University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content COMMUNICATING VIA ELECTRONIC MAIL: PATTERNS AND PREDICTORS OF USE IN ORGANIZATIONS by Charles William Steinfield 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) December 1983 UMI Number: DP22378 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 com plete m anuscript and there are missing pages, th ese will be noted. Also, if material had to be removed, a note will indicate the deletion. UMT Dissertation Publishing UMI DP22378 Published by ProQ uest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQ uest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United S tates Code ProQ uest LLC. 789 E ast Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 4 8 1 0 6 -1 3 4 6 U N IV ER SITY O F S O U T H E R N C A L IFO R N IA T H E G R A D U A T E S C H O O L U N IV E R S IT Y P A R K L O S A N G E L E S , C A L IF O R N IA 9 0 0 0 7 TV!). CH This dissertation, written by ....... CHARLES _WILLIAM__STEINFIELD ........ under the direction of hi.?..... Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of requirements of the degree of D O C T O R OF P H IL O S O P H Y Dean Date.... .. A 2 ? . DISSERTATION COM Chairman ACKNOWLEDGEMENTS There are many colleagues, faculty, family members and friends I wish to thank for their part in helping me achieve this goal. It would be impossible to thank everyone personally who played some role in keeping me on the path to this dissertation. To all of you whom I neglected -to mention, your help was no less appreciated. First, I gratefully acknowledge the efforts of my dissertation committee in all phases of the research on which this dissertation is based. Their counsel was valuable and their suggestions always made significant improvements. Dr. William Dutton chaired the committee, and took great care in both reviewing all of my work and keeping me on track. Dr. Janet Fulk gave constructive counsel throughout the research, and enhanced my learning experience by not letting me take the easier, but lower quality solutions to problems. Dr. Jon Miller introduced me to much of the theoretical literature, and stimulated my interest in organizational sociology. I would also like to acknowledge the assistance of the Annenberg School of Communications, which provided financial support, learning resources, and a stimulating intellectual environment. All of these factors contributed to the completion of this document. The guidance of other present and former faculty at the Annenberg School deserve is also recognized. Dr. Fred Williams and Dr. Richard Byrne served on my guidance committe and provided helpful counsel. Herb Dordick is thanked for his role making life at the school more personal, and making me feel like part of the family. Two former faculty members, James Danowski and Thomas Martin were instrumental in ii guiding my early learning and research experiences at Annenberg. Jim's friendship through those first years will always be remembered. None of this would have been possible without the enthusiastic support of the host organization for this study. Having promised company anonymity, I will not mention last names or the company name. My gratitude is extended to Dudley, Chuck, and Bill for their sponsorship of this research. The many people from this company who participated in the study are also thanked. Two colleagues merit special attention for their role in the production of this dissertation. Dr. John Ruchinskas and Dr. Lynne Svenning always had valuable advice and our collaboration on many exciting projects throughout our graduate careers was among the most influential of my educational experiences. Finally, friends and family were always there when I needed them. Richard Beada shared in the painful experiences of early graduate school, and his friendship and good spirits provided much needed emotional support. I am grateful to Susan Russick, who spent many long hours helping me with the final production of this document. Her emotional support made life bearable when the going got tough. My mother ard father instilled in me the value of education, and never doubted my ability to finish. Telephone calls from my brother and sister kept me going when my spirits were low. My family's quick infusions of cash, of course, made life much more bearable during poverty stricken graduate school days. Howie and Bebe are thanked as well for keeping me well fed through the years in Los Angeles. iii TABLE OF CONTENTS ACKNOWLEDGEMENTS LIST OF TABLES LIST OF FIGURES CHAPTER ONE. Statement of the Problem The Acceptance of Electronic Mail The Attributes and Impacts of Electronic Mail Limitations of Prior Research Study Overview Research Questions Findings - Purposes of Use Findings - Predicting Patterns of Use Theoretical Significance Practical Significance CHAPTER TWO. Review of the Literature and Development of Hypotheses The Dependent Variable - Purposes of Electronic Mail Use Independent Variables Task Characteristics/Task Environment Technology, Task Interdependence, and Coordination An Information Processing Model of Organization Design Communication Media and Inter-unit Communication Perceived Channel Characteristics Objective Channel Attributes Perceived Channel Attributes Individual/Demographic Characteristics System Access Variables Page ii vii ix 1 2 3 6 8 8 9 9 12 13 15 16 20 22 24 27 31 34 35 37 42 46 iv CHAPTER THREE. Methods 51 Research Context 51 Qualitative Analyses 54 Interviews 54 Observation 56 The Survey 57 Sampling 57 Data Gathering 59 Variables and Operationalizations 61 Dependent variables 61 Independent Variables 70 Task and Environment Variables 72 Channel Charactristics 83 Individual/Demographic Variables 86 System Access Variables 90 Analysis Strategy 92 CHAPTER FOUR. Analysis 94 Types of Uses 94 Dimensions of Uses 97 Multiple Regression Analyses 99 Task Use Regression Results 99 System Access Variables 99 Task Environment Variables 104 Perceived Channel Characteristics 105 Individual/Demographic Variables 106 Summary of Task Regression Results 106 Social Use Regression Results 107 System Access Variables 107 Task Environment 109 Perceived Channel Characteristics 109 Ind ividual/Dsmographics 110 Summary of Social Use Regression Results 111 Summary of Regression Analysis 111 Categorizing User Types 112 Discriminant Analysis Findings 117 Classification Analysis 121 Summary of Discriminant Analysis 124 V CHAPTER FIVE. Conclusions and Discussions 125 Question 1: The Uses of Electronic Mail 125 Types of Uses 125 Dimensions of Uses 131 Question 2: Factors That Lead to Different Patterns of Use 135 Discussion of Multiple Regressions 135 Predicting Task Use 135 System Access 135 Task/Environment 136 Perceived Channel Characteristics 138 Individual/Demographic Variables 139 Predicting Social Use 140 System Access 141 Task/Environment 141 Perceived Channel Characteristics 142 Individual/Demographics 143 Summary - Predicting Purposes of Use 143 Discriminating Among Types of Users 144 Theoretical Implications 145 Practical Implications 148 Limitations 149 Future Research 152 REFERENCES 154 APPENDICES 162 APPENDIX A: MOTIVATIONAL LETTERS 163 APPENDIX B: SAMPLE QUESTIONNAIRE 166 .vi_ LIST OF TABLES Table 2.1 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 Summary of Expected Relationships of Independent Variables to Task and Non-Task Related Uses of Electronic Mail Descriptive Statistics for Electronic Mail Use Frequency Job Types of Surveyed Electronic Mail Users Summary Statistics for EM Purposes of Use Items Factor Analysis Results for Purposes of Use Items Inter-item Correlations, Scale Reliability and Summary Statistics for Purposes of Use Scales Surtmary Statistics for Task Characteristic Items Factor Analysis Results for Task Characteristic Items Inter-item Correlations, Reliability, and Summary Statistics for Task Characteristics Perceived Environmental Uncertainty and Cross Boundary/Location Communication Need Summary Statistics Surtmary Statistics for Task Interdependence Items Factor Analysis Results for Task Interdependence Items Inter-item Correlations, Reliability and Summary Statistics for Intra and Inter Unit Task Interdependence Scales Summary Statistics for Perceived Channel Characteristics Factor Analysis Results for Perceived Channel Characteristics Inter-item Correlations, Reliability and Summary Statistics for Perceived Channel Characteristic Scales Summary Statistics for Individual/Demographic Variables Page 49 53 60 62 69 71 73 75 76 78 80 81 82 84 85 87 89 vii- 3.17 Summary Statistics for System Access Variables 91 4.1 Correlations of All Independent and Dependent 100 Variables 4.2 Task Use Regression Results 102 4.3 Social Use Regression Results 108 4.4 Summary of Expected and Obtained Relationships of 113 Independent Variables to Task and Non-Task (Social) Related Uses of Electronic Mail 4.5 Job Types of Electronic Mail User Groups 116 4.6 Number of Messages Sent and Received By User Groups 116 4.7 Canonical Discriminant Function Sunraary Statistics 118 4.8 Rotated Standardized Discriminant Function 118 Coefficients 4.9 Varimax Rotated Matrix of Correlations of Independent 120 Variables with Discriminant Functions 4.10 Group Centroids in the Discriminant Space 122 4.11 Classification Analysis Results 123 5.1 Proportion of Respondents Who Report at Least Some Use 127 of Electronic Mail for Each Purpose viii LIST OF FIGURES Figure Page 4.1 Plots of Group Centroids for Function 1 (X Axis) 122 by Function 2 (Y Axis) CHAPTER ONE Statement of the Problem Over the past several decades, we have witnessed dramatic technological advances in both the communications and computing fields. At the intersection of these areas can be found many "convergent" technologies, so named because they utilize both the capabilities of the computer to process information, and telecommunication systems to transmit information. Emerging videotex systems are one example, using the computer to store and process requests for information, and using a variety of transmission media, ranging from ordinary telephone lines to coaxial cable and fiber optics, to deliver information to consumers. Perhaps one of the more widely accepted "convergent" technologies is the use of computer and telecommunication systems for the purpose of facilitating human interaction. Most commonly known as electronic mail (1), these systems generally allow someone to type some form of message using a microcomputer or terminal, and send it to another computer that routes the message to the appropriate recipient or recipients. These messages then remain stored in an "electronic mailbox" until the recipient is ready to display them on his or her terminal. This short description of electronic mail sounds innocuous enough. However, research from a social science perspective on this new 71) The term electronic mail has been used to represent a wide variety of electronic forms of communication, including telex and twx services, telegrams, and computer-based message systems. In this document, electronic mail will be used to represent a computer based communication system, where a computer is used to enter, store, transmit, and display text between individuals or groups of individuals using terminals or microcomputers connected to a mainframe or network. 1 communication technology is important at this time for several reasons. 1) Electronic mail systems are gaining acceptance in a large number of organizations. 2) The attributes of electronic mail hold important implications for the human communication process in organizations, and lave lead to predictions and occasional findings of a wide range of social impacts. 3) Communication theorists have largely ignored Llectronic mail, particularly in the development of theory and research on organizational communication processes. When research on computer- mediated communication is conducted, it suffers from important limitations, particularly with regard to conceptualization of the "uses" of electronic mail. Each of these points is elaborated upon below. The Acceptance of Electronic Mail Many organizations, including most members of the Fortune 1000, seeking to increase productivity among employees whose work revolves around the manipulation and communication of information, either plan to or now employ some form of electronic mail (Today's Office, August, 1982). Over 70 electronic mail systems are commercially available; many used by some of the country's largest corporations (Hannagan et al., 1982). Mded to this are private, in-house systems that have been developed on the organizations' mainframe computer system. A look at the product offerings of most word processing and microcomputer vendors highlights the interest in electronic mail, as most offer "communications" capability, with some explicitly containing a "mail" option. Market forecasts can now be found that predict - exponential growth of electronic mail equipment sales, reaching the multi-billion 2 dollar mark over the next fifteen years (Predicasts Research Group, *1982) . Universities and private research institutions also contain many electronic mail users. Several years ago, for example, the Advanced Research Projects Agency of the Defense Department funded the creation of a network linking together mainframe computers at research institutions around the country. The idea was to expand the computing power of any one particular computer by allowing users access to programs on other computers in the network. Called Arpanet, it was soon discovered that the primary use of the network was for interpersonal communication among users at different installations (Marvin, 1983). Even home computer users now have access to electronic mail systems other than those available on their workplace computer. Publicly available information utilities like The Source and CompuServe offer a variety of information services to consumers for a fee. Included in their offerings are such services as access to remote databases, teleshopping, telebanking, and electronic mail. Even telephone and data communications companies like General Telephone and MCI now offer electronic mail. The Attributes and Impacts of Electronic Mail Communication via computer has been characterized as the cornerstone of the "office of the future," with the potential for a wide variety of impacts on communication processes, people, and organizations (Bair, 1979; Evans, 1979; Hiltz and Turoff, 1978; Nilles, 3 1982; Rice, 1981; Olson and Lucas, 1982; Uhlig et al., 1981; Williams, 1982). It is useful to examine some of the attributes of electronic mail and see how they might influence communication processes, leading to subsequent impacts on individuals and organizations. The list below is not meant to be exhaustive; rather the intention is to illustrate some of the current thinking about electronic mail and build a case for the importance of conducting research in this area. Communication is asynchronous. The store and forward capability of computer mail systems allows people to send and receive their messages when convenient. Generally people do not interact via electronic mail in real time. This attribute has a number of implications for the communications process. First, it has been heralded as the answer to "telephone tag," since people are not bound by their schedules. A sender can type and mail correspondence when he or she thinks of it, no matter what time it is, and can be reasonably certain that the message will be delivered. Recipients will find the message waiting for them the next time they access the mail system. The primary impact often mentioned here is one of increased productivity, as people communicate when they feel the need and avoid wasting time trying to reach people by telephone (Bair; 1979; Uhlig et al, 1981). Also mentioned are impacts on the time spent using the telephone and other print media, as more correspondence is transmitted via electronic mail (Tapscott, 1981). A second implication of asynchronous communication is that it allows people to communicate at times outside the normal workday, leading to shifts in working patterns (Johansen and DeGrasse, 1979). Finally, since messages are composed in 4 I ; ----- ; --- ^---- ------------ ---- -— —---- ---------------- --- --- ---—— — private with no time constraints or distraction from receivers, more care can be taken in wording, supposedly enhancing the accuracy of Lommunication (Hiltz and Turoff, 1978). Communication is unaffected by distance. Because the electronic signals are transmitted immediately, people at opposite ends of the ^country can send messages to each other as easily as those in the same office (given the appropriate network configuration). This attribute, in combination with asynchrony, has lead some to predict impacts on organizational structure. In particular, increased geographic lispersion and greater cross-locational collaboration on various projects are viewed as potential outcomes of the implementation of electronic mail (Johansen and DeGrasse, 1979; Nilles, 1982). Nilles (1982) further predicts the increased frequency of "teleworking," where people remain at home or in neighborhood work centers rather than commuting to some central location to work. Electronic mail can be both an interpersonal and a group communication medium. Messages can be sent privately to one or a few persons, or broadcast to a large number of recipients. Many systems, in fact, allow the definition of distribution lists, whereby a list of people can be given some name reflecting the nature of the affiliation, and messages can then be addressed to the distribution list. This attribute can be related to predictions of increased span of control among management as it eases the burden of communicating with subordinates (Price, 1979), and to further impacts on organizational communication structure. Wynn (1980) and Kiesler (1981) have noted the formation of "electronic communities of interest" made possible by 5 distribution list capabilities. Ibis feature, in fact, blurs traditional distinctions between electronic mail and computer conferencing (Hiltz and Turoff, 1978). Communications are in the text mode. Communication researchers jfeel that the limited nonverbal interaction, and lack of immediate feedback make this medium less appropriate for emotional or dnterpersonally involving activities (Short et al., 1976) A related impact hypothesized here concerns the depersonalization of the workplace (Olson and Lucas, 1982), although Hiltz and Kerr (1981) report that computer-mediated communication enhances feelings of cohesiveness among and between work groups. It is clear from this sampling that many feel the diffusion of computer-mediated communication portends a wide range of impacts on organizational life. In addition, many of these potential consequences have not been anticipated by the designers of computer-mediated jcommunication systems. While this heightens the importance of conducting empirical research on the social impacts of the technology, intermediate research on the nature of electronic mail use is a necessary precursor to impact analysis. This point is developed further in the next section. Limitations of Prior Research Despite its relatively widespread acceptance, particularly in organizational settings, only a handful of organizational communication researchers have focused upon electronic mail. Indeed, most research on the nature of communication in organizational settings ignores the 6 notion of medium as an important variable in the communication process (Porter and Roberts, 1976). When the concept of medium does enter the organizational communication picture, it typically is constrained to face-to-face, telephone, and print channels. Further gaps in the literature are exposed when specific research on electronic mail is examined. Generally, researchers have been concerned primarily with factors that help predict acceptance and use of electronic mail (Hiltz and Kerr, 1981), and how use of electronic mail affects the use of other media; particularly the telephone (Bair, 1979; Rice and Case, 1981; Tapscott, 1981). The notion of "use" typically means how many messages a person sends or receives (Craig, 1981; Hiltz, 1978; Hiltz and Turoff, 1981; Lippitt, 1980; Tapscott, 1981). This unfortunately tells us very little about what people actually "do" when using electronic mail, or perhaps more importantly, for what applications they are willing to employ this new medium. The few studies that do examine different "purposes" of use typically employ a typology derived from the literature on teleconferencing (e.g. Rice and Case, 1981); a typology that is not reflective of the ongoing use of electronic mail for day-to-day communication in organizations. This lack of interest in how people use electronic mail is somewhat surprising, as specific impacts of electronic mail use might well be predicated on the nature of use. Indeed, taking a usage perspective provides a realistic buffer against which the popular notions of powerful impacts can be evaluated. For example, someone who only employs the system for occasional greetings to friends is not likely to experience drastic shifts in working patterns. On the other 7 hand, they may experience greater perceived cohesiveness among organizational members. This argument has been convincingly advanced by Blumler (1979), in his rationale for employing a uses and gratifications perspective to the study of media effects. In a similar fashion Dutton et al., (1983) have argued for the careful study of how home computers are used, in order to approach the study of effects in a more informed manner. Study Overview Research Questions This study was designed to address the "uses" of electronic mail gap, by examining more closely the purposes for which people use the technology. As in the uses and gratifications approach to studying media, however, an analysis of the uses is only a start. To paraphase [proponents of this perspective, a more complete model would investigate the social and psychological origins that give rise to needs, as it is these independent factors that lead to patterns of media use (Katz, Blumler and Gurevitch, 1973). A large and diverse body of literature, arising from several different disciplines can be brought to bear on the question of social and psychological origins of the purposes for electronic mail use in organizational settings. The goals of this research can thus be characterized by the following broad questions. 1. What are the purposes for which people employ electronic mail in organizational settings? 2. What independent factors lead to different patterns of use? 8 Findings - Purposes of Use — — A large office products vendor served as a research site. The electronic mail system had been in place for seven years at the time of data collection, and had over 2000 users. Interviews with two dozen users, coupled with a several month observation period lead to the development of a list of over 30 purposes of use. A survey of 220 users using self-administered questionnaires built upon interviews and observations addressed both research questions in a more quantitative 'fashion. The electronic mail system was used for a variety of purposes; some anticipated by prior studies of electronic mail, but many bearing little resemblence to the standard teleconferencing-based typologies. A search for underlying dimensions of use uncovered task and socio- emotional use clusters. Task-related uses were reported as occurring more frequently than socio-emotional uses, but this might be a reflection of the frequency with which these communication situations arise in the day-to-day organizational life rather than a clear bias against using electronic mail for more interpersonally involving applications. Findings - Predicting Patterns of Use Theoretical and empirical literature from the organizational theory, organizational communication, teleconferencing, and computer- mediated communication areas were brought to bear on the development of factors that would predict patterns of electronic mail usage. Four major factors were identified: task and environment, perceived channel 9 characteristics, demographic, and system access. Hypotheses specified the direction of relationship of individual variables within each of these areas to task and non-task related uses, in general, greater task non-routineness, interdependence, environmental uncertainty, and time pressures, along with an increased need for communication across locations were expected to be associated with greater task use, but were not expected to relate to the non-task uses. On the other hand, perceived channel characteristics, including perceived utility, ease of use, social presence, and privacy were expected to be positively associated with both task and non-task uses. The third cluster of hypotheses expressed an expected positive relationship of educational level and experience with electronic mail systems to both task and non task uses of electronic mail, while age, level in the organization, and organizational tenure were expected to show a negative relationship. Finally, a set of system access variables, reflecting access to both relevant others via electronic mail, and to terminals were considered necessary, but not sufficient precursors to all types of electronic mail use. Multiple regression analyses were used to test these hypotheses, with indices of the amount of task and socio-emotional use serving as dependent variables. The amount of task use was primarily a function of a well developed electronic mail infrastructure, reflected by the importance of having coworkers outside the work group on the system, and not having to share terminals with others. The importance of a clear communication need, reflected by the existence of coworkers in other organizational locations was also significantly related to task use. Perceived utility showed a significant relationship to task use, that was probably indicative of an interdependence of use and attitudes. That is, the more one uses the system for task uses and finds it effective, the more likely he or she is to perceive it as .useful, leading to further task usage. Other perceived characteristics showed significant zero-order correlations, but appeared to influence usage by further affecting perceived utility. Task and environment variables did not relate significantly to task usage. The amount of socio-emotional usage was primarily a function of demographic characteristics. In particular, newer and younger employees were more likely to employ electronic mail for socio- emotional purposes. The perceived social presence and utility of electronic mail also related significantly here. One explanation for this pattern of findings is that the electronic mail system was viewed as an effective means of becoming integrated into the informal social networks within the organization by new employees. A followup analysis attempted to categorize users into either high or low task and socio-emotional use groups. By using mean scores on the task and socio-emotional indices, respondents were placed into one of four groups: 1) light users, low on both types of use, 2) primarily social users, 3) primarily task users, and 4) task and social users, heavy on both types. The largest group was this fourth one, demonstrating the integration of task and socio-emotional communication in organizations. Furthermore, a discriminant analysis supported prior regression analysis findings, highlighting similar variables as important in discriminating between the four categories. 11 This research represents an important first step in the study of computer-mediated communication systems in organizations. The relevance of much prior organizational and communication theory to the study of emerging communication technologies was highlighted. The importance of this study for communication practice should also be noted. Both theoretical and practical significance are discussed below. Theoretical Significance Theoretical significance is derived from this research in many ways. At a broad level, the organizational communication area is advanced through the study's focus on applications or functions of communication in an organizational setting via electronic mail. The nature of both task-related and socio-emotional communication uncovered in this research adds to our understanding of the functions of communication in organizations (cf. Farace, Monge, and Russell, 1978; Rogers and Agarwala-Rogers, 1979). Theories which focus on the choice among media, and the appropriateness of different media for varying tasks are also enhanced (Irving, 1982; Short, Williams, and Christie, 1976). Constructs such as social presence, which in the last decade have dominated theories of media appropriateness for interpersonal communication activities are examined in a new light. A final theoretical area informed by this research concerns the influence of the organization context on communication. A number of authors have focused on aspects of the task environment and 12 characteristics of individuals' tasks as they influence the need for communication (Galbraith, 1969, 1973; Ruchinskas, 1982; Thompson, 1967; rushman and Nadler, 1979). Others focus on an individual's role or position in the organization as an important influence on communication (Lawler, Porter, and Tennenbaum, 1968; Mintzberg, 1973; Ruchinskas, 1982). Both sets of variables are assessed in conjunction with individuals' perceptions of electronic mail attributes as predictors of types of electronic mail use. Finally, theoretical perspectives on the adoption and use of innovations are indirectly informed. Electronic mail is still a relatively new communication technology, even for many users participating in this study. Examining factors that might lead to varying uses of this new medium provides important insights into the innovation utilization process. Practical Significance Important contributions to the management and use of cmmunication technologies are derived from this study. Individual users of electronic mail learn of the many ways in which they may take advantage of its unique attributes. Current and future managers of electronic mail gain insights into how their system may be used. Furthermore, understanding the factors which influence different kinds of uses is an important first step toward creating policies which encourage desired uses. Finally, this research explores some uses which initially may be thought of as undesirable, but upon closer inspection may indeed be functional applications. Such counterintuitive research findings are 13 valuable as a guard against the formation of policies based on uninformed judgements. 