full matrix models” ) and again for covariances based separately upon high and low levels of task anaiyzability and routinism. Section 4.3 extends the regression-based tests through a series of structural equation models using LISREL estimation. With the exception of the model identification hypothesis calling for a recipro- 1 cal (nonrecursive) relationship between sending and receiving voice messages (Hypothesis 4.1), all hypotheses were supported. 4.1 - Descriptive Statistics Table 4 contains descriptive statistics for the sample. The average respon- ; dent/subject in this study has been employed by organization for about 11 years, has some college education, and is at the midpoint on the organizational level I measure. He or she spends 28 percent of their messaging time using the active : [messaging capabilities, has 8 messages stored, records 1.5 and sends 1.8 ■ j 26-second messages per week. The difference between the recording and . sending numbers reflect the degree to which one “ broadcasts” a single mes sage to multiple others. He or she receives 1.8 38-second voice messages and 9.8 30-second answering machine messages per week. As mentioned earlier, : the average user has been on the system for about 11 weeks. TABLE 4 SUMMARY STATISTICS VARIABLE N MIN MAX MEAN STD DEV STD ERROR Self-Report Education 283 1.00 5.00 3.11 1.18 0.07 Years at Company 318 1.00 36.00 10.70 8.50 0.48 Organizational Level 283 1.00 9.00 3.63 2.04 0.12 Mean Innovativness 270 3.00 7.00 5.63 0.83 0.05 Mean Analayzability 275 1.00 5.00 2.89 0.88 0.05 Mean # of Exceptions 276 1.00 4.80 2.53 ' 0.83 0.05 % VM use for Messaging 221 0.00 100.00 27.76 33.31 2.24 Change In Ability/Obtain info 121 --4 .0 0 4.00 0.37 1.59 0.14 Change In Ability/Distrib. Info 120 -6.0 0 5.00 0.13 1.86 0.17 Monitored Raw Messages Stored 450 0.10 100.00 7.89 9.80 0.46 Raw messages Recorded 289 0.05 24.67 1.47 3.16 0.19 Raw Message Length 187 0.01 4.35 0.43 0.61 0.04 Raw Message Sent 289 0.05 24.67 1.84 3.98 0.23 Raw Tel. Ans. Messages 411 0.05 55.82 9.75 10.33 0.51 Raw Tel. Ans. Mess. Length 374 0.01 2.33 0.50 0.42 0.02 Raw Messages Received 430 0.05 23.00 1.36 2.13 0.10 Raw Length of Mess Recv. 385 0.01 5.42 0.64 0.84 0.04 # Weeks System Usage 479 1.00 21.00 12.30 5.86 0.27 Normed Messages Stored 450 -2.74 3.00 0.02 0.99 0.05 Normed Messages Recorded 289 -2 .7 0 2.86 0.00 1.00 0.06 Normed Message Length 187 -2 .5 6 2.72 0.00 0.98 0.07 Normed Messages Sent 289 -2 .7 0 2.86 0.00 1.00 0.06 Normed Tel. Ans. Length 411 -2.8 2 2.97 0.00 1.00 0.05 Normed Tel. Ans. Length 374 -2.94 2.94 0.00 1.00 0.05 Normed Messages Received 430 -2 .9 8 2.98 0.01 1.00 0.05 Normed Length Mess. Recv. 385 -2 .9 5 2.80 0.01 0.99 0.05 Vertical Innovativeness 241 3.00 7.00 5.82 0.87 0.06 Vertical % Active VM Use 228 0.00 100.00 41.56 36.04 2.39 Vertical Weeks Used VM 437 1.00 21.00 12.43 5.94 0.28 Vertical Messages Stored 414 -2.74 2.33 0.07 0.98 0.05 Vertical Messages Recorded 288 -2 .2 6 2.54 0.59 1.03 0.06 Vertical Message Length 238 -2 .5 6 2.72 0.47 0.93 0.06 Vertical Messages Sent 288 -2 .2 6 2.37 0.61 1.03 0.06 Vertical Tel. Ans. Mess. 377 -2.82 2.22 -0.1 4 0.84 0.04 Vertical Tel. Ans. Length 352 -2 .0 6 2.26 0.04 0.90 0.05 Vertical Mess. Received 381 -2 .1 0 2.51 0.40 0.98 0.05 Vertical Recv. Mess. Length 353 -1 .9 5 2.80 0.30 0.86 0.05 Horiz. Mess. Stored 385 -2.11 1.70 0.09 0.63 0.03 Horiz. Mess. Recorded 158 -2 .7 0 1.78 0.03 0.73 0.06 Horiz. Mess. Length 96 -2.22 2.61 -0 .0 6 0.76 0.08 Horiz. Mess. Sent 158 -2 70 2.07 0.03 0.73 0.06 Horiz. Tel. Ans. Mess. 300 -2 .1 8 2.05 0.12 0.81 0.05 Horiz. Tel. Ans. Length 264 -1.7 9 1.88 0.12 0.81 0.05 Horiz. Mess. Received 340 -2 10 1.94 0.05 0.78 0.04 Horiz. Recv. Mess. Length 291 -2.5 9 2.28 0.05 0.78 0.05 65 ! The system usage measures prefixed with “ normed," "vertical,” and "hori zontal” are based on mean zero, unit deviation scales (transformation processes were described in the preceding methods chapter), The communication per formance measures “ change in ability to obtain info" and “ change in ability to distribute info" are the differences between T1 values and T2 values. A zero for the performance measures indicate no change. Systems users reported a sig nificant increase in ability to obtain information from others (mean .37, std. error .14, p < .05) regardless of usage level or task environment. Table 5 contains a zero-order correlation matrix of the most important vari- ; ables. 4.2 - Tests o f Hypotheses 4.2.1 - Hypotheses 2.1 and 2.2 Table 6 displays breakdowns of voice messaging use variables by dichoto mized measures of task anaiyzability and routinism. Initial tests of Hypotheses 2.1 and 2.2 at the broadest level are conducted by examining mean-based t-tests. .Hypothesis 2.1: To the degree that task environment is less routine, it will be ( characterized by a greater reliance upon active voice messaging. ; Hypothesis 2.2: To the degree that task environment is less analyzabte, it will be 1 characterized by a greater reliance upon active voice messaging. Both routinism and anaiyzability at Time 1 (see note 1) significantly affect ! usage of active voice messaging features. Both high exception task environ ments and low anaiyzability task environments average about 33 percent of mes- i j saging behavior using the active messaging features (on table - % use messag- j ing). Conversely, low exception and high anaiyzability environments average i 66 TABLE 5 ZERO-ORDER CORRELATIONS 2 -.3 8 c 3 -.11 -.1 5 a 4 -.1 5 -.0 7 .590 5 -.0 5 -.0 8 ,52 c .300 6 -.0 9 .02 .570 ,28b ,54c 7 -,3 4 c -.0 6 .04 .00 .11 - .07 8 -.0 6 .05 - .08 .08 .04 - .08 - .01 9 -.0 8 .06 .05 - .16 .1 5 a - .02 - .04 .19a 10 -.1 6 a -.0 2 .40c ,59c ,27c ,19b ,15a ,14a .14 11 -.0 7 -.0 9 ,73c ,45c ,48c ,37c .230 ,16b .03 ,66c 12 -.13 -.21 a .11 ,16 ,25b .11 ,25b .26b -.0 4 .27b ,33c 13 .09 -,2 6 b ,36c .31a ,28b ,22a .13 ,20a -.0 5 .17 ,30c .47c 14 ,25c -.1 0 - .11 - .07 - ,23c .05 - ,27c --,16b -.0 2 .00 --.17b - 2 5 b - -.19a 15 .240 -.0 9 - .04 - .09 - ,12 .09 - ,24c - -.15a -.11 - -.06 --,16b -.0 7 --.09 67c 1 2 3 4 5 6 7 8 9 10 11 12 13 14 N = 65 to 256 1 — ORGANIZATIONAL LEVEL 2 = TENURE 3 = MEAN # OF MESSAGES RECEIVED BY POSITION 4 = MEAN # OF MESSAGES SENT BY POSITION 5 = MEAN # OF MESSAGES RECEIVED BY VERTICAL ADJACENCY 6 = MEAN # OF MESSAGES SENT BY VERTICAL ADJACENCY 7 - EDUCATION 8 = INNOVATIVENESS 9 = INNOVATIVENESS OF VERTICAL ADJACENCY 10 = # OF MESSAGES SENT BY OBS. (WEIGHTED BY POSITION SIZE) 11 = # OF MESSAGES RECEIVED BY OBS. (WEIGHTED BY POSITION SIZE) 12 = T1 TO T2 CHANGE IN ABILITY TO OBTAIN INFORMATION 13 = T1 TO T2 CHANGE IN ABILITY TO DISTRIBUTE INFORMATION 14 = ANALYZABILITY OF TASK ENVIRONMENT 15 — ROUTINISM (NUMBER OF EXCEPTIONS) OF TASK ENVIRONMENT a= e.c.05 b= p.<.01 c= p_<.001 Correlations are based on pair-wise deletions weighted by position Sample size varies - see Table 4 for available N List-wise deletions were used for Lisrei models - correlations may vary 67 TABLE 6 T-TESTS OF MEAN DIFFERENCES BASED UPON TASK ENVIRONMENT Part A: Routinism (Lack of Exceptions) in Task Environment few exceptions many exception variable m ean std. dev m ean std. dev. T self-rep o rt mess, sent/recv 7.27 5.75 6.26 5.27 -1.25 % messaging 21.64 29.43 33.61 35.97 2.51 * * m onitored mess, stored 6.74 7.82 7.68 7.99 1.41 mess, recorded 0.82 1.92 1.24 2.20 2.71 * * mess, length 0.18 0.50 0.21 0.33 2.69** mess, sent 1.17 3.08 1.38 2.59 2.41** mess, received 1.10 1.59 1.49 1.87 2.28** Part B: Anaiyzability in Task Environment high anaiyzability low anaiyzability variable m ean std. dev. m ean std. dev T self-rep o rt mess, sent/recv 6.79 5.58 6.81 5.27 .30 % messaging 22.58 30.20 32.92 35.76 2.16** m onitored mess, stored 6.35 6.94 8.31 9.03 2.15* mess, recorded 0.87 1.86 1.21 2.31 0.73 mess, length 0.18 0.47 0.22 0.36 1.73* mess, sent 1.17 2.81 1.40 2.97 0.46 mess, received 1.09 1.35 1.54 2.14 1.30 ...... .......................... Monitored variables have been transformed from negative exponential to normal distributions for significance tests. T values and probability statistics represent transformed values. Transformation process is described in section 3.2.1. * pc.05 ** £ > <.01 68 only about 22 percent of messaging behaviors using these active capabilities (g < .01 for both t-tests). Since no significant reported differences exist between the total number of messages either set of groups sends or receives (in Table 6 - messages sent/recv.), these differences should reflect both relative and absolute differ ences in behaviors. To further analyze these associations, more explicit tests were made using computer-monitored, behavioral measures of active messaging behaviors. Five ' measures of active voice messaging behaviors (number of messages stored, number of messages recorded, number of messages sent, length of messages sent, and number of messages received) were tested across the two measures of task environment. Tests between environments with high and low numbers of exceptions universally found the high exception environments to use the active messaging features more heavily (g <.01; see Part A). ; System usage for low anaiyzability task environments exceeded that for high anaiyzability environments for all measures, but only differences between stor age, message length, and numbers of messages received were statistically sig- . nificant across the two task environments. The most general hypotheses relating task to messaging behaviors (2,1 and 2.2) received substantial support from i these preliminary analyses. 69 4.2.2 - Hypotheses 1.1 and 1.2 Hypotheses 1.1 and 1.2 will receive a series of tests in both multiple regres sions and LISREL models. Support of Hypothesis 1.1 is essential for the pro posed model. Hypothesis 1.1: To the degree that actors in a system are structurally equiva lent, they will manifest similar voice messaging behaviors (position effect). Hypothesis 1.2: Due to positional adjacencies created by vertical differentiation, superiors’ voice messaging behaviors influence subordinate positions' voice mes saging behaviors. Table 7 A reports the effects of horizontal and vertical differentiation on active voice messaging behaviors (number of messages sent, length of messages sent, number of messages received, and length of messages received). Table 78 reports the effects of vertical and horizontal differentiation on passive voice messaging behaviors (number and length of "answering machine” messages received). In all situations, the coefficients and explained variances represent each ob servation’s dependent variable value regressed upon the values of the corre sponding position and vertical adjacency variables weighted by the inverse of the number of observations in each position. Both tables report effects for five classifications: full matrix (all observations), I low anaiyzability, high anaiyzability, many exceptions, few exceptions. Hypothe ses 1.1 and 1,2 are tested at the full matrix level. Structural equivalence (posi- j tion effect) is a significant and powerful determinant of active and passive mes- j saging behaviors in all instances. Position explains approximately 30 percent of ’ the variance in sending messages, and 50 percent of length of messages sent. 70 TABLE 7A ■IMPACTS OF STRUCTURAL EQUIVALENCE ON- ACTIVE MESSAGING BEHAVIORS----- - Sending - de pendent variable = # of messages sent task environment N Position Adjacency full matrix low anaiyzability high anaiyzability many exceptions few exceptions .29* * * .59*** .16 40* * * .20 62 26 20 29 17 .54 * * * .85*** .24 .61*** .44* .02 -.2 0 .31 .05 . .03 dependent variable - length of messages sent task environment R 2 N Position Adjacency full matrix .50*** 62 .64*** .15 low anaiyzability 47* * * 26 .74*** -.14 high anaiyzability ^43* * 20 .53** .42* many exceptions 42* * * 29 .68*** .08 few exceptions 44* * 17 .56** .39* - Receiving - dependent variable = # of messages received task environment R N Position Adjacency full matrix low anaiyzability high anaiyzability many exceptions few exceptions .49**’ .47**’ 4 2 * * . .39**’ .47* *' 253 67 78 77 77 .60*** .46*** .65*** .51 ** * ,69*** .17** .31** .02 . 20 * -.01 i dependent variable = length of messages received task environment R 2 N Position Adjacency full matrix low anaiyzability high anaiyzability many exceptions few exceptions .56*** 253 .70*** .10* .29*** 67 .51*** .10 .44*** 78 .66*** .02 .17*** 77 .42*** .01 .55*** 77 .73*** .04 * p<,05 * * D< 01 *** (beta coefficients are standardized; t-tests are one-tail) 71 TABLE 7B - — IMPACTS OF STRUCTURAL EQUIVALENCE ON----- PASSIVE MESSAGING BEHAVIORS----- deipendent variable = # of messages sent task environment R 2 N Position Adjacency full matrix .50'* * 230 .64* * * .13** low anaiyzability 4 ■ ] * * * 56 54* * * .01 high anaiyzability ^52* * * 75 *56* * * .28** many exceptions .54*** 59 .39*** .45*** few exceptions .51*** 72 .02 j dependent variabie = length of messages sent task environment R2 N Position Adjacency full matrix .32*** 230 .35** * .35* ** low anaiyzability .24* * * 56 4 4 * * * .22* high anaiyzability ^34* * * 75 '22* 4 7 * * * many exceptions .4 1 * * * 59 .38** * .45*** few exceptions ^25* * * 72 .28** .36*** iH H iH i I , , p < 05 * * Q * | *** (beta coefficients are standardized; t-tests are one-tail) 72 1 Vertical differentiation, however, rarely achieved statistical significance, Sam ple: i size for these regressions range from 62 to 253 with (p < .001) in all cases. | Hypothesis 1.2 received moderate support at the full-matrix level. Vertical ! adjacency behaviors were significant predictors of all messaging behaviors ex cept number of messages sent, and the length of messages sent. Only in the : passive system usage equation for length of message recorded did it equal the coefficient and significance level achieved by the horizontal differentiation value (b=.35, p< .001). In the six regression equations, position (horizontal differentiation) and adja cency (vertical differentiation) explained an average of 44 percent of the vari- ‘ ance in messaging behaviors. The average standardized beta coefficients were .58 for horizontal differentiation and .15 for vertical differentiation. Horizontal differentiation accounted for about 75 percent of the explained variance across the six equations (i.e., vertical differentiation accounted for 25 percent of the explained variance). LISREL is used later for more comprehensive tests of these and other relationships. 4.2.3 - Hypotheses 2.3 and 2.4 ! Hypotheses 2.3 and 2.4 basically combine hypotheses 1.1 with 2.1 and 2.2. . They call for explicit relationships between structural equivalence, task environ ment and system usage. ! I I Hypothesis 2.3: To the degree that task environment is less routine, horizontal ! differentiation (position effect) will be a stronger determinant of using active voice messaging features which facilitate coordination of workflows. ' Hypothesis 2.4: To the degree that task environment is less analyzable, horizon- | tal differentiation (position effect) will be a stronger determinant of using active j voice messaging features which facilitate coordination of workflows. 1 Table 7 A contains multiple regression coefficients tests for Hypotheses 2.3 and 2.4. In all cases, the coefficients for low analyzable environments were higher than those for high analyzable environments. Similarly, coefficients for environments with many exceptions to routine universally were larger than those for environments with fewer exceptions. Coefficients for sending messages: = • low analyzable (.85, p <.001); high analyzable (.24, p = n.s.) • many exceptions (.61, p <.001); few exceptions (.44, p <.05). Coefficients for length of messages sent: • low anaiyzability (.74, p <.001); high anaiyzability (.53, p < .01); • many exceptions (.68, p <.001); few exceptions (.56, p <.01), Table 7C contains Z ' tests of beta coefficients for number of messages sent and length of messages sent for high and low analyzable task environments, and for high and low routinism environments (see note 3). Position effect on sending messages was significantly stronger for low analyzable environments compared ' to high analyzable environments (.84 vs. .24, Z' = 4,5, p <.001). Although the difference was quite pronounced, position effect on length messages sent was not significantly stronger for low analyzable environments compared to high ana- ! lyzable environments (.74 vs. .54, Z' = 1.67, p = n.s). Differences for position effects were not as strong across routine and non routine environments as those differences between the coefficients for high and 1 low anaiyzability environments. Position effect on sending messages was not ; significantly stronger for nonroutine environments compared to routine environ- i j ments (.61 vs. .44, Z! - 1.07, p = n.s.). Similarly, position effect on length 74 TABLE 7C Z' TEST OF REGRESSION COEFFICIENTS - COMPARISON OF POSITION EFFECT ON MESSAGING - — Sending— task environment / behavior coeff. N f Z analyzabitilty low analyzable / # messages sent .85 26 high analyzable / # messages sent .24 20 4 .5 *** low analyzable / length of messages .74 26 high analyzable / length of messages .53 20 1.67 routinism many exceptions / # messages sent ,61 29 few exceptions / # messages sent .44 17 1.07 many exceptions / length of messages .68 29 few exceptions / length of messages .56 17 .88 g<.05 ^ "0 0 1 (beta coefficients are standardized) 'messages sent was not significantly stronger for nonroutine environments com- [pared to routine environments (.68 vs. .56, Z' = .88, p = n.s.). Hypothesis 2.4 received strong support for sending messages, but only direc tional support for length of messages sent, Hypothesis 2.3 only received direc tional support for sending messages and length of messages sent. Tests are conducted later with LISREL models which help explain these asso ciations and more adequately test relationships. j 4.2.4 - Hypotheses 3.1, 3.2, and 3.3 These hypotheses are concerned with the effect of voice messaging behav- liors upon communication performance. Tables 8A and 8B contain multiple re- i gression models used to test these hypotheses. Hypothesis 3.1: Passive use of voice messaging does not possess as much capacity for enhancing communication performance as does active use of voice messaging. Hypothesis 3.2: Active use of voice messaging in less analyzable task environ ments will result in greater increases in communication performance than will use of active voice messaging features in more analyzable task environments. Hypothesis 3.3: Active use of voice messaging in less routine task environments will result in greater increases in communication performance than will use of active voice messaging features in more routine task environments. ■ Table 8B shows the results of regressing the two, over-tim e communication i j performance measures— 1) change in the ability to obtain information in a timely < manner, and 2) change in ability to distribute information to groups of peo ple— on measures of passive system use. Passive use of voice messaging had no significant effect on the measures of communication performance. Across the ten equations, passive use of messaging averaged an explained variance of 6 TABLE 8A — IMPACTS OF SYSTEM USAGE ON PERFORMANCE----- -Active System Usage and Performance Change- - Obtaining Information - dependent variable = ability to obtain info, from others on time | o # of mess. message I task environment R 2 N received length full matrix .18*** 103 .51 * * * -.1 5 low anaiyzability .23** 48 .66** -.27 high anaiyzability .09 54 .34* -.11 many exceptions 24* * * 56 .64*** -.2 4 few exceptions !08 46 .33* -.11 - Distributing Information - j dependent variable = ability to distribute info, to groups of people | task environment # of mess. message R N sent length full matrix low anaiyzability high anaiyzability many exceptions few exceptions .04 68 .08 .15 .26** 31 -.11 .58** .05 36 .19 -.2 3 .06 40 -.11 -.3 0 .18 27 .45 -.01 n n j r~; E< 05 1 * * O ' ' 01 ' * ** on* (beta coefficients are standardized; t-tests are one-tail) 77 TABLE 8B IMPACTS OF SYSTEM USAGE ON PERFORMANCE ! -Passive System Usage and Performance Change- I - Obtaining Information - dependent variable = ability to obtain info, from others on time # of mess. message task environment R 2 N taken length full matrix .03 109 .05 .13 low anaiyzability .10 49 .11 -.2 3 high anaiyzability .01 59 -.11 .11 many exceptions .04 55 -.0 3 .22 few exceptions .05 53 ■24 -.0 3 - Distributing Information - dependent variable = ability to distribute info, to groups of people task environment # of mess. message R N taken length full matrix low anaiyzability high anaiyzability many exceptions few exceptions .01 109 .12 -.0 3 .11 49 .18 .20 .01 59 .01 -.11 .05 55 .08 .17 .01 53 .06 -.14 p<.05 D<.01 (beta coefficients are standardized; t-tests are one E<.001 78 percent, In no test did any of the coefficients or any of the equations reach ; statistical significance, p ■ Table 8A depicts the effects of active use of voice messaging on the same i .performance measures. Number of messages received was a significant predic tor of an increased ability to obtain information in all equations (betas ranged .from .33 to .66; significance ranged from p < ,05 to p < .001). Length of messages received was not a significant predictor of ability to obtain information in any equation. The multiple regression models for the full matrix, for low ^'anaiyzability, and many exceptions, all reached statistical significance. Length of messages sent was a significant predictor of change in ability to distribute information (beta=.58, p < .0 1 ) for low analyzable environments, Num- : ber of messages sent was a significant predictor in the case of environments with few exceptions (b e ta -45, p <.05). Average explained variance across all equa tions was 14 percent. The significant R-squares ranged from .18 to .26. Hy pothesis 3.1 received substantial support. Hypothesis 3.2 calls for stronger associations in environments characterized by low anaiyzability. Table 8A is used to test this hypothesis. The R-squares for obtaining information are: low anaiyzability (.23, p <.01); high anaiyzability (.09, p=n.s.). The R-squares for distributing information are: low anaiyzability (.26, p <.01); high anaiyzability (.05, p=n.s.). Hypothesis 3.2 was