14 CHAPTER TWO Review of the Literature and Development of Hypotheses Literature from many disciplines may be brought to bear to help understand how people use electronic mail and to uncover factors which influence usage. Research on the kinds of communication activities that are associated with communication in organizations via different media played an important role in the development of the dependent variable (Short et al., 1976). Further refinements in thinking about the purposes of electronic mail use were possible upon examination of literature that more specifically addresses communication activities appropriate to this medium (Hiltz and Kerr, 1981; Johansen, Vallee, and Spangler, 1979; Kiesler, 1982; Phillips, 1982; Rice, 1980; Rice and Case, 1981). Independent variables, or factors which influence types of electronic mail uses, also have been culled from a diverse range of literature. At the most basic level, variables which have been associated with the need for communication receive attention. This research tends to be non-medium specific, and focuses on task-related communication. However, it is argued that without adequate need for communication, particularly of a task-related nature, little use of any medium, including electronic mail, will occur (O'Reilly and Pondy, j 1979). Therefore, factors which influence communication requirements are relevant to the question of electronic mail usage. Research which outlines factors which influence the use of different media generally and electronic mail specifically is, of 15 course, also important~in the selection of “independent variables. Thfs research examines both of the above sets of independent variables jointly in an attempt to explain variation in electronic mail usage. The Dependent Variable - Purposes of Electronic Mail Use To speak of purposes of electronic mail use, a functional orientation is evoked. The relevant issue, then, is not how many messages a person sends or receives, but rather what he or she tries to accomplish through the use of electronic mail, and how often. At the most basic level, this implies an intentional or purposeful definition of communication as opposed to definitions which simply involve any transfer of patterned matter or energy (Berio, 1960). It further injects some notion of content, albeit in a very general way, into the use of electronic mail or other computer-mediated communication systems. Many prior studies have treated usage simply from the perspective of the number of messages sent and received (Craig, 1981; Hiltz, 1978; Lippitt, 1980; Tapscott, 1980), or in terms of the pattern of linkages created by use (Freeman and Freeman, 1979; Hiltz, 1978; Rice, 1982). In fact, Rice (1980), in a comprehensive review of the research on computer-mediated communication, concluded that "it would be fruitful to determine just what it is that people do when they interact via computer." (p. 238) Research examining the communication activities occurring over different media was initiated to a large extent by the Communication Studies Group in Britain. In one study known as DACOM (Description and Classification of Meetings), Pye, Champness, Collins, and Connell 16 (1973) clustered various functions and activities associated with business meetings. Such activities and functions as negotiation, presentation of reports, delegation of work, information seeking, problem solving, forming impressions of others, and giving information to keep someone in the picture joined together in various combinations to form 12 unique meeting clusters. A followup study by Connell (1974), using a larger and more representative sample of business people, arrived at a similar classification of meeting clusters. Once they had developed the typology of meetings, the Communication Studies Group attempted to determine the appropriateness or substitutability of different teleconferencing alternatives for various communication activities. In this fashion, they hoped to arrive at estimates of the proportion of all business meetings for which teleconferencing technologies were a viable alternative to travel. The eventual outcome of much of this research was the development of a theory of social presence, which focused on the level of interpersonal involvement various communication tasks or activities require, and the inherent abilities of different media to support interpersonal involvement (Short et al., 1976). 'She crux of this theory postulates that people recognize that media differ in the amount of social presence they afford, and choose to use a medium based on the degree to which social presence is necessary for the particular communication task. Other researchers followed the CSG's lead, and examined time spent on various communication activities in teleconferences (Dutton, Fulk and Steinfield, 1982; Noll, 1977). Most of this research dealt ---------- _17_ primarily with audio, video, and face-to-face interaction in a group meeting situation. Those that considered computer-mediated communication typically focused on computer conferencing rather than electronic mail (Johansen et al., 1979). In a review of research on the appropriate applications for computer conferencing, Rice (1980) concludes that it is good for such activities as: 1) exchanging information, 2) asking questions, 3) exchanging opinions, 4) staying in touch, and 5) generating ideas. Following the theoretical developments of Short et al., (1976), these were felt to be tasks that do not require high interpersonal involvement. Tasks that do require interpersonal involvement and thus were felt to be inappropriate for computer conferencing included: 1) bargaining, 2) resolving disagreements, 3) getting to know someone, and 4) tasks requiring constant focused discussion. The everyday use of electronic mail, however, covers a variety of situations other than normal business meetings, and therefore should include a different set of purposes of use. Interviews and observations in the organization hosting this research revealed communication activities which occur in group as well as dyadic situations, and which reflect non-task as well as task communication. Some group uses not found in prior typologies include broadcasting information requests, opinion polling, and asking questions in a public setting to force a response or action. Non-task uses uncovered in interviews include advertising, organizing social activities, playing games, learning about interesting things or events, taking breaks from work, and filling up free time. The full range of uses employed in 18 this research are described in the next chapter on methods. In contrast to this diverse range of uses, in research examining the purposes of electronic mail usage, typologies based upon the teleconferencing literature are inevitably used. Rice and Case (1981), for example, employed the following ten "tasks": 1) exchanging information, 2) asking questions, 3) exchanging opinions, 4) staying in touch, 5) generating ideas, 6) decision making, 7) exchanging confidential information, 8) resolving disagreements, 9) getting to know someone, and 10) bargaining/negotiating. While such task typologies are useful for making media comparisons, particularly between potential group meeting substitutes, they do not capture the full richness of applications of an ongoing organizational electronic mail system. In addition, they include few more socially oriented uses beyond staying in touch or getting to know someone, both of which seem quite utilitarian. A few authors have observed people using computer- mediated communication for "social" purposes (Hiltz, 1978; Kiesler, 1982; Phillips, 1982; Wynn, 1980). Organizational theorists have also noted that much communication in organizations is non-task related (O'Reilly and Pondy, 1979). Another criticism that may be leveled at prior computer conferencing and electronic mail research indirectly relates to types of applications. That is, small numbers of subjects representing primarily academic and research personnel were often placed into artificially created communication situations with specific goals (Freeman and Freeman, 1979; Hiltz and Turoff, 1981; Johansen and DeGrasse, 1979). This again limits generalizability to the broader 19 population of electronic mail users in organizational settings. In summary, a review of prior theory and research on the purposes of use of telecommunication technologies in general and computer- mediated communication systems in particular reveals important ^weaknesses. First, the teleconferencing communication typologies based on the efforts of the Communication Studies Group, while useful for stimulating the focus on communication activities, are inadequate for the purpose of describing an ongoing organizational electronic mail system. Second, the bias has been somewhat toward looking at meeting ^situations, and task-related communication. Finally, it is difficult to generalize from much of the past research given the limitations with regard to sampling and atypical settings. Independent Variables Variables that are hypothesized to affect the nature and extent of electronic mail use are both those which have been associated with organizational communication in general and computer-mediated communication in particular. Four sets of independent variables have 'been chosen from previous literature for inclusion in this study. First, a cluster of variables dealing with the task environment and individuals' task characteristics form one set. A second is comprised of perceived attributes of electronic mail. Demographic variables constitute the third set of independent variables. Finally, system access variables reflecting access to both terminals and relevant people are deemed necessary but not sufficient for electronic mail use. These are considered as control variables. 20 The multivariate model predicting uses of electronic mail is consistent with recent trends in communication research (Goldhaber, jYates, Porter, and Lesniak, 1978; Monge, Edwards, and Kirste, 1978; <Ruchinskas, 1982; Svenning, 1982). It avoids the more univariate tendencies of system designers and vendors, who often seem to focus entirely on the capabilities of their system as the primary determinant of use. While the system feature or physical attribute argument has some merit in explaining use, it should not be employed in isolation. In essence, this approach may be considered "technologically deterministic" or "rationalistic." However, a number of authors assessing the implementation and use of computing in organizations have criticized this approach, noting the importance of social and political influences in shaping the use of new computer and communication technologies (Dutton, 1982; Dutton, Kovaric, and Steinfield, 1983; Kling, 1980; Kraemer, Dutton, and Northrop, 1981). Wynn (1980), for example, calls the unique ability of electronic mail systems to form distribution lists one of their most powerful communication features. However, she also notes that use of this capability can easily be counteracted by organizational policy. An approach which more closely approximates the multiple sources of influence on the use of computers and communication is recommended by Dutton et al., (1983). For these authors, use will be influenced conjointly by system, social environment, and personal factors. These three factors correspond closely with the task and environment (social), system access and perceived channel characteristics (system), and individual/demographic (personal) variable sets chosen for this research. 21 Hypotheses for each of the independent variable clusters will be presented at the beginning of each section. Although it is believed that certain variables might differentially influence usage depending on which type of use in question, hypotheses cannot be very specific prior to the empirical analysis of usage dimensions. A general task vs. non-task use dichotomy is employed since many independent variables rave only been related to task communication in the literature. ,Task Characteristics/Task Environment HI: Task routineness will be negatively associated with task uses of electronic mail. H2: Task interdependence will be positively associated with task uses of electronic mail. H3: Environmental uncertainty will be positively associated with task uses of electronic mail. H4: Cross location communication need will be positively associated with task uses of electronic mail. Organization theorists in the past have tended to implicitly rather than explicitly recognize communication as a key area of study. It seemed to go without saying that some form of communication is necessary to provide coordination if organizations were to capitalize on the economies and efficiencies of the division of labor. Explicit references to communication did arise in the human relations school of thought, primarily focusing on the development of informal 22 communication networks among lower level workers and its effects on worker morale (Mayo, 1945; Roethlisberger & Dickson, 1939). These informal networks were often thought of as existing in opposition to the formal structure, however, and conceptualizations integrating formal structure and informal networks into a common cooperative framework were rare. Several modern approaches to organizations do integrate the formal and informal structure, explicitly treating informal or lateral commmunication as a tool for coordination and problem solving. They are for the most part concerned with the design of organizations; i.e. •how an organization should be structured. These approaches are bound together by a number of common assumptions. First, they view organizations as open systems; the organization interacts with its environment by transporting inputs and outputs across its boundaries (Katz and Kahn, 1978). Second, organizations are faced with uncertainty from a variety of sources, both internal and external. This uncertainty must be processed if the organization is to survive and prosper. Third, different organizational structures seem to be more effective for varying levels of uncertainty (Burns and Stalker, 1961; Lawrence and Lorsch, 1967; Thompson, 1967). In particular, these and other researchers have found that organismic structures are more effective for coping with very uncertain environments, while mechanistic structures are more effective in certain or stable environments. Organismic structures are characterized by highly interconnected units. They are often thought of as being less formalized and less centralized in terms of power and control. The 23 smphasis is more on lateral communication than vertical. On the other land, mechanistic structures are more formalized and centralized, with the emphasis on vertical communication. This notion will be expanded upon in the next two sections by focusing upon two specific approaches to organization design. )Technology, Task Interdependence, and Coordination. Technology theorists focus on the techniques or transformation processes through which the organization changes inputs to outputs. An important premise of the technology school is that technology constrains organization structure in predictable ways. There are several influential typologies of organizational technology (cf. Perrow, 1967; Thompson, 1967; Woodward, 1965). A typology developed by Thompson (1967) has generated a great deal of empirical research and will form the focus of this discussion. Thompson outlined three types of technologies employed by modern organizations; mediating, long linked, and intensive. They are treated separately for analytical purposes; in most modern complex organizations all three types may be found. The mediating technology is one where the organization mediates between clients in its environment in a standardized fashion. Banks bring together borrowers and lenders of money. Similarly, telephone companies connect individuals and organizations wishing to communicate with each other. Long linked technologies reflect a serial transformation process. Assembly lines are most often used as an example. The product flows serially from one unit or individual to the 24 next on its way to completion and subsequent release into the ^environment. Finally, intensive technologies occur when organizations operate on inputs in a customized way. The transformation process is complex and tailored to each input. An example used by Thompson to lepict an intensive technology is the treatment of mental illnesss by psychologists and psychiatrists. Technology constrains structure, according to these theorists, because of the nature of the interdependence it creates among organizational units. As the technology becomes more complex, differenct forms of interdependence occur. Different forms of interdependence then have implications for the most appropriate coordination strategies. Three types of interdependence noted by Thompson are pooled, sequential, and reciprocal. With pooled interdependence, organizational units are dependent upon each other only by virtue of their individual impact on the overall well being of the organization. An example of pooled interdependence is the case of separate branch offices. Inputs do not pass from one branch to the next. Each branch operates on its own inputs, transforming them to the final product. Dependence exists, however, as poor performance of one unit affects the availability of resources for other units. Sequential interdependence occurs when the outputs from one unit become the inputs of another. It is a one directional flow. Lastly, reciprocal interdependence occurs when the outputs of one unit are another's inputs, and vice versa. There is a bi- or multi-directional work flow. These categories are conceptualized to form a Guttman scale. That is, each new type 25 represents a more complex situation and contains the prior form of interdependence. In addition, pooled interdependence can be related to mediating technologies, sequential to long linked technologies, and reciprocal to intensive technologies. When the actions of one individual or group affect or are (dependent upon another, some method of coordinating their activities is necessary. March & Simon (1958) describe three ways to achieve coordinated action between organizational units; standardization, programming or planning, and feedback. Thompson relates these to the three forms of interdependence; with pooled interdependence most efficiently coordinated by standardization, sequential by planning, and reciprocal by mutual adjustment or feedback. These three coordination modes are mentioned in order of increasing cost. More complex interdependence requires a more costly coordination strategy. Returning to implications for organizational communication patterns, this conceptualization leads us to expect that as the transformation process becomes more complex, the use of communication as a tool for coordination between organizational units increases. It is important to note that the preceding discussion has been directed at the organizational level. Organizations were conceptualized to employ a core technology, creating a consistent pattern of interdependence and requisite coordination modes. Thompson realized, as have other researchers, that in one organization a variety of technologies are to be found, with each creating different patterns of interdependence. This has led some researchers to apply Thompson's conceptualization to the department or division level of analysis (Mahoney & Frost, 1974). 26 T h u s , the pattern of communication between organizational units may not be distributed uniformly throughout the organization. The conclusion remains, however, that when units in some subsection of an organization employ complex technologies to transform inputs to outputs and thus are linked by reciprocal interdependence, a more decentralized communication structure is necessary. An Information Processing Model of Organization Design. Theorists in this area regard the organization as an information processing entity (Galbraith, 1973; 1977, Tushman and Nadler,1978). Organizational members are faced with varying levels of uncertainty in the pursuit of task execution, which must be reduced if the organization is to function effectively. Uncertainty is conceptualized as the difference between the information available and the information required to do a task (Galbraith, 1973). According to this approach, under conditions of higher task uncertainty, more information must be processed by decision makers in order to achieve a given level of performance. Galbraith (1973, 1977) outlines in considerable detail strategies which organizations may undertake to handle greater task uncertainty. They may act to reduce the need for information processing by creating slack resources or creating self-contained units. Several problems are inherent in either of these strategies. Creating slack resources often involves lowering the level of performance required, which may not be acceptable to customers or clients. Creating self-contained units reduces coordination needs by incorporating all necessary functional specialties in each task group, but is a highly inefficient use of 27 Jscarce resources. There may not be enough work to keep a trained specialist busy in just one work group, yet costs of such specialists prohibit their remaining unoccupied. Rather, their skills are best utilized when time-shared between groups. Coordination of schedules and activities is then necessary, defeating the purpose of creating self- contained units. The alternative strategies serve to increase an organization's capacity to process information. These are to implement or improve vertical information systems, or to create lateral relations at critical points of interdependence. Vertical information systems follow centralized control and decision making, but essentially are restricted to information that is capable of being systematically collected, stored and retrieved. This may not be possible or practical for many types of information used in solving problems or making decisions. The creation of lateral relations is a costly but effective method of increasing information processing capacity. Capacity is increased due to a number of reasons. First, it enables people to directly access the source of needed information. This entails fewer indirect links in the transfer of information, thereby potentially improving accuracy and timeliness. Second, rather than sending information up the hierarchy to a common supervisor, problems due to task interdependencies are dealt with by direct horizontal communication. These both serve to reduce managerial overload by lowering the number of problems or requests sent up the hierarchy. Third, the increased connectedness among organizational units implied by greater lateral - 28 relations means that there are more possible paths by which information is transmitted. The total amount of information available to an organization member is not only increased, but distribution and error detection is improved (Tushman and Nadler, 1978). Building upon the work of Galbraith (1973, 1977), Tushman and Nadler (1978) discuss three possible sources of work related uncertainty. Their analysis is directed at the subunit level based on the realization that subunits performing different kinds of tasks face and have control over varying levels of uncertainty (Hickson et al., 1971). The first source of uncertainty consists of the subunit's task characteristics in terms of task complexity and intra-unit interdependence. Tasks which are more complex are those considered to be nonroutine, unpredictable, and not well understood. This formulation borrows a great deal from conceptions of technology and task characteristics (Perrow, 1967; Hrebiniak, 1974; Hage and Aiken, 1968) . A second source, reflecting an open systems approach is the task environment. A variety of dimensions have been used to characterize the environment, and a debate over whether perceptual or objective indices should be used has continued (Downey et al., 1975; Downey and Slocum, 1975; Tosi, Aldag, and Storey, 1973). Weick has brought the notion of enacted environment to bear on this issue, suggesting that it is important to focus on those aspects of the environment attended to or enacted by organization members if we are to understand their actions. Tushman and Nadler (1978), following Duncan (1972), settle on a static/dynamic dimension of perceived environmental uncertainty, _____________________ 29 reasoning that if the environment is stable then fixed rules or standard operating procedures can be used as modes of coordination. When the environment is unstable or changing, information about the environment needs to be disseminated through the parts of the organization affected by those changes. In addition, as Tushman (1977) notes, increased cross boundary communication enhances innovation and problem solving, so necessary if the organization is to remain viable in the face of a changing environment (Rogers and Agarwala-Rogers, 1976) . Inter-unit task interdependence forms the third contributor to work related uncertainty. Based on the work of Thompson (1967), these theorists also note how increasing interdependence has implications for the need for information exchange and collaboration between subunits. Some empirical evidence exists to support this relationship (Lawrence and Lorsch, 1967; Van de Ven et al., 1976). These sets of factors affect work related uncertainty which in turn affects the need for information. Thus as work related uncertainty increases, information processing requirements increase. And, like Galbraith, the establishment of horizontal communication relationships is seen as an effective means of increasing information processing capacity. The present research diverges from much of this prior literature in one important way. An individual level of analysis as opposed to subunit, department, or organization level is employed. The notions of Galbraith (1973, 1977) of the importance of people directly accessing sources of needed information, and avoiding hierarchal channel overload 30 by direct horizontal communication seem consistent with an individual level approach. Furthermore, as will be discussed below, electronic mail is uniquely suited to enable direct cross boundary communication. Another difference from the research reviewed above should be noted. The above variables are proposed to affect the amount of task communication, particularly between organizational units. More specific to the use of telecommunication alternatives is a direct measure of the need for communication across locations. Proxemics has been found to be a primary influence on the use of mediated communication (Conrath, 1973; Irving, 1981; Ruchinskas, 1982; Steinfield, 1981; Steinfield, 1983). Hiltz and Kerr (1980) list geographic dispersion as one of the more important determinants of computer-mediated communication. Others have found that through the use of computer-mediated communication systems, geographically dispersed people are able to function as a group (Johansen and DeGrasse, 1979; Freeman and Freeman, 1979). Therefore, electronic mail use is proposed to be a function of the need for cross locational or boundary communication. Communication Media and Inter-unit Communication. In the preceding discussion, the notion of cross boundary communication was presented as a design strategy that can enable organizations to deal more effectively with complex and interdependent tasks and uncertain environments. Unfortunately, establishing communication relationships across organizational boundaries is often problematic. Researchers have noted that this type of interaction is difficult and prone to .............. ; ____________________ 3i distortion (Dearborn and Simon, 1958; March and Simon, 1958; Wilensky, 1967). There are several reasons why this is so. March and Simon have observed that people in different organizational subunits often create their own language or jargon in tthe interests of communication efficiency. Although this improves communication within the unit, it makes it more difficult to communicate with those in other units having other languages. A second problem leading to distortions in the communication process is that different units may have different and potentially ^conflicting subgoals (Simon, 1964; Wilensky, 1967). For example, they may be competing for the same resources or control over the same area. It is also important to consider that traditional media suffer ■from various constraints to the flow of information. Interactive media such as the telephone or face-to-face communication are subject to scheduling constraints. Communicating individuals have to be available at the same time. The percent of unsuccessful calls reported in business setting attests to the difficulty in coordinating schedules (Uhlig, Farber, and Bair, 1979). In addition, telephone, print, and face-to-face communication often are screened or filtered by secretaries. Face-to-face interaction is further hindered by geographic separation, often necessitating substantial transportation costs. Print also is hindered by geographic separation, leading to delays in delivery. Finally, an important precondition to creating a lateral relation when seeking needed information is awareness of where the information resides. A great deal of an organization's interpersonal resources may ___________________________________ 32 jgo under-utilized because there is no mechanism for bringing together people in need of some expertise and people capable of providing it (Malone, 1981a). It can be argued that attributes of electronic mail help make it relatively free from many of the above constraints. First, many mail systems allow for the creation of groups based upon common interests. These distribution lists often transcend normal organization boundaries. This feature can go a long way toward overcoming problems of distortion since distribution lists can evolve their own languages, goals, and norms (Wynn, 1980). (2) Electronic mail is both instant and asynchronous, freeing it from scheduling and distance constraints. Finally, the use of distribution lists and the broadcast capability of electronic mail systems help overcome the awareness problems. The lists can help direct seekers of information to a group of people who may be likely to possess relevant information. Otherwise, a message broadcast to the organization or subpopulations may also elicit a response from someone containing the needed expertise. Based on the above discussion, it is clear that a simple model predicting organization communication patterns from task and environment characteristics is not adequate. Rather, the communication media available must be considered in any model. In particular, the attributes of electronic mail may make it uniquely responsive to these task and environment characteristics hypothesized to shape communication patterns in an organization. J2) It may, however, lead to a new problem of cross distribution list communication. — --------- . _______ 33. Task and environmental characteristics have only been related to task communication in this discussion. Therefore it is proposed that these variables will not influence non-task uses of electronic mail. In summary, characteristics of the task and environment, in the form of itask routineness, environmental uncertainty, task interdependence, and jcross location communication need will be positively related with task related uses of electronic mail, but will have no association with non- jtask uses. Perceived Channel Characteristics H5: Perceived utility of electronic mail will be positively associated with both task and non-task uses. H6: Perceived ease of use of electronic mail will be positively associated with both task and non-task uses. H7: Perceived social presence afforded by electronic mail will be positively associated with task and non-task uses. H8: Perceived privacy afforded by electronic mail will be positively associated with task and non-task uses. A growing body of literature assesses the extent to which particular attributes of media affect usage. Researchers in this area (recognize that media have different characteristics which may make them more or less desirable in certain situations. This literature can be (organized into two streams of thought; those who emphasize the inherent characteristics of a medium and, more appropriate to this study, those who focus on perceived attributes. __________________________________________________________________ 34 pbjective Channel Attributes. Researchers focusing on inherent characteristics recognize that media have very different physical properties, affecting their appropriateness for some communication activities. For example, Schramm (1973) has pointed out important differences between interpersonal and mass communication channels. Among those mentioned are that channels differ in: 1) the senses they stimulate, with face-to-face channels providing the greatest range of sensory input, 2) their opportunities for feedback, 3) the ability to control the pace of communication (e.g. readers may set their own pace, but film viewers cannot), 4) the extent to which non-verbal communication is transmitted, 5) the multiplicative power (i.e. the number of people to which a message can be addressed, 6) the ability to preserve a message, 7) the power to be selective (i.e. it is easier to Jchange a television signal than tune out a face-to-face conversation, and 8) the power to meet specialized needs. Some authors have attempted to incorporate differences among teleconferencing media into one broad concept, usually referred to as social presence (Short et al., 1976). Essentially, these theorists argue that characteristics of media affect the extent to which people perceive others to be psychologically present when interacting with them over teleconferencing media. Social presence, while subjectively perceived by participants, is thought of by these researchers as a physical characteristic inherent in the medium. They have attempted to demonstrate that people do indeed perceive these differences, and that this affects their choice of media. While social presence is supposed to incorporate the range of differences among media, it typically ends 35 up arraying media in a linear fashion ranging from low to high bandwidth requirements. That is, print is usually lowest in social presence, followed by audio, then video, audio plus video, and finally face-to-face. The channel characteristic argument has also been applied to the use of computers, for communicating as well as for other purposes. Analogous to the notion of channel characteristics affecting use are arguments that differences in system features are responsible for the use of computers. Paisley and Chen (1982) have concluded that certain hardware and software features are more likely to encourage use in their review of a number of different text-oriented computer-based systems. A number of other authors have taken this "human factors" approach to explaining computer usage (see Dutton, Kovaric, and Steinfield, 1983 for a discussion of human factors approaches to the use of computing). For some, the complexity of software and hardware inhibits use (Vallee, 1982). Another approach treats the attributes of computer games as baseline features that promote use, with the argument made that the extent to which other software contains gamelike features will be an important determinant of use (Carroll, 1982; Malone, 1981b). In terms of computer-mediated communication, the system feature approach also has wide currency. In their wide ranging review and synthesis of research on the acceptance, use, and impacts of computer mediated communication systems, Hiltz and Kerr (1981) devote considerable time to discussions of system features that have been found to affect acceptance and use. They list 17 general interface factors of interactive systems and 19 system factors of computer- 36 mediated communication systems in their chapter on system features and usage. Most of the factors reflect built-in characteristics of various Lystems such as the availability of guidance and self-documentation, or I « the inclusion of text editors and formatters. More sophisticated features, such as the ability of the system to allow the formation of special interest communities, are noted here. In one sense, this approach has limited utility for predicting use ■in the present setting, since not only one channel, but one specific system is considered. Objective differences are basically held constant. This literature is relevant to the present research in two ways, however. First, most discussions of the importance of channel characteristics do not differentiate between perceived and objective attributes, and many attributes discussed as fixed or inherent properties can be thought of and measured as subjective factors. Second, it provides a context for the evaluation of the features of the system in use in the surveyed organization. For example, supporters of the objective channel difference approach would argue that electronic mail, as a low bandwidth medium, should not be used for in ter personally involving tasks. However, the existence of a distribution list capability allows certain group communication uses not possible on systems which do not support the formation of distribution lists. Perceived Channel Attributes. Supporters in this camp argue that how people perceive different media will determine use. People vary in how they perceive the same medium; e.g. what is complex for some is easy for others. This argument over perceptual vs objective indices is by 37 -no means constrained to channel characteristics. An analogous debate may be found in the environmental uncertainty literature discussed earlier (Aldrich, 1979). Much as different members of the same organization perceive varying levels of uncertainty in the enviroment (Pfeffer and Salancik, 1978), so do people differ in their ratings of the same channel on various attributes. Several perceived channel characteristics have been explored in past field and experimental work, including utility, accessibility, convenience, ease of use, and personalness or richness of a channel. Studies of information seeking among scientists and engineers provided the context for some of the initial work in this area. With the development of computerized information and retrieval systems, information scientists attempted to understand influences on information source selection. Initial work by Allen (1966) found that the perceived accessibility of a channel was a primary factor determining its use. In fact, he found no relation between the perceived effectiveness of information sources and use. In further research, Gerstberger and Allen (1968) arrived at the interesting conclusion that engineers chose more readily accessible sources of information over those rated to be higher in technical quality. Others also concluded that accessibility or perceived ease of use were the critical dimensions affecting channel choice (Rosenberg, 1966; Werner, 1965). In research on other populations, accessibility and ease of use remain consistent predictors of channel use. In a study of information channel use by farmers, Lee (1968) found convenience to be the most 38 important factor, while managers' use of speakerphones was found to be a function of accessibility (Champness, 1973). Further, availability or accessibility is considered by Edelstein (1973) to be an important determinant of mass media use. Effort required to access sources of information was found to affect consumers' use of information channels (McEwen and Hempel, 1977). More recent research by Steinfield (1981) found perceived availability of face-to-face channels to affect managers’ choice of communication mode for monitoring performance. Finally, literature on the adoption and diffusion of innovations is relevant, given the relative newness of electronic mail as a general communication medium. Often mentioned in this literature is the complexity of the innovation, clearly related to notions of ease of use (Rogers and Shoemaker, 1971; Brief et al., 1976). More specific to teleconferencing, intentions to use video conferencing were found to be a function of perceived ease of use by Svenning (1982). Importantly, ease of use was found to be comprised of perceived accessibility, convenience and ease of use in this study and a companion study by Ruchinskas (1982). Notions of the ease of use of electronic mail may be found in the frequent inclusion of typing skill as an independent variable. Typing ability does not define ease of use, however. Johansen et al., (1979) found it of limited utility in explaining use of computer conferencing and electronic mail. Hiltz and Kerr (1981) report that findings are mixed for this variable, but also note that the friendliness of the interface will be a critical determinant of the use of computer- mediated communication systems. 39 While the accessibility/ease of use/convenience factor often supersedes perceived utility, research has shown that peoples' expectations about the effectiveness of using a channel impacts on the choice to use it. Perceived utility affected consumers' choice of media in research by Fett (1975) and McEwen (1978), while Atkin (1973) employs this concept to explain patterns of information seeking. Innovation theorists also note the importance of perceived utility, as evidenced by the observation by Downs and Mohr (1979, p. 393) that "a substantial proportion of the variation in adoption can often be explained simply by how useful an innovation is perceived to be in carrying out a job." Similarly, Rogers and Shoemaker (1971) employ the analogous concept of "relative advantage" in explaining adoption and use of innovations. Finally this concept has proved fruitful in research on the adoption and use of information and communication technology. Notions of perceived utility were a strong correlate of acceptance and use in a study of the NLS system by Edwards (1977). The personalness or richness of a channel has received considerable attention, particularly in recent work on new communication technologies. Communication Studies Group researchers, while theorizing social presence to be an inherent attribute of media, usually measure it in terms of peoples' subjective ratings. They attempted to quantify differences in social presence among media using subjects' ratings on semantic diffential scales defined by such adjectives as sociable, personal, warm, and sensitive (Champness, 1972, 1973; Christie, 1973; 1974; Short et al., 1976). They propose that tasks differ in the degree to which they are interpersonally involving. 40 Highly interpersonal tasks require a communication medium with a high degree of social presence; i.e. those being communicated with are perceived as "real people". This concept, while criticized by some for its differential composition in separate studies (Albertson, 1980), has been employed in much current communication research. Research on actual use or intentions to use video conferencing have been informed by notions of jsocial presence (Dutton et al., 1982; Noll, 1977; Svenning, 1982), as have studies of the use of more traditional media (Ruchinskas, 1982). It is important to note here that findings with regard to the value of the social presence concept are not all positive. Ruchinskas (1982) did not find social presence to be a significant predictor of channel use, while the results of Korzenny (1978) and Korzenny and Bauer (1979) only partially supported their theory of "electronic propinquity." With regard to electronic mail, researchers have tended to assume that it is a limited bandwidth medium, and hence is low in social presence (Rice, 1980). However, it is more likely that people vary in low much social presence they ascribe to electronic mail (Johansen et al., 1979). It is therefore considered an independent variable, with variations in perceived social presence related to variations in patterns of use. A final characteristic which has received relatively little attention in studies of mass media use is the perceived privacy ■afforded by a medium. Hiltz and Kerr (1981), however, note the importance of security in the acceptance and use of electronic mail. Wynn (1980) has commented on the unique ability of electronic mail to _____________________________________________________________________ 4!_ allow the private composition of group directed messages. This attribute may not be enough for certain applications, though. Popular notions of computer piracy point to the need to examine perceptions of the security of messages. The ability of electronic mail both to retrieve and to forward messages without the sender's consent further highlights the potential importance of this concept. As Hiltz and Turoff (1978) note, privacy can be constrained by unscrupulous executives, who can have a system programmed to enable their monitoring. However, these authors further point out that " From a technological point of view the security of such systems presents the same problems as corporate data bases, files of internal memos or correspondence, or staff reports. Although a reasonable amount of physical and hardware security can be bought, the ultimate security is provided by the people involved as users and suppliers of the services." (p. 152) This highlights the importance of relying on perceived privacy over built in system security features as a predictor of use. The discussion of perceived channel attributes and electronic mail uses can be summarized by the following statement. Perceived channel attributes, including perceived utility, ease of use, social presence, and privacy will be positively associated with task and non- task related uses of electronic mail. Individual /Demographic Characteristics. H9: Level in the organization will be negatively associated with task and non-task uses of electronic mail. .42J H10: Tenure in the organization will be negatively associated with task and non-task uses of electronic mail. Hll: Educational level will be positively associated with task and non-task uses of electronic mail. H12: Age will be negatively associated with task and non-task uses of electronic mail. H13: Prior experience with electronic mail will be positively associated with task and non-task uses of electronic mail. An important class of variables which have been shown to influence communication patterns are those which represent the role or position of an individual in the organization, and which reflect the background characteristics of the individual. Organizational role has been called a primary influence on organizational behavior by Katz and Kahn (1979) and prior research has demonstrated the importance of both role and background on communication channel use (Goddard, 1973; Hunt, 1970; Mintzberg, 1973; Rosenbloom and Wolek, 1970; Ruchinskas, 1982; Steinfield, 1981). In one study of time spent using various media, individual role variables were among the most influential predictors (Ruchinskas, 1982). Two role variables receiving attention in prior organizational communication research are level (status) and organizational tenure. (Both Goddard (1973) and Hunt (1970) found that higher level people not only spend more time in communication related activities, they were more inclined to use oral channels, in addition, increased tenure is 43 typically associated with greater use of oral channels, as people both learn their way around the organization and develop a network of contacts upon whom they can call. In Mintzberg's (1973) observational analysis of five executives, he characterized them as preferring verbal (oral) media affording faster feedback. Similarly Ruchinskas (1982) found that as level increases, total time communicating increases while use of written channels decreases. These findings pose somewhat of a dilemma for hypothesizing the relationship of level to electronic mail uses. Past research clearly indicates that as one moves up the organizational ladder, communication becomes a more prominent aspect of organizational life. However, based on research which specifies the different channels used, it appears that a large part of the increase in communication is due to greater amounts of time spent on the telephone and in face-to-face meetings. It is not clear whether increased level, then, would be associated with increased electronic mail usage. In fact, much of the popular literature on executives' anxiety over the use of terminals would lead to an expectation of decreased usage at the higher levels. On the other hand, the relative ubiquitousness of the electronic mail system in this organization, the friendliness of its interface, the more rapid feedback electronic mail affords over other print media, and the oft- noted problems associated with telephone tag (Uhlig et al., 1979), might lead to greater use with increasing level and tenure. However, the bulk of the evidence to date calls for predicting a negative relationship between level and tenure and use of electronic mail. Beyond a person's role in the organization, two demographic ________________________________________________________________ 44 characteristics are proposed to influence electronic mail use. Age and educational background are two variables familiar to social scientists. Greater education is often associated with increased use of written channels among organization communication researchers (Rosenbloom and Wolek, 1970; Ruchinskas, 1982). Ruchinskas also found telephone use to be negatively related to educational level. A moderately positive relationship found by Hiltz and Kerr (1981) in their review of computer mediated communication research is concluded to be a result of more educated users feeling less threatened by system complexity. The ability to compose complex messages without constraints on time may also be a factor leading to the appeal of electronic mail for higher educated people. Age, on the other hand, generally was found to be inversely related to the use of computer-mediated communication systems. Hiltz and Kerr explain this by reasoning that older people may feel more threatened by new ways of doing things. These authors also point out that younger people may have had more exposure to computing, and hence may be more receptive to communicating via this technology. Of course, the amount of prior experience using electronic mail, on this system or other systems should influence the amount of use, for either task related or non-task purposes (Hiltz and Kerr, 1981). It is therefore included as the final individual/demographic independent variable. The relationship of individual/demographic variables to uses of electronic mail can be summarized in the following way. Age, educational level, and prior experience with electronic mail will be 45 positively associated with task and non-task uses of electronic mail, while level in the organization and organizational tenure will be negatively associated with both types of uses. System Access Variables H14: Access to a electronic mail terminals will be positively associated with both task and non-task uses of electronic mail. H15: The number of work group coworkers who are accessible via electronic mail will be positively associated with task and non task uses of electronic mail. H16: The number of relevant coworkers outside the work group who are accessible via electronic mail will be positively associated with task and non-task uses of electronic mail. A number of prior studies of electronic mail and computer conferencing has uncovered a set of variables considered so necessary to system use that they are treated here as control variables. That is, a goal of the present research is to account for variance in electronic mail usage over and above that which can be explained by these control variables. The critical component of these variables is access. Access to a terminal, and access to relevant people on the system are essentially considered to be necessary, but not sufficient determinants of the use of electronic mail (Edwards, 1977; Hiltz and Kerr, 1981; Hiltz and Turoff; 1978; Johansen et al., 1979). Edwards, in a study of the use of the NLS system, found strong evidence that the more people had to share terminals in order to access ___________________________________________ 46 the system the more use was depressed. Similarly, Johansen et al., (1979), conclude from their own research and a review of others' experiences that access is crucial, preferably with each participant provided with their own terminal in their office. Uhlig et al., (1979) also recommend providing users with their own terminals placed in their workspace in order to ensure effective implementation. It appears that if a person has to leave his or her workspace, or must wait for someone else to finish before using the system, they will search for other means of accomplishing the interaction. Of course, even with a terminal, if there is no one else with whom a person needs or desires to communicate, the question of electronic mail use is moot. Like any network, value is a function of the number and relevance of the nodes (Dordick et al., 1981). In the present case, two groups of relevant others are included: members of a person's work group, and contacts outside the work group. Given the likely proximity of work group members, and hence the greater availability of face-to-face interaction with these individuals, this aspect of accessible people may not be as influential as the number of non-work group contacts using electronic mail. Therefore, these two organizational locations are differentiated in the present research. In summary, relative access to people on the system, and access to terminals will be positively associated with task and non-task uses of electronic mail. Summary A diverse body of literature, linked by its relevance to 47 jorganization communication provides the framework for this research. Teleconferencing research spurred thinking about the uses of electronic mail by focusing on communication activities. However, the usual typologies of communication activities are more relevant to the analysis of business meetings than to the day-to-day applications of an organizational electronic mail system. Therefore, substantial efforts are made in this research to identify the many kinds of uses of electronic mail. Variables that predict patterns of electronic mail usage are derived from organization theory, organizational communication research, teleconferencing research, and prior studies of computer- mediated communication systems. Variables associated with the task and environment, perceived channel characteristics, individual/demographic characteristics, and system access are predicted to influence the ways in which people use electronic mail. Patterns of usage, prior to empirical analyses of the many usage items, are simply conceived of as either task or non-task related. A summary of expected relationships is provided in Table 2.1. 48 Table 2.1: Surrmary of Expected Relationships of Independent Variables to Task and Non-Task Related Uses of Electronic Mail DEPENDENT VARIABLES INDEPENDENT VARIABLES Task Uses Non-Task Uses Task/Environmental Variables Task Routineness - 0 Time pressures + 0 Task Interdependence + 0 Envirormental Uncertainty + 0 Cross Location Communication Need + 0 Perceived Channel Characteristics Utility + + Ease of Use + + Social Presence + + Privacy + + Individual/Demographic Variables Age - - Education + + Experience with Electronic Mail + + Organizational Level - Organizational Tenure - 49 Table 2.1 (continued): Summary of Expected Relationships of Independent Variables to Task and Non-Task Related Uses of Electronic Mail DEPENDENT VARIABLES INDEPENDENT VARIABLES Task Uses Non-Task Uses System Access Variables Access to Terminals + + Coworkers outside group on system + + Coworkers within group on system + + 50 CHAPTER THREE Methods Research Context The research questions were examined in a large, decentralized corporation offering a wide array of office products. An electronic mail system developed by the research wing of the organization had been in place for approximately 7 years at the time of the research. People in this research group began using the system, but it quickly diffused to other units within the company. At the time of the study, there were over 2000 users in six cities around the U.S. as well as in overseas locations. The electronic mail system was integrated with word processing and other functions in a work station concept. People retained most computer files, including those associated with their personal messages, locally on their own disks, but since all workstations were connected to a network, large files could be stored remotely. Also on this network were computers whose main function was the routing and processing of mail. Anyone with a "mailbox" recognized by the network could send mail to any other user on the system. Messages could be sent to either single individuals, or to groups of people; and groups could be more economically addressed through the use of "distribution lists." That is, a set of individual addresses could be associated with some list name, and using this name when addressing mail would ensure that all members of the group received the same message. Indications of the extent to which the system is used are provided 51 by usage statistics gathered in self-report fashion from a sample of ^company employees. The sample, which is described in more detail below, indicated how many messages they had sent and received in the .previous week. They further broke these down into either personal or distribution list messages. Although self-report estimates of communication frequency are questionable from a reliability standpoint, they do provide a measure of how often people feel they make use of the mail system. The host organization did not utilize the system to automatically gather usage statistics at the individual level, considering this to be a potential invasion of user privacy. The typical user reports sending 12.2 messages per week to other individuals, or just over 2 per day (Table 3.1). This mean is highly skewed, however, by some very heavy users, and the median figure drops to 4.9. All of the usage statistics, in fact, are skewed by heavy users, and the median figures are probably better measures of central tendency. Users send fewer measures to distribution lists (median=1.4 per week), but a multiplicative effect is evident, as these few distribution list messages sent are compared to the number received (median=49.7 per week). The number of individual messages received compares closely with the number sent to individuals (median=5.2). A further indication of the important role electronic mail plays in this organization is obtained by estimates of how much time people spend using it. Subjects estimated using electronic mail an average of nearly 50 minutes in a typical workday. Use of the mail system was free to employees; a policy which clearly affects how a system is used. Given this unrestricted usage 52 J TABLE 3.1: Descriptive Statistics for Electronic Mail Use Frequency dumber in Previous Week Mean Median S.D. Sent to individuals 12.2 4.9 18.4 Sent to distribution lists 4.1 1.4 7.6 Received - individual 16.3 5.2 29.3 Received - distribution list 75.5 49.7 94.0 minutes per day using electronic mail 49.1 39.0 44.0 53 policy, the length of organizational experience, and the diverse user population, the setting appeared an ideal one for exploring japplications of this new medium, and examining important social and psychological factors which help explain patterns of use. Qualitative Analyses The uses of electronic mail and factors which might predict usage patterns were uncovered through a program of field research, incorporating a combination of qualitative and quantitative methodologies. Qualitative methods included interviews with users and observation of system usage in its natural organizational environment jOver a period of several months. Quantitative methods included a cross-sectional survey using a self-administered questionnaire. Interviews Approximately 25 interviews were conducted with people having different amounts of experience with the company electronic mail system. Several of these users had been involved in designing and implementing the system. Interview questions were basically open- ended, although specific topics formed the core of each interview. Essentially, interviewees commented on the nature of their work, their communication patterns, their attitudes towards the electronic mail system, and their use of the system. System designers were also able to elaborate upon how certain features of the mail system had been adopted and utilized by the user community. A case in point is the proliferation of distribution lists on the system, reflecting a vast array of task-related and socio-emotional interest groups. In this 54 fashion, insights were provided regarding the nature of electronic mail usage in the organization and how usage is tied to the organizational and individual context. It became clear during this stage of the research program that any simple typology of electronic mail applications based on past teleconferencing studies could not adequately capture the complexity of usage. For example, a typical communication activity found in teleconferencing research that focused on computer-mediated communication is information exchange (Rice, 1980). Interviewees, however, noted that information exchange over the electronic mail system took many forms. On the one hand, a user might send a personal message to one or a few acquaintances when seeking information. On the other hand, users might take advantage of the broadcasting capability of electronic mail, and send a request for information to a multitude of system users, including many unknown recipients. These are strikingly different means of obtaining information. Many other unique purposes of use were identified through interviews, leading to the beginnings of an expanded list of electronic ■mail applications. The interview process yielded additional benefits, however. Divisions within the company using electronic mail were identified, helping to guide the survey sampling strategy. Equally important, the interviews provided a historical and contextual perspective on system use, and the overall communication climate of the organization. Most felt that the company encouraged open communication and sharing of information. The policy of unrestricted use of the electronic mail system is one manifestation of this climate. No 55 attempts were made by management to discourage non-task related communication, either through direct edict or indirect means such as charging for system use. During the interviews, and subsequent observations of system use, the only controls on content observed dealt with either security of company data or individual users attempting to alter the course of discussion on distribution lists. Many commented on the quality of the communication climate, particularly with regard to the electronic, viewing it as a benefit of their workplace. Some even