supported, i Hypothesis 3.3 calls for stronger associations in environments characterized by low routinism. Table 8A is used to test this hypothesis. The R-squares for obtaining information are: many exceptions (.24, p <.001); few exceptions (.08, < !p=n.s.). The R-squares for distributing information are: many exceptions (.06, i !p=n.s.); few exceptions (.18, p=n,s.). Hypothesis 3.3 was supported only for obtaining information. The preceding tests offer general support for the model proposed in this j dissertation. However, multiple regressions such as the above tend to isolate ! associations and fail to provide a comprehensive test of the model as a system i of covariances. The following section uses LISREL to provide a much stronger i and more comprehensible test. 1 1 r i 4.3 - Structural Equation Models A series of LISREL models based upon the theoretical model shown in Figure 2 were analyzed first using the full matrix and then the two matrices based upon median splits of task anaiyzability. Models were based only upon anaiyzability due to several conceptual and empirical considerations. Numbers of exceptions and anaiyzability manifested similar results in the regression analyses, with numbers of 1 exceptions yielding slightly higher coefficients in several instances. With similar findings, parsimony would dictate the use of a single indicator for a series of structural equation models. Also, the samples based upon median splits and list-wise deletions of cases left the number of exceptions groups with vastly dif fering samples for several key equations (40 and 27 for messages received). Conceptually, anaiyzability offers an additional appeal as it more nearly cap- : tures the substance of information richness and needs of collaborative work. As mentioned in the section on information richness, anaiyzability is nearly identical to Daft and Lengel’s concept of equivocality. J To thoroughly test the hypotheses proposed in this research, only three mod els would have been used: one for the full matrix, and one each for high and low anaiyzability. As explained in the analysis section, however, cases were depleted I too rapidly to test the full theoretical model with half the available sample (i.e., based upon the median splits). Because of this limitation imposed by sample size, the full theoretical model is tested using a full matrix, and smaller LISREL models are used to examine hypotheses relating specifically to task environment. Figure 3 depicts the estimated model based upon the full matrix of system users. All observations in this matrix are users of active voice messaging fea tures. This is to say, any observation included in this model must have sent and received messages during the five month duration of behavioral data collection. The full theoretical model (Figure 2) was submitted for LISREL estimation. An iterative process of "fixing” non-significant paths trimmed the model to the state depicted in Figure 3 (see note 2). This theory trimming process involved running the model, detecting and fixing a single path with the lowest absolute t-statistic, then re-running the reduced model. This process was repeated until only paths significant at the e < .05 level (2-tail t-test) remained. The figure indicates residual terms as left pointing, unconnected arrows di rected at the endogenous variables. All path coefficients are maximum likelihood estimates in standardized form. The total explained variances for endogenous variables are indicated in parentheses directly above the variable. Goodness of fit statistics for the entire model are depicted in the lower left corner of the figure. Variables are numbered from top to bottom, from left to right to facilitate path notation. Standard path notation is used (i.e., a path from innovativeness to number of messages received is called path 3,6). Though not shown on the figure, all exogenous variables are free to correlate with one another (i.e., a free, symmetric phi matrix with ones on the diago nal— Table 9 depicts all paths and total effects for the model). As no nonrecur sive links are modeled, the error terms are uncorrelated (i.e., a free, diagonal FIGURE 3: STRUCTURAL EQUIVALENCE, VM BEHAVIOR & COMM. PERFORMANCE FULL MATRIX CO NJ 10 11 ORGANIZATIONAL LEVEL .39 INNOVATIVENESS .23 HORIZONTAL DIFF. RECEIVE MESS. VERTICAL DIFF SEND MESSAGES .31 .65 GOODNESS OF FIT FOR ENTIRE MODEL Chi square = 25.85 (fi= -26) degrees of freedom = 22 adjusted goodness of fit = .81 coeff. of determination = .87 HORIZONTAL DIFF. SEND MESSAGES ,49 VERTICAL DIFF. RECEIVE MESS. .32 SUPERVISOR’S INNOVATIVENESS -.27 •79) (.37) # OF MESSAGES RECEIVED .29 3 ♦ | .46 (.60) | # OF MESSAGES SENT 4 k .40 CHANGE IN ABILITY TO OBTAIN INFO (.39) .63 .63 CHANGE IN ABILITY TO DISTRIBUTE INFO .61 hypothesized paths control paths N = 33 (R-2’s for endogneous variables are in parantheses) TABLE 9 j FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 3 i J Phi Matrix vert. send vert. recv horiz. send horiz. recv. org. level innov. super innov. vsend 1.00 vrecv .83 1.00 hsend .54 .63 1.00 hrecv .66 .67 .73 1.00 orglvl -.3 0 -.3 2 -.1 7 -.41 1.00 innov -.0 8 -.1 4 -.0 4 -.0 2 -.3 5 1.00 sinnov -.0 4 .09 -.0 5 -.0 7 .10 .22 1.00 TOTAL EFFECTS Total effects of X on Y vert. send vert. recv. horiz. send horiz. recv. org. level innov. super. innov. obtain .08 .04 .06 .08 .39 .63 -.0 4 distribute .06 .03 .04 .05 .25 .40 -.0 2 sending 0 .32 .48 0 0 ,23 -.2 7 receiving .31 .15 .22 .26 0 -.0 5 -.12 Total effects of Y on Y obtain info. distrib. info. send mess. receive mess. obtain 0 0 .13 .29 distribute .63 0 .08 .18 sending 0 0 0 0 receiving 0 0 .46 0 83 [psi matrix). i I The model has a nonsignificant chi square statistic (28.85, g=.26) indicating i that the associations proposed by the model are not significantly different from the observed associations. The adjusted goodness of fit index is .81, and the coefficient of determination is .87, i 1 4.3.1 - Hypotheses 4.1 and 4.2 These two hypotheses specify relationships between endogenous variables. Hypothesis 4.1: A nonrecursive relationship exists between sending and receiv ing voice messages. Hypothesis 4.2: The ability to obtain information will result in an increased ability to distribute information, Hypothesis 4.1 states that the number of messages sent will partially deter mine the number of messages received, and that the number of messages re ceived will partially determine the number of messages sent. This hypothesis was rejected. Sending messages partially determines the number of messages re ceived (path 3,4 - beta .46, g < .001), but receiving messages had no signifi- : cant effect on sending. i | Hypothesis 4.2 states that an increase in ability to obtain information will in- ! crease one's ability to distribute information. This hypothesis was supported (g i ! ! 2,1 - beta .63, g < .001). i I 4.3.2 - Analyses of Hypotheses 1.1 and 1.2 ; Hypothesis 1.1: To the degree that actors in a system are structurally equiva- l lent, they will manifest similar voice messaging behaviors (position effect). Hypothesis 1.2: Due to positional adjacencies created by vertical differentiation, superiors’ voice messaging behaviors influence subordinate positions' voice mes saging behaviors. Hypothesis 1.1 is tested through path 4,9 and path 3,7. Both paths are signifi cant. Horizontal differentiation is a powerful determinant of sending messages (path 4,9 - gamma .49, g < .001), and a moderate determinant of receiving messages (path 3,7 - gamma ,26, g < ,001). Hypothesis 1.1 is supported. ; Hypothesis 1.2 also is supported, but in a different manner than theoretically postulated. The theoretical model (Figure 2) indicated that sending behaviors by a superior would predict subordinate sending behaviors, and that superior receiv ing behaviors would predict subordinate receiving behaviors (i.e., parallel process of vertical interaction). ! Path 3,8 and path 4,10 represent the influence of vertical differentiation on voice messaging behavior. The effect for receiving messages (path 3,8 - gamma .31, g < .001) and the effect for sending messages (path 4,10 — gamma .32, g < .01) are significant, but reversed from the paths specified in the theoretical model. Superior sending determines subordinate receiving and supe rior receiving determines subordinate sending (i.e., sequential processes of inter action) . A degree of vertically oriented communication is indicated in these asso ciations. These associations will be explored further in following models based j upon dichotomized task environments. i i 4.3.3 - Communication Performance Though no performance hypotheses are testable based upon this full matrix model (all hypotheses regarding communication performance were based upon differences across task environments), the number of messages sent was a 85 significant determinant of change in ability to obtain information "(path T,3 - beta ; , .29, p <.01). The number of messages sent only exerted indirect influences on I change in ability to distribute information through the number of messages sent land change in ability to obtain information (path 2,4 - total effect .13). 4.3.4 - Control Variables Organizational level directly affected change in ability to obtain information at * any usage level (path 1,3 - gamma .39, p < .001). As the scale for organiza- i : tionai level is inverted (1=management, /^secretarial, etc.), this association indi- ' cates that lower level individuals receive greater performance enhancement at ' any usage level. j !-->• Similar to organizational level,- innovative persons reported substantially ' greater improvement in ability to obtain information at any usage level (path 1,6 — ffa m m a .65, p < ,001). Innovativeness is positively associated with number of ^messages received (path 3,6 - gamma .23, p < .01) and negatively associated :*with number of messages sent (path 4 ,6 — gamma -.1 5 , p < .01). Supervisor! i : innovativeness also was negatively associated with number of messages sent by the subordinate (path 4,11 - gamma -.2 7 , p < .001). Full Matrix Model Explained Variance ! i The full matrix model of active users explained approximately 7G percent o f ; j the variance in sending and receiving behaviors (60 percent of sending and 79 percent of receiving) . It also expiained about 38 percent of change in communi cation performance (37 percent of ability to obtain and 39 percent of ability, i 1 1 distribute information). 86 1 4.3.5 - Task Environment and Voice Messaging Behaviors t i Separate matrices were created based upon a median split of task i I analyzability. Figure 4 depicts the reduced theoretical model (no performance [ 'measures) used to analyze differences in voice messaging behaviors in high and low analyzable task environments. The reduced models allow sufficient sample size for stable LISREL estimates when working with split samples. As in the original model, these models were estimated by beginning with all posited links free, followed by iterative theory trimming based upon fixing paths with lowest absolute t-values. Because these reduced models include an additional 29 observations ex cluded from the full model, an initial model based upon the full matrix version was estimated in Figure 5. 4.3.6 - Full Matrix Model f ' Most of the control variables which contributed significantly to the original full model were not significant in this reduced model. Education, however, was asso- ; ciated significantly with number of messages sent (path 2,7 - gamma .22, q < ! .01). i I Differences in significant paths of control variables could be attributable to i ; removal of indirect paths to the performance variables, more stable estimates i ; achieved by a greater N with fewer variables, and of course, due to the inclusion ; of the 29 additional observations. FIGURE 4: THEORETICAL MODEL OF STRUCTURAL EQUIVALENCE & VM BEHAVIORS ORGANIZATIONAL LEVEL TENURE HORIZONTAL DIFF. RECEIVE MESSAGES VERTICAL DIFFERENTIATION RECEIVE MESSAGES HORIZONTAL DIFFERENTIATE SEND MESSAGES VERTICAL DIFF. SEND MESSAGES EDUCATION INNOVATIVENESS SUPERVISOR’S INNOVATIVENESS # OF MESSAGES RECEIVED # OF MESSAGES SENT hypothesized paths control paths FIGURE 5: STRUCTURAL EQUIVALENCE & VM USE - FULL MATRIX VERTICAL DIFF. SEND MESSAGES HORIZONTAL DIFF. RECEIVE MESS. VERTICAL DIFF. RECEIVE MESS. HORIZONTAL DIFF SEND MESSAGES EDUCATION # OF MESSAGES RECEIVED N = 62 GOODNESS OF FIT MEASURES FOR ENTIRE MODEL Chi square = 3.19 (fi= .67) degrees of freedom = 5 adjusted goodness of fit = .94 coefficient of determination = .69 .33 # OF MESSAGES .45 hypothesized paths control paths (R-2’s for endogneous variables are in parantheses) TABLE 10 FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 5 Phi Matrix vert, send vert. recv. horiz. send horiz. recv. educ. vsend 1.00 vrecv .66 1.00 hsend .39 .66 1.00 hrecv .65 .66 .71 1.00 educ -.1 5 -.1 3 .07 -.0 3 1.00 i TOTAL EFFECTS Total effects of X on Y vert. send vert. recv. horiz. send horiz. recv. educ. sending receiving 0 .28 .28 .15 .49 .26 0 .20 .22 .12 Total effects of Y on Y send mess. receive mess. sending receiving 0 .53 0 0 WvWuV--:; 90 i The estimated paths from the hypothesized differentiation variables and the I ; voice messaging variables, however, are nearly identical across the two models. . Comparisons are as follows for the full and reduced models respectively: ■ • vertical differentiation on receiving (.31, .28); • horizontal differentiation on receiving (.26, .20); • vertical differentiation on sending (.32, .28); and • horizontal differentiation on sending (.49, .49). I i 4.3.7 - Low Analyzable Environments Figure 6 depicts the trimmed model for persons in less analyzable task envi ronments. Horizontal differentiation is a powerful determinant of number of mes sages sent (path 2,6 gamma .75, p < .001), and a substantial determinant of number of messages received (path 1,4 - gamma .31, p < .01). Education is • positively associated with number of messages sent (path 2,6 - gamma .27, p < i .01) and innovativeness is negatively associated with number of messages re- » ceived (path 1,3, - gamma -.2 0 , p < .01). Sending messages is a powerful : determinant of receiving messages (path 1,2 - beta .63, p < .001). Vertical differentiation (supervisor messaging behaviors) did not affect significantly posi tion’s messaging behaviors. Approximately 70 percent of the variance in active voice messaging behav iors is explained by the model for low analyzable task environments (71 percent, of sending behaviors and 68 per percent of receiving behaviors). The total: I effect of horizontal differentiation on receiving messages is .78 and the total effect of horizontal differentiation on sending messages is .75. ; 91 I FIGURE 6: STRUCTURAL EQUIVALENCE & VM USE - LOW ANALYZABLE ENVIRONMENTS 3 INNOVATIVENESS .33 -.20 (.68) HORIZONTAL DIFF. RECEIVE MESS. 4 # OF MESSAGES RECEIVED .31 .29 ,63 (.71) # OF MESSAGES SENT HORIZONTAL DIFF SEND MESSAGES 5 6 EDUCATION .27 hypothesized paths control paths GOODNESS OF FIT MEASURES FOR ENTIRE MODEL Chi square = 5.65 (p = .28) |M =: 3 4 degrees of freedom = 4 adjusted goodness of fit = .77 coefficient of determination = .78 (R-2’s for endogneous variables are in parantheses) TABLE 11 FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 6 i I \ TOTAL EFFECTS Total effects of X on Y horiz. send horiz. recv. educ. innov. sending receiving .75 .47 0 .27 0 .31 .17 -.2 0 Total effects of Y o n Y send mess. receive mess. sending receiving 0 ,63 0 0 93 Phi Matrix horiz. send horiz. recv. educ. innov hsend hrecv educ. innov. 1.00 .69 .21 .12 1.00 .14 .12 1.00 -.01 1.00 ! 4.3.8 - High Analyzable Task Environments Figure 7 depicts the estimated model for high analyzable task environments. Vertical differentiation is a powerful determinant of receiving messages (path 1,3 - gamma .62, p < .001), as is number of messages sent (path 1,2 - beta .45, P < .001). Horizontal differentiation partially determines number of messages sent (path 2,4 - gamma .50, p < .001). Innovativeness also is associated signifi- : cantly with sending messages (path 2,5 - gamma .39, p < .01). Approximately 56 percent of the variance of active messaging behavior in high analyzable task environments is determined by the model (76 percent of receiving messages and 36 percent of sending messages). The total effect of horizontal differentiation on sending messages is .50, and the total effect of horizontal differentiation on receiving messages is .23. i 4.3.9 - Comparison of Task Environments Relationships between measures of differentiation and active voice messaging behaviors differ both in magnitude and in character in the low versus high 1 analyzability task environments. While vertical differentiation (influence of superi or’s messaging behaviors) had no influence on messaging behaviors in low ana- Ivzable environments, it was the only exogenous determinant of receiving mes- ; saaes in the high analyzable environments. Horizontal differentiation was the I most powerful predictor of sending and receiving in low analyzable environments. i ; i Additionally, horizontal differentiation generated substantially more total effect on endogenous variables in the low analyzable environments (see Table 13, Part A). t 94 FIGURE 7: STRUCTURAL EQUIVALENCE & VM USE - HIGH ANALYZABLE ENVIRONMENTS ! VERTICAL DIFF. SEND MESSAGES HORIZONTAL DIFF SEND MESSAGES INNOVATIVENESS (■76) # OF MESSAGES RECEIVED (.36) I .45 # OF MESSAGES SENT .23 .64 hypothesized paths control paths GOODNESS OF FIT MEASURES FOR ENTIRE MODEL N = 28 Chi square = 1.27 (g = .74) degrees of freedom = 3 adjusted goodness of fit = .91 coefficient of determination = .75 (R-2’s for endogneous variables are in parantheses) TABLE 12 FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 7 Phi Matrix vert. send horiz. send innov vsend 1.00 hsend .40 1.00 innov. .14 -.10 1.00 TOTAL EFFECTS Total effects of X on Y vert. send horiz. send innov. sending 0 .50 .39 receiving .62 .23 .17 Total effects of Y on Y send receive mess. mess. sending 0 0 receiving .45 0 96 TABLE 13 HORIZONTAL DIFFERENTIATION AND MESSAGING----- PART A Total Effects Comparisons of Horizontal Differentiation by Analyzability analyzability behavior high low sending .50 .75 receiving .23 .78 j PART B Z Tests of Path Coefficients Across Environments task environment/ path coeff. N / Z Position on Sending low analyzable environments .75 34 high analyzable environments .50 28 2 .2 * Position on Receiving low analyzable environments .31 34 high analyzable environments ns 28 * * * Sending on Receiving low analyzable environments .63 34 1.33 high analyzable environments .45 28 ! * £<.05 I * * D < 0 1 ** * rt/n m (beta coefficients are standardized; t-tests are one-tail) 97 • high analyzable / sending messages: total effect = .50 • low analyzable / sending messages: total effect = ,75 • high analyzable / receiving messages: total effect = ,23 • low analyzable / receiving messages: total effect = ,78 Similar relationships between high and low analyzable environments were evi dent for the individual coefficients (see Table 13, Part B). Z' tests were used to determine if the coefficients of the direct and indirect paths between horizontal differentiation and messaging for high and low analyzability environments were ■significantly different (see note 3). The direct effect of position on sending i messages (low analyzable = .75, high analyzable = .50, Z' = 1.33, g <.05) and the direct effect of position on receiving messages (iow analyzable — .31, high ; analyzable = n.s) were significantly higher for low analyzability environments than 'for high analyzability environments. The hypothesis of stronger effect of position on voice messaging behaviors in less analyzable environments is strongly sup ported (i.e., Hypothesis 2,3). Though not directly posited by a hypothesis, the relationship between sending and receiving messages was examined using a Z' test. Although the effect of sending messages on receiving messages was greater for low analyzable envi ronments, the difference was not significant at g <.05 (low analyzable = .63, high analyzable = .45, Z’ = 1.33). 4.3.9.1 Messaging and Performance Similar to the preceding models of horizontal differentiation and active mes saging behaviors, a series of LISREL models was used to examine differences between relative communication performance gains in high and low analyzable i environments. Figure 8 depicts the theoretical model used in the analyses, In 98 FIGURE 8: THEORETICAL MODEL OF VM USE & COMM. PERFORMANCE # OF MESSAGES RECEIVED CHANGE IN ABILITY TO OBTAIN INFORMATION LENGTH OF MESSAGES RECEIVED # OF MESSAGES SENT CHANGE IN ABILITY TO DISTRIBUTE INFORMATION LENGTH OF MESSAGES SENT hypothesized paths control paths ;addition to numbers of messages sent and received, the lengths of the mes sages also are analyzed. An initial model analyzed the full matrix and is portrayed in Figure 9. Number of messages received was the only significant exogenous variable (path 1,3 - gamma .37, g < .001). Change in ability to obtain information was a significant predictor of change in ability to distribute information (path 2,1 - beta .42, g < .001). Much like the full model (Figure 2), no messaging behavior had a direct ieffect on change in ability to distribute information. The model is relatively weak, explaining only 14 and 18 percent of the variance in ability to obtain and ability to distribute information respectively. 4.3.9.2 - Low Analyzable Environments Figure 10 depicts the estimated model for active messaging behaviors and communication performance in low analyzable environments. Number of mes sages received is a significant determinant of change in ability to obtain informa tion (path 1,3 -gam m a .42, g < .01). Additionally, length of messages sent is a significant predictor of change in ability to distribute information (path 2,4 - I j gamma .39, g < .01). The low analyzability model is the only model capable of I j generating a link from messaging to distributing information (an important finding I discussed further in Chapter 5). Similar to all preceding performance models, ; change in ability to obtain information is a significant predictor of change in ability to distribute information (path 2,1 - beta .35, g < .01). The low analyzable task environment model explains about 27 percent of the ; variance in change in communication performance (17 percent of the variance in i ability to obtain information and 36 percent of the variance in ability to distribute FIGURE 9: VM USE & COMM. PERFORMANCE - FULL MATRIX (.14) # OF MESSAGES RECEIVED CHANGE IN ABILITY TO OBTAIN INFORMATION .37 .42 (.18) CHANGE IN ABILITY TO DISTRIBUTE INFORMATION .86 .82 GOODNESS OF FIT MEASURES FOR ENTIRE MODEL Chi square = .70 (g = .40) ju _ go degrees of freedom = 1 adjusted goodness of fit = .96 coefficient of determination = .14 hypothesized paths control paths (R-2’s for endogneous variables are in parantheses) TABLE 14 FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 9 Phi Matrix - none - : . -..