mentioned its importance in making this organization an attractive place to work. Thus it can be postulated that the communication environment in the study organization set the context for a particular evolutionary path of electronic mail applications. The general importance placed on free and open interaction may have fostered a more varied set of uses then would be found elsewhere. Understanding this context enhances the interpretation of quantitative data, and wherever possible, discussion of types of uses is supported by anecdotes from interviews. Observation The extensive use of "public" distribution lists (i.e. available to any who wished to join) afforded the opportunity to observe a variety of group applications. One task-related and one non-task related distribution list were joined, allowing receipt of all public messages addressed to these lists. This unobtrusive observation of system use enabled the identification of additional types of uses not found in earlier 56. typologies and not mentioned in interviews. Furthermore, examples of unique kinds of uses that were mentioned in interviews were observed, giving meaning to them that only first-hand experience allows. Some applications observed include advertising products, opinion polling, policing a distribution list to ensure appropriate usage, organizing social activities, learning about interesting things or events, finding out about what is happening around the company, and participating in entertaining conversations or games. What distinguishes these uses •from other kinds of applications traditionally thought of as belonging to the realm of electronic mail is their group and/or socio-emotional nature. During the interview and observation period, further insights into the organizational context were gained by virtue of daily contact with employees and system administrators. This experience not only aided in questionnaire design, but also in interpretation of findings. The Survey Sampling The sampling frame for this research consisted of employees using the organization electronic mail system. Divisions which contained a substantial number of electronic mail users were included in the Lampling frame. As noted in the discussion on limitations of prior electronic mail and computer conferencing research, few studies incorporate a wide Lange of job types. A goal then, of this research, was to enhance generalizability through the inclusion of respondents representing the .57J diverse range of white collar jobs. However, job classification information was not available prior to the selection of the sample. The use of multiple divisions and the inclusion of a large enough sample were strategies which helped ensure diversity in the sample. The sample of 400 users was randomly drawn from lists of employees from the included divisions on which electronic mail users had been lighlighted. Users were identified from a variety of sources, including distribution lists of members of various organization units which were stored online, an electronic mail users directory, and several divisional telephone directories which also included electronic mail addresses of any users in the division. Of the 400 questionnaires mailed to users, 220 were returned leading to a 55% resonse rate. Response did not seem to be affected by division as the 180 non-respondents were fairly evenly distributed across divisions. No other information on non-respondents was available. The 220 respondents came from several different job categories, although more programmers, engineers, and research personnel are included than other types of jobs (Table 3.2). Together they account for over two-thirds of the sample. Given the random sampling technique, this is most likely indicative of the relative distribution of job types among the electronic mail user population. An additional 30% of the sample is comprised of various white collar workers, including planning, clerical, personnel, program office/management, testing, facilities, finance, marketing, and technical writing employees. From this distribution it appears that employees in more 58 technical or "hard science" capacities receive access before those in the "softer" or less technical areas. Data Gathering The questionnaire was constructed using an iterative process, with each version the subject of pretesting among selected users. These pretest respondents commented on wording and appropriateness of items, as well as time required for instrument completion. Questionnaires were sent via conventional rather than electronic mail for several reasons. First, a companion study, which compared attitudes of users and non-users was conducted at the same time as this research effort, using the same user sample. Non-users could not receive questionnaires via electronic mail, and utilizing a different method of distribution might bias response rates and possibly response content. Second, the questionnaire was fairly large, and might have caused problems for users with limited working space on their disks. Third, a hardcopy was deemed necessary so that respondents could bring the questionnaire home to fill it out during non-working hours. Although users could have had a copy printed, this represented an extra behavioral step which might have reduced the response rate. Several steps were taken to ensure an adequate response rate. First, accompanying each questionnaire was a motivational letter from the company manager sponsoring the research (Appendix A). Second, the questionnaire was designed for ease of reading and completion. A self- addressed envelope was included with each questionnaire, and respondents were instructed to simply use the company internal mail 59 Table 3.2: Job Types of Surveyed Electronic Mail Users Job Type Number Percent jProgramming 59 27% •Engineering 47 22 •Research 46 21 'Planning 12 6 jJlerical 11 5 •Personnel 8 4 ■Program Mgmt 8 4 Testing 8 4 Other * 21 10 220 100% * Other includes facilities, marketing, finance, and technical writing. 60 service, thereby avoiding any mail charges. In addition, respondents were guaranteed anonymity. Finally, approximately two weeks after (questionnaires had been sent out, a followup letter encouraging response was mailed to all non-respondents using the electronic mail Lystem (Appendix A). Variables and Operationalizations Dependent Variables The dependent variable - the uses of electronic mail - was created in a multi-step process. Respondents were asked to rate on a 5 point scale the extent to which they used electronic mail for a series of purposes. The scale ranged from never (1) to very often (5). In addition, a not applicable category was provided to avoid having people ■rate items for tasks which they do not do at work anyway, even via other media. Thirty-two individual items were culled from a number of sources: 1) lists of communication activities used in prior research on teleconferencing and electronic mail (Dutton et al, 1982; Pye et al, 1973; Rice and Case, 1981; Short et al., 1976), 2) interviews with users, and 3) personal observations of uses of the mail system. Usage items were then factored to uncover a set of fewer dimensions of electronic mail uses. Prior to this, however, several items were removed from subsequent analyses because of their low variance and high number of missing cases. Generally, when the system was virtually not used for some purpose, or the variance was so limited on a purpose as to inhibit further analysis, it was removed. Summary statistics for all usage items are provided in Table 3.3. 61 Table 3.3: Summary Statistics for EM Purposes of Use Items * Variables Retained for Further Analyses ** Mean S.D. N Send a message in place of a phone call 3.5 1.24 214 Distribute/provide information 3.4 1.17 216 Learn about events/things I'm interested in 3.4 1.20 217 Leek task information from people I know 3.3 1.28 211 Keep track of company happenings 3.3 1.24 213 jGive/receive feedback on reports/ideas 3.2 1.13 209 jCoordinate activities of projects I'm on 2.9 1.33 208 Keep a record of agreements/interactions 2.8 1.38 199 Send/receive pointers to large files 2.7 1.30 203 Schedule meetings/appointments 2.7 1.33 212 Monitor progress of projects I work with 2.6 1.42 193 Broadcast requests for information 2.5 1.10 211 Keep in touch/maintain relationships 2.4 1.18 211 Take a break from my work 2.4 1.19 215 3rainstorm/generate ideas 2.3 1.08 208 Participate in entertaining events (games) 2.2 1.19 208 or conversations Organize/coordinate a social activity 1.8 1.08 200 Fill up "free" time 1.7 1.01 206 62 Table 3.3 (continued): Summary Statistics for EM Purposes of Use Items Variables Removed from Further Analyses *** Mean S.D. N Forward messages to someone who is not on some distribution list 2.1 .97 213 Poll opinions on a topic 1.8 .98 210 Advertise/respond to products for sale 1.8 .91 203 Resolve conflicts/disagreements 1.8 .97 203 Ask questions in a public setting (eg sending many copies) to force a response or action 1.7 .92 203 Carry on negotiaions/bargaining 1.7 .97 179 Get a message directly to someone whose calls are normally screened 1.7 .98 197 Get to know someone 1.6 • -j 00 207 Monitor performance of my subordinates **** 1.6 1.05 140 Ensure people use DLs for appropriate topics 1.5 .82 156 Discuss confidential matters 1.5 .75 208 Find out about job/promotion openings 1.3 00 in . 208 * Respondents were provided the stimulus phrase "I use electonic mail to...". They circled a nimber ranging from l=never to 5=very often. A "not applicable/not something I do at work" option was provided and treated as missing data. ** Variables with a standard deviation < 1.00 were not retained • *** Two additional usage items - "send a message requiring an immediate response" and "send a message requiring a response the same day" were not included for conceptual reasons. They reflect timing rather than purpose of use. **** This item was dropped because of the number of missing cases • 63 Respondents most frequently use electronic mail for task-oriented, information exchange purposes. Distributing or providing information is a common application (mean=3.4) and based on system observation includes such activities as sending announcements, or answering people's requests for information. Similarly, seeking task information from acquaintances (mean=3.3), giving and receiving feedback on reports or ideas (mean=3.2), and telling people where to access large files (mean=2.7), are somewhat frequent information exchange-oriented uses. A quite different method of seeking information, not really possible via other communication media, is the use of electronic mail to broadcast requests for information (mean=2.5). A message might be addressed to a large distribution list, beginning with "Anyone know anything about...". This application's moderately high score demonstrates the different information flows that are possible once electronic mail systems are in place. Brainstorming over electronic mail is done, but not as often (mean=2.3). The actual incidence of brainstorming might be a limiting factor here. The most common application reflected the use of electronic mail to substitute for the telephone (mean=3.5). The oft-mentioned ability of electronic mail to help avoid "telephone tag" appears to translate into frequent use for this purpose. It is not clear from this category what the nature of calls substituted might be, although subsequent factor analyses show this use loading with other task-related uses. This use, coupled with the use of electronic mail to keep a record of agreements or interactions (mean=2.8) demonstrates a benefit beyond simple telephone tag avoidance. Substitution not only helps avoid 64 wasting time, but allows documentation of the communication activity for future reference. In addition, the use of the system to keep in bouch with others seems to also reflect a use that traditionally is nandled by telephone (mean=2.4). Another set of more frequent applications reflects the use of electronic mail for coordination purposes. Project coordination (mean=2.9), project monitoring (mean=2.6), and scheduling (mean=2.7) are task-related examples of coordination uses. On the other hand, socio-emotional coordination applications, such as organizing social ^activities, are less frequent (mean=1.8). The opportunities for ^organizing social activities, however, might be less common than the task-related uses. Respondents reported using electronic mail for several non-task related activities. Some distribution lists served as general information utilities, with people sending messages to members about interesting events or bits of information related to the list topic (mean=3.4). This use, along with the keeping track of company happenings (mean=3.3) highlighted a newsletter function of the system. (Further entertainment functions of the electronic mail system are evidenced by its use for taking breaks from work (mean=2.4), participating in games (mean=2.2), and filling up free time (e.g. something to read during lunch or breaks) (mean=1.7). These are not frequent applications, but do illustrate how electronic mail integrates with socio-emotional needs of employees. Many of the uses identified in interviews and observations are not frequent applications of the electronic mail system, although they 65 might be occasional uses (see Table 5.1 for an alternative method of representing use frequency). These usage variables further exhibited little variation, essentially reducing to a dichotomous situation, where a respondent either had never used the system for the purpose in question, or did so on a very infrequent basis (i.e. scored 2 on the 5 point scale). Respondents did not often poll other users for their opinions (mean=1.8), an activity that arguably would not occur often via any medium. The fact that it occurs at all is a significant finding, serving to differentiate the communication environment of electronic mail organizations from other organizations without this technology. Advertising via electronic mail is relatively infrequent (mean=1.8), but most individuals probably do not have much frequent need to advertise or respond to products for sale. This is further evidence that a widely diffused electronic mail system might eventually substitute for other company information systems like bulletin boards or newsletters. Conflict resolution and negotiation are two applications that often appear in teleconferencing research. It is generally concluded that face-to-fact interaction is preferred for handling these activities, and their relatively low use here supports these findings (means=1.8 and 1.7, respectively). However, without knowledge of the incidence of the need for either of these activities, no conclusion is justified. Asking questions in a public setting to force a response or action is another relatively infrequent use (mean=1.8). It may be the case ■ 66 that people are concerned about maintaining good relationships with other users, and forcing them to take action is not one way of accomplishing this. Hence it is not anticipated that this would occur frequently. Similarly, using the mail system to get through to someone whose calls are screened (mean=1.7) is another way in which the mail system can be a resource as long as it is not abused and done frequently. Getting to know someone over the mail system is not at all a common application (mean=1.6), although 46% percent mentioned it doing it at least once. These situations probably do not arise very often. Monitoring performance is virtually not done (mean=1.6); nor is the use of the system for discussing confidential matters (mean=1.5). The former may reflect attempts to maintain a more personal work environment in the face of office automation. No generally accepted method of monitoring white collar performance exists, particularly via electronic mail, so this might reflect the difficulty of performance monitoring over the system. Concern for the adequacy of security is implicitly voiced by the lack of use for confidential matters. Finally, other infrequent uses include policing distribution lists to ensure people stick to appropriate topics (mean=1.5) and finding out about job and promotion openings (mean=1.3). The former seems to be left up to the founders of distribution lists, while the latter is somewhat of an infrequent situation in general, relative to other kinds of communication that occurs in offices. Forwarding messages to people who are not on a distribution list, who may have some interest in the topic, is done occasionally (mean=2.1), but exhibits little variance. 67 A principal factors analysis with coxnmunality estimates placed in :he diagonal was performed on the use items along with a varimax rotation scheme. Because of the limited variation on many of the items, a decision was made to exclude them from the factor analysis, as they would not serve to differentiate among factors. An objective criterion of a standard deviation of less than 1.0 was used to determine which items to leave out of the analyses. Those items that remain, however, do constitute the bulk of applications for most users, and therefore the dimensional structure obtained seems a more appropriate description of actual usage styles. Because a large number of cases were dropped when listwise deletion was used, a substitution strategy recommended by Rummel (1970) jwas employed. Each usage item was regressed on all other uses, and for those cases missing on only one of the items, the predicted score was substituted for the missing value on that item. Two factors emerged, explaining 48.9% of the variance: one with primarily task-related uses loading highly and one with primarily non-work or socio-emotional uses loading highly (Table 3.4). The task vs. socio-emotional dichotomy at first does not seem to capture the complexity of usage patterns. However, this factor structure bears a close resemblance to a number of prior theoretical approaches to communication in organizations which differentiate set ween task and socio-emotional communication (Bales, 1950; Farace et al., 1977; Katz and Kahn, 1978). It is therefore both conceptually ■appealing and parsimonious. Two items did not load on either dimension; broadcasting requests 68 Table 3.4: Factor Analysis Results for Purposes of Use Items * Factor 1: Factor 2: Variables Task Usage Social Usage Coordinate activities of projects I'm on .84 -.01 Schedule meetings/appointments .70 .13 Monitor progress of projects I work with .70 .04 Send a message in place of a phone call .66 .24 Distribute/provide information .66 .10 Iteep a record of agreements/interactions .65 .03 Seek task information from people I know .64 .12 Gi've/receive feedback on reports/ideas .63 .15 Brainstorm/generate ideas .59 .23 Send/receive pointers to large files .58 .32 Take a break from my work .01 .70 Fill up "free" time -.10 .69 [»earn about events/things I'm interested in .05 .67 Participate in entertaining events (games) .20 .62 or conversations Keep in touch/maintain relationships .27 .56 jOrganize/coordinate a social activity .32 .52 proadcast requests for information .39 .32 Keep track of company happenings .22 .25 * Principal factors analysis using communality estimates on the dia gonal. Table entries are varimax rotated factor loadings. Two factor solution explained 49 percent of variance. 69 for information, and keeping track of company happenings. These might reflect additional dimensions, such as broadcasting and surveillance, that would arise if more individual items had been included. High loading items on each dimension were subjected to a reliability analysis with task use items achieving an alpha of .90 and socio-emotional items an alpha of .80. The two dependent variables - hereafter referred to as task use and social use - were therefore created by taking the mean of high loading items (Table 3.5). The high reliabilities, coupled with the conceptual fit of task vs. socio- emotional communication purposes, increases confidence in the validity of these dependent variables. One point of clarification is necessary with regard to the task/social dichotomy. These purposes are not meant to be mutually exclusive, and are rarely so in the real world. That is, much of our socio-emotional communication is supportive of task communication. It may be the case that any single electronic mail message can contain both task and social elements. In fact, a moderate correlation exists between these dependent variables in support of this notion (r=.33). Independent Variables Four sets of independent variables were included: 1) task characteristics and other contextual or environmental variables, 2) perceived channel characteristics, 3) individual/demographic variables, and 4) system access variables. Measurement of each set and scale constructions are described below. 70 Table 3.5: Inter-item Correlations, Scale Reliability and Summary Statistics for Purposes of Use Scales TASK USES 1 2 Correlation Matrix (N=187) 3 4 5 6 7 8 9 1. Seek info - 2. Distribute info .48 - 3. Give feedback .45 .53 - 4. Brainstorm .38 .45 .54 - 5. MDnitor projects .49 .41 .50 .44 - 6. Keep records .38 .44 .37 .36 .52 7. Substitute for phone .49 .46 .44 .39 .46 .45 8. Point to files .41 .39 .35 .41 .43 .46 .54 9. Schedule .45 .41 .41 .35 .46 .43 .52 .51 - 10. Coordinate projects .47 .52 .48 .50 .59 .58 .57 .44 .72 Cronbach's alpha = .90 Scale Mean = 3.0 Standard Deviation = .91 SOCIAL USES Correlation Matrix (N=201) 1 2 3 4 5 1. Keep in touch - 2. Learn about events .41 - 3. Participate in events .36 .49 - 4. Organize social activity .43 .32 .48 - 5. Take breaks .34 .43 .49 .26 6. Fill free time .36 .40 .38 .35 .62 Cronbach's alpha = .80 Scale Mean = 2.3 Standard Deviation = .81 71 Task and Environment Variables. A series of 7 point scales were used to measure most items in this area, with subjects rating the extent to which their job involved or was governed by specific characteristics. Measures of task characteristics, including items reflecting standardization, routineness, and pressure were drawn from Ruchinskas (1982) and Svenning (1982). Environmental uncertainty items were derived from Tushman (1977). Task interdependence items were based on Thompson's (1967) conceptualization, and following Tushman and Nadler (1978), included both intra and inter-unit task interdependence. A cross boundary/locational communication need scale was constructed from one item asking the number of work group members located in another city, and a second yes/no question regarding membership on project teams outside the work group. Summary statistics for all individual task and environment items used in constructing scales are provided in Table 3.6. An examination of the means of individual task characteristic items shows respondents more likely to feel that their jobs involve time pressures (mean=4.9) and unexpected problems (mean=5.4), and somewhat likely to involve crises and urgent matters (mean=4.3). Less often are jobs governed by rules, policies and regulations (mean=3.3), with an occasional need to deal with new people on the job reported (mean=3.7) . Jobs are not seen as very standardized (mean=3.3), and are not seen as involving very well defined subject matter (mean=3.5). In addition, respondents only perceived their tasks to have clearly defined outcomes to a moderate extent (mean=4.0). 72 Table 3.6: Summary Statistics for Task Characteristic Items Variables * Mean S.D. Crises/urgent matters 4.3 1.66 Time pressures 4.9 1.39 Unexpected problems/situations 5.4 1.16 Rules, policies, and regulations 3.3 1.62 Working with people I'm not acquainted with 3.7 1.54 Well defined subject matter 3.5 1.31 Tasks with clearly defined outcomes 4.0 1.57 Standard operating procedures 3.3 1.33 Routine repetitive tasks 2.4 1.28 Finding novel solutions to problems 5.6 1.41 * Scales ranged from l=not at all phrases "My job involves..." or were used. to 7=very much. The stimulus "My job is governed by..." 73 Generally, tasks are not seen as routine (raean=2.4), and seem to require a certain amount of innovative thought (mean=5.6). Overall, these task characteristics seem quite appropriate as a description of the professional white collar task situation in the host organization. The 10 task characteristic items were factored, producing three dimensions (Table 3.7). A pressure dimension emerged, primarily defined by the presence of time pressures and crises, unexpectedly, the extent to which subjects' jobs were governed by rules, policies, and regulations also loaded on this dimension. One possibility here might stem from the non-routine nature of work, and the relative frequency of unexpected problems. Use of rules and regulations might be viewed as a means of coping with more frequent exceptions. A second dimension containing items reflecting how well defined or standardized jobs were also emerged. The third dimension, routineness, contained just two high loading items; the routineness or repetitiveness of tasks, and a negative loading for the extent to which jobs involved finding novel solutions to problems. On the basis of these analyses, three scales were constructed and subjected to a reliability analysis. Cronbach's alpha for the three scales constructed from the means of high loading items were: pressures (.73), well defined tasks (.67), and routineness (.52). Hie inclusion of only two items contributed to the relatively low alpha on routineness (Table 3.8). These task characteristics provide a view of respondents' jobs as being somewhat likely to involve pressure (mean=4.3), less likely to deal with standardized activities or well defined outcomes (mean=3.6), and generally not routine (mean=2.4). 74 Table 3.7: Factor Analysis Results for Task Characteristic Items * Variables Factor 1: Pressures Factor 2: Standard Factor 3: Routine jCrises/urgent matters .79 i • o .07 Time pressures .72 .13 -.05 Unexpected problems/situations .56 -.09 -.12 Rules, policies, and regulations .49 .24 .36 Working with people I'm not acquainted with .47 -.07 .09 Wall defined subject matter -.17 .73 .08 Tasks with clearly defined outccmes -.06 .57 .21 Standard operating procedures .28 .54 .42 Routine repetitive tasks .11 .12 .66 Finding novel solutions to problems .12 -.15 -.52 * Principal factors analysis using communality estimates in diagonal. Factor entries are varimax rotated loadings, factor solution explained 57 percent of variance. N=200. the Three 75 Table 3.8: Inter-item Correlations, Reliability, and Summary Statistics for Task Characteristic Scales Correlation Matrix PRESSURE SCALE 1 2 3 4 1. Crises/urgent matters - 2. Time pressures .62 - 3. Unexpected problems/situations .44 .34 4. Rules, policies, and regulations .34 .38 .27 5. Working with people I'm not .41 .37 .18 .22 acquainted with Cronbach's alpha = .73 Scale Mean = 4.3 Standard Deviation = Correlation Matrix STANDARDIZED TASKS SCALE 1 2 1. I/fell defined subject matter - 2. Tasks with clearly defined outcomes .46 3. Standard operating procedures .39 .37 Cronbach's alpha = .67 Scale Mean = 3.6 Standard Deviation = Correlation Matrix ROUTINENESS SCALE 1 1. Routine repetitive tasks - 2. Finding novel solutions to problems .35 (reverse coded) Cronbach's alpha = .52 Scale Mean = 2.4 Standard Deviation = 1.01 1.07 1.10 76 Two environmental uncertainty items drawn from Tushman (1977) were used. Each measured a different aspect of instability in the environment, The first involved a measure of whether inputs (techniques, skill, or information) required to handle tasks are changing. The second, worded so that a high score reflected stability (in order to reduce the likelihood of response set bias) measured whether products were output into a stable or changing environment. A look at the means shows that respondents perceived a fair amount of environmental uncertainty surrounding their jobs (means=5.2 and 3.1, respectively; Table 3.9). The environmental uncertainty items were moderately correlated (-.36). The second measure was reverse scored, and then both were combined into one scale (alpha=.53). The scale mean (5.1) reflects this perception of environmental uncertainty. A three point cross boundary/location communication need variable was constructed in the following manner. Subjects were scored 1 if they did not have any co-workers in other locations and were not members of any project teams outside their work group. A score of 2 was assigned to those who either had co-workers in other locations or were a member of an outside project team. Finally, if an individual had co-workers in other locations and was on an outside project team, they were scored 3 (Table 3.9). The mean score of this scale is 1.7, reflecting a moderate amount of cross locational work patterns in the sample. Task interdependence items were constructed to reflect Thompson's notions of pooled, sequential, and reciprocal interdependence. Mean ___________________________________________________________________77 Table 3.9: Perceived Environmental Uncertainty and Cross Boundary/ Location Communication Need Summary Statistics PERCEIVED ENVIRONMENTAL UNCERTAINTY Mean * S.D. 1. Techniques, skills, or information required to 5.2 1.42 produce my group's products are always changing 2. Marketplace,technology and regulations for my 4.9 1.41 work group's products are generally stable Correlation = .36 Cronbach's alpha = .53 Scale Mean = 5.06 Standard Deviation = 1.14 Mean S.D. CROSS BOUNDARY COMMUNICATION NEED ** 1.7 .68 * Scale ranged from l=not at all to 7 = very much ** Scale ranged from 1= no work group members in other cities and not a member of an outside project team, 2=one or other true, and 3=both true. 78 scores show a tendency for respondents to characterize their work as interdependent in a reciprocal fashion with other work group members (Table 3.10). The products of their work feeds into someone else's work (mean=5.1) and their work requires input from another member (mean=5.3). Work is less likely to be completed in an isolated and independent fashion (mean=3.3). In addition, reciprocal interdependence is demonstrated by the degree to which tasks are completed in a team approach (mean=4.6). As Thompson proposed, reciprocal interdependence incorporates sequential. The responses to measures in inter-unit interdependence parallel those for within the unit, with the exception that reciprocal interdependence is lower (mean=4.0). Work groups are less likely to work in a team approach with other groups, but do exhibit sequential interdependence (means=5.1 and 4.9). These sets of items were factored, producing a within group and between group dimension (Table 3.11). On both dimensions, low or pooled interdependence loaded negatively, while sequential and reciprocal interdependence items loaded positively. Because of their unidimensional nature, the items for each were combined into a within-group and between-group interdependence scale by reverse scoring the pooled items and averaging. The two scales; within-group (alpha=.73) and between-group interdependence (alpha=.75) demonstrated sufficient reliability (Table 3.12). Scale means reveal a fairly substantial degree of interdependence is perceived by respondents (within-group mean=4.9; between-group mean=4.7). 