---------— ...........—-------- —---- ..... . . --- : : --- : -------------- _ --„ TOTAL EFFECTS Total effects of X on Y # mess, received obtain info, distribute info. .37 .16 - */7 | Total effects of Y on Y obtain distribute info. info. obtain info. 0 0 distribute info. .42 0 -A' 7 • •• • ■ 102 103 FIGURE 10: VM USE & COMM. PERFORMANCE - LOW ANALYZABLE ENVIRONMENTS # OF MESSAGES RECEIVED (.17) .42 CHANGE IN ABILITY TO OBTAIN INFORMATION .69 .35 (.36) CHANGE IN ABILITY TO DISTRIBUTE INFORMATION LENGTH OF MESSAGES SENT .39 GOODNESS OF FIT MEASURES FOR ENTIRE MODEL -------------------- hypothesized paths Chi square = 1.37 (e = .51) ------------------- control paths degrees of freedom = 2 ki _ oo adjusted goodness of fit = .91 hi - coefficient of determination = .32 (R-2’s for endogneous variables are in parantheses) { TABLE 15 i FULL EFFECTS AND PHI IDENTIFICATION FOR FIGURE 10 Phi Matrix send length # mess, received send length # mess. recv. 1.00 .69 1.00 - . l -/ ?.?■; . . • . I TOTAL EFFECTS Total effects of X on Y send length # mess, received obtain info, distribute info, 0 .39 .42 .15 Total effects of Y on Y obtain info. distribute info, obtain info, distribute info. 0 .35 0 0 104 information). The total effect of messaging on change in ability to distribute infor- : mation is .54 and the total effect of messaging on change in ability to obtain information is .42. 4.3.9.3 - High Analyzable Environments Figure 11 depicts the estimated model for active voice messaging behaviors in high analyzable environments. No measure of messaging behaviors signifi cantly affected either measure of communication performance. Ability to obtain information was a marginally significant predictor of ability to distribute information (beta .25, g < .05). Clearly, none of the variance in ability to obtain information was explained, and only 6 percent of the variance in change in ability to distribute information. 4.3.10 - Summary It is not necessary to compare Z' statistics across high and low analyzable task environments to test Hypothesis 3,2. The low analyzable communication performance model explained 27 percent of change in communication perform ance over time, while voice messaging failed to reach significance in the high analyzable model. The hypothesis of stronger effects of active voice messaging , upon communication performance in low analyzable environments (Hypothesis i ! 3.2) received very strong support. i I j 105 106 FIGURE 11: VM USE & COMM. PERFORMANCE - HIGH ANALYZABLE ENVIRONMENTS .25 (.06) CHANGE IN ABILITY TO OBTAIN INFORMATION CHANGE IN ABILITY TO DISTRIBUTE INFORMATION .94 N = 36 hypothesized paths I i control paths j GOODNESS OF FIT MEASURES FOR ENTIRE MODEL (R-2’s for endogenous vars. are in parentheses) (NONE AVAILABLE) I CHAPTER 5: DISCUSSION AND CONCLUSIONS i I i This chapter is divided into five sections: 1) a restatement of the disserta tion’s purpose, 2) a summary of the outcomes of the hypotheses tested and their implications, 3) an empirical comparison of the model tested with a conven tional model of the determinants and outcomes of voice messaging use, 4) spe cific strengths and weaknesses of the approaches used in this study, and 5) implications of this model for organizational communication research. 5.1 - Purpose of the Dissertation The central motivation of this study was to explicate and test a comprehen sive mode! of media usage and outcomes from the perspective of organizational differentiation. As described in the work of several contingency theorists re viewed, an organization's structure is a product of the forces of differentiation and integration (Lawrence and Lorsch, 1967). Differentiation enables organizations to accomplish tasks too large or com plex for a single organizational work unit. It provides a means for an organization to assign portions of tasks to work units to be addressed by specific organiza tional members. Differentiation also allows personnel to be organized around ; projects or to work collaboratively on various portions of organizational tasks. |This “ specialization” creates work units with specific task environments which are i 'reflected in differing communication needs. i As described in the section on media usage and attributes, media possess j differing capabilities to transmit information (both quantity and quality of informa tio n ), Additionally, newer forms of organizational media often are com puter-me- ; 107 idiated. It was argued that computer-mediation offers capabilities with potential i jto enhance collaborative and less analyzable communication tasks. It was the ^ specific effort of this research to examine voice messaging as the state of the i jart in computer-mediated organizational media. ; The media usage model proposed in this dissertation argues that task envi- ; ronments exist at the work group level. Efforts to understand the effects of task environment upon media usage should be conceptualized and operationalized at i : the work group level. The model also argues that, in general, network analysis approaches are well I 'suited to the task of detecting and measuring organizational work groups. It i 'further argues that the structural equivalence concept of a jointly occupied net work position can both conceptualize and operationalize at the level of the work j group task environment. By using structural equivalence approaches to operationalize positions as I j ; manifestations of horizontal organizational differentiation, this dissertation tested specific hypotheses relating differentiation, task environment, media attributes, :and communication performance. 5.2 Implications of Hypotheses Tests i i i t 5.2.1 - Position Effect i i Hypothesis 1.1: Structurally equivalent actors in a system will manifest similar | voice messaging behaviors (position effect). Hypothesis 1.1 is the conceptual cornerstone for this dissertation. This study “ mapped" the formal organizational chart and created structurally equivalent po- 1 108 'sitions based upon series of superior-subordinate linkages. It was not known if actors occupying a position actually communicated (were linked in a network analysis relational sense), In fact, some of the positions were occupied by per sons in different cities. They were, however, structurally equivalent in their formal linkages to others in the system. The position, then, represents a specific place in the organization’s system of vertical and horizontal differentiation. The actors jointly occupy a position based upon organizational delegation of authority and responsibility. Hypothesis 1.1 was supported strongly in a variety of tests. In a series of regression equations, position universally was a significant predictor of both pas sive and active messaging (Tables 7 A and 7B). Across six regression models examining numbers and lengths of messages sent and received, position ex plained an average of 33 percent of the variance in messaging behaviors. In two LISREL models (one with communication performance measures and one with out— see Figures 3 and 5) an average of 66 percent of the variance in voice massaging behaviors was explained. The general implication of these tests is that task environment, as operationalized by structural equivalence, is a powerful determinant of voice mes- Isaging behaviors. Efforts to understand media usage in organizations will be handicapped unless considerations are made for structural measures of task environment. The power and parsimony of a structural equivalence approach is | considered preferable to models mixing levels of analysis or those dependent ; solely upon an individual-level of analysis. 5 .2.2 - Positional Adjacencies i 'Hypothesis 1.2: Due to positional adjacencies created by vertical differentiation, superiors’ voice messaging behaviors influence subordinate positions’ voice mes saging behaviors. Though it doesn't directly examine structural equivalence, Hypothesis 1.2 is , derived from Hypothesis 1.1. The basic notion is that positions are linked through systems of vertical adjacencies. Each position reports to one actor in the next vertical position. Information flows are determined by network position, and con joint use of media is necessary for communication to occur. As a superior in a superior-subordinate link possesses discretionary authority, a superior’s media use patterns should influence a subordinate’s media use patterns. Hypothesis 1.2 was tested in a variety of regression and structural equation models. Unlike Hypothesis 1.1, vertical adjacency was not a universally signifi cant predictor of messaging behaviors. In the regression models (Tables 7A and 7B), vertical adjacency was a moderate predictor of all messaging behaviors other than the number and length of messages sent. A potential problem with the regression models of vertical adjacency became .apparent in the LISREL models, however. The original model specification re gressed subordinate sending behaviors on superior sending behavior, and subor dinate receiving behaviors on superior receiving behavior. Modification indices generated by the LISREL models indicated that, in fact, supervisor sending be- ! [havior determined subordinate receiving behaviors (see Figures 3 and 5), and i |that supervisor receiving behavior was associated with subordinate sending be- ! haviors. Clearly, these associations reflect vertical, sequential communication across 'the adjacency link between superiors and subordinates, The association between I 1 110 i superior sending behavior and subordinate receiving behavior became very clear in the structural equation models for highly analyzable tasks (Figure 7). Superior sending was a significant influence on the number of messages received. For positions characterized by analyzable tasks, the number of messages one’s su perior sends is the best determinant of the number of messages the position receives. Moreover, it is the single best predictor of high analyzable position messaging behaviors overall (see Figure 7). The high analyzable task positions had the lowest explained variance for send in g messages (36 percent) and the weakest link between their own sending behaviors and receiving behaviors. If the task is analyzable, the influence of the ; supervisor is greater than if the task is less analyzable. These associations could ;be a reflection of less collaboration and less messaging within the highly analyz- iable positions and of greater vertical communication down the hierarchy, This , point will be returned to in the discussion of hypothesized relationships based iupon the analyzability of the task environment. 5.2.3 - Task Environment and Active Voice Messaging Use I Hypothesis 2.1: To the degree that task environment is less routine, it will be characterized by a greater reliance upon active voice messaging. [Hypothesis 2.2: To the degree that task environment is less analyzable, it will be j characterized by a greater reliance upon active voice messaging. * I i These general hypotheses are based on an assumption that task characteris tic s can be a determinant of media usage. The social presence and information j richness models specify that optimal performance depends upon an appropriate I I “ fit ” between task demands and media attributes. Assuming tasks are different!- j jated and collaborative in nature, as the number of exceptions to routine in- | creases, and as task analyzability decreases, effective organizational perform ance dictates that increased coordination of workflows is necessary. ' It is argued that use of active voice messaging features can enhance one’s capacity to coordinate workflows and that the active features will be used more i 'heavily in less analyzable and less routine environments, i Both of these hypotheses were supported (see Table 6). Low analyzable and I | nonroutine positions reported using the active messaging features 33 percent of ! the time as compared to routine and analyzable positions which used it about 22 percent of the time. Additionally, individuals in nonroutine and low analyzable environments univer sally sent, stored, recorded, and received more and longer messages (Table 6). All of these behavioral measures were significantly different across the high/low ! routinism dimension, and number of messages stored, message length, and num ber of messages received were significantly different across the high/low analyzability dimension. These tests tend to support the notion of a "fit” between task environment and media characteristics (e.g., social presence and information richness). , Since all persons in this study had access to and used the active features, these findings indicate that persons in positions tended to seek out and use the active messaging features for tasks which were less analyzable or less routine. In addi tion to lending support to the notion of a task characteristic/media attribute fit, theses findings also indicates that positions collectively recognize and use the capabilities of media in regard to their tasks. i ! i 112 5.2.4 - Task Environment and Position Effect .Hypothesis 2.3: To the degree that task environment is less routine, horizontal ! differentiation (position effect) will be a stronger determinant of using active voice [messaging features which facilitate coordination of workflows. ■Hypothesis 2.4: To the degree that task environment is less analyzable, horizon tal differentiation (position effect) will be a stronger determinant of using active voice messaging features which facilitate coordination of workflows. These two hypotheses basically combine Hypothesis 1.1 with 2.1 and 2.2, 'As tasks become less analyzable or less routine, the effect of position should 'become a more efficient predictor of messaging behaviors. This is to say that routine tasks or analyzable tasks do not prohibit the use of active messaging .features, they simply do not require it as much. But as tasks become less ■ analyzable or less routine, greater communication performance is required and less discretion in media choice would be permissible. Hypothesis 2,4 was strongly supported in regression-based tests for sending messages (average R-square for low analyzable environments = .53) and re jected for messages length, and receiving messages (Tables 7A and 7C). Simi larly, Hypothesis 2.3 received strong support for sending message (average iR-square for nonroutine environments = .41), but was rejected for message r i j length and receiving messages. LISREL models also were used to test the hypotheses in a more comprehen- i ;Sive and comprehensible manner (Figures 6 and 7, and Table 13). When mod- i | eled in a system of covariances, the average total effect of position on active messaging was .76 in low analyzable positions, and .37 in high analyzable envi- i ronments (Table 13). Also, every path coefficient for low analyzable positions I | was significantly greater than those for high analyzable positions (Table 13). 113 I Hypothesis 2.4 received strong support from the structural equation models. , Hypothesis 2.3 was not tested with LISREL for reasons discussed in the Methods . chapter. The implications of these findings are that, although position is a generally strong determinant of active messaging behaviors, it is a more powerful determi nant in less analyzable or less routine environments where it is hypothesized to offer greater potential benefit. It also indicates that task environment is not a universal determinant of media usage. Position had no direct effect on receiving messages in high analyzable environments. For task environment to act as a determinant of media usage, task characteristics should require or potentially benefit from particular attributes of the media. 5.2.5 - Communication Performance Hypothesis 3.1: Passive use of voice messaging will not enhance communication performance as much as active use of voice messaging. A distinction was made between active and passive use of voice messaging. Passive use simply entails use of the system as an answering machine while active use involves asynchronous “ sending,” storage and manipulation of voice messages (see section 2.5 on computer-mediated communication and voice ;messaging). Hypothesis 3.1 posits that greater communication performance en hancement is obtainable through active messaging. 1 Although passive use of the system was much greater than active use (e.g., 1 ' users averaged about ten passive messages per week and less than two active messages), the effect of passive messaging failed to reach statistical signifi- : cance in ten models of change in communication performance regressed on , I passive messaging behaviors (Table 8B). Passive messaging had no significant i 1 114 'effect on the ability to obtain or distribute information. An "answering machine” , .may be a greater convenience to the caller than the recipient of the message. Active messaging behavior, however, explained 18 percent of the change in ! ability to obtain information for all environments, and 23 and 24 percent for low analyzable and nonroutine positions respectively (Table 8A). It also explained 26 percent of the change in ability to distribute information in low analyzable environ m ents. Hypothesis 3.1 was supported, i | Though Hypothesis 3,1 is rather general, it has particular implications for i communication researchers. The measurement of organizational media requires specific conceptualizations and operationalizations. A generic system usage vari able probably would have obscured the findings of this research. Media, espe- ' ciaily computer-mediated media, can be used in a variety of ways. As was ■ made clear in these findings, not all uses possess the same capacities for com munication performance. Building upon several of the preceding hypotheses, it was hypothesized that use of active voice messaging features in less analyzable or less routine task ■ environments would result in greater performance benefits than would use of the same features in highly analyzable or more routine environments. This argument I is rooted in the idea of a “ fit” between task characteristics and media attrib- j utes. I : Hypothesis 3.2: Active use of voice messaging in less analyzable task environ- j ments will result in greater increases in communication performance than will use ; | of active voice messaging features in more analyzable task environments. j Hypothesis 3.3: Active use of voice messaging in less routine task environments will result in greater increases in communication performance than will use of active voice messaging features in more routine task environments. i | : 115 The average regression-based R-square for position effect on change in ; {ability to obtain and distribute information in low analyzable positions was about !.25. Position effect was not a significant determinant of change in ability to I {obtain and distribute information in high analyzable environments, Hypothesis 3.2 :was supported, ! The regression-based position effect for obtaining information in nonroutine ' | environments was .24. Position effect was not a significant determinant of dis- i I ! tributing or obtaining information in routine environments, nor was it a significant i determinant of distributing information in nonroutine environments. Hypothesis 3.3 was supported only for obtaining information. Further testing was conducted using LISREL (Figures 10 and 11). The low | analyzable structural equation model explained 17 percent of the change in ability . ,to obtain information and 36 percent of the change in ability to distribute informa- i Tion. Active voice messaging failed to reach significance in the high analyzable I i model. Hypothesis 3.2 again received strong support. 1 An implication of these findings is that voice messaging, as a form of com puter-mediated media, can significantly improve communication performance in dess analyzable environments. With this fact in mind, implementation of voice messaging systems is most appropriate for tasks involving collaborative and less analyzable task requirements (if active use is encouraged and supported). i Another implication is that the unidimensional information richness scale (Daft j i ■ and Lengel, 1984) is too limited to offer much insight into computer-mediated ! i | communication (i.e., by the parameters of the information richness continuum, j voice messaging offers relatively low information richness and should be inappro- i priate for less analyzable tasks). While voice messaging ranks low on informa tion richness, the present research found it to be most beneficial for communica- ; 116 Ttion in less analyzable task environments (analyzability parallels Daft and Lengel’s concept of equivocality). Researchers of computer-mediated communication systems need to create -a more comprehensive classification scheme incorporating the abilities of com puter-enhanced and computer-mediated systems. The single dimension of in formation is inadequate. ^ These associations also are indicative of the progress that com puter-m edi ated systems have made in a few years. While computer-mediated systems (generally text-based) are considered appropriate primarily for simple informa tion conveyance and tasks not involving high personal involvement (Rice, 1980), voice messaging was found to be used more heavily and offer greater communi- : cation performance advantages in less analyzable environments. Clearly voice messaging offers a much broader bandwidth than text-based systems, Additionally, conventional text-based, computer-mediated systems 'have, in a sense, adapted users to the system (i.e., users sit at conventional computer terminals and use conventional computer-oriented commands and syntax). Voice messaging has adapted the computer to the user through a device familiar to all - the telephone. 5.2.6 - Additional Hypotheses and Discussion of the Model i ; Hypothesis 4.1: A nonrecursive relationship exists between sending and receiv- 1 ing voice messages (sending and receiving messages creates a mutually causal (relationship). 'Hypothesis 4.2: An increased ability to obtain information will result in an in- [ creased ability to distribute information. I These two hypotheses were not of critical theoretical interest to either the j structural equivalence or contingency-oriented focus of this dissertation. They i 117 were, however, suggested by and necessary for the model proposed in Figure 2, |Hypothesis 4.2 was supported in all tests. Hypothesis 4,1, however, was not, and i ■ its rejection leads to several interesting observations. : As communication generally involves feedback, sending messages was pos ited to cause messages to be returned. Similarly, the receipt of a message was posited to cause a message to be sent. The structural equation term for this association is nonrecursiveness (mutual causality). This link, however, was proven to be directional from sending to receiving (Figure 3). This directionality has several potential implications both conceptually and em pirically. Conceptually, the direction of this link indicates that messaging gener ally is initiative oriented— sending messages is a powerful determinant of receiv- i ing messages. This factor could indicate that sending often is done with a re- Iquest for information (e.g., a superior requesting information from a subordinate <or co-workers requesting information from one another). This direction of this link also could help to explain the much stronger link 'between receiving messages and increased ability to obtain information, than the link between sending messages and increased ability to distribute information (see note 4). Additionally, regardless of task environment, users of the active messaging features reported a significant increase in ability to obtain information , (see Table 4). ■ Based upon the preceding observations, a somewhat subjective scenario fol- . lows based upon the assumption that people generally are more concerned with j obtaining information than with distributing information (i.e., the ability to obtain information potentially possesses more direct relevance for one’s own job per- ; form ance). 118 I Voice messaging is used asynchronously to request information from others i within the position and across the vertical adjacency. The information returned i may be in the form of a voice message, or in other more conventional forms ;(e.g., synchronous phone call, memo, etc.). Responses to a voice message in i 'means other than a voice message helps to explain the less than perfect correla tion between sending and receiving messages. Combined with an ability to broadcast messages, it also helps to explain why approximately 190 more peo- !pie received messages than sent them, As the model in Figures 2 and 3 depicts, sending messages causes mes sages to be returned. The returned message increases the ability to obtain information which, in turn, leads ultimately to an increased ability to distribute information. The system is used as a sophisticated information gathering tool. One additional element in this scenario concerns users in low analyzable posi tions. Only in low analyzable positions did a direct link exist between sending messages and increased ability to distribute information. It is suggested here that, due to the collaborative tasks in low analyzable task positions, an ability to .distribute information is highly salient. While persons in more analyzable positions may be concerned primarily with obtaining information, low analyzable and col laborative tasks require that information is obtained and distributed rapidly and i j effectively. In view of the relatively high explained variance of change in commu nication performance for the low analyzable positions (27 percent), use of active . voice messaging features appears to facilitate the intensive information sharing required by collaborative work. An empirical implication of this directional link lies primarily in levels of ex- | plained variance. Because sending and receiving are quite highly correlated, number of messages usually manifested higher explained variance than sending messages (i.e., receiving universally was regressed on sending). Again, how- ■ lever, the single exception is that of the low analyzable positions (Figure 6). Here i the position effect was so powerful, that slightly more of the variance in sending . i !was explained than receiving (71 and 68 percent, respectively). The relationship ' i ■between position effect, sending messages, and increased ability to distribute | information is uniquely prominent in the low analyzable task environments. ; The LISREL model for low analyzable environments best represents the theo- jretical model depicted in Figure 2. Position strongly determines sending and i ■ receiving behaviors. The messaging behaviors then directly determine both in creased abilities to obtain and distribute information. The model fits. t 5.3 - Empirical Comparison of the Proposed Model 1 With a Conventional, Individual-level Model i i j Although the structural equivalence model of media usage presented in this ■ , dissertation has proven to be rather powerful and parsimonious, it would be infor- I mative to know how a more conventional, individual-level model of media usage ; i ; ; would fit the same data. This type of comparison is a clear way to make a direct ; empirical evaluation of the proposed model. Up to this point, all the analyses in this dissertation have been conducted with ! the work group as the unit of analysis. All cases were weighted by the inverse of j the group size so that each work group (position) weighed evenly in the analyses. From a contingency perspective, it rs argued that organizations are best under- I stood and measured as systems of differentiated units. An organization is far 120 i more than a simple aggregation of ail of its members. Individual-level analyses ! ibasically strip from the organization what makes it “ organized.” ! 1 A more conventional approach to this study, however, would be to maintain an individual-level of analysis (each observation represents its own values with a weight of one) and enter the task environment variables (analyzability and num- , bers of exceptions) as exogenous variables. Figure 12 presents such a theoreti cal model for the data analyzed in this research, The model contains the same ■i control variables as the model in Figure 2 and has the additional variables of task i analyzability and numbers of exceptions. Analyzability and number of exceptions are based on same scales which were, dichotomized for the models portraying effects of task environment (see Table 3). This model, of course, does not Rave the structural equivalence variables (horizontal and vertical differentiation), b- The same approaches were used in testing this mode! as all LISREL models ^presented previously (i.e., starting with a full model and sequentially fixing links until only significant paths remain), Figure 13 depicts the resultant model. '4 Task analyzability and numbers of exceptions were not significantly associ ated with either usage or communication performance, Additionally, voice mes saging behaviors were not significant determinants of either communication per formance variable. Organizational level explained 10 percent of the variance in sending messages, and only sending messages was a significant determinant of number of messages received (beta ,66, fi < .001). Innovativeness explained 17 percent of the variance in change in ability to obtain information which, in turn, explained 15 percent of the variance in change in ability to distribute Information, Although the model does fit the data (adjusted goodness of fit = .88) , it; provides a very weak model of voice messaging usage and communication performance (coefficient of determination = .25). FIGURE 12: INDIVIDUAL-LEVEL ANALYSIS OF TASK ENV., VM USE & COMM. PERF. ORGANIZATIONAL LEVEL TENURE ANALYZABILITY # OF EXCEPTIONS EDUCATION INNOVATIVENESS SUPERVISOR INNOVATIVENESS # OF MESSAGES RECEIVED n CHANGE IN ABILITY TO OBTAIN INFORMATION # OF MESSAGES SENT CHANGE IN ABILITY TO DISTRIBUTE INFORMATION hypothesized paths control paths 123 FIGURE 13; INDIVIDUAL-LEVEL ANALYSIS OF TASK ENVIRONMENT, VM USE & COMM. PERFORMANCE - FULL MATRIX MODEL .41 (.17) (-44) .56 .39 .66 .83 (.15) (.10) CHANGE IN ABILITY TO DISTRIBUTE INFORMATION -.32 .90 CHANGE IN ABILITY # OF MESSAGES SENT ORGANIZATIONAL LEVEL INNOVATIVENESS # OF MESSAGES RECEIVED CHANGE IN ABILITY TO OBTAIN INFORMATION GOODNESS OF FIT FOR ENTIRE MODEL Chi square = 9.74 (p = .46) degrees of freedom = 10 adjusted goodness of fit = .88 coeff. of determination = .25 N = 55 hypothesized paths control paths (R-2’s for endogneous vars. are in parentheses) Clearly, it is not only a much weaker model of the same data, it arrives at altogether different conclusions from the model in Figure 3. Task environment (has no significant effect on active use of voice messaging and use of voice i messaging has no significant effect on changes in ability to obtain or distribute information. i j Aside from the absence of the relatively powerful horizontal and vertical differ- ; lentiation variables in Figure 3, the relative weakness of this model can be seen in I |two distinct areas: 1) task environment operationalized at the individual-level of i I analysis is a non-significant predictor of messaging behaviors, and 2) effects of messaging behaviors on communication performance are not detectable using individual level measures. Again, it is argued that these low associations are an artifact of measurement 'approaches which simply remove any form of “ organization” from the organiza tion- Individual-level measurement and modeling result in aggregate analysis, not l 'organizational analysis. It is noted, however, that Figure 13 should represent a good, conventional .attempt to model messaging and communication performance. It uses LISREL estimation and the same over-tim e behavioral and performance change meas- i ures as the model in Figure 2. These comparisons provide powerful support for the model presented in Fig ures 2 and 3. Using identical measures and estimation procedures, the compari- ,son model in Figures 12 and 13 proved to be far weaker and, perhaps, even , misleading. I 5 .4 S p e c ific S tren g th s and W e a k n e s s e s o f th e P rop osed M o d e l { I This study has several important potential weaknesses. One clear weakness 'is the time period over which the data were collected. The average user was on the system for only 12,5 weeks. A longer period of usage may have uncovered ; stronger associations between messaging and changes in communication per form ance. Also, a critical mass of users is necessary for optimal effectiveness ; (Markus, 1987; Rice et al., 1988), It is not certain whether a critical mass was achieved over the duration of this study. • Although an actual network analysis of the informal structure of the organiza tion was planned, it was not possible to collect the necessary data. An ideal : model of organizational structure, media usage, and performance would have 'included measures of both formal and informal organizational structure. A potential problem with this study, as well as with most network analy- !sis-based studies, is the fact that observations are not independent for the 'measures of position. The value of the position is generated by extracting the value for a single observation from that of the sum of the rest of the observations • in the position. This procedure creates a “ unique” value for each observation. The problem of non-independent observations occurs when, for example, each person in a ten person position has a value based upon eight of the same ■ observations. The values will not be "identical” as a different observation is I extracted each time. But in the case of a large position, the extraction of one i I value may make minimal differences in the values generated for each observa- | tion in the position. Clearly, the larger the position (the greater the number of ‘ observations linked to a single superior), the greater the threat of inflated asso- ; ciations stemming from non-independent observations. : 125 The other extreme, of course, is a position with only two observations. These observations would be truly independent as no observations are shared. In order to test for the influence of non-independent observations, a series of regressions was conducted based upon the size of the position (an unweighted regression of an observation’s values for the number of messages sent and received regressed on the position value for each variable). In other words, one regression was run for all positions with two observations, another regression for ; all positions with three observations, etc. if there were an effect of non-independent observations, explained variances should become progressively higher as the size of the position increases. Table ■ 16 depicts the effect of position size on explained variances. Part A indicates The size of the position, the number of observations at each specific position size (N), the R-square, and the significance level of the R-square. The values from the table in Part A were used to generate a correlation matrix of the R-squares and sizes of positions. The N was used to count each position size/R-square once for each occurrence. As can be seen in Part B, position size and R-square are significantly correlated, but the correlations are negative. Smaller positions are better predictors of messaging behaviors than the larger positions. Instead of the potential non-independence of the observa tions artificially inflating the explained variance, the smaller, relatively more cohe sive positions create stronger measures of task environment. This association is * j conceptually appealing and the statistical threat of non-independent observations t was not manifested in the procedures used in this research. i 126 TABLE 16 TEST OF THE EFFECT OF NON-INDEPENDENT OBSERVATIONS ON MEASURES OF POSITION Part A - sending messages - - receiving messages - position size N R2 2 46 ,43* * * 3 55 * * * 4 51 ^52* * * 5 57 .77 6 44 .07 7 17 .69 * * * 8 24 .11 9 8 36 11 21 .01 15 7 .37 position size N R2 2 28 .55*** 3 20 .10 4 29 29 * * * 5 19 !21 6 29 .12* 7 10 .10 8 9 .20 11 2 .31 15 3 .22 Part B Correlations Between Size of Position and Explained Variance variable N corr. sending messages 140 “ 49 * * * receiving messages 331 -.4 3 * * * i A conceptually unimportant, but logistically significant consideration of the I methods used in this research is the time, organizational cooperation, informa- i ■ tion, computer resources, and computing skills required to assemble and analyze ! ; the data. Aside from the normal collection of questionnaires at two points in time, organizational members from the various sites were required to capture the be- , havioral messaging data on a weekly basis. These collections required both cooperation and computing skills on site for the five month duration of the study. As these data originally were “ captured” and printed, a massive amount of data had to keyed manually (approximately 800,000 characters— see Appendix 4 for ■ an example of the computer-generated data). The data-entry process required considerable time and financial support. Approximately 20 computer programs, large and small, had to be written to I read in and process the data into formats usable by standard statistical pro- i grams. The control file which simply reads in the raw data contains over 2,000 Mines of code (see note 5). I Additionally, weeks of manual, detail-oriented labor were required to convert ! organizational charts into data usable by the structural equivalence program de picted in Appendix 1. Although none of the above has any conceptual impor- i I tance, the time, organizational cooperation, and resources required to conduct this type of research may make it unattractive or impractical for some research ers to conduct. Clearly, these factors could limit the applicability of this form of research. Strengths of this research lie in several areas. Data collection involved multi- I pie sources of data including archival, self-report, and behavioral (Cook and * Campbell, 1983). Usage measures were taken through self-report, and more 128 ; importantly, through observed behavioral measurements in the form of com- i I ! puter-monitored data (Rice and Borgman, 1983). ! A considerable body of research of has accumulated around the apparent inability of respondents to accurately report their communication behaviors (Ber nard and Killworth, 1977; Bernard, Killworth, and Sailer, 1980, 1981, 1982; Bernard, Killworth, Kronenfeld, and Sailer, 1984; Romney and Weller, 1984). Behavioral data avoids measurement error associated with inaccuracies of self-reports. The use of behavioral data in this research, along with self-report ’ and archival data, also enables a degree of triangulation of measurement, Additionally, the research used multiple methods to analyze and model the data (Cook and Campbell, 1983). Use of multiple methods provides a check against method bias (observed associations are a partial product of the method used). The use of structural equation models and LISREL also allows for simulta neous evaluation of equations and a variety of test statistics for verifying the fit of the model to the data. • The combination of over-tim e measurement with structural equation ap proaches enables the construction and testing of valid causal models. I 5.5 Implications of the Present Research for Communications Applications ! i I l ‘ i | I The findings of the present research suggest that voice messaging does not fit "neatly" into conventional classifications of organizational media. It is rela tively low in information richness compared to fa ce -to -fa ce conversation (i.e., j asynchronous, no visual cues, etc.), but findings indicate it enhances communi- ! 129 cation performance most in less analyzable and less routine task environments. As conventional theory suggests that “ less rich” media are inappropriate for equivocal (less analyzable) tasks (Daft and Lengel, 1984, 1986), this rather ■unique confluence of voice and computer-mediation suggests several important considerations for voice messaging applications. Voice Messaging and Collaboration Voice messaging enhances abilities to obtain and distribute information in less analyzable and less routine environments (the association was not significant in high analyzable task environments). This ability is particularly important for or ganizational work units conducting collaborative work. Collaboration basically entails “ breaking-down” complex tasks into smaller tasks to be addressed by persons with specific expertise, Once the individual "pieces” are completed, the task must be “ re-assem bled.” Collaboration requires substantial flows of information between participants, and this communication can be "costly" in terms of time and resources devoted to it. The asynchroneity of voice messaging makes it less disruptive or "intru sive” for teams of workers in collaborative projects to exchange information. It , is argued here that voice messaging is a potential substitute for a portion of the i imore costly fa ce -to -fa ce communication characteristic of collaborative work. Additionally, voice messaging removes constraints related to time zones, differ- ; ences in working hours, and geographical distances, and could enable collabora tio n among participants which otherwise would not be practical or possible. The generally universal access (telephone handset) and low technical profi- ! ciencv demands (no computing skills, simply use telephone touch-tone pad) of voice messaging make the system very accessable and convenient to use in i ! 130 | comparison to conventional computer-mediated systems. Additionally, less po- | tential exists for the system to be rejected by managers and professionals who are accustomed to using the telephone to conduct "business” (i.e., compared ;to typing on computer keyboards). Typing on keyboards also may be associ ated stereotypically with clerical work, and may make text-based systems less attractive for persons higher in the organizational structure (i.e., typically persons involved with less analyzable and less routine work). It is argued here that the , interface to the system is appropriate for the current working styles of profes sionals involved in collaborative and less analyzable work (i.e., the “ target group” for whom the present research indicates will benefit the most from the system). The processing features of voice messaging (e.g., future delivery of mes- i sages, time shifting, broadcasting single messages to multiple others, automated storage and retrieval of messages, forwarding pertinent messages to others in a work group, etc.) also are seen as especially relevant to collaborative work. These features potentially can improve both the efficiency and the effectiveness of collaborative work by facilitating coordination of group communication flows. A key to the application of voice messaging in less analyzable task environ- ; ments may be, however, that voice is used rather than text. Persons generally » | are able to speak more quickly and comprehensively than they can “ type” an ' | equivalent text-based message. Additionally, voice supplies many subtle cues r : important to human communication and understanding (Short et al., 1976). It is ■ I argued here that these cues are particularly important for less analyzable or less routine collaborative work where finer lines of distinction, more complex, or more equivocal situations are common (Daft and Lengel, 1984, 1986; Short et al., 1976), I 131 , , Passive Versus Active Messaging The present findings also suggest, however, that all voice messaging behav iors “ are not created equal.” Although passive use of the system (simply used as a powerful answering machine) was five times more frequent than active use of the system (actually sending, receiving, and processing asynchronous mes sages), passive use of the system did not improve users’ communication per- : formance. Passive use does little to modify or enhance extant communication behaviors. Active use, on the other hand, involves adopting new approaches to communication. Although active messaging can improve communication per formance, it will require the user to modify his or her behaviors— a modification which may not occur readily without explicit support and motivation from superi ors or colleagues. Clearly, a critical mass of users also is necessary before active use of voice messaging can become a viable communication alternative (Markus, 1987). The distinction between passive and active use of voice mes saging may be critical to the successful implementation of a voice messaging system. Implementation Considerations The current findings suggest that voice messaging possesses the potential to significantly enhance organizational communication, in order to maximize com munication performance benefits Of such a system, however, two implementation factors are suggested: ® Active voice messaging behaviors must be encouraged and sup ported, Although use of active voice messaging resulted in significant increases in abilities to obtain and distribute information, use of voice messaging as an answering machine resulted in no significant gains in communication performance. : • Voice messaging system implementation should be directed at col laborative task environments, or environments characterized by less analyzable or less routine tasks. Although voice messaging resulted in significant increases in communication performance for less analyzable and less routine tasks, the associations were either much weaker or not significant for analyzabie and routine tasks. Attempts to implement voice messaging in an organization without specific attention to the target user groups, or without attention to the types of voice messaging behaviors to support and encourage, will not reap the benefits of the . findings of the present research. 