79 Table 3.10: Surrmary Statistics for Task Interdependence Items Intra-unit Task Interdependence Variables Mean ** S.D. In my group, the products of my work: Are independent of other's work 3.3 1.8 Feed into someone else's work 5.1 1.4 Require input from someone else's work 5.3 1.4 Are completed with others in a team approach 4.6 1.7 Inter-unit Task Interdependence Variables Mean ** S.D. The products of my work group: Are independent of other groups 3.3 1.8 Feed into another work group 5.1 1.6 Require input from another work group 4.9 1.6 Are completed as a team with another group 4.0 1.9 ** Scales ranged from l=not at all to 7=very much 80 Table 3.11: Factor Analysis Results for Task Interdependence Items * Factor 1: Variables Inter-Unit Factor 2: Inter-Unit In my group, the products of my work: Are independent of other's work -.41 -.19 Feed into someone else's work .70 .20 Require input from someone else's work .69 .27 Are completed with others in a team approach .61 .32 The products of my work group: Are independent of other groups -.16 -.55 Feed into another work group .35 .61 Require input from another work group .29 .56 Are completed as a team with another group .24 .70 * Principal factors analysis using communality estimates in the diagonal. Factor entries are varimax rotated loadings. Two factor solution explained 58 percent of variance. N==208 81 Table 3.12: Inter-item Correlations, Reliability and Summary Statistics for Intra and Inter Unit Task Interdependence Scales INTRA-UNIT TASK INTERDEPENDENCE Correlation Matrix 1 2 3 1. Independent (reverse coded) - 2. Feed into other's work .35 - 3. Require other's input .28 .54 4. Completed as a team .40 .44 .52 Cronbach's alpha = .73 Scale Mean = 4.9 Standard Deviation = 1.14 INTER-UNIT TASK INTERDEPENDENCE Correlation Matrix 1 2 3 1. Independent (reverse coded) - 2. Feed into other's work .40 - 3. Require other's input .34 .47 4. Completed as a team .42 .51 .47 Cronbach's alpha = .75 Scale Mean = 4.7 Standard Deviation = 1.27 82 Channel Characteristics. Subject's perceptions of electronic mail attributes were measured by a series of 7 point semantic differential scales similar to those employed by Ruchinskas (1982) and Svenning (1982). Seventeen pairs of bipolar adjectives reflecting such attributes as ease of use, utility, personalness, and confidentiality were rated by respondents. Summary statistics (Table 3.13) demonstrate that users have a high opinion of the electronic mail system, perceiving it to be effective, useful, and efficient. It is further viewed as easy to use, simple rather than complex, comfortable, flexible, and accessible. Users were less adamant about the personalness or warmth of electronic mail (means=4.4), but tended to rate it as sociable and informal. They were equally mixed with regard to the publicness and confidentially of this medium (means=3.8 and 4.2, respectively). The combined use of electronic mail for interpersonal messaging (a private use) as well as distribution list participation (public use) probably helped foster this uncertainty. Finally, electronic mail is perceived by respondents to be convenient, reliable, and fast. All seventeen items were factor analyzed, and the four expected dimensions emerged (Table 3.14). An ease of use factor was defined‘by the adjectives easy, comfortable, simple, flexible, and accessible. The second factor contained the effective, useful, and efficient items and was termed perceived utility. The third dimension consisted of the social presence items; personal, warm, sociable, and informal. A final factor was comprised of the confidential and private items, and thus was termed perceived privacy. 83 Table 3.13: Summary Statistics for Perceived Channel Characteristics 1 Variables * Mean S.D. lEasy/difficult 1.6 1.2 Uncomfortable/com fortable 6.2 1.1 jCamplex/simple 5.8 1.3 Flex ible/inflex ible 2.3 1.4 Lccessible/inaccessible 1.9 1.4 Effective/ineffective 1.8 1.0 Useful/not useful 1.5 .9 Efficient/inefficient 1.8 1.1 Impersonal/personal 4.4 1.5 Cold/warm 4.4 1.2 Sociable/unsociable 2.7 1.4 Informal/formal 2.6 1.5 Not confidential/confidential 3.8 1.8 Pr ivate/public 4.2 1.7 Convenient/inconvenient 1.7 1.1 Unreliable/reliable 6.2 1.1 Slow/fast 5.8 1.4 * Respondents were given the stimulus phrase "Typically, using electronic mail is...". The endpoints of the 7 point semantic differential scales are presented (e.g. easy=l; difficult=7). 84 Table 3.14: Factor Analysis Results for Perceived Channel Characteristics Factor 1: Factor 2: Factor 3: Factor 4: (Variables Ease of Use Utility Social Presence Privacy jEa sy/d ifficult .64 .31 -.11 .00 Uncomfortable/comfort. -.62 -.31 .21 .01 ^Complex/simple -.62 -.18 .15 .15 Flex ible/inflex ible .58 .34 -.36 -.02 Accessible/inaccess. .47 .37 .08 -.04 Effective/ineffective .25 .67 -.19 -.10 Useful/not useful .27 .65 -.16 .05 Efficient/inefficient .46 .58 -.13 -.22 Imper sonal/per sonal -.12 -.11 .77 .20 Cold/warm .01 -.20 .69 .11 Soc iable/unsoc iable .18 .32 -.54 -.01 Informal/formal .23 -.04 -.40 .20 Not confidential/conf. -.03 -.10 .09 .88 Pr ivate/public .11 .09 .04 -.74 Convenient/inconvenient .49 .48 -.10 -.07 Unreliable/reliable .22 -.37 .11 .14 Slow/fast -.20 -.33 .22 .09 85 Three items did not load cleanly on any factor and were discarded. Perceived convenience cross-loaded on both the utility factor and the ease of use factor, a finding that seems intuitively meaningful. Perceived reliability and perceived speed also did not load on any factor. Four scales were created using the mean of the high loading items on each factor. Internal reliabilities for these scales were assessed using Cronbach's alpha. Each of the scales achieved an internal reliability in excess of .70, paralleling the findings of Ruchinskas and Svenning {ease of use alpha=.78, utility alpha=.78, social presence alpha=.71, and privacy alpha=.80; Table 3.15). Generally scale means reflect perceptions of electronic mail as easy to use (mean=6.1) and useful (mean=6.2), but show respondents mixed feelings with regard to its social presence and privacy (means=3.9). Individual/Demographic Variables. Age, organizational tenure, and electronic mail experience were captured using ratio scales. Experience was the sum of the number of months using the current system, plus months using any prior systems. The age of respondents ranged from 20 to 65 years, averaging approximately 36 years. The average respondent had been with the organization for 7 years. Prior experience with electronic mail ranged from one month to over 10 years, with average falling at just over three years. Thus this was a fairly experienced sample. Level was measured on a five-point scale ranging from upper management (1) to clerical people (5). Because of the few upper level 86 Table 3.15: Inter-item Correlations, Reliability and Summary Statistics for Perceived Channel Characteristic Scales Correlation Matrix EASE OF USE SCALE 1 2 2 1 1. Easy * - 2. Comfortable .56 - 3. Simple .52 .43 4. Flexible * .46 .50 .46 5. Accessible * .40 .39 .30 .35 Cronbach's alpha = .78 Scale Mean = 6.1 Standard Deviation = .93 Correlation Matrix UTILITY SCALE J L 2 1. Effective * - 2. Useful * .58 - 3. Efficient * .52 .51 Cronbach's alpha = .78 Scale Mean = 6.2 Standard Deviation = .86 (continued) Table 3.15: (cont inued) Inter-itan Correlations, Reliability and Summary Statistics for Perceived Channel Characteristic Scales Correlation Matrix SOCIAL PRESENCE SCALE 1 2 3 1. Personal - 2. Warm .60 3. Sociable * .42 .42 4. Informal * .30 .20 .34 Cronbach's alpha = .80 Scale Mean = 3.9 Standard Deviation = 1.0 Correlation PRIVACY SCALE 1 1. Confidential - 2. Private * .67 Cronbach's alpha = .80 Scale Mean = 3.9 Standard Deviation = 1.49 * Items were reverse coded. 88 Table 3.16: Summary Statistics for Individual/Demographic Variables * ** Variables Mean S.D. Age 35.9 8.35 Organization Tenure (in years) 7.0 5.27 EM Experience (in months) 38.6 35.66 Level * 1.99 .71 Education ** 3.5 1.05 4 point scale with l=manager, 2=professional, 3=technician and 4=clerical 5 point scale of highest level of education attained with l=high school, 2=some college, 3=college, 4=masters and 5=Ph.D. 89 managers, however, this group was combined with other managers, [producing four levels: managers, professionals, technicians, and clerical people. A majority of the sample fell in the professional category (144), both limiting variance on this variable and creating a mean 2.0. Education was measured on a five point scale ranging from thigh school (1) to Ph.D. (5). Most (83%) had college degrees, with 50% going on to receive graduate degrees. Demographic averages reveal a somewhat young, but highly educated sample who have had a substantial amount of exposure to electronic mail. This fits well with other independent variables that show tasks to be non-routine, and show ipositive endorsement of electronic mail. System Access Variables. Two of the three access variables - number of (people with whom subjects share a terminal or work station, and the number of electronic mail users in subjects' work groups were measured using ratio scales. The third, the extent to which work contacts outside the work group were likely to be electronic mail users was captured using percentages. It was felt that this type of response format would be easier for respondents then asking them to estimate the number of non-group contacts who are also electronic mail users. Most people had their own workstation (60%), although some shared with as many as 10 or more people (8%). Hie number of group members using electronic mail varied considerably around a mean of 15 (S.D.=18.91). This implies that a variety of situations exist, including groups where no one but the respondent uses the system, and where everyone used it. in addition, the proportion of contacts outside the workgroup who use electronic is substantial (mean=56.29%). 90 Table 3.17: Summary Statistics for System Access Variables Variables Mean S.D. Number of people sharing terminal * 2.0 3.25 Number of people in group using EM ** 14.78 18.91 Percent of non-group contacts using EM 56.29 35.70 * Outliers (10 or more) recoded to 10 ** Outliers (100 or more) recoded to 100 91 Analysis Strategy The first research question dealt with identifying the uses of electronic mail. In addition the dimensions of use were of interest. Earlier in this section, the factor analysis methods used to uncover dimensions of use were discussed. Multiple regression analysis was used to address the second research question; assessing the relationship of task environment, perceived channel characteristics, individual and demographic variables, and electronic mail system access variables to the patterns of electronic mail use. These four sets of independent variables were entered in stepwise, hierarchical regressions as predictors of task and social use of electronic mail. The system access variables were treated as control variables and forced in first, with other variable sets allowed to enter subsequently in any order. Because there were so many variables, a minimum F of 2.0 was selected to keep variables with little added contribution out of the analysis. Furthermore, in order to maintain an acceptable number of cases, means were substituted for missing data on independent variables. Listwise deletion of cases missing on the dependent variable was employed. An F test of the change in variance explained recommended by Kerlinger and Pedhazur (1973) assessed the significance of the 3 other variable sets, once the system access variables had been entered. The multiple regression analysis was used to assess the extent to which independent variables predict the amount of task use and social use. A supporting analysis involved clustering people into groups according to their usage style. Four groups, reflecting low task and 92 social use, low task and high social use, high task and low social useJ and both high task and social use were created by splitting the sample at use means. A discriminant analysis was then performed to determine which variables best differentiated among the groups. Again, because of the number of variables, a stepwise procedure was employed. The ability of the discriminating variables to classify people accurately into groups was also assessed. The results of these analyses are presented in the next chapter. 93 CHAPTER FOUR Analysis Analysis of the survey data provides an interesting picture of the nature of electronic mail use in this organization. The analysis jsegins with further interpretation of the types of electronic mail juses. Relevant examples from the qualitative phase of the research will be interspersed with quantitative results wherever possible. The interpretation of findings then focuses on the dimensionality of usage types obtained in the factor analysis. Later sections describe results of the predictive analyses for both task and social use. These findings highlight the interdependence of the technological system and social system, demonstrating the importance of the communication infrastructure of the organization. The interdependence of attitudes and electronic mail use are also illustrated. Finally the usefulness of electronic mail as a tool of social integration is exhibited in the analysis of factors that predict social use. The multiple regression findings are then supplemented with results of a discriminant analysis. A preliminary attempt to categorize users by their style of use, and then examine factors that differentiate among the styles reemphasizes the importance of infrastructure and attitudes. Types of Uses Electronic mail users in this organization clearly applied their system in many novel and interesting ways. Each type of use, however, did not occur with the same reported frequency and some interesting 94 comparisons of the mean ratings on individual items are discussed below. Telephone substitution was rated as the most frequent individual use (mean=3.5), corresponding to a company finding that electronic mail users spend half as much time on the telephone as non-users (Steinfield, 1983). Information exchange types of uses such as distributing/providing information, information seeking, and giving/receiving feedback were also rated as occurring frequently. In addition, two uses not given much attention in prior research - learning of interesting things or events and keeping track of what is happening around the company - received high ratings. Many of the more interpersonally involving applications received low mean ratings. Negotiating, resolving conflicts, and getting to know someone all had mean scores of less than 2 on the 5 point scale. This cannot be interpreted as support for the social presence notion, however, since ratings are not relative to other media. That is, someone who rarely uses electronic mail for negotiating may not be avoiding this particular medium, but rather might simply not do much negotiating on the job. Future research can address this issue by ^assessing the frequency of these interpersonally involving activities, Lnd then noting their relative distribution among all media. Whether infrequent or not, the finding remains that these are occasional uses of electronic mail for some, and research on how this medium might alter communication activities, such as negotiating processes, is necessary. A few activities did not receive particularly high ratings, yet 95 highlight potentially important and unique applications of this new medium for organizations in the future. Perhaps most interesting is the use of the electronic mail system for ad hoc opinion polling. Through the use of distribution lists, users had instant access to virtually all other electronic mail users. One particularly innovative application observed during the course of this research demonstrates the power of this medium. A small design team facing a critical decision point surveyed the user population for their reactions to jseveral design choices. Within a day they had received a couple of hundred responses, with an overwhelming majority favoring one design option over another. This team then fed back the survey responses to users, and opted for the majority's choice. No other communication technology could possibly have allowed such rapid access to this organization's remote technical expertise. Of course, the effectiveness of opinion polling is dependent upon the limited extent to which people are subjected to surveys. A pool of experts may quickly tire of frequent efforts of others to tap their knowledge; a fact social scientists have long known. Therefore the limited occurrence of opinion polling may be evidence of necessary self regulation among the users in order to preserve a precious resource. A few activities virtually were not done by electronic mail. The monitoring of subordinate performance, and discussion of confidential [natters are good examples here. The former may reflect a desire to (maintain a more personal environment and avoid "1984-like" monitoring (or surveillance. In discussions with users a clear bias against any form of management monitoring of system use was apparent, and this more 96 than likely affected managers' willingness to use electronic mail for performance monitoring. Despite this "hands-off" policy by management, nowever, people still avoided using the system for confidential natters, reflecting a concern for the adequacy of electronic mail security. Dimensions of Uses With the reduction of items due to limited variance, a factor analysis yielded a simple task and socio-emotional use dichotomy. Although a more varied set of usage dimensions was expected, containing Luch clusters of items as information exchange, scheduling, monitoring, Locial uses, etc., this outcome fits with many existing conceptions of ■organizational communication. In a review of the organizational communication area, O'Reilly and Pondy (1979, page 123) note that "the expression of emotion and the maintenance of relations may be as important to a sender and receiver as communication for coordination and control. The ten items making the task use factor include such varied purposes as telephone substitution, information exchange items, project coordination and monitoring, keeping records of agreements, sending pointers to files, scheduling and brainstorming. Respondents reported using the system often for these purposes, implicitly voicing support for the electronic mail system as a task communication tool. Social use was made up of six items with the most frequent use - learning of interesting things or events - reflecting an entertainment and news gathering function. Some people use electronic mail as an 97 electronic newsletter or magazine. Taking a break from work and filling up free time also fit well here. The other social uses include keeping in touch, participating in games or entertaining discussions, and organizing social activities. All these fit well with the ability of electronic mail to allow the formation of electronic communities of interest or clubs (Hiltz and Turoff, 1978; Wynn, 1980). That social uses occur less often again does not necessarily imply that people avoid the system for interpersonal use due to its low social presence. Two equally plausible explanations are possible. First, the frequency of social communication may be lower than task communication; highly probable in an organizational setting. Second, the system itself is designed and promoted as a tool for improving productivity. This might negatively influence motivations to use the system for personal reasons. Given the likelihood of the uneven absolute frequency of task and socio-emotional communication (O'Reilly and Pondy, 1979), a better test of the importance of social presence is to assess the extent to which it predicts the amount of each type of use. This analysis is described in the next section. Finally, the two dimensions of use are theoretically as well as empirically related. Much social communication is supportive of task- related activities. Farace et al. (1977), for example, consider socio- emotional communication as necessary for system maintenance. Here the two dimensions exhibited a moderate correlation of .33. That this relationship is not very strong is a function of the nature of social use items loading on this dimension. Many are purely entertainment- 98 oriented, and might not overlap with social communication aimed at the creation and maintenance of task-oriented relationships. 'Multiple Regression Analyses The four sets of independent variables did not exhibit equal ability in predicting task and social use. These results are discussed below. In order to assist in the interpretation of regressions, and particularly to sort through relationships among the independent variables, an intercorrelation matrix is provided (Table 4.1). Task Use Regression Results Using an F of 2.0 as the inclusion criterion, eight of the independent variables entered into the regression predicting the amount of task use (Table 4.2). These eight variables explained 52% of the variance in task use, significant at the .001 level (F=24.2). The contribution of the four independent variable sets will be discussed separately. System Access Variables. A positive relationship was expected between task use and access to relevant others outside the work group, access to work group co-workers, and access to terminals. Since access to terminals was indexed by the number with whom a terminal is shared, operationally a negative beta would support the this hypothesis. Two of the three hypotheses concerning system access factors were supported by the task regression analysis. Both the extent to which work contacts outside a person's work group were electronic mail users (beta=.29; p<.001) and the sharing of 99 Table 4.1: Correlations of All Independent and Dependent Variables 1 2 2 A 1. TASKUSE - 2 . SOCUSE .33 - 3. SHARE .36 - . 1 0 - 4. GROUPEMS .16 - . 0 2 - . 1 2 - 5. OUTGRPEMS .51 .19 -.29 .23 6 . EASEUSE .43 .29 - . 2 1 . 1 2 7. UTILITY .46 .30 -.23 . 1 0 8 . SOCPRES .33 .31 -.14 .07 9. PRIVACY .19 . 0 2 -.18 -.05 1 0 . ROUTINE -.14 .05 .09 -.07 1 1 . PRESSURE .07 .03 . 1 0 - . 0 2 1 2 . STANDARD .03 .17 .09 .08 13. ENVUNC .17 -.05 .07 .18 14. INTDFW . 1 0 . 1 0 -.04 .19 15. INTDPB .03 .09 .14 .07 16. CROSSLOC .35 -.08 -.04 .15 17. EDUC . 0 1 . 0 1 -.06 . 0 2 • 00 AGE .06 -.37 . 0 2 .07 19. CMPYRS - . 0 2 .40 - . 0 1 . 0 0 • o (N EXPER .35 . 0 2 -.18 .17 2 1 . LEVEL .09 .23 .05 - . 0 2 5 6 7 8 9 10 36 - 28 .70 - 13 .40 .41 - 2 0 .19 .19 . 1 2 - 15 I • o 00 -.07 . 0 0 . 0 1 - 19 - . 1 0 -.08 . 1 2 -.04 . 1 2 16 .15 .07 .06 .06 .36 1 0 .04 in o • - . 0 1 -.09 -.27 04 .07 . 1 0 -.06 - . 1 2 -.07 08 .03 .07 - . 0 2 I • o -j - . 1 2 1 0 -.04 . 0 0 .16 .03 - . 1 0 03 I • o uo O • 1 -.06 . 0 1 -.31 09 - . 1 0 -.07 -.06 -.07 . 0 0 .12 -.10 -.05 -.11 -.02 .00 .36 .19 .09 .15 .07 -.16 .04 .13 .15 .09 .05 .42 100 Table 4.1 (continued): Correlations of All Independent and Dependent Variables 1 1 1 2 13 14 15 16 17 18 19 2 0 L2 . STANDARD .09 - •13. ENVUNC .04 -.25 - | 14. INTDPW - . 0 2 . 0 2 .09 - 15. INTDPB .17 .05 .05 .53 - 16. CROSSLOC .27 -.06 . 1 0 . 0 2 .16 - 17. EDUC - . 1 1 -.23 .16 . 0 1 - . 0 1 .13 - 18. AGE .07 -.08 .18 - . 0 1 .05 . 1 0 . 1 0 - 19. CMPYRS .15 -.07 .14 -.07 -.03 . 2 0 - . 1 1 .65 - 2 0 . EXPER - . 1 1 -.03 .13 .04 -.06 .29 .23 .08 . 0 0 - 2 1 . LEVEL -.13 . 2 2 -.26 - . 1 1 - . 1 2 - . 2 1 -.37 -.25 -.32 - . 1 1 101 Table 4.2: Task Use Regression Results Variables * R-Square R Beta F Prob. Out-Group EM Contacts .26 .51 .29 24.37 < . 0 0 1 Share Terminal .31 -.36 -.19 11.90 < . 0 1 Cross Boundary Comm. .40 .35 C M • 18.30 < . 0 0 1 Utility .49 .46 .24 10.95 < . 0 1 Environmental Unc. .50 .17 .09 3.37 NS Pressure .51 .07 . 1 1 4.31 <.05 Ease of Use .52 .43 . 1 1 2.15 NS EM Experience .52 .35 .08 2.13 NS Multiple R = .72 R-Squared = .52 Standard Error = .64 Degrees of N = 187 Freedom = 8 , 178 F Prob. = 24.2 = < . 0 0 1 * Variables are presented in order of entry, with an F of 2.0 required to enter. Control variables were forced in first. ** Standardized regression coefficients are presented since independent and dependent variable scale ranges differed. 102 terminals (beta=.19; p<.001) were significant predictors of task use. These two variables alone accounted for 31% of the task use variance, lending credence to the proposition that both access and availability of relevant others are necessary precursors to task-related uses of electronic mail. The number of group members using electronic mail did not enter the equation, and had relatively small simple correlations with task use (.16), out-group electronic mail contacts (.23), and access (-.11). It is unlikely then, that the group electronic mail user variable would have accounted for variance in task use had the other system access variables not entered first. One explanation is that the likely proximity of group members allows easy access to face- to-face communication, reducing reliance on electronic mail. On the other hand, contacts outside the work group might be less accessible to face-to-face interaction, increasing the potential value of electronic mail. These results correspond to frequent findings of distance as a critical variable in explaining use of different communication media (Steinfield, 1981, 1983; Irving, 1981; Ruchinskas, 1982; Conrath, 1973) . The primary importance of system access fits well with the theoretical expectation that access is preliminary requirement for electronic mail usage. That is, no matter what the nature of a person's job or orientation to electronic mail, without access to the system and to relevant people, task uses of electronic mail are quite unlikely. System access variables are not strongly related to other independent variables, but do show small to moderate positive correlations with perceived utility and ease of use, and with prior 103 jexperience using electronic mail. Consistent with the hypothesis of access being the precursor, the flow of influence probably is from access to perceptions, although the analysis is purely correlational and cannot determine causal flows. However, it may be argued that only with access will people perceive the system to be easy to use and useful. Finally, those with prior experience are both more likely to have access, and more likely to have electronic mail contacts. iTask Environment Variables. A negative association was predicted between task related uses of electronic mail and task routineness. Positive relationships were expected between task use and time pressures on the job, task interdependence, perceived environmental uncertainty, and cross boundary communication need. Only two of these hypotheses were supported by the regression findings. The cross boundary communication need variable was the most prominent of the task environment variables (beta=.24; p<.001). Use of electronic mail for task related purposes is thus increased by this type of communication need, reflecting users' perceptions of electronic mail as a good tool for spanning work group as well as geographic boundaries. The two other task characteristics which entered the equation were less influential, with only job pressures reaching the .05 level of significance. Because of the overall number of significance tests, a probability level of p<.05 should be regarded with caution. One out of every 20 significance tests could be significant at the .05 level just by chance. Note that the standardized coefficient for this variable 104 exceeds its simple correlation with the dependent variable; evidence of suppression (Cohen and Cohen, 1975). Other task characteristics had little relationship to task usage. One explanation may be that electronic mail is used for both routine and non-routine task purposes, making it equally useful for either standardized and routine jobs or non-standard and non-routine jobs. Perceived Channel Characteristics. Four perceived attributes of electronic mail were hypothesized to relate positively to task use; utility, ease of use, social presence, and privacy. Not surprisingly, the perceived utility of using electronic mail was the most important of the four channel characteristic scales (beta=.24; pC.Ol). However, none of the other channel characteristic hypotheses were supported. It is clear from Table 4.1, however, that both perceived ease of use and perceived social presence strongly correlate with perceived utility (.70 and .40, respectively). Both also correlate with task use, but to a lesser extent (.43 and .33, respectively). One interpretation may be that perceived utility is an intervening variable; electronic mail is perceived to be more useful if it is thought to be easy to use and to a lesser extent personal. Then perceptions of utility influence decisions to use the system for task purposes. Also likely is the possibility of an interdependent relationship between perceived utility and task use. That is, if someone uses the system effectively for some task, then their perceptions of electronic mail utility should increase. This in turn would increase the likelihood of future task use. 105 The perceived privacy or confidentiality of electronic mail was not related to task use. Given the reluctance to use the system for confidential matters, perceived privacy is clearly an issue. However, since this usage item had so little variance, it was not included in task use. In fact, most of the task uses might involve information that is not sensitive; hence whether or not the system was perceived as private would not affect task usage. Individual/Demographic Variables. Educational level and experience were predicted to positively relate to task use, while a negative relationship was hypothesized for age, level, and organizational tenure. Demographic variables, however, appear to play an insignificant role in predicting task use. Only experience with electronic mail exhibited a moderate zero order correlation with task use (.35), but once the system access variables had entered, it offered no new explanatory value (beta = .08, NS). Quite possibly, with more experience also comes a greater number of contacts on the system as well as a greater likelihood of having a private terminal. Level, tenure, age, and education were unrelated to task use. Task applications of electronic mail are seemingly equally relevant to respondents of different demographic categories. Summary of Task Regression Results Access to relevant others as well as to a terminal or workstation were clearly critical factors in explaining task use. However, a full 2 1% of additional variance is accounted for, primarily by the need for communication across boundaries, and by perceived channel attributes. 106 This finding is consistent with the expectation that access variables are necessary, but not sufficient predictors of task use. Since, the access variables were forced into the equation first, a test of the significance of variance accounted for by the remaining variables was performed. This F test of the difference in R-squared is significant at the pC.OOl level (F = 13.0). Social Use Regression Results A very different pattern of findings is obtained when predicting the amount of social use of electronic mail. Seven variables entered using the F of 2.0 criterion, accounting for 31% of social use variance (F=12.2, pC.OOl; Table 4.3). As before, the four variable sets are discussed separately. System Access Variables. As with task use, access to coworkers both in and outside the work group were predicted to relate positively to social use, while increased sharing of terminals was expected to be associated with less of this type of use. However, in contrast to the task use equation, none of the system access variables significantly related to social use. The lack of influence of both the number of work group members as well as work contacts outside the work group using electronic mail may be illustrative of the differential composition of task and social networks. This is analogous to the finding that formal and informal networks do not overlap (Farace et al., 1977; O’Reilly and Pondy, 1979). A possible explanation for the unimportance of the sharing of terminals variable might lie in the occasions for social use. Should 107 Table 4.3: Social Use Regression Results Variables * R-Square R Beta F Prob. Out-Group EM Contacts .04 .19 .05 .71 NS Organizational Tenure .18 -.40 -.23 8.73 < . 0 1 Social Presence .25 .31 . 2 1 10.51 < . 0 1 Utility .27 .30 .14 4.86 <•05 Age .29 -.37 -.17 4.79 <.05 Well-Defined Tasks .30 .17 . 1 0 2.98 NS Intra-unit Interdepend. .31 . 1 0 .09 2.19 NS Multiple R = .55 Degrees of Freedom - 1, 193 F = 1 2 . 2 R-Squared = .31 N = 201 Prob.= < . 0 0 1 Standard Error = .69 * Variables are presented in order of entry, with an F of 2.0 required to enter. Control variables were forced in first. 108 people confine non-task use to non-working hours or lunch breaks, a shared terminal may be more accessible. In any case, these findings for system access variables and social use demonstrate the importance of taking into account the type of use when attempting to predict acceptance and use of electronic mail. Different types of use are influenced by different variables. Task Environment. Task characteristics and other contextual variables also did not relate significantly to social use, supporting the expectations from Chapter 2. The organizational communication literature relates task communication to the reduction of uncertainty created by non-routine tasks, task interdependence, or environmental uncertainty, but does not address socio-emotional communication in this regard. It could be argued that as tasks become more complex and non routine, people would use social communication to establish relationships with others who might help them deal with uncertainty. However, an equally plausible argument can be made for greater social communication in routine task situations. More time might exist for socializing, and the intrinsic interest in the job might be less. Social relationships then become important, making up for boring tasks. The two effects might cancel each other out, resulting in no relationship between task attributes and social use of electronic mail. Perceived Channel Characteristics. Perceived utility, ease of use, social presence, and privacy were all hypothesized to bear positive associations with social use. Two of the perceptions of electronic 109 mail did significantly predict social use. The perceived social presence scale was foremost among these (beta=.21; pC.Ol), followed by perceived utility (beta=.14, p<.05). Ease of use did not make an independent contribution once social presence and utility had entered in the equation. Here again, the strong interrelationships among the perceived channel attributes influence which variables enter into the equation. The primacy of one perception over another in influencing social use is difficult to assess with much confidence as each showed a similar positive correlation with the dependent variable (social presence .31, utility .30, and ease of use .29). Social presence, however, had the largest standardized coefficient, a parameter that reflects the influence of other independent variables. The importance of social presence for the socio-emotional use of electronic mail has considerable theoretical merit. The privacy scale once again was not related to social usage, again due to respondents' general avoidance of electronic mail for confidential matters. Individual/Demographics. As with task use hypotheses, a positive relationship was predicted between non-task uses and educational level and experience. Similarly, a negative association was expected for age, level, and organizational tenure. However, unlike the previous task use findings, two demographic variables, organizational tenure (beta=-.23; p<.01) and age (beta=-.17; p<.05), were among the most influential determinants of the social use of electronic mail. The newer to the company and the younger an individual was, the greater his or her social use. Tenure and age were of course highly correlated 110 (.65), since new employees also tend to be younger. However, the fact that age made an independent contribution even after tenure had entered the equation supports the distinction between these variables. New employees may use the system as a socializing mechanism to help make contacts and develop a social network. Those who have been around the company longer may not need electronic mail for this, their social network has already been developed. An explanation for the independent importance of age may lie in the greater exposure younger people have had wth computer games. They may be more acclimated to the use of computers for entertainment, and hence may be more likely to use electronic mail for social or recreational purposes. Education, level, and experience with electronic mail did not enter into the regression equation, and do not appear to be a factor in determining social uses of electronic mail. Summary of Social Use Regression Results A significant amount of the variance in social use is explained by demographic variables and by perceptions of electronic mail attributes. In particular, it appears that social use is more prevalent among newer and younger employees, who view using electronic mail as sociable and easy to use. System access and task variables had little influence. The F test of the difference in R-squared once system access variables had entered was significant (F=12.6; pC.OOl). Summary of Regression Analysis A summary of hypothesis tests regarding the direction of relationship between independent variables and the two types of use is 111 presented in Table 4.4. Hie overall pattern of findings seems to indicate that task use is dependent upon the existence of a task communication infrastructure that is accessible via electronic mail. It further requires a positive orientation to electronic mail. 'Finally, to the extent that the system can satisfy communication needs not met via other media (more specifically, timely and accurate interaction with coworkers in other locations), it will be utilized. On the other hand, social use is not dependent upon the task communication infrastructure. Rather, if low tenure and age are thought of as an indication of an undeveloped social network, then a different kind of communication need is met via electronic mail. It is the need to become integrated into the social network of the organization. However, as with task communication, a positive orientation to electronic mail is important, particularly with regard to the perceived social presence afforded by the medium. Categorizing User Types In order to provide a more descriptive picture of who uses electronic mail for social and/or task purposes, a typology of user types was developed. The first group is comprised of 49 people who are relatively light users of the system for both task and social purposes. Their scores on these two dependent variables are less than the mean (or median, as these measures of central tendency were equal). The second group, with 36 respondents, are above the mean for social use, but below on task use. Primarily task users form the third group, containing 37 people. Finally, the largest group is made up of 61 people who scored above the mean on both types of uses. 112 Table 4.4: Sunmary of Expected and Obtained Relationships of Independent Variables to Task and Non-Task (Social) Related Uses of Electronic Mail DEPENDENT VARIABLES INDEPENDENT VARIABLES Task Uses Non-Task Uses Expected Obtained Expected Obtained iTask/Environmental Variables Task Routineness - 0 0 0 * Time Pressures + -j. * * 0 0 * Task Interdependence + 0 0 0 * Environmental Uncertainty + 0 0 0 * Cross Location Communication Need + , kkkk 0 0 * Perceived Channel Characteristics Utility + 4. kkk . + + kk Ease of Use + 0 + 0 Social Presence + 0 + 4. * * * Privacy + 0 + 0 Individual/Demographic variables Age - 0 - _ * * Education 0 + 0 Experience with Electronic Mail + 0 + 0 Organizational Level - 0 - 0 Organizational Tenure — 0 — ... * * * 113 Table 4.4 (continued): Sunmary of Expected and Obtained Relationships of Independent Variables to Task and Non-Task (Social) Related Uses of Electronic Mail DEPENDENT VARIABLES INDEPENDENT VARIABLES Task Uses Non-Task Uses Expected Obtained Expected Obtained System Access Variables Sharing of Terminals (operational opposite of access) _ *** 4. 0 Coworkers outside group on system + 4 **** 4 Q Coworkers within group on system + 0 + 0 * Confirmation of the hypothesis of no by an insignificant beta. relationship is provided ** Significant at the p<.05 level. *** Significant at the p<.01 level. **** Significant at the p<.001 level. 114 Further insights into the types of people falling into these groups, as well as support for the validity of this typology are provided by examining certain descriptive information broken down by these categories (Table 4.5). People in different job categories are pot equally likely to belong to each group. Engineers tend to fall into the light user (32%) and primarily social user group (38%). Research people are spread across all types, but most are either heavy users (33%) or light users (28%). Programmers are the most likely to be heavy users (46%), perhaps because they spend so much time working with computers anyway. Finally, when all the other job categories are combined to form a type of non-technical job classification, many fall into the heavy (33%) and primarily task groups (27%), although another quarter are light users. The actual frequency of messages sent and received appears to support the classification of groups (Table 4.6). That is, users who are above average on both task and social use consistently report sending and receiving more electronic messages; of both the personal and distribution list variety. Likewise, light users report the fewest number of messages sent and received. Lastly, primarily task users report more messages sent and received than social users, but less than the heavy users. This follows from the greater amounts of task use than social use. An analysis of variance shows that for each type of message measure, the differences among groups are significant. These linear relationships lend a certain amount of validity to the user group typology. 115 Table 4.5: Job Types of Electronic Mail User Groups [USER TYPE Light Primarily Social Primarily Task Heavy Engineers 12 (32%) 14 (38%) 5 (14%) 6 (16%) JOB TYPE Research Programmers Other 12 (28%) 12 (23%) 13 (25%) 8 (19%) 8 (15%) 7 (14%) 9 (21%) 8 (15%) 14 (27%) 14 (33%) 24 (46%) 17 (33%) 39 (100%) 43 (100%) 52 (100%) 51 (100%) Table 4.6: Number of Messages Sent and Received by User Groups USER TYPES PERSONAL Sent mean S.D. MESSAGES Received mean S.D. DISTRIBUTION LIST Sent Received mean S.D. mean S.D. Light Primarily Social Primarily Task Heavy 4.8 (5.9) 7.4 (18.6) 17.9 (24.6) 21.7 (20.0) 4.5 (6.4) 15.3 (42.4) 21.5 (24.6) 26.2 (31.8) 1.3 (2.3) 36.9 (31.9) 2.3 (5.1) 71.5 (92.8) 5.3 (8.7) 75.4 (74.3) 8.1 (10.2) 124.2 (129.4) * An analysis of variance shows user groups to be significantly different on each message frequency measure. (pC.OOl for each, except personal messages received, where pC.Ol) 116 Discriminant Analysis Findings The four sets of independent variables were entered into a step wise discriminant analysis in order to assess their ability to differentiate among the four user types. In order to enter the analysis, a probability of .05 was required. Five variables proved to be significant discriminators, to a large extent paralleling earlier regression results. It should be noted that discriminant analysis is generally reserved for situations in which groups exist based on truly discrete measures. Artificially creating groups based on median splits of continuous variables heightens the chances of error in classification as individuals who are just above or below medians may rightfully belong in the opposite group (McLaughlin, 1980). Thus the possibilities of incorrect classification are increased for this analysis. However, this analysis is useful in that it provides a preliminary attempt to isolate specific types of users rather than just uses, although as such it should be regarded as exploratory. As McLaughlin notes, multiple regression is a more appropriate analysis when faced with continuous dependent measures. Therefore the discriminant analysis should be viewed only as supplemental to the earlier regression analysis. Two of the three functions obtained account for a significant amount of variation among groups (Table 4.7). Five discriminating variables entered into the analysis at the p<.05 criterion level. The standardized canonical discriminant function coefficients were varimax rotated to assist in interpretation and are presented in Table 4.8. 117 Table 4.7: Canonical Discriminant Function Summary Statistics Wilks' Lambda Percent of Canonical Prior to Function Variance Correlation Removal Chi-Square df P 1 76.27 . 6 6 .46 139.24 15 . 0 0 0 1 2 20.35 .41 .80 38.74 8 . 0 0 0 1 3 3.39 .18 .97 5.90 3 . 1 2 Table 4.8: Rotated Standardized Discriminant Function Coefficients Sanction 1 Function 2 Function 3 Out-Group EM Contacts .74 .08 o o • Cross Boundary Comm. .29 .47 .41 Organizational Tenure . 0 1 - . 2 2 .85 Perceived Utility r- o • I . 8 8 - . 0 1 Share Terminal -.55 .34 .08 118 The first function reflects a strong positive contribution for the system access variables, both the number of out-group electronic mail contacts, and access to a terminal. The second function contains large coefficients for perceived utility and cross boundary communication need. The third function, however, does not explain a significant amount of group variation once the variation from the first two functions has been removed. Therefore, any interpretations should be viewed as tentative, although the pattern of coefficient weightings and correlations with variables appears to make theoretical sense. This function shows positive contributions from organizational tenure and to a lesser extent, cross boundary communication need. These same variables were among the most important in predicting the amount of task use and social use in earlier regression analyses. Interpretation of the functions is aided by an examination of the rotated structure matrix, containing the correlations of variables with discriminant function scores (Table 4.9). The first function appears to be a combination of infrastructure or access variables. As such, it resembles the important predictors of task use. Indeed, the group centroids on this function show separation of the two groups having high task use from the two groups with low task use (Table 4.10). Thus, someone who has many contacts around the organization who use electronic mail, and has his or her own terminal seems more inclined to be a heavy task user. The second function, however, correlates most strongly with perceptions of the utility and ease of use of electronic mail. Group centroids indicate that this function serves to separate people who are 119 1 Table 4.9: Varimax Rotated Matrix of Correlations of Independent Variables with Discriminant Functions Function 1 Function 2 Function 3 Out-Group EM Contacts .80 .13 -.15 Share Terminal -.64 .25 . 1 2 EM Experience .33 . 1 0 .07 Group EM Users .19 . 1 0 .03 Routineness -.17 -.07 -.03 Privacy .14 . 1 0 -.06 Education .14 - . 0 2 . 0 2 Utility .06 .79 -.09 Ease of Use .14 .50 -.13 Social Presence . 0 2 .33 - . 0 2 Inter-unit Interdepend. . 0 1 .19 .09 Intra-unit Interdepend. .08 . 1 0 -.03 Organizational Tenure -.03 -.19 .91 Age -.03 - . 1 0 .54 Cross Boundary Comm. .19 .34 .54 Level - . 1 0 .15 -.30 Pressure -.17 . 0 2 .23 Environmental Unc. . 0 1 .07 .13 Well-Defined Tasks . 0 2 .07 -.08 1 2 0 light users from all other groups. Thus, people who do not have a positive orientation to electronic mail do not use it for either task or social purposes. A plot of group centroids for the two significant functions (Figure 4.1) graphically demonstrates these spatial relations among the four groups. Clearly the two groups exhibiting high task use are most similar on these dimensions, while the light group is the most dissimilar to the other three. Separation between high and low social user group would have been achieved by the third function, had it accounted for a significant amount of group variation. It correlates most highly with the tenure and age variables found earlier to explain variation in social use, and with the existence of coworkers in other locations. Classification Analysis A final component of the discriminant analysis involves the extent to which the five discriminating variables are able to correctly classify people into the four user groups. The probability of belonging to each group was adjusted by the group sizes, and group membership was predicted from the five variables. Predicted membership is then compared to actual membership, providing an estimate of the accuracy of classification. Acknowledging that the percentage of correctly classified might be inflated (McLaughlin, 1980), results of the classification analysis are presented in Table 4.11. Slightly more than 57% of respondents were correctly classified. It appears that these variables do quite well in classifying light users and heavy users, but run into problems distinguishing the 121 Table 4.10: Group Centroids in the Discriminant Space GROUPS Function 1 Function 2 Function 3 Light Users -.71 -.76 . 2 1 Primarily Social I 00 I • o C T l -.59 Primarily Task .53 .37 . 6 6 Heavy Users .79 .43 - . 2 1 Figure 4.1: Plots of Group Centroids for Function 1 (X Axis) by Function 2 (Y Axis) .8 .6 .4 GROUP 4 .2 GROUP 3 .10 — . 8 -- . 6 ---.4---.2----0----.2----.4----. 6---.8— .10 GROUP 2 .2 .4 .6 GROUP 1 . 8 122 Table 4.11: Classification Analysis Results No. of Predicted Group Membership Actual Group Cases 1 2 3 4 1. Light Users 49 30 (61.0%) 7 (14.3%) 4 ( 8 .2%) 8 (16.3%) 2. Mostly Social 37 13 (35.1%) 15 (40.5%) 0 ( 0 %) 9 (24.3%) 3. Mostly Task 36 2 (5.6%) 4 ( 1 1 .1%) 13 (36.1%) 17 (47.2%) 4. Heavy Users 61 2 (3.3%) 6 (9.8%) 6 (9.8%) 47 (77.0%) Percent Correctly Classified = 57.38% 123 primarily social users and the primarily task users. The high task/low social use group was often mistakenly placed in the high task/high social use group, further illustrating the closeness of these groups. The primarily social use group was most often incorrectly assigned to the light user group, their nearest neighbor in the discriminant space according to Figure 4.1. Summary of Discriminant Analysis Conclusions from this analysis remain tentative, based on the method of determining group membership. However, a picture of potential user types is provided, with other usage frequency data supporting the typology. Based on the discriminant analysis, it appears that heavy task users, whether using electronic mail also for social purposes or not, are quite dissimilar from low task users. Variables indicative of the electronic mail communication infrastructure differentiate well between the high and low task users, paralleling earlier regression findings. The orientation toward electronic mail discriminates the light users from all others. Finally, tenure and age, possibly indicative of the need for integration into the organization’s social network, and communication need weakly differentiate between high and low social users. This also substantially parallels the social use regression results. 124 : CHAPTER-FIVE Conclusions and Discussions Question One: The Uses of Electronic Mail Types of Uses The first research question asked what the purposes of electronic mail use in organizational settings are. Based on the qualitative findings, as well as survey results, it is concluded that the electronic mail system in the study organization is used for a wide assortment of purposes, demonstrating a greater diversity of usage purposes than past research would predict. As Hiltz and Turoff (1978) noted, people have underestimated the potential uses of this medium and computer mediated communication allows many forms of interaction that otherwise could not take place without great difficulty. Over thirty different purposes were extracted from interviews with users and observations, with a large proportion not anticipated in much of the literature on computer mediated communication. A conclusion in much of the earlier literature is that computer mediated communication systems, as a low bandwidth channel which allows limited nonverbal communication, would be used more for tasks which do not require much interpersonal involvement (Heimstra, 1982; Rice, 1980). Without a basis on which to compare among media, which is beyond the scope of this research, the question of whether the electronic mail system is used less for interpersonally involving tasks cannot be answered. To be sure, the mean scores are higher for purposes which seem to fit the "low interpersonal involvement" 125 description. These include the various task related information exchange uses. On the other hand, activities generally thought of as requiring a great deal of interpersonal involvement, including negotiating, conflict resolution and getting to know someone (Rice, 1980; Short et al., 1976), received very low scores on the usage scale. The alternative hypothesis, however, that these communication situations are less frequent is quite plausible. Another way of viewing the use data, taking into account that some people simply do not do much negotiation or conflict resolution, is to examine the proportion of respondents who never use electronic mail for these purposes, or conversely, the proportion who have at least occasionally done so. Since people were given the option of noting that a particular activity is not something they do at work, (i.e. the question was not applicable) those who checked never might be viewed as avoiding the mail system for some purpose. As shown in Table 5.1, a substantial number of people do occasionally use electronic mail for these kinds of activities. Social presence is perhaps more appropriately thought of as a perceptual variable that may be used to predict the use of various communication media, rather than as an inherent attribute of media. In the discussion of regression analyses, where use is predicted from subjective perceptions of social presence, a more illuminating measure of the influence of this variable is provided. The picture becomes even murkier when the assumption of no emotional or interpersonal content in an activity such as giving feedback or distributing information is questioned. While typically 126 Table 5.1: Proportion of Respondents Who Report at Least Seme Use of Electronic Mail for Each Purpose * PURPOSES OF USE Distribute/provide information Learn about events/things I'm interested in Send a message in place of a phone call Give/receive feedback on reports/ideas Seek task information from people I know jKeep track of company happenings Broadcast requests for information Coordinate activities of projects I'm on Schedule meetings/appointments Keep a record of agreenents/interactions Send/receive pointers to large files Keep in touch/maintain relationships Br a i nstorm/g enerate ideas Take a break from my work Forward messages to someone who is not on seme distribution list i 94 93 92 92 91 91 84 81 80 77 77 77 75 74 73 127 Table 5.1 (continued): Proportion of Respondents Who Report at Least Some Use of Electronic Mail for Each Purpose * PURPOSES OF USE % Monitor progress of projects I work with 69 Participate in entertaining events (games) or conversations 64 Advertise/respond to products for sale 56 Poll opinions on a topic 52 Resolve conf1icts/disagreements 52 Organize/coordinate a social activity 49 Ask questions .in a public setting (eg sending many copies) to force a response or action 48 Get to know someone 46 Fill up "free" time 46 Get a message directly to someone whose calls are normally screened 43 Carry on negotiations/bargaining 41 Discuss confidential matters 36 Ensure people use DLs for appropriate topics 30 Monitor performance of my subordinates 29 Find out about job/promotion openings 22 * Same use was defined as 2 or more on a scale from where 1 equals never, and 5 equals all the time. 1 to 5 128 though of as not requiring interpersal involvement, providing feedback, for example, might involve highly sensitive criticism. Others researchers have observed that people do imbed socio-emotional content within task related messages in computer mediated communication (Hiltz, 1978; Hiltz and Turoff, 1978; Kiesler, 1982; Phillips, 1982). Perhaps more significant issues are raised when examining uses not typically found in past empirical research. First, a number of uses which take advantage of the distribution list and carbon copy capabilities of electronic mail are noteworthy. Broadcasting information requests and opinion polling are examples of group communication for task purposes. These illustrate the ability of electronic mail to serve as a medium for accessing remote sources of information, transcending both organizational and geographic boundaries. Significantly, users may not be acquainted with people who respond to requests; they may only know the particular distribution list which is likely to contain potential information sources. The political consequences of certain uses should not be overlooked. Two uses identified in this research, asking questions in a public setting to force a response or action, and getting through to someone whose calls are normally screened, provide innovative examples of how a new technology can be adapted and used to serve political ends. Further political implications arise purely from the ability to create distribution lists. Interest groups can be formed having their own sets of goals. The process of coalition formation can occur in this fashion, and cross-locational coalitions might arise. The system allows distribution list founders to specify whether or not lists are 129 "open to the public" or not, making this medium more viable as a means of communication for politically-minded interest groups. This supports the importance of examining the influence of computer and communication technologies on power relationships in organizations (Danziger et al., 1982; Dutton, forthcoming; Dutton and Kraemer, 1980; Kraemer and Dutton, 1982) . Another set of uses demonstrates the multi-faceted social side of electronic mail. For instance, learning of interesting things or events, keeping track of what's happening around the company, and advertising all suggest that the electronic mail system was used much like a magazine or company newsletter. Participating in games or entertaining conversations, organizing social activities, taking breaks, filling up free time; keeping in touch, and getting to know someone further emphasize the many social and entertainment functions of electronic mail. That electronic mail was not perceived to be appropriate for very private applications was reflected in the lack of use for discussing confidential information. The inability to control where a message, once sent, could be forwarded, the permanence and retrievability of stored messages, and the potential ability of someone with computer programming skills to tap into the computers responsible for message transfer all are probable contributing influences here. Other uses were occasional applications for a minority of the respondents, but were generally not a part of the typical user's repertory. The avoidance of electronic mail for monitoring employee performance might be an attempt to maintain a personal work environment 130 jandT-based on interviews, appears to reflect a more general-hands-off (policy that management has taken with regard to peoples' use of the system. Equally likely is the avoidance of performance monitoring due to the lack of easy or accepted strategies for accomplishing this technically difficult activity over electronic mail. Monitoring distribution lists to ensure that topics are appropriate are rare. This use received low scores, most likely because founders, and occasionally members, of distribution lists each helped regulate the discussion on their list, and any one person would rarely have to send a message to this effect. Finally, the lack of use for finding about job or promotion openings most likely reflects the relatively infrequent number of times a person would be looking for a different job. Dimensions of Uses The clustering of uses into a task and social dimension is conceptually appealing. It fits with nicely with prior taxonomies of organizational communication, which usually include a task or production function and a socio-emotional or maintenance function (Katz and Kahn, 1978; Farace et al, 1977). The moderate positive correlation between the created indices of task and social use is accounted for by viewing some social communication as instrumental in nature. This is also consistent with the view of socio-emotional communication as fulfilling a maintenance function. One additional use, keeping track of what's happening around the company, did not load on either the task or social factor and may be evidence of a surveillance function. 131 Taxonomies of the functions of communication in society typically include surveillance (e.g. Lasswell, 1948; Schramm, 1971), and the attributes of electronic mail in this setting appear to make it an ideal medium for fulfilling this function. Broadcasting also did not load on either dimension, and may be evidence of an additional class of communication activities. It may be that certain other uses not in the factor analyses due to limited variance would have loaded with this application to form a broadcasting dimension. Such uses as advertising and opinion polling seem to include the one-to-many style of communicating of this application. The identification of a social use dimension, particularly one with several more entertainment-type applications, opens up many additional issues. One in particular is the question of whether this recreational use is functional or dysfunctional from an organizational standpoint. In much of the popular, as well as empirical literature on electronic mail, social uses other than keeping in touch or getting to know someone are rarely mentioned. In fact, as Marvin (1983) notes about the values surrounding telecommunications systems in general, a productivity bias often pervades this literature. Proponents of computer mail talk about the "communication payoffs," reasoning that electronic mail can make the large proportion of time devoted to communication activities more efficient and effective. A common example often mentioned here is the potential of electronic mail to help reduce time wasted on unproductive "shadow functions". These include such activities as time spent dialing busy telephone numbers or playing telephone tag (Bair, 1979 , Uhlig et al., 1979). 