5.6 - Implications of the Present Research for Communication Research and Theory This research has united central components of contingency theory with structural equivalence to create a reasonably powerful model of organizational voice messaging usage and communication performance. Though the focus of ■ this research was voice messaging, the model proposed and approaches used should be applicable to the study of most organizational media. The use of or- i aanizational position as the unit of analysis is considered the critical conceptual and empirical component of this model. This study has demonstrated that organizational position is a significant deter- : minant of messaging behaviors and that the position effect is stronger in less i i analyzable environments where communication is most critical. It also has shown that low analyzable and nonroutine positions accrue significantly greater commu- i r nication benefits from active use of voice messaging than do highly analyzable ! and routine positions. 133 This research argues that individual-level analysis "rem oves” the organizing i I effect inherent in organizations, thus creating a vastly different and much weaker model of media usage and performance (Figure 13). It is suggested that, unlike most text-based computer-mediated systems, voice messaging is particularly well suited for collaborative and less analyzable tasks. Moreover, it was argued that the expanding range of computer-mediated ; organizational media requires a more comprehensive classification scheme than is afforded by the social presence or information richness continua. Future or ganizational communication theory and research should be directed toward the creation of a more powerful media classification scheme. As mentioned, the next logical step in the progression of the model of organ izational media proposed and tested in this dissertation is to combine measures of informal structure (based upon reported or observed patterns of communica tion) with the formal structural measures used in this research. In conclusion, it is argued that the model tested in this research contributes ' considerable conceptual and empirical knowledge to our understanding of organ izational media use as it relates to organizational structure. I i 134 NOTES Routinism is defined as the relative number of exceptions to routine procedures one encounters. Given an exception to routine, analyzability is the relative difficulty (analyzability) of the search neces sary to determine the appropriate response. These two constructs are drawn directly from the work of Charles Perrow (1972) and are operationalized through scales created by Withey, Daft and Cooper (1983). “ Fixing,” as used in LISREL model trimming, distinguishes a potential structural equation path with is not “ free.” Fixing a path prohibits LIS REL from estimating a direct path coefficient between two variables. As all gamma and beta path coefficients are directional, two paths must be fixed between each pair of variables to prevent a direct ef fect. Z' prime tests compare standardized coefficients using a Z table (i.e., a Z' statistic of greater than 1.96 is significant at p <.05). Sending and receiving messages are strongly, associated as are ob taining and distributing information (see Figure 3). The resultant direc tional indirect path from sending messages through receiving mes sages through obtaining information to distributing information (path 2,4) may have prevented a direct link from sending messages to dis tributing information from reaching statistical significance. Moreover, the existence of these two strong, paired, directional associations re sult in relatively higher explained variances for both messages received and ability to distribute information. Al! programs used in this research are available from the author. 135 I References t i i I Alien, T. (1977). Managing the Flow o f Technology. Cambridge, Mass: ! MIT Press. I ! Argyle, M. (1969). Social interaction. London: Methuen. < Bell, D. (1973). The Coming of the Post-Industrial Society. New York: Ba- i sic Books. | Bernard, R., and Killworth, P. (1977). Informant accuracy in social network data II, Human Communication Research, Vol. 4, No. 1, pp. 3-18. . Bernard, R,, Killworth, P., and Sailer, L. (1980). Informant accuracy in so cial network data IV: a comparison of clique-level structure in behav- : ioral and cognitive data. Social Networks, Vol. 2, pp. 191-218, ! Bernard, R., Killworth, P., and Sailer, L. (1981). A review of informant ac curacy in social network data. In Hummel, W, (Ed.) Modelle fur Ausbreitungsprozesse in Socialen Strukturen. Duisberg: Sozisleis- ; senschaftlichen Kooperative. i Bernard, R., Killworth, P., and Sailer, L. (1982). Informant accuracy in so- ; ciai network data V: an experimental attempt to predict actual commu nication from recall data. Social Science Research, Vol. 11, pp. : 30-60. I Bernard, R., Killworth, P., Kronenfeld, D., and Sailer, L. (1984). The prob- | lem of informant accuracy: the validity of retrospective data. The An- I nuai Review o f Anthropology, Vol. 13, pp. 495-517. i ; Beshicn, R., and Palmer, A. W. (1972), Studying managerial behavior, In- j ternational Studies o f Organization and Behavior, Vol. 2, No. 1.@@ Blau, P., and Meyer, M. (1971). Bureaucracy in Modern Society. New York: Random House. Blau, P., and Alba, J. (1982). Empowering nets of participation. Adminis trative Quarterly, Vol. 27, pp. 3 6 3 - 379. 136 1 Brass, D. (1981). Structural relationships, job characteristics, worker satis faction, and job performance. Administrative Science Quarterly, Vol. 26, pp. 331-348. . Burgleman, R., and Sayles, G. (1986). Inside Corporate Innovation: Strat egy, Structure and Managerial Skills. New York: Free Press. : Burns, T., and Stalker, G. (1961). The Management of Innovation. Lon don: Tavistock. Burt, R. (1976). Positions in networks. Social Forces, Vol. 55, pp. 93-122. Burt, R. (1977). Positions in multiple network systems. Part One: a general conception of stratification and prestige in a system of actors cast as a social typology. Social Forces, Vol. 57, pp. 106-131. Burt, R. (1980). Models of network structure. Annual Review of Sociology, Vol. 6, pp 79-141, Byrne, K. (1984). Impact of a Computer-based Message System on an Or ganization. M.S. Thesis. Ottowa, Canada: Carleton University. Coleman, J., Katz, E., and Menzel H. (1966). Medical Innovation: A Diffu sion Study. Indianapolis: Bobbs-Merrill. Conrath, D. (1978). Organizational communication behavior: description and prediction. In M. Elton, W, Lucas, and D. Conrath (Eds.) Evaluat ing New Telecommunication Services. New York: Plenun. ; Cook, T., and Campbell, D. (1983). Quasi-Experimentation: Design and Analysis Issues for Field Settings. Boston, MA: Houghton Mifflin. Daft, R,, and Lengel, R. (1984). Information richness: a new approach to manager information processing and organization design. In B, Staw and L. Cummings (eds.), Research in Organizational Behavior, Vol. 6, pp. 191-233. Daft, R., and Lengel, R. (1986). Organizational information requirements, media richness, and structural design. Management Science, 32, 5, 554-571. 137 Daft, R,, and Wiginton, J, (1979). Language and organization. Academy of Management Review, Vol. 4, No. 2, pp. 179-191. Davis, K. (1967). Human Relations at Work: The Dynamics of Organiza tional Behavior. New York: McGraw-Hill, Dewhirst, H. (1971). Influence of perceived information-sharing norms on communication channel utilization. Academy o f Management Journal, Vol. 4, pp. 305-315. Donaldson, L. (1976). Woodward, technology, organizational structure, and performance-a critique of universal generalization. Journal of Manage ment Studies, Vol. 13, pp. 255-273. Dormois, H., Fioux, F., and Gensollen, M. (1978). Evaluation of the future potential for various future communication modes. In M. Elton, W, Lucas, and D. Conrath (Eds.) Evaluating New Telecommunications Services. New York: Plenum. Downey, H,, and Slocum, J. (1975). Uncertainty: measures, research, and sources of variation. Academy of Management Journal. 18, pp. 562-577. Duncan, R. (1974). Characteristics of organizational environment and per ceived environmental uncertainty, Administrative Science Quarterly, Vol. 17, pp. 313-327. Fayol, H. (1937). The administrative theory in the State. In Gulick, L. and Urwick, L. (eds.) Papers on the Science o f Administration. New York: Institute of Public Administration, pp. 99-134. Fayol, H. (1949). General Industrial Management. London: Pittman, Freeman, L. (1980). Q-analysis and the structure of friendship networks. International Journal of M an-M achine Studies, Vol. 12, No. 3, pp. 381-387. Addison-Wesley Co. Fulk, J., Power, G., Schmitz, J., and Steinfield, C. (1986). A social infor mation processing model of media use in organizations. Paper pre sented at the annual International Communication Association Confer ence in Chicago. 138 j Galbraith, J. (1973). Designing Complex Organizations. Reading, MA: Ad- dison-Wesley. Galbraith, J. (1977). Organization Design. Reading, Mass.: Addison-Wes- ley Publishing Co. Galegher, J., Kraut, R., and Egido, C, (1987). Proceedings of the confer ence on Technology and Cooperative Work. Tucson: University of Ari zona. Gerwiri,. D. (1979). Comparative analysis of structure and technology: a critical appraisal. Academy of Management Review, Vol. 4, pp 41-51. Giuliano, V. (1982). The mechanization of office work, Scientific American, 149-164. Goddard, J. (1973). Office Linkages and Location. London: Pergamon Press. Hall, D., and Lawler, E. (1970). Job characteristics and pressures and or ganizational integration of professionals. Administrative Science Quar terly, No. 15, pp. 271-281. Hardy, A. (1982). The selection of channels when seeking information: cost/benefit vs. least-effort. Information Processing and Manage ment, Vol. 18, No. 6, pp. 289-293. Hiltz, S. (1983). Online Communities: a case study of the office of the fu ture. Norwood, NJ: Ablex. 1 Hiltz, S., and Turoff, M. (1978). Network Nation: Human Communication via Computer. Menlo Park, CA: Addison-Wesley. i Huber, G. (1982). Organizational information systems: determinants of their performance and behavior. Management Science, Vol. 28, No. 2, pp. 138-155. i Huber, G. (1984). The nature and design of post-industrial organizations. j Management Science, Vol. 30, pp. 928-951. ! 139 \ Hurt, H., Joseph, K., and Cook, C. (1977). Scales for the measurement of J innovativeness. Human Communication Research, 4, 1, 58-65. Innis, H. (1951). The Bias of Communication. Toronto: University of Toronto Press. Johansen, R., De Grasse, R., and Wilson, T. (1978). Group Comunication Through Computers (Volume 5): Effects on Working Patterns. Menlo Park, CA: Institute for the Future. Johansen, R., Valiee, J., and Spangler, K. (1979). Electronic Meetings: | Technical Alternatives and Social Choices. Reading, MA: Ad- [ dison-Wesley. Katz, P., and Tushman, M. (1979). Communication patterns, project per formance, and task characteristics: an empirical evaluation in an R&D setting. Organizational Behavior and Human Performance, 23, 139-162. | Kelley, G. (1955). The Psychology of Personal Constructs. New York: Norton. Keller, R. (1986). Predictors of the performance of project groups in R&D organizations. Academy o f Management Journal, 29, 715-726. ; Kerr, E., Hiltz, S. (1982). Computer-Mediated Communication Systems. New York: Academic Press. Khandwalla, P. (1977). The Design of Organizations. New York: Harcourt, J Brace, Jovanovich. i Killworth, P., and Bernard, R. (1976). Informant accuracy in social network data. Human Organization. Vol. 35, pp, 269-286. i Killworth, P., and Bernard, R. (1979). Informant accuracy in social network data III: a comparison of triadic structures in behavioral and cognitive data. Social Networks. Vol. 2, pp. 19-46. Knoke, D., and Kuklinski, J. (1982). Network Analysis. Beverly Hills: Sage Publications. 140 j Kroeber, D., and Watson, H. (1984). Computer-Based Information Sys- | terns. New York: Macmillan Publishing. ' Lawrence, P., and Lorsch, J. (1967). Organization and Environment: Man- , aging Differentiation and Integration. Boston: Graduate School of Business Administration, Harvard University. Leduc, N. (1979). Communicating through computers. Telecommunica tions Policy, September, pp, 235-244. Lorraine, F., and White, H. (1971). Structural equivalence in social net works. Journal of Mathematical Sociology, Vol. 1, pp: 49-80. Markus, L. (1987). Toward a critical mass theory of interactive media, uni versal access, interdependence, and diffusion. Communication Re search, Vol. 14(5), pp. 491-511. McLuhan, M. (1960). Explorations in Communications. Boston: Beacon Press. McLuhan, M. (1962). The Gutenberg Galaxy. Toronto: University of Toronto Press. 1 Miller, J., and Nichols, M. (1981). An update in a longitudinal study of the adoption of an electronic mail system: the effects of system use. In G. Reeves and J. Sweigart (Eds.) 13th Annual Proceedings of the American Institute for Decision Sciences, Miner, J. (1982). Theories of Organizational Structure and Process. New York: Dryden Press. I Mintzberg, H. (1970). Structured observation as a method to study mana gerial work. Journal of Management Studies, No. 2. i i Mintzberg, H. (1971). Managerial work: analysis from observation, Man agement Science, 18, No. 2. ; Mintzberg, H. (1973). The Nature o f Managerial Work. New York: Harper and Row. Palme, J. (1981). Experience with the use of COM computerized con ferencing system. FOA Rep, C10166E-M6(H9). Swedish National Defense Research Institute, Stockholm, Sweden. 141 Panko, R. and Panko, R. (1981). A survey of EMS at DARCOM. Computer Networks, Vol. 5, pp. 19-34. Perrow, C. (1972). Complex Organizations: A Critical Essay. Glenview, IL: Scott, Foresman. Picot, A., Klingensberg, H., and Kranzle, H, (1982). Office technology: a report on attitudes and channel selection from field studies in Ger many. In M. Burgoon (Ed.) Communication Yearbook 6. Beverly Hills, CA. Pool, I. (1977). Social Impact of the Telephone. London: MIT Press. Pool, I. (1983). Technologies of Freedom. London: Belknap Press. Randolph, W. (1978). Organization Technology and the media and purpose of organizational communications. Journal of Business Research, Vol. 8, pp. 237-259. Reid, A. (1977), Comparing telephone with fa ce -to -fa ce contact, in I. Pool’s (ed.) The Social Impact of the Telephone. London: MIT Press. Reimann, B.: and Inzerilli, G. (1979), A comparative analysis of empirical research on technology and structure. Journal o f Management, Vol. 5, 167-192. Rice, R. (1980). The impacts of computer-mediated organizational and in terpersonal communication. In M. Williams (Ed.) Annual Review o f In formation Science and Technology, Vol. 15. White Plains, N.Y.: Knowledge Industries Publications, pp. 221-2 49. Rice, R. (1987). Computer-mediated communication systems and or- ganizatonal innovation. Journal of Communication, Vol. 37, 4, pp. 65-94. Rice and Associates. (1984). The New Media: communication, research, and technology. Beverly Hills: Sage Publications. Rice, R., and Bair, J. (1984). New organizational media and productivity. In Rice and Associates The New Media: communication, research, and technology. Beverly Hills, CA: Sage. 142 [ Rice, R., and Borgman, C. (1983). The use of computer-monitored data in i information science and communication research. Journal of the j' American Society for information Science, Vol. 34, pp. 247-256. Rice, R., and Case, D. (1983). Electronic messaging systems in the univer sity: a description of use and utility. Journal o f Communication, Vol. 3, No. 1, pp. 131-152. Rice, R., Grant, A., Schmitz, J., and Torobin, J. (1988). Organizational in formation processing, critical mass, and social influence: a network approach for predicting use and outcomes of electronic messaging. Communication Research, (in press). . Rice, R., and Richards, W. (1985). An overview of network analysis meth- | ods and programs. In Dervin, B. and Voigt, M. (Eds.) Progress in Communication Sciences. Norwood, NJ: Ablex. Rice, R., and Shook, D. (1987). Understanding organizational media us- ! age: an information processing and media bandwidth approach, Pa- ! per presented at the annual Academy of Management Conference in New Orleans. f Rice, R., and Shook, D. (1988). Integrated office systems: impacts of ac- I cess and usage on organizational communication. ACM Transactions > on Office Information Systems, in press, Rogers, E. (1983). Diffusion of Innovation. 3rd Edition. New York: Free Press. ' Rogers, E,, and Rogers, R. (1976). Communication in Organizations. New j York: Free Press. Rogers, E., and Kincaid, D. (1981). Communication Networks: Toward a I New Paradigm for Research. London: Free Press. Romney, A. and Weller, S. (1984). Predicting informant accuracy from pat terns of recall among individuals. Social Networks. Rosenberg, V. (1967). Factors affecting industrial personnel for information gathering methods. Information Storage and Retrieval, Vol. 3, pp. 119-127. 143 Saris, W., and Stronkist, H. (1984). Causal Modelling in Nonexperimental Research. Amsterdam: Sociometric Research Foundation. ! Shook, D., and Rice. R. (1988) Task Analyzability and Routinism as Deter minants of Voice Messaging Behaviors: A Contingency Theory Ap- ; proach. Paper accepted for presentation at the annual International ' Communication Association Conference in New Orleans. Short, J., Williams, E., and Christie, B, (1976). The Social Psychology of j Telecommunications. London: John Wiley & Sons. i | Steinfield, C., and Fulk, J. (1986). Task demands and manager’s use of ! communication media: an information processing view. Paper pre- j sented to the Organizational Communication Division, Academy of j Management Conference. i Steinfield, C., Jin, B., and Ku, L. (1987). A preliminary test of a social infor mation processing model of media use in organizations. Paper pre sented at the annual Internation Communication Association Confer ence in New Orleans. ' Svenning, L., and Ruchinskas, J. (1984). Computer-mediated communica tion in an organizational setting: explaining task-related and socio-emotional uses. In Rice and Associates The New Media: Com munication, Research, and Technology. Beverly Hills: Sage Publica tions. ' Taylor, F. (1911). The Principles of Scientific Management. New York: Har per and Rowe. ; Thompson, J. (1967). Organizations in Action. New York: McGraw-Hill. i J Thompson, J. (1973). Society’s frontiers for organizing activities. Public < Administration Review, Vol. 33, pp. 327-335. Tushman, M. (1979). Work characteristics and subunit communication structure: a contingency analysis. Administrative Science Quarterly, Vol. 24, pp. 82-98. Tushman, M. and Nadler, D. (1978). An information processing approach to organizational design. Academy Management Review, Vol. 6, No. 3, pp. 613-624. 144 Uhlig, R. (1977). Human factors in computer messaging systems. Datama tion, Vol. 23, No. 5, pp. 120-126. Weber, M. (1946). Max Weber: Essays in Sociology. Translated and ed ited by Mills, C. New York: Oxford University Press. Weber, M, (1947). The Theory o f Social and Economic Organization. Translated and edited by Parsons, T., and Henderson, A, New York: Free Press. Weick, K. (1979). The Social Psychology of Organizing. New York: Ran dom House. White, H,, Boorman, A., and Brieger, R, (1976). Social structure form mul tiple networks 1. Block models of roles and positions. American Jour nal of Sociology, Vol. 81, pp. 730-780. Withey, M., Daft, R. and Cooper, W. (1983). Measures of Perrow’s work unit technology: An empirical assessment and a new scale. Academy of Management Journal, 26, 45-63. Woodward, J. (1956). Industrial Organization: Theory and Practice. Lon don: Oxford University Press. Woodward, J. (1958). Management and Technology. Problems of Pro gress in Technology No. 3. London: Her Majesty’s Stationary Office. Woodward, J, (1965). Industrial Organization: theory and practice. Lon don: Oxford University Press. Wright, P. (1974). The harassed decision-maker: time pressures, distrac tions, and the use of evidence. Journal of Applied Psychology, No. 59, pp. 555-561. 145 APPENDIX 1 -ALGORITHMS USED TO CREATE— — STRUCTURAL EQUIVALENCE VARIABLES----- i ■ The Next Step S um m arizes Vm Usage Variables Based Upon S upervisor Id. M ean Values \For The R esultant G roups A re Then M erged Into The M aster Dataset W here G roup Values A re C reated By S u btracting The Individual Values O f Each O bservation. The R esultant Values R epresent M easures O f S tru ctu ra lly Equivalent Behaviors From Fiorizontal D ifferentiation DATA MASTER2; SET xxxxx. MASTER; !ATTRiB RSTOR RFUTURE RREC RLEN RSEN RTELNO RTELEN ' RVMNO RVMLEN PERF01-PERF13 LENGTH-6; Step Sets 0 Values O f M essaging Equal To M issing * A nd Creates P e rform ance Change Variables ARRAY ASPEN[*] STOR— VMLEN; ARRAY RASPEN[*] RSTOR— RVMLEN; DO 1 = 1 TO DIM (ASPEN); IF ASPEN[l]=0 THEN ASPEN[I]=.; \ RASPEN[I]=ASPEN[I]; END; ARRAY T1 [* ] CONDIT01-CONDIT13; ARRAY T2 [* ] COND2_01-COND2_13; . ARRAY PERFORM [ * ] PERF01-PERF13; DO 1 = 1 TO DIM (PERFORM); PERFORM [I] =T1 [I] — T2 [I]; END; LABEL STOR=’NORMED ASPEN MESSAGES STORED’ RSTOR-’RAW ASPEN MESSAGES STORED’ FUTURE-’NORMED ASPEN MESSAGES FUTURE DELIVERY’ RFUTURE-’RAW ASPEN MESSAGES FUTURE DELIVERY' REC=’NORMED ASPEN MESSAGES RECORDED/WEEK’ ! RREC-'RAW ASPEN MESSAGES RECORDED/WEEK’ RLEN-’RAW ASPEN MESSAGE LENGTH" LEN-’NORMED ASPEN MESSAGE LENGTH' I SEN=’NORMED ASPEN MESSAGE SENDT/WEEK" : RSEN-’RAW ASPEN MESSAGE SENT/WEEK’ j RTELNO-'RAW ANS MACHINE MESSAGES/WEEK’ TELNO-’NORMED ANS MACHINE MESSAGES/WEEK’ TELEN-’NORMED ANS MACHINE MESSAGE LENGTH" RTELEN-'RAW ANS MACHINE MESSAGE LENGTH’ RVMNO-’RAW ASPEN MESSAGES RECEIVED/WEEK’ VMNO-’NORMED ASPEN MESSAGES RECEIVED/WEEK’ ! VMLEN-’NORMED LENGTH ASPEN MESSAGES RECEIVED’ RVMLEN-’RAW LENGTH ASPEN MESSAGES RECEIVED’ ; 146 APPENDIX 1 (CONT.) * Step Rank O rders A nd N orm alizes Aspen Behavioral Data * PROC RANK DATA=MASTER2 OUT=MASTER2 NORMAL=BLOM; ! VAR STOR— VMLEN; * Step S um m arizes Data Set To C reate M ean Values For Positions * PROC SUMMARY DATA=MASTER2; CLASS LEADER; VAR USEWEEKS— VMLEN MESPER2; OUTPUT OUT=SUMM MEAN=GPWEEKS GPSTORE GPFUTURE GPRECORD GPLENGTH GPSEND GPTELNUM GPTELEN GPVMNUM GPVMLEN GPMESPER; PROC SORT DATA=MASTER2; BY LEADER; PROC SORT DATA-SUMM; BY LEADER; * Step Creates Position Value * DATA POSITION; MERGE MASTER2 SUMM; BY LEADER; DROP _TYPE_; ’IF . TYPE_ *=0; ■ARRAY GROUP[*] GPWEEKS— GPMESPER; ; ARRAY IN D IV[*] USEWEEKS— VMLEN MESPER2; IF _FREQ_ >1 THEN DO; DO 1=1 TO DIM (GROUP); GROUP [I] = ( ( GROUP [I] * FREQ ) -INDiV [I]) / (_FR E Q _-1); END; END; ELSE DO; DO 1=1 TO DIM (GROUP); GROUP[l]=.; ! END; ! END; !WEIGHT=1/_FREQ_; : LABEL GPWEEKS=’GROUP MEAN: WEEKS USED VM ’ i GPSTORE='GROUP MEAN: VM MESSAGES STORED' | GPFUTURE=’GROUP MEAN: FUTURE DELIVERY VM MESSAGES’ ■ GPRECORD=’ GROUP MEAN: VM MESSAGES RECORDED' ! GPLENGTH=’GROUP MEAN: LENGTH OF VM MESSAGES' ; GPSEND=’GROUP MEAN: VM MESSAGES SENT’ j GPTELNUM=’GROUP MEAN: TEL. ANS. MESSAGES' GPTELEN='GROUP MEAN: TEL. ANS. MESSAGES LENGTH' | GPVMNUM=’ GROUP MEAN: VM MESSAGES RECEIVED’ ! GPVMLEN=’ GROUP MEAN: RECEIVED VM MESSAGES LENGTH’ GPMESPER='GROUP MEAN: % VM USE ACTIVE MESSAGING' W EIGHT=’1NVERSE OF GROUP SIZE'; APPENDIX 1 (CONT.) The Follow ing Steps S trip O ff Invdividual-ievel M easures O f M edia Usage By Id, Renam e dd=Leader A nd M erge Variables Into Individual Obs As M easures O f S upervisor Behaviors. i The Resultant Values R epresent S tru ctu ra l Equivalence M easures O f Vertical D ifferentiation 1 i ■PROC DATASETS; SAVE POSITION; IdATA VERTICAL; SET POSITION; ! KEEP ID INOVMEAN USEWEEKS— VMLEN MESPER2; I RENAME ID=LEADER USEWEEKS=LDWEEKS STOR=LDSTORE j FUTURE=LDFUTURE REC=LDRECORD LEN=LDLENGTH i SEN=LDSEND TELNO=LDTELNUM TELEN=LDTELEN VMNO=LDVMNUM VMLEN=LDVMLEN MESPER2=LDMESPER INOVMEAN-LDINOVAT; I I LABEL USEWEEKS='SUPERVISOR WEEKS USED VM’ i STOR='SUPERVISOR VM MESSAGES STORED’ FUTURE^’SUPERVISOR FUTURE DELIVERY VM MESSAGES' REC=’SUPERVISOR VM MESSAGES RECORDED’ LEN=’SUPERVISOR LENGTH OF VM MESSAGES’ SEN=’SUPERVISOR VM MESSAGES SENT' TELNO=’SUPERVISOR TEL. ANS. MESSAGES’ TELEN=’SUPERVISOR TEL. ANS. MESSAGES LENGTH’ VMNO=’ SUPERVISOR VM MESSAGES RECEIVED' | VMLEN=’SUPERVISOR RECEIVED VM MESSAGES LENGTH’ ; MESPER2=’SUPERVISOR % VM USE ACTIVE MESSAGING’ INOVMEAN=’SUPERVISOR INNOVATIVENESS' ID='SUPERVISOR ID’; PROC SORT; BY LEADER; ‘DATA MASTER2; MERGE VERTICAL POSITION(IN=C); BY LEADER; IF C=1; IPROC DATASETS; SAVE MASTER2; J PROC RANK DATA=MASTER2 OUT=DISS GROUPS=2; ! VAR ANALYZE EXCEPT MESPER2; RANKS ANALYZE2 EXCEPT2 ACTIVE2; i PROC DATASETS; SAVE DISS; MODIFY DISS; LABEL ANALYZE2=’DICHOTOMIZED ANALYZABILITY' EXCEPT2=’DICHOTOMIZED # OF EEXCEPTIONS’ ACTIVE2=’DICHOTOMIZED USE OF MESSAGING’; RUN; i 148 APPENDIX 2 - T1 QUESTIONNAIRE - Co. X V oice M a il P r o je c t Survey T h is q u e s tio n n a ire examines your c u rre n t com m unication p r a c tic e s and work s it u a t io n as p a r t o f an e v a lu a tio n o f th e v a lu e o f v o ic e messaging a t Co. X. v o ic e messaging is a co m bin a tion o f te le p h o n e and computer te c h n o lo g ie s th a t s u p p o rt people who work to g e th e r though th e y have d i f f i c u l t y m eeting p e rs o n a lly o r re a c h in g each o th e r by phone. One person sends a message to one or a l i s t o f people; th e system s to re s th e message e le c t r o n ic a lly ; r e c ip ie n t s l is t e n to t h e i r te le p h o n e messages a t t h e i r convenience. They have o p tio n s such as r e p ly in g im m e d ia te ly a t th e touch o f a b u tto n , re sen d ing th e message, o r sa vin g messages fo r la t e r re fe re n c e . The q u e s tio n n a ire is b e in g a d m in is te re d to s e v e ra l groups who a re about to re c e iv e v o ic e messaging s e rv ic e , have been u s in g v o ic e messaging s e rv ic e s , and to some who a re n o t, so th a t we can compare how our com m unication may change w ith t h is new system. T h is is n o t a t e s t . There a re no r i g h t or wrong answers. The U n iv e r s ity o f S outhern C a l i f o r n ia 's School o f Communication is p a r t ic ip a t in g w ith us in g a th e rin g in fo rm a tio n . Your responses w i l l be kept, co m p le te ly c o n fid e n tia l by U n iv e r s it y re s e a rc h e rs . Your answers w i l l be combined w ith o th e rs so t h a t no in d iv id u a l responses w i l l be re p o rte d o r made a v a ila b le to anyone. You w i l l re c e iv e a copy o f th e summary r e p o r t in e a rly F a ll. In o rd e r to match t h i s q u e s tio n n a ire w ith th e one you w i l l re c e iv e th re e months from now, we must have yo u r name on th e q u e s tio n n a ire . Please in d ic a te your name, address, and phone on th e space p ro v id e d . Please l e t us know y o u r o p in io n s by p u ttin g th e envelope in th e m ail today. You may want to keep t h i s cover sheet f o r your in fo rm a tio n . Thank you again f o r p a r t ic ip a t in g in t h i s su rvey. I f you have any q u e s tio n s about, t h is p r o je c t, p le a se c a ll xxxxxxxxxxxx a t x x x -x x x x . i f you have any a d d itio n a l th o u g h ts or su g ge stio n s about the f o llo w in g q u e s tio n s , or about your e xp erie nce s w ith v o ic e messaging s e rv ic e s , please leave those comments on th e v o ic e m ailbox g iv e n to you when t h is q u e s tio n n a ire is p ro v id e d . You may want to w r it e i t h e re : __________________________________________ Co. X v o ic e M a il P ro je c t survey 149 APPENDIX 2 (CONT.) . 1 1 _________ 1 - 6 Throughout th e q u e s tio n n a ire , please w r it e your answer in th e box or c i r c l e 'th e number. Ig n o re th e numbers to th e r i g h t ; they are f o r coding purposes. Please p r i n t your name. T h is is ve ry im p o rta n t, as i t makes it. p o s s ib le to match t h i s q u e s tio n n a ire to o th e r ones. Your responses w i l l be c o m p le te ly c o n f id e n t ia l. No Co. X perso n n e l w i l l have access to your answers. P rin te d Name ( F ir s t , L a st Name) what is your Co. X I | Have you ever used v o ic e messaging s e rv ic e s ? ! 1 Yes 2 No 11 25 te le p h o n e number? Area Code Local Number (E xte n sion ) _____ How long have you been w orkin g at Co. X?______________ years Where do you work? 1 xxxxxxxxxx 7 xxxxxxxxxxxxx 2 xxxxxxxx — Telecom m unications 8 xxxxxxxxxxxxxxxxxxxxxxxx 3 xxxxxxxx — A u d it 9 xxxxxxxxxxxx__________________________________ 4 xxxxxxxx — R e c ru itin g 27 5 xxxxxxxxxx — Pension 6 xxxxxxxxxx — Insurance | What is your job t i t l e : __________________________________ _ ____________________ ___ 29 'What ty p e o f jo b do you do? 1 Management 2 P ro fe s s io n a l 3 Agent , 4 Sales or M a rketin g 5 T e ch n ica l 6 A d m in is tra tiv e • 7 S e c r e ta r ia l or C le r ic a l________________________________________________________________________ _ S O t h e r : __________________________ 31 What is your h ig h e s t le v e l o f fo rm a l education? 1 High school 4 Some g raduate school 2 Some c o lle g e 5 Graduate degree 3 C o lle g e degree_______________________________________________________________________________________ 32 33 | I f NO, go to th e n e xt page. I f YES, fo r a p p ro x im a te ly how many months have you used or d id you use these s e rv ic e s ? ___________ months_______________________________________ _ 34-35 do you c u r r e n t ly use v o ic e messaging s e rv ic e s ? 1 Yes 2 No I I | I f NO, go to th e n e x t page.__________________________________________ _ I f YES, 36 How many v o ic e messages do you send and re c e iv e in an average day? ___________ messages____________________________________ 150 APPENDIX 2 (CONT.) 1 37-39 These q u e s tio n s ask about yo u r e x p e c ta tio n s co n ce rn in g v o ic e messaging s e rv ic e s to be implemented in t h is p i l o t : Please c i r c l e th e e x te n t to which you agree w ith each o f th e fo llo w in g statem ents about your e x p e c ta tio n s c o n cern ing v o ic e messaging. I w i l l have more tim e to get in fo rm a tio n th a t c a lle r s ask me to p ro v id e them. Voice messaging is a p p ro p ria te fo r Co. X 's b u sin e ss. I w i l l waste le s s tim e in te le ph o n e ta g . (1) s tr o n g ly agree (2) agree (3) somewhat agree (4) undecided (5) somewhat d isa g re e (6) d isa g re e (7) s tr o n g ly d isa g re e s tr o n g ly agree n e u tra 1 3 4 5 6 3 4 5 6 3 4 5 6 s tro n g ly disa g re e 40 I w i l l o fte n c a ll a c o lle a g u e 's v o ic e box number to leave a request, fo r in fo rm a tio n in s te a d o f c a ll i n g to reach th a t person d ir e c t l y . 1 2 3 4 5 6 I w i l l have s h o rte r phone c a lls . 1 2 3 4 5 6 I w i l l g e t more in fo rm a tio n about what a c a ll e r wants w ith v o ic e messaging than w ith a message s l ip . 1 2 3 4 ‘ ■W hen t r a v e lin g , I w i l l be a b le to use m .y tim e more e f f i c i e n t l y . ^O v e ra ll c o s t savings w i l l r e s u lt from my use o f v o ic e messaging. I w i l l have few er phone c a lls . 46 ,1 w i l l use v o ic e messaging in th e evenings 1 and on week-ends when I th in k o f messages I need to send to c o lle a g u e s . 3 4 5 6 V oice messaging is a p p ro p ria te fo r th e k in d o f work I do. 3 4 I w i l l g e t more work done because I w i l l be a b le to manage my tim e b e tte r . 5 6 i I w i l l o fte n use a v o ic e message in s te a d ! o f a s h o rt w r it t e n message. I am c o n fid e n t about my a b i l i t i e s to use a l l th e fe a tu re s o f a v o ic e messaging system. s tr o n g ly agree 5 6 n e u t r a 1 5 3 s tro n g ly d isa g re e 151 APPENDIX 2 (CONT.) Suppose th a t th e c o s t o f re g u la r phone s e rv ic e is $100 a month fo r each p e rs o n 's phone a t Co. X. Based upon any e x p e c ta tio n s you m ight have about v o ic e 'm essaging, how much a d d itio n a l would you be w i l l i n g to pay per month fo r v o ic e messaging s e rv ic e s ? $ __________/month _ _ 57-59 T h is s e c tio n p ro v id e s you w ith an o p p o r tu n ity to t e l l us, in your own words, about your e x p e c ta tio n s . Please w r it e a few sentences fo r each q u e s tio n . How w i l l v o ic e messaging change th e way you communicate w jth o th e rs? _________________ 60-61 In what s p e c if ic a p p lic a tio n s o r o p p o r tu n itie s m ight v o ic e messaging be e s p e c ia lly u s e fu l? 62-63 How m ight v o ic e messaging a f f e c t Co. X 's r e la tio n s h ip w ith custom ers or agents? 64-65 Please note any o th e r e x p e c ta tio n s o r o p in io n s you have about v o ic e messaging. 66-67 (use th e back i f you need more room, o r c a ll th e "comments" v o ic e messaging box g iv e n to you d u rin g th e t r a in in g session) 152 APPENDIX 2 (CONT.) T h is s e c tio n asks q u e s tio n s about yo u r w orkin g c o n d itio n s . ; how v o ic e messaging m ight a f f e c t any o f these c o n d itio n s . (1) s tro n g ly Please in d ic a te the e x te n t to which you agree w ith each o f th e fo llo w in g statem ents about c o n d itio n s o f your work. we are in te re s te d to o b se rve ( 2 ) (3) (4) ( 5 ) ( 6 ) (7) s tr o n g ly agree am a b le to g e t th e in fo rm a tio n I need from o th e rs on tim e . 1 ; can d i s t r ib u t e in fo rm a tio n to groups o f people q u ic k ly and e a s ily . 1 ; agree agree somewhat agree n e u tra l somewhat, d isa g re e d is a g re e s tr o n g ly d isa g re e s tro n g ly n e u tra l disa g re e 68 The people in my w orkgroup a re a v a ila b le enough f o r me to accom plish my jo b . 74 My work c o n ta c ts o u ts id e Co. X are a v a ila b le enough f o r me to accom plish my jo b . 1 2 3 4 5 6 7 I have s u f f i c i e n t o p p o r tu n ity to th in k through p r o je c ts w ith o th e rs in riiy w orkgroup. 1 2 3 4 5 6 7 I am a b le to focus on p r i o r i t y ta s k s . l 2 3 4 5 6 7 My p r o d u c t iv it y s u ffe r s because o f phone c a lls th a t in t e r r u p t me. 1 2 3 4 5 6 7 When I'm n o t a b le to g e t in fo rm a tio n needed fo r a p r o je c t/w o r k ta s k , Co. X loses money d i r e c t l y ( p o lic ie s n o t w r it t e n ; cla im s no t p a id q u ic k ly , e tc .) 1 2 3 4 5 6 7 I am ve ry s a t i s f i e d w ith my a b i l i t y to contact. a s p e c if ic person to o b ta in in fo rm a tio n . 1 2 3 4 5 6 7 I am a b le to tr a c k my p r o je c ts and know d e a d lin e s fo r what needs to be done. 1 2 3 4 5 6 7 I o fte n do n o t answer my phone because o f o th e r work I am d o in g . 1 2 3 4 5 6 7_________________ _ I e xp erie nce a g re a t deal o f work s tre s s . 1 2 3 4 5 G 7 M eetings w ith my workgroup p r im a r ily serve to d i s t r ib u t e in fo rm a tio n . 1 2 3 4 5 6 7_________________ _ s tr o n g ly s tro n g ly 80 agree n e u tra l d isa g re e 153 APPENDIX 2 (CONT.) T h is s e c tio n and th e n e xt one ask about th e o v e r a ll n a tu re o f your com m unications and work. We a re in te re s te d in how p a tte rn s o f com m unication, a v a i l a b l i t y , and th e n a tu re o f th e work you do are r e la te d to th e u s e fu ln e s s o f v o ic e messaging s e rv ic e s . To answer these q u e s tio n s , c o n s id e r th a t your "d e p artm en t" c o n s is ts o f a l l th e people who do your fu n c tio n , in yo u r lo c a tio n , in your d iv is io n . You are a ls o a s s o c ia te d w ith a p rim a ry fu n c tio n , such as c la im s o r u n d e rw ritin g . W ith in your departm ent, you may work on a " u n it " or "te a m ." Please id e n t if y your departm ent, fu n c tio n and u n it : 12 1 ~3 D epartm ent: F u n c tio n : U n it or "team" : w r it e here th e f i r s t i n i t i a l and l a s t name o f yo u r u n i t 's le a d e r or s u p e rv is o r: Please e s tim a te th e percentage o f yo u r o v e r a ll com m unication w ith people 1 w it h in your u n it/te a m /w o rk g ro u p % 2 o u ts id e your unit., b u t in your fu n c tio n % 3 o u ts id e o f.y o u r fu n c tio n , b u t w it h in your lo c a tio n 3 o u ts id e o f yo u r Co. X lo c a tio n b u t w it h in any o th e r Co. X lo c a tio n 4 to agents o r agency personnel o u ts id e any Co. X lo c a tio n 5 to custom ers o u ts id e any Co. X lo c a tio n 6 to non-custom ers o u ts id e any Co. X lo c a tio n TOTAL: 100 % E stim a te th e average number o f phone c a lls you make a day. . 2.3-24 'E s tim a te th e average number o f phone c a lls you re c e iv e a day ; E s tim a te th e number o f your phone c a lls th a t in v o lv e t r y i n g to reach someone you d id n 't get th e f i r s t tim e. j E s tim a te the average number o f phone messages in a day th a t you ta k e f o r o th e rs in yo u r workgroup or u n it . J though "phone answering fo r o th e rs " is not. your jo b . ! t I f you a re on "phone d u ty ," in d ic a te th e average hours per < week i t is your r e s p o n s ib ilit y to answer phones fo r o th e rs whether or n o t t h i s is in your jo b d e s c r ip tio n . IF YOU ARE SALES AND MARKETING PERSONNEL, what percentage o f an average day is spent in c o n ta c t w ith c u rre n t o r p o te n t ia l customers? 9-10 15-16 2 1 - 2 2 27-28 32-32 3 3 - 3 4 154 APPENDIX 2 (CONT.) T h i s s e c t i o n a s k s a b o u t t h e n a t u r e o f y o u r w o r k . Please c i r c l e th e number th a t b e st in d ic a te s (1) to ve ry l i t t l e e x te n t what you a c tu a lly do in yo u r work. (2) to l i t t l e e x te n t (3) to some e x te n t (4) to g re a t e x te n t (5) to ve ry great, e x te n t To what e x t e n t . . . Is your work ro u tin e ? ve ry l i t t l e e x te n t Do people in yo u r departm ent do about the same jo b in th e same way most o f the tim e? Do people in your departm ent perform r e p e t it iv e a c t i v i t i e s in doing t h e i r jobs? Is th e re a c le a r ly known way to do th e major typ e s o f work 3rou n o rm a lly encounter? Is th e re a c le a r ly d e fin e d body o f knowledge m a tte r which gu id e s you in doing j'our work? Are your d u tie s r e p e t it io u s ? i I s th e re an u n de rsta n d ab le sequence o f steps th a t can be fo llo w e d in d oing yo u r work? Are th e ta s k s you perform th e same from day to day? To do your work., do you a c t u a lly r e ly on e s ta b lis h e d procedures and p ra c tic e s ? Do you communicate w ith w orkers in o th e r lo c a tio n s ? very g re a t e x te n t 35 40 Do you p a r t ic ip a t e on p r o je c t teams o u ts id e yo u r departm ent? ve ry l i t t l e e x te n t some e x te n t 45 ve ry g re a t e x te n t Im agine a . t y p ic a l day a work; then e s tim a te the approxim ate p e rc e n t o f the tim e you spend in th e f o llo w in g a c t i v i t i e s (please sum th e p e rce n ts to 100%): re a d in g or w r it in g in t e r n a l r e p o r t s / f i l e s re a d in g or w r it in g l e t t e r s or memos re a d in g o th e r m a te ria l ( jo u r n a ls , books, e tc .) u sin g th e tele ph o n e u s in g v o ic e m ail u s in g a computer te rm in a l (PC o r mainframe) in fa c e - to - fa c e c o n v e rs a tio n s (not m eetings) in m eetings a l l o th e r a c t i v i t i e s TOTAL 100 % 50-51 56-57 6 2-63 155 APPENDIX 2 (CONT.) . T h i s s e c t i o n a s k s y o u r o p i n i o n s a b o u t u s i n g v a r i o u s w a y s t o c o m m u n i c a t e . Please r a te th e a p p ro p ria te n e s s o f (1) a p p ro p ria te m e m o s /le tte rs and te le ph o n e com m unication (2) somewhat, a p p ro p ria te fo r th e fo llo w in g a c t i v i t i e s . (3) n e u tra l (4) somewhat in a p p ro p ria te (5) in a p p ro p ria te a p p ro p ria te n e u tra l in a p p ro p ria te exchanging m e m o s /le tte rs 1 2 3 4 5 in fo rm a tio n te le ph o n e l 2 3 4 5 b a rg a in in g and m e m o s /le tte rs l 2 3 4 5 n e g o tia tin g tele ph o n e l. 2 3 4 5 g e ttin g to know m e m o s /le tte rs 1 2 3 4 5 someone tele ph o n e 1 2 3 4 5 asking m e m o s /le tte rs 1 2 3 4 5 q u e s tio n s tele ph o n e 1 2 3 4 5 s ta y in g m e m o s /le tte rs 1 2 3 4 5 in touch tele ph o n e 1 2 3 4 5 exchanging m e m o s /le tte rs 1 2 3 4 tim e - s e n s itiv e te le p h o n e 1 2 3 4 in fo rm a tio n g e n e ra tin g m e m o s /le tte rs 1 2 3 4 5 ideas te le ph o n e i 2 3 4 5 r e s o lv in g m e m o s /le tte rs 1 2 3 4 5 disagreem ents te le ph o n e 1 .2 3 4 5 making m e m o s/le tte rs l 2 3 4 5 d e c is io n s telephone a 2 3 4 5 exchanging m e m o s /le tte rs ] 2 3 4 5 c o n f id e n tia l in fo rm a tio n te le ph o n e 1 2 3 4 r> making commitments m e m o s /le tte rs 1 2 3 4 5 tele ph o n e 1 2 3 4 5 a p p ro p ria te n e u tra l in a p p ro p ria te APPENDIX 2 (CONT.) T h is s e c tio n and th e n e xt one ask about your a t t it u d e s tow ard in n o v a tio n and th e o v e r a ll r e p u ta tio n o f yo u r departm ent. We are in te re s te d here in how e xp erie nce s w ith v o ic e m a il is r e la te d to your a tt it u d e s and p o s s ib ly to th e o v e r a ll e ffe c tiv e n e s s o f th e departm ent. (1) s tr o n g ly agree Please in d ic a te th e e x te n t (2) agree to which you agree w ith (3) somewhat agree each o f th e fo llo w in g statem ents (4) undecided about y o u r s e lf and Co. X. (5) somewhat, d isa g re e (6) d isa g re e (7) s tr o n g ly d isa g re e s tr o n g ly un- s tr o n g ly agree decided agree I o fte n f in d m yse lf s k e p tic a l o f new ide a s. 1 2 3 4 5 6 7 I am g e n e ra lly c a u tio u s about a c c e p tin g new ide a s. 1 3 3 4 5 6 7 I r a r e ly t r u s t new ideas u n t i l I can see w hether th e v a s t m a jo r ity o f people accept them. I 2 3 4 5 6 7 I f in d i t s tim u la tin g to be o r ig i n a l in my t h in k in g and b e h a v io r. 1 2 3 4 5 6 7 I am aware th a t I am u s u a lly one o f the la s t people in my group to accept som ething new. 1 2 3 4 5 6 7 I tend to fe e l th a t th e o ld way o f l i v i n g and d oing th in g s is th e b e s t way. 1 2 3 4 5 6 7 I am r e lu c ta n t about a d o p tin g new ways o f doing th in g s u n t i l I see these them w orkin g fo r people around me. 1 2 3 4 5 6 7 I must see o th e r people u s in g new in n o v a tio n s b e fo re I w i l l c o n s id e r them. 1 2 3 4 5 6 7 I am ch a lle n g e d by unanswered q u e s tio n s . 1 2 3 4 5 6 7 I am ch a lle n g e d by a m b ig u itie s and unsolved problem s. 1 2 3 4 5 6 7 iCo. X is always moving tow ard th e development, o f new ways o f d oing b u s in e s s . 