132 Despite this emphasis on the potential improvements to productivity afforded by computer mediated communication, not all use is so utilitarian in this strict sense (Marvin, 1983). For example, Marvin points out the social dialogue often found among users of the ARPANET, even though such use is officially frowned upon. Other authors have observed socio-emotional message content among users of computer mediated communication systems (Hiltz, 1978; Kiesler, 1982; Phillips, 1982; Siegel, Kiesler, and McGuire, 1981; Wynn, 1980) illustrating an apparent contradiction with the purely utilitarian views. The many social uses uncovered in this research support the argument advanced by Marvin (1983) that people will find a way to. subvert rigid productivity-oriented frameworks imposed on telecommunications systems, and build into them a "play" or "pleasure" component. Are these ostensibly non-productive uses functional in some way? Several persuasive arguments have already been advanced in support of this position. Marvin argues for the incorporation of play in telecommunication systems in order to provide an escape from strict work related communication. Play can be relaxing, and can be a good mechanism for reducing stress. Furthermore, it serves as an outlet for creative expression not typically found in task communication situations. There are fewer rules or restrictions in communication in play situations, allowing freer use of the imagination. An anecdote provided by one of the interviewees during the qualitative research illustrates how this creative outlet can be beneficial in work. A group of programmers were devising a game to 133 play using the organizational electronic mail system. In order to create the game, a difficult logic problem had to be solved. They were able to develop the code, and later realized that it contained the solution to a problem they had been having with the product-oriented software the group had been developing. Wynn (1980) points out another benefit of social use, particularly with regard to the formation of distribution lists based on common interest. Participation in these electronic communities performs a socialization function, helping to integrate individuals into the organization. Such personal and community use also helps to create communication networks, which, once in place can be used to support task-related information flow. In effect, social uses may be a valuable means of establishing the lateral relations deemed so necessary for processing work-related uncertainty by Galbraith (1973, 1979) and Tushman and Nadler (1978). As March and Simon (1958) proposed, the probability of communication over some link increases each time it is used. Several additional benefits may result from this socialization process. Cohesiveness among employees may be increased. Fulfilling this maintenance function of communication may lead to improved morale, with other side effects like decreased absenteeism and turnover following. Finally, it must be remembered that social use, like task use still requires knowledge of the system, and may be a useful vehicle for teaching newcomers various system functions and commands. Just as game software has been touted for its potential to involve more people 134 with computers, so may social and entertainment uses ease the acclimation to electronic mail. jQuestion Two: Factors That Lead to Different Patterns of Use Discussion of Multiple Regressions A substantial amount of variation in the task use factor is accounted for by the proposed independent variables. Scxne very clear patterns emerge, paralleling much recent work assessing influences on communication mode choice. In general, hypotheses relating system access, perceived attributes, and cross boundary communication needs to task use were supported. Significantly, a different set of predictors emerge for the task use factor than for the social use factor, further underscoring the distinction between these types of uses. Hypotheses relating social use to perceived attributes and to certain demographic characteristics were supported. Clearly, researchers need to take into account the nature of communication activities when attempting to identify influences on the use of media. The patterns or relationships uncovered in the regression analyses are quite informative, and point to many interesting issues. The pattern of findings for each analysis are discussed below, with speculations about the reasons for lack of influence of certain variables provided. Predicting Task Use jsystem Access. Variables reflecting access to both people and terminals proved to be highly predictive of task use of electronic 135 mail~crearlyr“easy~access~to-term inaIs-affected~t±ie"amount-of" task---- use, supporting earlier findings by Johansen et al (1979) and Edwards (1977). Sharing a terminal increases the likelihood of it being either busy or located in another office, and a user may simply find other means of accomplishing task communication in these situations. Interestingly, the proportion of work contacts outside the work group who also use electronic mail was a far more important variable than the number of work group members using electronic mail. Two conclusions are implied by this finding. First, enough people must be on the system before it becomes useful for task purposes. Second, it appears that the system becomes even more useful when it connects people from diverse organizational and geographic locations. The proximity of work group members increases the availability of face-to- face interaction for users, a luxury perhaps not as readily available when communicating outside the work group. This supports the notion of electronic mail as an effective medium for establishing and maintaining lateral relations. Task/Environment. Oily one contextual variable accounted for a significant amount of variance in task use; the measure of cross boundary/location communication need. In effect, however, this variable reflects the influence of proxemics more than the work environment. In one sense, it bears conceptual similarity to the access to people variable discussed above. It reflects the existence of relevant others either outside the work group or in another location, while the control variable measures whether these people are 136 also on the systerru Ifiaf~~it”is a signTficant predictor of “task use jparallels the importance of proxemics on communication mode choice noted in other research (Conrath, 1973; Irving, 1981; Ruchinskas, 1982; Steinfield, 1981; Steinfield, 1983). It supports the proposition that the value of electronic mail for task communication is enhanced when others are either geographically or organizationally dispersed (Hiltz and Kerr, 1981; Johansen and DeGrasse, 1979). The lack of a relationship of task characteristics, environmental uncertainty, or interdependence to task-related electronic mail use casts some doubt on their importance in understanding use of electronic mail. It may be that people with jobs that are characterized by low task uncertainty still find frequent task related uses of electronic mail. Examples are clerical people routing memos, scheduling appointments for others, or other tasks where the mail system eases the burden of routine paper flow. However, before discounting the influence of task uncertainty on communication patterns in general or electronic mail use in particular, several rival explanations must be examined. One possibility results from the Tushman and Nadler (1978) notion that cross boundary communication does not necessarily increase with higher task-related uncertainty. Rather, they propose a congruency approach, tying these variables together in affecting performance. That is, when there is a fit between cross boundary communication and task uncertainty (i.e. both are low or both are high), performance is high. When there is no fit, performance will suffer. Galbraith (1973, 1977) also notes that creating slack, which may involve decreased performance is one avenue for dealing with task 137 uncertainty. However, these authors are positing a group level effect on performance. Since it is unknown how the notion of congruency applies to individual performance, and individual performance measures were not available, this research only examined the relationship of task uncertainty to communication. Several other reasons are equally worthy of exploration. It may jbe the case that other media have picked up the slack in communication. Task use of electronic mail may not be positively related to these measures of task uncertainty, but task use of other media may indeed bear a strong association. Another possibility is that different media may have different value for processing work related uncertainty. One electronic mail message may be worth more in terms of uncertainty reduction than a telephone call. One obvious example of this is when using the broadcasting capability; one message does indeed equal many calls or face-to-face episodes. Thus the amount of communication may not be the critical issue, but rather some measure of the quality of communication. Finally, the correlation between these variables may have been constrained by a lack of variance among work-related uncertainty in this sample. Because the sample was predominantly professionals and managers in the same company, only a truncated distribution of task uncertainty may have been included. Perceived Channel Characteristics. Subjective perceptions of the attributes of electronic mail appear to play an important role in affecting task use. Logically, perceived utility had the strongest relationship with task use, given the utilitarian nature of these 138 purposes. Other channel characteristics were not significant, but the magnitude of zero-order correlations with task use as well as Lulticollinearity among themselves suggests that perceived utility suppressed any effect of these attributes. An explanation consistent with this data is that ease of use and perceived social presence influence how useful electronic mail is perceived to be, which in turn influences task use. Furthermore, subjective perceptions and use are most likely related through an interdependent process. The more useful electronic mail is perceived to be, the more it is used for task purposes, and if successful, perceived utility is further heightened. The conclusion remains, however, that perceived utility is positively associated with task use. A pattern seems to be emerging in the study of perceived channel characteristics and the use of interactive communication media. Ruchinskas (1982) found little relationship when assessing the use of traditional media, while for Svenning (1982), perceived attributes were highly significant predictors of intentions to use video conferencing. This lead Ruchinskas to conclude that such perceptions were probably more important for newer, less familiar and less routinized media, especially when use is somewhat discretionary. On the other hand, the use of older media is more constrained by organizational norms and policies, habits, and availability. Individual/Demographic Variables. Demographic differences among users did not influence the amount of task use. It did not covary with either age or education, contrary to expectations derived from Hiltz 139 and Kerr,s~(r98I) summary. TheToglc~]5eHif3“€he importance of education - that better educated people understand the system and are less threatened by its complexity - does not seem appropriate given the friendly and easy to use interface in this electronic mail system. Given the goal of software designers to enable novices to use computers, it seems likely that educational level will not be a factor in the future. A similar argument can be advanced for the unimportance of age. Older people need not be threatened by system complexity. Two additional demographic variables, tenure and level, also did not influence task use. Newer employees used the system as much for task related purposes as those who had been with the company longer. The limited variance in the level variable prohibits conclusions to be drawn about its effect on task use. Two-thirds of the sample were professionals, and virtually no higher management were surveyed. The amount of prior experience with electronic mail did not account for a significant amount of variance in task use once the system access variables had entered the equation. The fairly substantial zero-order correlation (r = .35) supports the proposition that task use increases with more experience. However, it is also likely that more experienced users would also have their own terminal, and would have more contacts on the system. Once these are controlled for, other facets of experience such as better knowledge of system features, add little to variance in task use. Predicting Social Use As with task uses, a substantial amount of variance in social use is accounted for by the proposed independent variables. Unlike this 140 former usage factor, however, system access variables did not influence the social use of electronic mail. An altogether different set of predictors emerged, providing a pattern of findings that highlight exciting new directions for research in this relatively unexplored area. Discussion of the influence of the four independent variable clusters is provided below. System Access. The unimportance of access to people - both in and beyond the work group - can be explained most readily by the operationalization of these variables. Both were defined in some way to be work-related contacts. Prior organizational research has demonstrated that task and socio-emotional networks do not necessarily overlap (Farace et al, 1977; Mayo, 1930). Thus it is not unreasonable to find that task and social electronic mail networks do not overlap. In addition, much social use has the flavor of magazine readership; 'people browse on different socially-oriented distribution lists. Therefore who sends a particular message is not important; how entertaining that message is assumes greater importance. Sharing a terminal also did not appear to inhibit social use. Given the likelihood that social use occurs in breaktime or after normal work hours, busy terminals are not an issue. Furthermore, the group nature of some social uses such as game playing may actually be facilitated by having several people around a terminal. Task/Environment. Characteristics of the task and work environment were of little consequence to social electronic mail use. Like the unimportance of 141 access to work related contacts, the cross boundary/location communication need variable has little influence. Again, the non- overlap of task and social networks is probably a factor here. While people in routine, standardized, and/or low pressure jobs might have Lore time for social interaction, or value social interaction more since tasks are less intrinsically motivating, those in the opposite task environment might need the escape provided by social interaction. These competing trends might cancel each other out, resulting in no relationship between the task/environment variables and task use. The present research cannot sort out these interpretations. Perceived Channel Characteristics. Some support for the importance of perceived social presence in predicting social use is provided. Social presence can be thought of as varying among individuals. For some users electronic mail is very personal and rich, while others perceive it to be cold and impersonal. These perceptions in turn influence willingness to apply the system for social purposes. Perceived utility was also a significant predictor of social use. This is consistent with the concept of social use as utilitarian in some way. Ease of use was insignificant, but had a moderately positive zero-order relationship with social use. Its influence might be one step removed from social use. That is, if using electronic mail is thought to be simple and easy, then attitudes about utility and perhaps social presence might be affected. Then, these in turn influence social use. The interdependence notion should be mentioned with regard to ease of use as well. The more one uses electronic mail for either task or social purposes, the easier it will become to use. __________________________________________________________ 142 Perceived privacy was not a good predictor of social use. The public nature of so many social distribution lists could affect this jrelationship. This value of this variable might show up for purposes jwhich are demonstrably sensitive. llndividual/Demographics. Two of these variables were among the most influential in predicting social use, highlighting some important possibilities for new conceptions of electronic mail. Both tenure and age significantly predict social use, with newer and younger employees aeing more likely to use the system for social purposes. Yet, these demographic attributes do not relate to task use as well. This leads to a conceptualization of electronic mail as a potentially valuable means of integrating into the social networks of the organization, much as Wynn (1980) predicted. Those with more tenure have less need to find social contacts, and most likely have developed patterns of social communication via other media prior to the introduction of electronic nail. Experience, level, and education did not appear to influence social use. Given the entertainment and recreational nature of much social use, it is conceivable that they attract inexperienced as well as less educated users. Limited variance on level constrains its explanatory value. Surtmary - Predicting Purposes of Use Regression analysis was successful in identifying important contributors to task and social use. Over 50% of the variance in task use was accounted for, while more than 30% of social use variance was __________________________________________________________________143 ‘ explained. Task use appears to depend on variables that involve the communication infrastructure, including access to relevant others via electronic mail, and access to terminals. It further depends on a positive orientation to electronic mail, as measured by the perceived channel attributes. Finally, increased task use results from a specific need that is met by electronic mail, that of communicating with geographically and organizationally dispersed coworkers. Social use depends on a very different set of variables. The importance of tenure in predicting social use can be interpreted by viewing low tenure as reflecting a need for social integration. Thus social use in part depends on the need for social integration that newer employees exhibit. It further depends on a positive orientation, particularly with reference to how personal the system is perceived to I be. Discriminating Among Types of Users By dichotomizing the amount of task or social use, a two by two categorization of user types was developed. Light users, who exhibit low task and social use, also report sending and receiving the fewest number of messages. Those who primarily use the system for social purposes, and are below average for task use form the second group. Another group of users report more task use, but less social use. Finally, the largest number of respondents make use of electronic mail for both task and social purposes. Clearly, then, these two kinds of uses are not mutually exclusive. The discriminant analysis primarily served to supplement earlier 144 regression findings. The same variables that predict a significant amount of variance in task and social use also help differentiate among the four user groups. Classification is only moderately successful, primarily due to similarities between the two groups with higher task use, and the two groups with two groups with lower task use. Some interesting implications are drawn from this analysis. First, the system access variables, characterized as task communication infrastructure above, separate heavy task users from light task users. However, having a positive orientation not only separates heavy task users, but heavy social users from the light user group. This highlights the importance of a positive orientation to any kind of application of electronic mail. A third discriminant function was not significant, but had an interpretable pattern of loadings. This function separated high social users from low social users, reemphasizing the potential importance of social integration needs of new employees. Thus a similar picture emerges from this analysis as that uncovered in the regressions. Furthermore, a potentially useful typology of user types, along with characteristics that differentiate among them are provided. Theoretical Implications These findings have numerous implications for existing theories of organizational communication and for computer-mediated communication. At the broadest level, this research has highlighted the need to incorporate organizational communication theory into our research on 145 emerging communication technologies. A number of hypotheses about electronic mail use were derived from a range of theoretical orientations in the organizational communication area. In particular, the organizational design and organizational technology and structure proved valuable in developing insights into likely task related uses. The lack of significance of task and environment variables in predicting task use should not be taken as evidence of their unimportance in theoretical development, however. More refined measurement strategies might be required. For example, the notion of task interdependency might be thought of and measured more specifically as information interdependency, as some communication researchers have suggested (Goldhaber et al., 1978). The relative importance of the cross boundary communication need variable seems to indicate the existence of information interdependencies. Electronic mail seems to be a valuable means for spanning these boundaries. In addition, the significance of cross boundary communication need, coupled with access to people outside the work group, supports the small but growing literature on proxemics and communication mode choice. These studies have shown physical distance to be an important predictor of the use of electronic media. Not only does research on emerging communication technologies stand to benefit from organizational communication theory, but the reverse is also true. Many concepts in organizational communication theory can be informed in some way by this study's focus on electronic mail uses. The use of the system for project coordination and monitoring might enable leaders, for example, to handle large groups 146 more efficiently. Thus, notions of span of control might be different depending on whether people have access to this information and communication technology. Returning to organizational structure, the freedom from time and space constraints afforded by electronic mail seems relevant to the question of decentralized vs. centralized operations in information related work. Another area of organizational communication informed by this research involves the study of communication networks. Many have touted the advantages of computer-mediated communication systems for automatically collecting communication data (Rice, 1980, 1982). However, our conceptions of what constitutes a linkage are strained by the vast, indirect information linkages created by large distribution lists. Similarly, when someone receives hundreds of messages per week, jguite often many are discarded without being read (Steinfield, 1983). An automated data collection strategy would count these messages as potential links, even though no information is transmitted. Further implications for sociometric approaches to organizational communication are derived from the social uses of electronic mail. Network theorists often stress the need to look at communication ^structures based on varying content (Goldhaber et al., 1978). The Leeming lack of overlap of task and social networks here supports this notion. With the expanding use of distribution lists, people may indirectly participate in large numbers of networks, complicating pictures of corrmunication structure. Finally, the importance of socio-emotional communication, particularly for new employees, has relevance to research on organizational communication climate and communication satisfaction. Those interested in the quality of communication relationships might question whether this socio-emotional interaction really can be an effective substitute for more face-to-face communication. Of course, the teleconferencing literature as well as more specific research on computer-mediated communication is informed by the broader view of communication purposes taken here. The distinction between typologies that focus on business meeting situations, and those that wish to categorize everyday use of electronic has important implications for researchers in this area. In addition, the focus on perceived channel attributes adds to the literature on the use of different communication media. One important finding is the value of viewing social presence as a perceived attribute that can vary among people for the same medium, not as an inherent characteristic of media. Furthermore, it appears fruitful to consider different attributes depending on the nature of the application. Task use is best predicted by perceived utility here, while social use appears more influenced by perceived social presence. Practical Implications This research advances our understanding of the nature of electronic mail use in organizations. As such, findings have practical utility for those interested in the implementation of new office information technology. Organizations can derive expectations of the types of uses which can arise in a relatively unrestricted setting from this research. In addition, the variety of innovative applications 148 identified here can be instructive to newcomers to this technology, providing a valuable lesson in how they can take advantage of the unique attributes of electronic mail. The pattern of relationships predicting task and use can also be useful in the implementation of electronic mail. Strategies to encourage use would 1) concentrate on creating an appropriate infrastructure. This might involve identifying communication relationships, particularly those that span locations, and ensuring system access for this people. In addition, it appears that having people share terminals, while saving some equipment expense, is an inappropriate strategy for system implementation. Task related usage is significantly reduced when people do not have their own terminals. 2) Attitudes toward the technology are important, and should be identified. A positive orientation to electronic mail could be fostered through an internal information campaign. The heavy use of electronic mail for socio-emotional purposes by new employees also has implications for policy. Since task use did not seem negatively affected, and in fact showed a moderate positive correlation with social use, these kinds of applications might be functional. This suggests that restrictive usage policies might deny these individuals an important tool for integrating into organizational social networks. Limitations inferences from this study are constrained by several important limitations. Foremost among these is the fact that the research was 149 carried out in only one organization. Strictly speaking, the random sample of respondents only allows generalization to this particular organization. Based on the size of the sample, and the several different kinds of jobs represented, a case can be made for the potential applicability of these findings to other organizations employing large numbers of white collar professionals. Ultimately, support for the external validity of this research must come from additional studies completed in a number of other organizations. Generalizations to other organizations must also be tempered by comparisons to their particular communication environments. Not all organizations have norms fostering open information sharing or social interaction. To a large extent, communication norms in this organization might be attributed to its composition of white collar professionals. Manufacturing organizations or others containing sizable blue collar style' employee populations may operate under an entirely different set of carnmunication norms. A second limitation is more methodological. That is, only cross sectional quantitative data was collected. Hence, the analyses reported here only present a snap shot of the organization. The extended duration of the qualitative analysis provides some confidence in the relative stability of purposes of use, however. Further the use of correlational analyses, combined with this cross-sectional approach cannot determine causal patterns. Therefore the direction of relationships, while hypothesized to flow from independent to dependent variables, is untested. The necessary concomitant variation has been established for a number of variables, however, highlighting areas 150 where methods more suited to causal analysis can be directed most fruitfully. Another methodological problem stems from the self-report nature of the data. Inaccuracy in self reports of communication behavior are recognized. However, the same can be said of all perceptual indicators. Furthermore, the internal consistency found between user group membership and other self reports of electronic mail use frequency are promising. Ideally in future research, the capabilities of the computer in automatically collecting usage data would be utilized. Finally, this investigation and the conclusions reached here essentially belong to a functional perspective where the reference point implicitly is what is good for the organization. Although it is difficult for one study to operate from more than one perspective, alternative approaches should be noted. Other theoretical orientations might take quite different approaches to the assessment of the use of computer-mediated communication systems. Some may, for example, wish to examine the use of electronic mail as a tool of power and influence. These researchers might ask who controls the system, and whether such systems are imposed on a workforce, hypothesizing that it is used to maintain the existing distribution of power. On the other hand, some might question whether unrestricted use can be a mechanization for potentially increasing employee participation in decision making. In this fashion, the electronic mail system would be viewed as a democratizing influence on organizational functioning. Still others, operating from a class or 151 status orientation, would focus upon how this technology closes or widens status differentiation between knowledge workers and other types of organizational participants. This is only a cursory sampling of alternative approaches, but it illustrates the potential importance of information technology across a wide variety of theoretical orientations. Future Research Several directions for future research are suggested by this study. First, an analysis of electronic mail use among other populations is one next step. These populations include other organizations of varying sizes and in varying industries. In the not too distant future, consumer use of electronic mail might flourish, making home users another potential population of interest. A more refined assessment of the dimensionality of uses is another possible research direction. Based on the kinds of uses identified here, a conceptualization of dimensions call be developed and tested with an instrument that is sensitive to expected dimensions. For example, surveillance uses did not form their own dimension in this research, primarily because there were not enough items to enable a factor to emerge. In addition, other methods might be employed in the measurement of purposes of use. Content analysis of electronic mail messages is one appropriate method. Future research should also attempt to isolate the influence of the specific organization policies governing use. For example, how does cost relate to task and social use, or management surveillance of message traffic? 