1 2 3 4 5 6 7 .Co. X can be d e s c rib e d as f l e x i b l e and c o n tin u a lly a d a p tin g to change. 1 2 3 4 5 6 7 i I People in Co. X are always se a rc h in g fo r i fre s h , new ways o f lo o k in g a t problem s. 1 2 3 4 5 6 7 , s tr o n g ly un- s tro n g ly agree decided agree 25 30 3 7 157 APPENDIX 2 (CONT.) In r e la t io n to o th e r departm ents a t Co. X, how do you t h in k yo u r departm ent ra te s on each o f th e fo llo w in g fa c to r s s in c e th e b e g in n in g o f th e year? (1) fa r below average (2) somewhat below average (3) about average (4) somewhat above average (5) fa r above average fa r below average average fa r above average R ep u ta tio n fo r work e x c e lle n c e In flu e n c e on o r g a n iz a tio n a l p o lic y 1 2 3 4 5 6 7 ] 2 3 4 5 6 7 38 A tta in m e n t o f w orkgroup p ro d u c tio n o r s e rv ic e g oals 1 2 3 4 5 6 7 d u a lit y or accuracy o f work produced 3 4 5 6 7 E ff ic ie n c y o f w orkgroup o p e ra tio n s 1 2 3 4 5 Number o f in n o v a tio n s o r new ideas in tro d u c e d by th e workgroup 1 2 3 4 5 6 7 fa r below fa r above average average average 43 Thank you! Your th o u g h ts and o p in io n s a re very im p o rta n t to t h is stu d y. Please seal and r e tu r n t h i s q u e s tio n n a ire in the enclosed envelope!, and m a il to Ron Rice a t th e U n iv e r s ity o f southern C a lif o r n ia . Please fe e l fr e e to leave any a d d itio n a l comments or thoughts at the s u g ge stio n v o ic e m a ilb o x l is t e d on th e cover page o f t h is q u e s tio n n a ire . We a p p re c ia te your h e lp ! 158 APPENDIX 3 - T2 QUESTIONNAIRE - Co. X V oice M a il P ro je c t Survey 21_______ 1 -6 Throughout th e q u e s tio n n a ire , please w r it e your answer in th e box or c i r c l e 't h e number. Ig n o re th e numbers to th e r i g h t ; they are fo r cod in g purposes. Please p r in t your name. Your responses w i l l be co m p le te ly c o n f id e n t ia l. No Co. X personnel w i l l have access to your answers. P rin te d Name ( F ir s t , L a st Name) 7 What is your Co. X - ______________ -____________ te le ph o n e number? Area Code Local Number _______ ________ 11 How long have you been w orkin g a t Co. X? _______ ye a rs _ _ 25 Do you c u r r e n t ly use v o ic e messaging s e rv ic e s ? 1 Yes 2 No I I | I f NO, go to the n ext page. _ _ I f YES, 27 How many v o ic e messages do you send and re c e iv e in an average day? _____________ messages________________________________________ ____ 28-30 V oice messaging can be used in two ways: "Telephone a nsw ering" — your phone " r o l l s o v e r" and a c a ll e r g ets your v o ic e message when you a r e n 't a v a ila b le . "Messaging" — someone e ls e on th e system d ia ls in to t h e i r box and then sends you a message. You use th e " r e p ly fu n c tio n " to respond. You can a ls o send a "group d is t r ib u t i o n message" to s e v e ra l people a t once. ’ A p p ro xim a te ly what percentage o f your t o t a l usage was (sum to 10 0 %): ___ te le ph o n e answ ering: % m essaging: % Please c i r c l e th e e x te n t to wh i c h (1) to ve ry l i t t l e e x te n t each o f th e fo llo w in g v o ic e messaging (2) to l i t t l e e x te n t v o ic e messaging fu n c tio n s (3) to some e x te n t have been b e n e fic ia l to your (4) to great, e x te n t in your work (5) to very great, e x te n t ve ry l i t t l e e x te n t some very g re a t e x te n t th e te le p h o n e answ ering fu n c tio n 1 2 3 4 5 _ _ th e messaging fu n c tio n 1 2 3 4 5 _ _ th e r e p ly fu n c tio n 1 2 3 4 5 _ _ th e group d is t r ib u t i o n fu n c tio n ] 2 3 4 5 _ _ 3 8 159 APPENDIX 3 (CONT.) 'These q u e s tio n s ask about your e xp erie nce s co n ce rn in g v o ic e messaging s e rv ic e s p ro v id e d in t h i s p i l o t . Please c i r c l e th e e x te n t to which you agree w ith each o f th e fo llo w in g statem ents about yo u r e xperiences w ith v o ic e messaging. I have more tim e to g e t in fo rm a tio n th a t c a lle r s ask me to p ro v id e them. V oice messaging is a p p ro p ria te fo r Co. X 's b u sin e ss. I waste le s s tim e in te le ph o n e ta g . {1) s tr o n g ly agree (2) agree (3) somewhat agree (4) n e u tra l (5) somewhat, d isa g re e (6) d isa g re e (7) s tr o n g ly d isa g re e 3 9 s tr o n g ly agree n e u tra l 3 4 s tr o n g ly d is a g re e 5 6 5 6 5 S 40 I o fte n c a ll a c o lle a g u e 's v o ic e box number to leave a re q u e s t fo r in fo rm a tio n in ste a d o f c a ll i n g to reach that, person d ir e c t l y . 1 2 3 4 5 6 7 I have s h o rte r phone c a lls . 1 2 3 4 5 6 7 I g e t more in fo rm a tio n about what a c a ll e r wants w ith v o ic e messaging than w ith a message s l ip . 1 2 3 4 6 7 When t r a v e lin g , I am a b le to use my tim e more e f f i c i e n t l y . O v e ra ll c o s t savings r e s u lt from my use o f v o ic e messaging. I have few er phone c a lls . 1 2 3 4 3 4 5 3 4 5 6 7 46 I use v o ic e messaging in th e evenings and on week-ends when I th in k o f messages I need to send to c o lle a g u e s . 1 2 3 4 5 6 V oice messaging is a p p ro p ria te fo r th e k in d o f work I do. 3 4 5 6 J . g e t more work done because I am a b le to manage my tim e b e tt e r . 3 4 5 6 o fte n use a v o ic e message in ste a d o f a s h o rt w r it t e n message. 1 2 3 4 5 6 I am c o n fid e n t about my a b i l i t i e s to use a l l th e fe a tu re s o f a v o ic e messaging system, I O v e r a ll, I am a r e lu c ta n t user. 3 4 3 4 54 strongly s tr o n g ly agree n e u tra l d isa g re e 160 APPENDIX 3 (CONT.) Suppose th a t th e c o s t o f re g u la r phone s e rv ic e is $100 a month fo r each p e rs o n 's phone a t Co. X. Based upon your e xp erie nce s w ith v o ic e messaging, how much a d d itio n a l would you be w i l l i n g to pay per month fo r v o ic e messaging se rv ic e s ? $ /month ____ 55 T h is s e c tio n p ro v id e s you w ith an o p p o rtu n ity to t e l l us, in yo u r own words, about yo u r e xp e rie n ce s. How has v o ic e messaging changed th e way you communicate w ith o thers? In what s p e c if ic a p p lic a tio n s o r o p p o r tu n itie s co u ld v o ic e messaging be e s p e c ia lly u s e fu l? ____ 61 How might, v o ic e messaging a f f e c t Co. X 's r e la t io n s h ip w ith custom ers or agents?__________________________________________________________________________________ ____ 63 what d id you l i k e b e s t about u sin g v o ic e messaging? ____ 64 What d id you l ik e le a s t about, usin g v o ic e messaging?________________________________________ 6 6 I f you have n o t used v o ic e messaging a t a l l , or ve ry seldom, ple a se b r i e f l y e x p la in why.________________________________________________________________________ ____ 68 1 6 1 APPENDIX 3 (CONT.) I These q u e s tio n s r e la t e to t r a in in g , s u p p o rt, and system fe a tu re s . Please p la c e a check b e sid e th e typ e s o f t r a in in g you re c e iv e d b e fo re you began u s in g th e system: T r a in in g w ith a group o f people ____ In d iv id u a l t r a in in g w ith a c o o rd in a to r_______________________ S e lf t r a in in g ____ Other (e x p la in : _________- ___________________________________ ) ____ E va lu a te how easy i t is f o r you to use each o f th e fo llo w in g v o ic e messaging fu n c tio n s . (1) Very Easy (2) Somewhat Easy (3) N e ith e r Easy Nor D i f f i c u l t (4) Somewhat D i f f i c u l t (5) Very D i f f i c u l t Very Easy N e ith e r ve ry D i f f i c u l t I n i t i a l i n g your m ailbox ' Sending messages . Forw arding your phone to v o ic e messaging Forw arding or r e - r o u tin g messages Sending messages to more than one person . S kip p in g to th e end o f a message Doing ANY fu n c tio n you had not done b e fo re G e ttin g a s s is ta n c e when you could n o t f ig u r e o u t what to do O v e ra ll use o f th e system Please c i r c l e th e e x te n t to which each o f th e fo llo w in g fa c to r s would in cre a se th e v a lu e o f v o ic e messaging to you. 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 !) 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 (.1) to ve ry l i t t l e e x te n t (2) to l i t t l e e x te n t (3) to some e x te n t (4) to g re a t e x te n t (5) to v e ry g re a t e x te n t ve ry l i t t l e e x te n t 14 A l i g h t on your phone th a t t e l l s you i f you have messages 1 2 3 I Being a b le to reach anyone in 1 2 3 ' Co. X th ro u g h v o ic e messaging Being a b le to reach anyone in i 2 3 yo u r d iv is io n th ro u gh v o ic e messaging j More s k i l l a t u s in g th e system 1 2 3 More encouragement from your l 2 3 I manager to use th e system 1 O th e rs : 1 2 3 some 4 ve ry g re a t e x te n t 2 0 162 APPENDIX 3 (CONT.) T h i s s e c t i o n a s k s y o u a b o u t t h e n a t u r e o f y o u r c o m m u n i c a t i o n s a t w o r k . ■E stim ate th e average number o f phone c a lls you make a day. ___________ _ _ i 2 1 -2 2 E stim a te th e average number o f phone c a ll s you re c e iv e a day. ___________ _ _ E s tim a te th e number o f your phone c a ll s th a t in v o lv e t r y i n g to reach someone you d id n 't g e t th e f ir s t , tim e ._____________________ _ _ 25-26 E stim a te th e average number o f phone messages in a day th a t you ta k e f o r o th e rs in your workgroup o r u n it , I though "phone answering fo r o th e rs " is not your jo b . ___________ I f you a re on "phone d u ty ," in d ic a te th e average hours per week i t is your r e s p o n s ib ilit y to answer phones fo r o th e rs w hether or no t t h i s is in your jo b d e s c r ip tio n . ___________ _ _ 29-30 IF YOU ARE SALES AND MARKETING PERSONNEL, what percentage o f an average day is spent in c o n ta c t w ith c u rre n t or p o te n t ia l customers?_______________________________ __ Im agine a ty p ic a l day a t work; then e s tim a te th e approxim ate p e rc e n t o f th e tim e you spend in th e fo llo w in g a c t i v i t i e s (please sum the p e rc e n ts to 100%): re a d in g or w r it in g in t e r n a l r e p o r t s / f i l e s __________________ ___ re a d in g or w r it in g l e t t e r s or memos ____ ___ re a d in g o th e r m a te ria l ( jo u r n a ls , books, e tc .) __ _ _ 37-38 u s in g the tele ph o n e ____________________________________________ _____ ___ u s in g v o ic e m a il______________________________________________________ ___ u s in g a computer te rm in a l (PC or mainframe)_______________ _ _ in fa c e - to - fa c e c o n v e rs a tio n s (not m eetings)______________ _ _ in m eetings_____________________________________________________________ _ _ a l l o th e r a c t i v i t i e s _ ._____ ___ TOTAL 100 % 49-50 163 APPENDIX 3 (CONT.) T h is s e c tio n asks q u e s tio n s about your w orking c o n d itio n s . We are in te re s te d to observe how v o ic e messaging has a ffe c te d any of these c o n d itio n s . Please in d ic a te th e e x te n t to which you agree w ith each o f th e fo llo w in g statem ents about c o n d itio n s o f your work. (1) s tr o n g ly agree (2) agree (3) somewhat agree (4) n e u tra l (5) somewhat d is a g re e (6) d isa g re e (7) s tr o n g ly d isa g re e s tr o n g ly d isa g re e s tro n g ly agree n e u tra l I am a b le to g e t th e in fo rm a tio n I need from o th e rs on tim e . 1 2 3 4 5 6 I can d is t r ib u t e in fo rm a tio n to groups o f people q u ic k ly and e a s ily . 1 2 3 4 5 6 The people in my w orkgroup are a v a ila b le enough fo r me to accom plish my jo b . 1 2 3 4 5 6 51 My work c o n ta c ts o u ts id e Co. X a re a v a ila b le enough fo r me to accom plish my jo b . 1 2 3 4 5 6 7 I have s u f f i c i e n t o p p o r tu n ity to th in k through p r o je c ts w ith o th e rs in my w orkgroup. 1 2 3 4 5 6 7 I am a b le to focus on p r i o r i t y ta s k s . 1 2 3 4 5 6 7 My p r o d u c t iv it y s u ffe r s because o f phone c a lls th a t in t e r r u p t me. When l-'m not a b le to g e t in fo rm a tio n needed f o r a p r o je c t/w o r k ta s k , Co. X lose s money d i r e c t l y ( p o lic ie s n o t w r it t e n ; c la im s no t p a id q u ic k ly , e tc .) 5 6 57 I am ve ry s a t is f ie d w ith my a b i l i t y to c o n ta c t a s p e c if ic person to o b ta in in fo rm a tio n . I am a b le to tr a c k my p r o je c ts and know d e a d lin e s fo r what needs to be done. I o fte n do n o t answer my phone because o f o th e r work I am d o in g . 1 2 3 4 5 6 7 _ _ | I e xp erie nce a g re a t deal o f work s tre s s . 1 2 3 4 5 6 7 ___ I I Meetings w ith my w orkgroup p r im a r ily serve to d is t r ib u t e in fo rm a tio n . 1 2 3 4 5 6 7 _ _ i s tr o n g ly s tr o n g ly 63 I agree n e u tra l d isa g re e 164 APPENDIX 3 (CONT.) . T h i s s e c t i o n a s k s y o u r o p i n i o n s a b o u t u s i n g v a r i o u s w a y s t o c o m m u n i c a t e . Please r a te th e a p p ro p ria te n e s s o f m e m o s /le tte rs and te le ph o n e com m unication f o r th e fo llo w in g a c t i v i t i e s . Exchanging in fo rm a tio n : (1) a p p ro p ria te (2) somewhat a p p ro p ria te (3) n e u tra l (4) somewhat, in a p p ro p ria te (5) in a p p ro p ria te a p p ro p ria te n e u tra l in a p p ro p ria te B a rg a in in g and n e g o tia tin g : G e ttin g to know someone: Asking q u e s tio n s : S ta y in g in touch: ■ Exchanging tim e - s e n s itiv e in f o r m a t io n : G enerating id e a s : R eso lvin g d is a g re e m e n ts : Making d e c is io n s : Exchanging c o n fid e n t ia l in fo r m a tio n : Making commitments: m e m o s/le tte rs 1 2 3 4 5 tele ph o n e 1 2 3 4 5 v o ic e messaging 1 2 3 4 5 m e m o s/le tte rs 1 2 3 4 5 tele ph o n e 1 2 3 4 5 v o ic e messaging 1 2 3 4 5 m e m o s/le tte rs 1 2 3 4 5 telephone 1 2 3 4 5 v o ic e messaging 1 2 3 4 5 m e m o s/le tte rs .1 2 3 4 5 tele ph o n e 1 2 3 4 5 v o ic e messaging 1 2 3 4 5 m e m o s/le tte rs 1 2 3 4 5 tele ph o n e :t 2 3 4 5 v o ic e messaging i 2 3 4 5 m e m o s/le tte rs i 2 3 4 5 telephone i 2 3 4 5 v o ic e messaging i 2 3 4 5 m e m o s/le tte rs i 2 3 4 5 tele ph o n e i 2 3 4 5 v o ic e messaging i 2 3 4 5 m e m o s /le tte rs i 2 3 4 5 tele ph o n e i 2 3 4 5 v o ic e messaging l 2 3 4 5 m e m o s /le tte rs i 2 3 4 5 tele ph o n e l 2 3 4 5 v o ic e messaging i 2 3 4 5 m e m o s/le tte rs i 2 3 4 5 tele ph o n e ] 2 3 4 5 v o ic e messaging i 2 3 4 5 m e m o s/le tte rs i 2 3 4 5 tele ph o n e i 2 3 4 5 v o ic e messaging i 2 3 4 5 appropr ia te n e u t r a 1 in a p p ro p ria te APPENDIX 3 (CONT.) In r e la t io n to o th e r departm ents a t Co. X, how do you th in k your departm ent ra te s on each o f th e fo llo w in g fa c to r s s in c e th e b e g in n in g o f th e year? (1) fa r below average (2) somewhat below average (3) about average (4) somewhat above average (5) fa r above average f a r below average average fa r above average R ep u ta tio n fo r work e x c e lle n c e 1 2 3 4 5 In flu e n c e on o r g a n iz a tio n a l p o lic y 1 2 3 4 I) A tta in m e n t o f w orkgroup p ro d u c tio n or s e rv ic e g oals 1 2 3 4 5 Q u a lity or accuracy o f work produced 1 2 3 4 5 E f f ic ie n c y o f w orkgroup o p e ra tio n s . 1 2 3 4 5 Number o f in n o v a tio n s or new ideas in tro d u c e d by th e workgroup 1 2 3 4 5 fa r below average average fa r above average 3 8 43 Please comment here on your reasons: 41-42 IF YOU COMPLETED AN EARLIER, SIMILAR QUESTIONNAIRE ABOUT VOICE MESSAGING, sto p here. Thank you! Your th o u g h ts and o p in io n s a re ve ry im p o rta n t to t h is s tu d y . Please r e tu r n t h i s q u e s tio n n a ire to your s it e c o o rd in a to r. IF YOU HAVE NOT COMPLETED THE EARLIER QUESTIONNAIRE, p le a se go on to th e n ext page. in n o v a tiv e n e s s : 13 item s 166 APPENDIX 3 (CONT.) IF YOU HAVE COMPLETED AN EARLIER, SIMILAR QUESTIONNAIRE, p le a se go to th e end o f th e la s t page. T h is s e c tio n asks about your a ttitu d e s toward in n o v a tio n . (1) s tr o n g ly agree Please in d ic a te th e e x te n t (2) agree to which you agree w ith (3) somewhat agree each o f th e fo llo w in g statem ents (4) undecided about y o u r s e lf and Co. X. (5) somewhat d isa g re e (6) d isa g re e (7) s tr o n g ly d isa g re e s tr o n g ly un- s tro n g ly agree decided d isa g re e 1 o fte n fin d m yse lf s k e p tic a l o f new ideas. 1 2 3 4 5 6 7 _ _ 43 I am g e n e ra lly c a u tio u s about a c c e p tin g new ideas. 1 2 3 4 5 6 7 _ _ I r a r e ly t r u s t new ideas u n t i l I can see w hether th e v a s t m a jo r ity o f people accept them. 1 2 3 4 5 6 7 I fin d i t s tim u la tin g to be o r ig in a l in my t h in k in g and b e h a v io r. 1 2 3 4 5 6 7 I am aware th a t I am u s u a lly one o f th e ' la s t people in nty group to accept som ething new. 1 2 3 4 5 6 7 I tend to fe e l th a t th e o ld way o f l iv i n g and doing th in g s is th e b e s t way. 1 2 3 4 5 6 7 _ _ 48 I am r e lu c ta n t about a d o p tin g new way's o f d o in g th in g s u n t i l I see them w orkin g f o r people around me. 1 2 3 4 5 6 7 1 I must, see o th e r people u sin g new | in n o v a tio n s b e fo re I w i l l c o n s id e r them. ' I am ch a lle n g e d by unanswered q u e s tio n s . 1 2 3 4 5 6 7 1 2 3 4 5 6 7 I am ch a lle n g e d by a m b ig u itie s and unsolved problem s. 1 2 3 4 5 0 7 'Co. X is always moving toward the development o f new ways o f d oing b usiness. | Co. X can be d e s c rib e d as f l e x ib l e I and c o n t in u a lly a d a p tin g to change. People in Co. X are always se a rch in g fo r fre s h , new ways o f lo o k in g a t problem s. 55 s tr o n g ly un- s tro n g ly agree decided d isa g re e 167 APPENDIX 3 (CONT.) w ork group ro u tin e n e s s : 11 item s T h is s e c tio n asks about th e 'n a tu r e o f your work. Please c i r c l e th e number th a t b e s t in d ic a te s what you a c tu a lly do in yo u r work. To what e x t e n t . (1) ( 2 ) (3) <4) (5) very l i t t l e e x te n t to very l i t t l e e x te n t to l i t t l e e x te n t to some e x te n t to g re a t e x te n t to ve ry g re a t e x te n t ve ry some g re a t e x te n t Is your work ro u tin e ? 1 2 3 4 5 Do people in your departm ent do about th e same job in th e same way most o f th e time? 1 2 3 4 5 Do people in your departm ent perform r e p e t it iv e a c t i v i t i e s in doing t h e i r jobs? 1 2 3 4 5 Is th e re a c le a r ly known way to do th e major types o f work you n o rm a lly encounter? 1 2 3 4 5 Is th e re a c le a r ly d e fin e d body o f knowledge m a tte r which guides you in doing your work? 1 2 3 4 5 Are your d u tie s r e p e titio u s ? 1 2 3 4 5 Is th e re an unde rsta n d ab le sequence o f steps th a t can be fo llo w e d in d oing your work? 1 2 3 4 5 Are th e ta s k s you perform the same from day to day? 1 2 3 4 5 To do your work, do you a c tu a lly r e ly on I e s ta b lis h e d procedures and p ra c tic e s ? 1 2 3 4 5 \ Do you communicate w ith w orkers in o th e r lo c a tio n s ? I 2 3 4 5 , Do you p a r t ic ip a t e on p r o je c t teams I o u ts id e your department? 1 2 3 4 5_________________ _ I 12 very some ve ry [ l i t t l e e x te n t extent g re a t e x te n t f I c ro s s -lc c & ti'a n , jo b t i t l e , demographics 168 ! APPENDIX 3 (CONT.) T h i s l a s t s e c t i o n s a s k s s o m e g e n e r a l q u e s t i o n s a b o u t y o u a n c l y o u r j o b . To answer t h i s n e xt s e t o f q u e s tio n s , c o n s id e r th a t your ''department." c o n s is ts o f a l l th e people who do your fu n c tio n , in your lo c a tio n , in your d iv is io n . You are a ls o a s s o c ia te d w ith a p rim a ry fu n c tio n , such as c la im s or u n d e rw ritin g . W ith in your departm ent, you may work on a "u n it." or "tea m ." Please e s tim a te th e percentage o f your o v e r a ll com m unication w ith people . . . 1 w it h in your u n it/te a m /w o rk g ro u p ft_______________________________ _ 13-14 2 o u ts id e your u n it , b u t in your fu n c tio n ft 3 o u ts id e o f your fu n c tio n , b u t w it h in yo u r lo c a tio n ft _ _ 3 o u ts id e o f yo u r Co. X lo c a tio n but w it h in any o th e r Co. X lo c a tio n ft _ _ 19-20 4 to agents or agency personnel o u ts id e any Co. X lo c a tio n ft _ _ 5 to custom ers o u ts id e any Co. X lo c a tio n ft _ _ 6 to non-custom ers o u ts id e any Co. X lo c a tio n _ ft _ _ 25-26 TOTAL: 100 ft What typ e o f job do you do? 1 Management 2 P ro fe s s io n a l 3 Agent 4 Sales o r M a rketin g 5 T e ch n ica l 6 A d m in is tr a tiv e 7 S e c r e ta r ia l or C le r ic a l 8 O ther: _________________________ What is your h ig h e s t le v e l o f form al education? 1 High school 4 Some graduate school 2 Some c o lle g e 5 Graduate degree 3 C olle g e degree _ _ 28 Thank you! Your th o u gh ts and opinions, are very im portant, to t h i s study. Please r e tu r n t h is q u e s tio n n a ire to your s it e c o o rd in a to r. 169 APPENDIX 4 - EXAMPLE OF COMPUTER-MONITORED DATA Page 1 Date: JUL 16 1987 Time: 15:43:00 ASPEN REPORTS In d iv id u a l S u b s c rib e r Usage S t a t i s t i c s DOE, JOHN M ailbox Number 30130 Weekly S t a t i s t i c s PERIOD 18 JUL 11JUL MESSAGE STORAGE T o ta l Message Storage 0 3 Storage fo r New Messages 0 2 Storage fo r A rchived Messages 0 0 Storage fo r F u tu re D e liv e ry Messages 0 1 MESSAGES RECORDED AND SENT Messages Recorded 0 7 T o ta l Length (m inutes) o 19 Avg Length (m inutes) 0.0 2.7 Messages Sent 0 14 ME T o ta l Message Storage 6 9 Storage f o r New Messages 2 8 Storage fo r A rch ive d Messages 4 1 Storage fo r F u tu re D e liv e ry Messages 0 0 MESSAGES RECORDED AND SENT Messages Recorded 8 o T o ta l Length (m inutes) 15 0 Avg Length (m inutes) 1.9 o.o Messages Sent 8 0 MESSAGES RECEIVED T o ta l Messages Received 10 7 170 APPENDIX 4 (CONT.) 18JUL 11JUL T o ta l Length (m inutes) 4 4 Avg Length (m inutes) 0.4 o.B Telephone Answering 5 7 T o ta l Length (m inutes) 2 4 Avg Length (m inutes) 0 .4 0.6 V oice M ail 5 0 T o ta l Length (m inutes) 2 0 Avg Length (m inutes) 0.4 o.o 3 Bad Passwords 0 o Page 6 Date: JUL 16 1987 Time: 15:43:00 ASPEN REPORTS In d iv id u a l S u b s c rib e r Usage s t a t i s t i c s Average Accesses Per Day 3.8 1 . 6 New Messages A u to -D e le te d o 0 A rchived Messages A u to -D ele te d o 0 171