152 The relationship of electronic mail use to other media provides still another important direction for future research. Only by assessing the frequency with which various communication behaviors occur, and then examining how they distribute among media, can we address questions such as the applicability of different media for interpersonally involving tasks. Finally, this brief look at possible research directions concludes with the need to relate the uses of electronic mail to impacts on individuals, groups, and organizations. In order to examine impacts on individual productivity, for example, researchers might first ask whether the system is used for task purposes, and if so, for which kinds. In this fashion, a contingency model of individual impacts could be developed. Similarly, impacts at the group and organizational level would be a function of the type of media available, and the nature of the applications for these media. The outcome of this future research would be a model depicting the influences of attitudinal, organizational, and individual on styles of use, which in turn lead to a differential pattern of impacts. 153 REFERENCES Albertson, L. Trying to eat an elephant. Communication Research, 1980, 7, 387-400. Aldrich, H. Organizations and Environments. Englewood Cliffs, N.J.: Prentice-Hall, 1979. Allen, T. Managing the flow of scientific and technological information. Unpublished PhD. dissertation, Massachusetts Institute of Technology, 1966. Atkin, C. Instrumental utilities and information seeking. In P. Clarke (Ed.), New Models for Communications Research. Beverly Hills: Sage, 1973. Bair, J. H. A communication perspective for identifying office automation payoffs. Invited paper. Symposium on Automated Office Systems, New York University, 1979. Bales, R. F. Interaction Process Analysis: A Method for the Study of Small Groups. Cambridge, MA: Addison-Wesley, 1950 Berio, D. The Process of Communication. New York: Holt Reinhart and Winston, 1960. Blumler, J. The role of theory in uses and gratifications studies. Communicatin Research, 1979, 6, 9-36. Brief, A., Delbecq, A., Filley, A., and Huber, G. Elite structure and attitudes: an empirircal analysis of adoption behavior. Administration and Society, 1976, 8, 227-248. Burns, T. and Stalker, G.M. The Management of Innovation. London: Tavistock, 1961. Carroll, J. M. The adventure of getting to know a computer. Computer, November, 1982, 49-58. Champness, B. Feelings toward media in group situations (Report No. E/7216000/CH). London: University College, Communications Study Group, 1972. Champness, B. Attitudes toward person-person communications media. Human Factors, 1973, 15, 437-447. Christie, B. Appendix M. in P. Goldmark (Ed.), The 1972-1973 New Rural Society Project. Fairfield, Conn: Fairfield University, 1973. 154 Christie, B. Semantic differential judgements of communications media and other concepts: differences between the media (Report No. E/74120/CH). London: University College, Communications Study Group, 1974. Cohen, J. and Cohen, P. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences, New York: Wiley, 1975. Connell, S. The 1973 office communications survey. Communications Studies Group paper no. p/74067/CN, (1974). Conrath, D. Communication patterns, organizational structure, and man: some relations. Human Factors. 1973, 15, 459-470. Craig, L. C. Office automation at Texas Instruments, Inc. In M. Moss (Ed.) Telecommunications and Productivity. Reading, Massachusetts: Addison-Wesley, 1981, 202-214. Danziger, J., Dutton, W. H., Kling, R., and Kraemer, K. Computers and Politics. New York: Columbia University Press, 1982. Dearborn, D. and Simon, H. Selective perception: a note on the departmental identifications of executives. Sociometry, 21, 1958, 140-144 Dordick, H., Bradley, H., and Nanus, B. The Emerging Network Marketplace. Norwood N.J.: Ablex, 1981. Downey, H.K., Hellriegel, D., and Slocum, J. W. Environmental uncertainty: the construct and it's application. Adminstrative Science Quarterly, 20, 1975, 613-629. Downey H.K. and Slocum, J. Uncertainty: measures, research and sources of variation. Academy of Management Journal, 1975, 18, 562-578. Downs, G.W. and Mohr, L. B. Toward a theory of innovation. Administrative and Society, 1979, 10, 379-408. Duncan, R.B. Characteristics of organizational environments and perceived environmental uncertainty. Administrative Science Quarterly, 1972, 17, 313-327. Dutton, W. H. Decision making in the information age: computer models and public policy. In Voigt and Dervin (Eds.), Progress in Communication Sciences, forthcoming. Dutton, W. H., Fulk, J., and Steinfield, C. W. Utilization of video conferencing. Telecommunications Pol icy, 1982,6^, 3. Dutton, W. H., Kovaric P., and Steinfield, C. W. The social dynamics of personal computing, unpublished Manuscript, Annenberg School of Communications, University of Southern California, 1983. 155 Dutton, W. H. and Kraemer, K. Automating bias. Society, 1980, 17. Edelstein, A. An alternate approach to the study of source effects in mass communications. Studies of Broadcasting, 1973, 9, 5-31. Edwards, G. An analysis of usage and related perceptions of NLS - a computer based text processing and communication system. Bell Canada Headquarters Business Development, Montreal, Canada, 1977. Far ace, R.V., Mange, P.R., and Russell, H.M. Communicating and Organizing, Reading, MA: Addison-Wesley, 1977. Fett, J. Situational factors and peasants' search for market information. Journalism Quarterly, 1975, 53, 429-435. Freeman S., and Freeman, L. The networker's network: a study of the impact of a new communication medium on sociometric structure. Paper presented to the Seminar on Communication Network Analysis, East-West Center Communication Institute, Honolulu, Hawaii, January,1979 Galbraith, J. Designing Complex Organizations. Reading, MA: Addison- Wesley, 1973. Galbraith, J. Organizational Design. Reading MA: Addison-Wesley, 1977. Gerstberger, P., and Allen, T. Criteria used by research and development engineers in the selection of an information source. Journal of Applied Psychology, 1968, 52, 272-277. Goddard, J. Office linkages and location. In D. Diamond and J. McLoughlin (Eds.) Progress in Planning, Oxford: Pergamon Press, 1973, 111-231. Goldhaber G., Yates, M., Porter, D., and Lesniak, R. Organizational communication: 1978. Human Communication Research, 1978, 5, 76- 96. Hage, J. and Aiken, M. Routine technology, social stucture and organizational goals. Administrative Science Quarterly, 1968, _4, 366-376. Hannagan et al The Survey of Electronic Mail Systems, Edition II. Chicago, IL, 1981. Heimstra, G. Teleconferencing, concern for face, and organizational culture. Communication Yearbook, 1982, 6. Hickson, D.J., Hinings, C.R., Lee, C.A., Schneck, R.E., and Pennings, J.M. A strategic contingencies theory of intraorganizational power. Administrative Science Quarterly, 1971, 16, 216-229. 156 Hiltz, S.R. Controlled experiments with computerized conferencing: results of a pilot study. Bulletin of the American Society for Information Science, 1978, 4 (5), 11-12. Hiltz, S. R., and Kerr, E. Studies of computer mediated communication systems: a synthesis of the findings. Final report submitted to the National Science Foundation, Division of Information Science and Technology, 1980. Hiltz, S.R. & Turoff, M. The Network Nation: Human Communication via Computer. Reading, MA: Addison-Wesley, 1978. Hiltz, S. R., and Turoff, M. The evolution of user behavior in a computerized conferencing system. Communications of the ACM, 1981, 24, 11. Hrebiniak, L. G. Job technology, supervision, and work group structure. Administrative Science Quarterly, 1974, 19, 395-410. Hunt, R. Technology and organization. Academy of Management Journal, 1970, 235-252. Irving, R. Mode choice in dyadic communications. Unpublished PhD dissertation, University of Waterloo, 1981. Johansen, R., Vallee, J., and Spangler, K. Electronic meetings: technical alternatives and social choices. Menlo Park: Addison- Wesley, 1979. Katz, D. and Kahn, R.L. The Social Psychology of Organizations (second ed.) . New York: John Wiley, 1978. Kerlinger, F. and Pedhazur, E. Multiple Regression in Behavioral Research. New York: Holt, Rinehart and Winston, 1973. Kiesler, S. Computer mediated communication. Proposal to the National Science Foundation, Program on Information Sciences and Technology, 1982. Kling, R. Social analyses of computing: theoretical perspectives in recent empirical research. Computing Surveys, 1980, 12, 1. Korzenny, F. A theory of electronic propinquity: mediated communication in organizations. Communication Research, 1978, 5, 3-23. Korzenny, F., and Bauer, C. A preliminary test of the theory of electronic propinquity: organizational teleconferencing. Paper presented to the international Communication Association, Philadelphia, PA., 1979. 157 Kraemer K., and Dutton, W. H. The interests served by technological reform: the case of computing. Administration and Society, 1979, 2, 80-106. Lasswell, H. D. The structure and function of communication in a society. In Bryson (Ed.), The Communication of Ideas, New York: Institute for Religious and Social Studies, 1948. Lawler, E., Porter, L., and Tennenbaum, A. Managers' attitudes toward interaction episodes. Journal of Applied Psychology, 1968, 52, 432-439. Lawrence, P.R. and Lorsch, J.W. Organization and Environment, Boston: Harvard Business School, 1967. Lee, R. The flow of information to disadvantaged farmers. Unpublished PhD Dissertation, University of Iowa, 1966. iLippitt, M. E. Patterns of use and correlates of adoption of an electronic mail system. Prepared for the Proceedings of the American Institute of Decision Sciences, Las Vegas, Nevada, 1981. McEwen, W. Bridging the information gap. Journal of Consumer Research, 1978, 4, 247-251. McEwen, W., and Hempel, D. How information needs and effort affect channel choice. Journalism Quarterly. 1977, 54, 149-153. McLaughlin, M. L. Discriminant analysis in communication research. Multivariate Techniques in Human Communication Research, New York: Academic Press, 1980. Mahoney, T. A., and Frost, P. J. The role of technology in organizational effectiveness. Administrative Science Quarterly, 1974, 19, 5, 338-353. Malone, T. W. Organizing information processing systems: parallels between human organizations and computer systems. Working paper, Palo Alto Research Center, Xerox, 1981a. Malone, T. W. Toward a theory of intrinsically motivating instruction. Cognitive Science, 1981b, 4, 333-369. March J., and Simon, H. Organizations. New York: Wiley, 1958. Marvin, C. Telecommunication policy and the pleasure principle. Telecommunications Policy, March, 1983. Mayo, E. The Social Problems of an Industrial Civilization, Massachusetts: Harvard University Press, 1945. 158 yiintzberg, H. The Nature of Managerial Work. New York: Harper and Row, 1973. Ylonge, P., Edwards, J., and Kriste, K. The determinants of communication structure in large organizations: a review of research. Paper presented to International Communication Association, Chicago, 1978. Noll, A. M. Teleconferencing communications activities. ikee Transactions on Communications, 1977, 8-14. Dlson, M. H. and Lucas, Jr., H. C. The impact of office automation on the organization: some implications for research and practice. Communications of the ACM, November,1982, 838-847. O'Reilly C., and Pondy, L. Organizational Communication. In Kerr (Ed.), Organizational Behavior, Columbus, Ohio: Grid Publications, 1979, 119-150. Paisley, W. & Chen, M. Children and electronic text: challenges and opportunities of the "new literacy": an exploratory study. Stanford, CA: Institute for Communicatons Research, 1982. Perrow, C. A framework for the comparative analysis of organizations. American Sociological Review, 1967, 32, 194-208. Pfeffer J. and Salancik, G. The External Control of Organizations. New York: Harper and Row, 1978. Phillips, A. L. Computer conferencing: success or failure? Systems, Objective, Solutions, 1982, 2, 203-218. Porter, L. W., and Roberts, K. Organizational communication. In Dunnette (Ed.), Handbook of Industrial and Organization Psychology, Chicago: Rand McNally, 1976, 1553-1589. Pye, R., Champness, B., Collins, H., and Connell, S. Description and classification of meetings. Unpublished Working Paper No. P/73160/PY, Communication Studies Group, London, England. Rice, R. The impacts of computer-mediated human communication. Annual Review of Information Science and Technology, 1980. Rice, R. Longitudinal role development and system structure of researchers using a computer conferencing system. In Burgoon (Ed.), Communication Yearbook 6, California: Sage, 1982, 925-944. Rice R., and Case, D. Electronic messaging in the university organization. Presented to the Speech Communication Association, Anaheim, California, 1981. 159 Roethlisberger, F. J., and Dickson, W. J. Management and the Worker. Cambridge, MA: Harvard University Press, 1939. Rogers, R, and Agarwala-Rogers, R. Communication in Organizations. New York, NY: Free Press, 1976. Rogers, E. and Shoemaker, F. Communication of Innovations. New York: Free Press, 1971. Rosenberg, V. The application of psychometric techniques to determine the attitudes of individuals toward information seeking. Unpublished manuscript, Lehigh university, 1966. Rosenbloom, R., and Wolek, F. Technology and Information Transfer: A Survey of Practice in Industrial Organizations. Boston: Harvard University, 1970. Ruchinskas, J. R. Communicating in Organizations: The Influence of Context, Job, Task, and Channel. Unpublished Dissertation, University of Southern California, 1982. Rummel, R. J. Applied Factor Analysis. Evanston: Northwestern University Press, 1970. Schramm, W. The nature of communication between humans. In Schramm and Roberts (Eds.), The Process and Effects of Mass Communications, Revised Edition, Urbana, Illinois: University of Illinois Press, 1971. Schramm, W. Men, Messages and Media: A Look at Human Communication. New York: Harper and Row, 1973. Short, J., Williams, E., and Christie, B. The Social Psychology of Telecommunications. London: J. Wiley & Sons, 1976. Siegel, J., Kiesler, S., and McGuire, T. Participation affect, and decision in computer-mediated groups. Unpublished manuscript, Carnegie Mellon University, 1981. Simon, H. On the concept of organizational goal. Administrative Science Quarterly, 1964, jj, 1-22. Steinfield, C. The uses and impacts of electronic mail. Working paper, Xerox, 1983. Steinfield, C. Explaining managers' use of communication channels for performance monitoring: implications for new office services. Paper presented to International Communication Association, Minneapolis, Minn., 1981. 160 Svenning, L. Explaining predispositions toward telecommunication innovations: The case of video conferencing. Unpublished PhD dissertation. University of Southern California, 1982. Tapscott, D. Investigating the office of the future. Telesis, 1981, 1, 2-6. Thompson, J. D. Organizations in Action. New York: McGraw-Hill, 1967. Today's Office Electronic mail: a new medium for the message. August, 1982, 3/7, 3. Tosi, H., Aldaq, R., and Storey, R. On the measurement of the environment: an assessment of the Lawrence and Lorsch environmental uncertainty subscale. Administrative Science Quarterly, 1973, 19, 183-197. Tushman, M. Special boundary roles in the innovation process. Administrative Science Quarterly, 1977, 22, 587-605. Tushman, M. and Nadler, D. Information processing as an integrating concept in organizational design. Academy of Management Review, July, 1978, 613-624. Uhlig, R., Farber, D., Bair, J. The Office of the Future: Communications and Computers. New York, NY: North-Holi and, 1979. Vallee, J. The Network Revolution. Berkeley, California: And/Or Press, 1982 Van de Ven, A., Delbecq, A., and Koenig, R. Determinants of coordination modes within organizations. American Sociological Review, 1976, 41, 322-338. Werner, D. A study of the information seeking behavior of research scientists. Unpublished MA Thesis, Northwestern University, 1965. Wilensky, H.L. Organizational intelligence. Nfew York: Basic Books, 1967. Woodward, J. (ed.), Industrial Organization: Theory and Practice. London: Oxford University Press, 1965. Wynn, E. Computers message systems as a unique medium of communications. Presented to the ICGC, Atlanta, Georgia, 1980. 161 APPENDICES 162 Appendix A MOTIVATIONAL LETTERS 163 Inter-Office Memorandum To Study participant (Electronic mail user) Date June 15,1982 From Location Subject Your help with our electronic mail study Organization Researchers firom the Annenberg School of Communications at USC are helping us conduct a study electronic mail, as well as examine any adverse side effects. We feel that this information will be valuable in planning future system development and marketing efforts. You have been selected to participate in this study through a random sampling of electronic mail users. Your experiences make you uniquely qualified to help us establish our database on uses and impacts of electronic mail. As a community, we have been using electronic mail for a number of years in a variety of ways not yet experienced by other users of this technology. Whether you have been using electronic mail for some time, or have just begun to use i t , please carefully fill out the enclosed questionnaire. The questionnaire gathers information about the nature of your job, communication practices, uses of and attitudes toward electronic mail, possible new electronic mail features, and your background. It has been designed for easy reading and response; in pretests people averaged between 20 and 30 minutes to f i l l i t o u l All responses are completely confidential; results will be presented in statistical summary only. I f you do not have an electronic mailbox at please contact Chip Steinfield at the address listed below and we will send you an alternate questionnaire form. After completing the questionnaire, return it in the enclosed envelope or send it via company mail to: Chip Steinfield, Mail Stop Please contact Chip Steinfield if you have any questions about the study. We would appreciate receiving vour questionnaire bv July 1. of electronic mail uses here at Our aim is to document the many uses and benefits of Inter-Office Memorandum To Study participant Daie July 7,1982 From Location Subject Your help with our electronic mail study Organization Dear Respondent, Recently you received a questionnaire as pan of our study of the uses and impacts of electronic mail. You were one of a select group of employees chosen on the basis of your work location, your division, and whether you use electronic mail in your work or not Both users and non-users of electronic mail "Were included in the survey in order to see how people accomplish their communication with and without the use of this technology. In order to obtain a representative cross section of opinion and to ensure the integrity of the information collected it is extremely important that everyone responds. If you have already completed and returned the questionnaire, we thank you for your prompt attention. However, if you have not yet had a chance to complete the survey, please do so at your earliest convenience. Your input will help us in planning future system development and marketing efforts based on the best available information. Thank you for your cooperation and let us assure you again that the information you provide will be held in the strictest confidence by the research team. If you have any questions, or need another copy of the questionnaire, please feel free to contact the research director, Chip Steinfield at 165 Appendix B SURVEY QUESTIONNAIRE 166 Electronic Mail Questionnaire 167 IF YOU CURRENTLY DO NOT USE ELECTRONIC MAIL, PLEASE CALL CHIP STEINFIELD AT TO OBTAIN A DIFFERENT VERSION OF THIS QUESTIONNAIRE. 1. J O B C H A R A C T E R I S T I C S ______________________________________________________________________________________ not at very all m uch 1 2 3 4 5 S 7 (circle below) MY JOB INVOLVES: ROUTINE REPETITIVE TASKS 1 2 3 4 5 6 7 TASKS WITH CLEARLY DEFINED OUTCOMES 1 2 3 4 5 6 7 WORKING WITH PEOPLE I’M NOT ACQUAINTED WITH 1 2 3 4 5 6 7 FINDING NOVEL SOLUTIONS TO PROBLEMS 1 2 3 4 5 6 7 MANY DIFFERENT KINDS OF TASKS 1 2 3 4 5 6 7 MY JOB IS GOVERNED BY: CRISES/URGENT MATTERS 1 2 3 4 5 6 7 STANDARD OPERATING PROCEDURES 1 2 3 4 5 6 7 WELL-DEFINED SUBJECT MATTER 1 2 3 4 5 6 7 TIME PRESSURES 1 2 3 4 5 6 7 RULES, POLICIES AND REGULATIONS 1 2 3 4 5 6 7 UNEXPECTED PROBLEMS/SITUATIONS 1 2 3 4 5 6 7 2. W O R K G R O U P C H A R A C T E R I S T I C S THE TERM WORK GROUP AS USED BELOW REFERS TO YOUR IMMEDIATE ORGANIZATIONALLY ASSIGNED UNIT not at all 1__ (circle below) very m u c h 3 Z DECISION M AKING IN MY WORK GROUP: ROUTINE OPERATING DECISIONS ARE MADE MAINLY BY 1 2 3 4 5 6 7 THE HEAD OF THE WORK GROUP STRATEGIC OR POLICY DECISIONS ARE MADE MAINLY 1 2 3 4 5 6 7 BY THE HEAD OF THE WORK GROUP WORK GROUP PRODUCTS: THE TECHNIQUES, SKILLS, OR INFORMATION REQUIRED TO 1 2 3 4 5 6 7 PRODUCE MY GROUP’S PRODUCTS ARE ALWAYS CHANGING THE MARKETPLACE, TECHNOLOGY, AND REGULATIONS FOR 1 2 3 4 5 6 7 MY WORK GROUP’S PRODUCTS ARE GENERALLY STABLE 168 3. W O R K G R O U P M E M B E R S H I P NUMBER OF PEOPLE IN MY WORK G R O U P... IN THIS CITY: (number) IN OTHER CITIES: (number) NUMBER OF PEOPLE USING ELECTRONIC MAIL IN MY WORK GROUP: (number) I PARTICIPATE ON A PROJECT TEAM(S) OUTSIDE MY WORK GROUP. yes no (circle one) IF SO, ARE PEOPLE ON THIS TEAM LOCATED IN ANOTHER CITY? yes no (circle one) 4. W O R K F L O W S not at very all m u c h 1 2 3 4 5 6 7 (circle below) IN MY GROUP THE PRODUCTS OF M Y W ORK : ARE INDEPENDENT OF OTHER’S WORK 1 2 3 4 5 6 7 FEED INTO SOMEONE ELSE’S WORK 1 2 3 4 5 6 7 REQUIRE INPUT FROM SOMEONE ELSE’S WORK 1 2 3 4 5 6 7 ARE COMPLETED WITH OTHERS IN A TEAM APPROACH 1 2 3 4 5 6 7 THE PRODUCTS OF M Y W O R K GROUP: ARE INDEPENDENT OF OTHER GROUPS 1 2 3 4 5 6 7 FEED INTO ANOTHER WORK GROUP 1 2 3 4 5 6 7 REQUIRE INPUT FROM ANOTHER WORK GROUP 1 2 3 4 5 6 7 ARE COMPLETED AS A TEAM WITH ANOTHER GROUP 1 2 3 4 5 6 7 5. D U R I N G A N A V E R A G E W O R K W E E K . I S P E N D . . . percent of mv time (fill in below) ON THE TELEPHONE % IN FACE-TO-FACE CONVERSATIONS AND/OR MEETINGS % READING (or looking at printed materials) % SENDING/RECEIVING/PROCESSING ELECTRONIC MAIL % PRODUCING WRITTEN MATERIALS ON A COMPUTER % PRODUCING WRITTEN MATERIALS BY HAND OR TYPEWRITER % USING OTHER OFFICE EQUIPMENT (calculator, copying machine, etc.) % OTHER (thinking, walking, traveling, etc.) % = 100% 169 8. TYPICALLY. USING ELECTRONIC MAIL IS: USING THE ADJECTIVES BELOW, PLEASE INDICATE HOW YOU FEEL ABOUT USING ELECTRONIC MAIL. FOR EXAMPLE. IF YOU FEEL THAT USING ELECTRONIC MAIL IS VERY DIFFICULT YOU WOULD MARK THE SCALE LIKE THIS: EASY_______ :_______ :_______ :_______ : _______:_______ : X DIFFICULT TYPICALLY, USING ELECTRONIC M AIL IS: INFORMAL. USEFUL. SOCIABLE. UNRELIABLE. ACCESSIBLE. UNCOMFORTABLE. EASY. EFFECTIVE. COLD. CONVENIENT. IMPERSONAL. FLEXIBLE. COMPLEX. PRIVATE. EFFICIENT. SLOW. NOT CONFIDENTIAL . FORMAL . NOT USEFUL . UNSOCIABLE . RELIABLE . UNACCESSIBLE .COMFORTABLE . DIFFICULT . INEFFECTIVE .WARM . INCONVENIENT . PERSONAL . INFLEXIBLE .SIMPLE . PUBLIC . INEFFICIENT .FAST .CONFIDENTIAL 9. M Y E L E C T R O N I C M A I L S Y S T E M E X P E R I E N C E (fill in blanks below) THE ELECTRONIC MAIL SYSTEM 1 USE NOW IS (eg Laurel). . . (name) 1 HAVE USED IT FO R ... (vears) AND (months) 1 SHARE A COMPUTER/TERMINAL WITH HOW MANY PEOPLE? (number) PERCENT OF PEOPLE 1 DEAL WITH OUTSIDE MY WORK GROUP WHO USE ELECTRONIC MAII____ % APPROXIMATE NUMBER OF HOURS PER DAY 1 WORK ON A COMPUTER IS .. . (hours) THE PERCENT OF THIS TIME 1 USE ELECTRONIC MAIL IS . . . % 1 USED ANOTHER SYSTEM PRIOR TO THIS FOR .. . (vears) AND (months) 10. D I S T R I B U T I O N L I S T (DL) M E M B E R S H I P APPROXIMATE NUMBER OF DISTRIBUTION LISTS I BELONG TO THAT ARE: ORGANIZATION DLs (LOCATION OR DEPT) WORK RELATED INTEREST DLs NON-WORK/SOCIAL DLs OTHER (number) (number) (number) (number) (please specify) NUMBER OF WORK RELATED DLs I FOUNDED: (number) NUMBER OF NON-WORK/SOCIAL DLs I FOUNDED: (number) I FIND IT DIFFICULT TO PUT MYSELF ON A DISTRIBUTION LIST: yes sometimes no (circle one) 171 11. E L E C T R O N I C M A I L U S E USING THE PREVIOUS WORK WEEK AS A POINT OF REFERENCE. PLEASE ESTIMATE: THE NUMBER OF TIMES LAST WEEK THAT I . . . SENT A MESSAGE TO: a distribution list(s)? (num ber) individual people? (num ber) RECEIVED A MESSAGE ADDRESSED TO: a distribution list(s)? (number) me personally? (number) ANSWERED SOMEONE’S PREVIOUS MESSAGE: (number) FORWARDED SOMEONE A MESSAGE: (number) COPIED MYSELF ON MESSAGES SENT: (number) DELETED A DISTRIBUTION LIST MESSAGE WITHOUT READING: (num ber) MADE A HARDCOPY OF A MESSAGE I RECEIVED (number) WAS LAST WEEK TYPICAL OF YOUR ELECTRONIC MAIL USE? yes no (circle one) IF NO, HOW WAS IT DIFFERENT? (eg. sent more, received more, etc.) HOW MANY MESSAGES ARE TYPICALLY IN YOUR ACTIVE MAIL FILE? (number) IF YOU HAVE A SECRETARY, DOES THIS PERSON: READ YOUR ELECTRONIC MAIL FOR YOU? never sometimes always (circle one) TYPE YOUR ELECTRONIC MAIL FOR YOU? never sometimes always (circle one) 172 12. MY USE OF ELECTRONIC MAIL I USE ELECTRONIC M AIL TO . . . INFORMATION EXCHANGE SEEK TASK INFORMATION FROM PEOPLE I KNOW BROADCAST REQUESTS FOR INFORMATION DISTRIBUTE/PROVIDE INFORMATION GIVE/RECEIVE FEEDBACK ON REPORTS/IDEAS KEEP TRACK OF COMPANY HAPPENINGS BRAINSTORM/GENERATE IDEAS FORWARD MESSAGES TO SOMEONE WHO IS NOT ON SOME DISTRIBUTION LIST DISCUSS CONFIDENTIAL MATTERS FIND OUT ABOUT JOB/PROMOTION OPENINGS PERFORMANCE MONITORING MONITOR PERFORMANCE OF MY SUBORDINATES MONITOR PROGRESS OF PROJECTS I WORK WITH ENSURE PEOPLE USE DLs FOR APPROPRIATE TOPICS SUPPLEMENTING OTHER MEDIA KEEP A RECORD OF AGREEMENTS/INTERACTIONS SEND A MESSAGE IN PLACE OF A PHONE CALL SEND/RECEIVE POINTERS TO LARGE FILES GET A MESSAGE DIRECTLY TO SOMEONE WHOSE CALLS ARE NORMALLY SCREENED very never often 1 2 3 4 5 (circle below) 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 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 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 5 not applicable not something I do at work 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 173 1 2. MY USE OF ELECTRONIC MAIL (continued)__________________________________ not applicable very not something 1 USE ELECTRONIC M AIL TO . . . never often I do at v SOCIAL EXCHANGE 1 2 3 4 5 (circle below) a KEEP IN TOUCH/MAINTAIN RELATIONSHIPS t 2 3 4 5 8 LEARN ABOUT EVENTS/THINGS I’M INTERESTED IN 1 2 3 4 5 S GET TO KNOW SOMEONE 1 2 3 4 5 8 PARTICIPATE IN ENTERTAINING EVENTS (GAMES) OR CONVERSATIONS 1 2 3 4 5 8 SCHEDULING SCHEDULE MEETINGS/APPOINTMENTS 1 2 3 4 5 8 COORDINATE ACTIVITIES OF PROJECTS I’M ON 1 2 3 4 5 8 SEND A MESSAGE REQUIRING AN IMMEDIATE RESPONSE 1 2 3 4 5 8 SEND A MESSAGE REQUIRING A RESPONSE THE SAME DAY 1 2 3 4 5 8 ORGANIZE/COORDINATE A SOCIAL ACTIVITY 1 2 3 4 5 8 OTHER USES CARRY ON NEGOTIATIONS/ BARGAIN 1 2 3 4 5 8 TAKE A BREAK FROM MY WORK 1 2 3 4 5 8 FILL UP "FREE" TIME 1 2 3 4 5 8 RESOLVE CONFLICTS/DISAGREEMENTS 1 2 3 4 5 8 ADVERTISE/RESPOND TO PRODUCTS FOR SALE 1 2 3 4 5 8 POLL OPINIONS ON A TOPIC 1 2 3 4 5 8 ASK QUESTIONS IN A PUBLIC SETTING (EG SENDING MANY COPIES) TO FORCE A RESPONSE OR ACTION 1 • 2 3 4 5 8 OTHER 1 2 3 4 5 8 (please specify) 174 15. BACKGROUND INFORMATION CATEGORY WHICH BEST DESCRIBES MY JOB? (check one) MARKETING/SALES PLANNING "PROGRAMMING/SOFTWARE TESTING TECHNICAL WRITING RESEARCH CLERICAL ENGINEERING PUBLIC RELATIONS FINANCE SYSTEM OPERATIONS FACILITIES PERSONNEL TRAINING OTHER________________ (please specify) ARE YOU ... (circle one) UPPER MANAGEMENT 1 MIDDLE MANAGEMENT 2 PROFESSIONAL/EXEMPT 3 TECHNICAL/NONEXEMPT 4 CLERICAL 5 OTHER 6 YEARS I HAVE WORKED FOR THIS COMPANY: YEARS IN MY CURRENT POSITION: NUMBER OF REPORTING LEVELS I AM FROM MY DIVISION HEAD NUMBER OF PEOPLE I SUPERVISE . .. IN THIS CITY: IN OTHER CITIES: MY AGE IS ...? MY SEX IS ...?('circle one) HIGHEST LEVEL OF EDUCATION ATTAINED IS ...? (circle one) male female Jyears) Jyears) Jnum ber) Jnum ber) _{number) Jyears) HIGH SCHOOL 1 SOME COLLEGE 2 COLLEGE 3 MASTERS 4 PH.D. 5 THANK YOU VERY M U CH FOR YOUR P A R TIC IPA TIO N ! PLEASE USE SPACE BELOW OR NEXT PAGE FOR ANY COMMENTS ON THIS STUDY. RETURN COMPLETED QUESTIONNAIRES TO: ATTN: CHIP STEINFIELD. MAIL STOP COMMENTS 175

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A system of participation and allocation preferences in 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

Cinematic concepts and techniques in the adaptation and staging of chamber theatre

PDF

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

PDF

Correlates of deceit: A cross-cultural examination

PDF

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

PDF

Communication and the implementation of information systems: The case of word processing in state and local governments

PDF

Causes, consequences, and moderating factors of career plateauing.

PDF

Communicative features of educational computer software

PDF

Children's judgments of the similarity of television series and the similarity of gratifications associated with television series

PDF

Electronic persuasion: The uses of electronic mail for interpersonal influence in organizations

PDF

A dynamic reformulation of perceptions of inequity: Their organizational antecedents and outcomes

PDF

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

PDF

Astrocyte regulation of endothelial haemostasis function via transforming growth factor-beta

PDF

Communications medium and product class: Their effect on negotiation outcome

PDF

An analysis of the effects of controls imposed on the production environment of a dramatic television series, "The Senator"

PDF

Career advancement in managerial hierarchies in the United States firms: A multi-theoretic model and empirical tests of the determinants of managerial promotion

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

Asset Metadata

Creator Steinfield, Charles William (author)

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

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 Multimedia Communications,OAI-PMH Harvest,Speech Communication

Language English

Advisor Dutton, William (committee chair), Fulk, Janet (committee