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Confirming media effects in voting behavior: A media dependency perspective

PDF

Bringing social worlds together: Information systems as catalysts for new interactions in health care organizations

PDF

Crisis management and narcissism: A Kohutian perspective

PDF

Communicative features of educational computer software

PDF

Communicative performance with media: Effects and implications of television and adult-child interaction on preschooler's language

PDF

Computers and political communication: Innovation in political campaign organizations

PDF

An organizational culture approach to the study of individual power and its use

PDF

Communicating in organizations: The influence of context, job, task, and channel

PDF

Controlling myths of the new religious right: A narrative analysis of Pat Robertson's 1988 presidential campaign

PDF

A system of participation and allocation preferences in organizations

PDF

Correlates of children's responses to an affective television and print package designed to reduce sex-role stereotyping

PDF

Beyond binary theories to agile learning: An empirical investigation of learning methods in warranty management.

PDF

Communication, consensus and commitment in collective action organizations

PDF

Computers simulation of particle transport processes in porous media

PDF

Accommodation in small groups: Patterns and consequences of adjustments in group member communication style over time

PDF

Communication during physical intimacy: A theoretical and methodological study of communication competence

PDF

An empirical investigation of the communicative rules which should guide the use of deceptive messages within non-interpersonal and interpersonal relationships

PDF

A structurational analysis of labor negotiations

PDF

Communicating via electronic mail: Patterns and predictors of use in organizations

PDF

A study on the content and determinants of close exchange relationships in channels of distribution

Asset Metadata

Creator Shook, Douglas Edward (author)

Core Title A structural equivalence and contingency theory perspective on media usage and communication performance: The case of voice messaging

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Communication Theory and Research

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag mass communications,OAI-PMH Harvest,Speech Communication

Language English

Advisor Rice, Ronald (committee chair), el-Sawy, Omar (committee member), Rogers, Everett M. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-715570

Unique identifier UC11343542

Identifier DP22434.pdf (filename),usctheses-c17-715570 (legacy record id)

Legacy Identifier DP22434.pdf

Dmrecord 715570

Document Type Dissertation

Rights Shook, Douglas Edward

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA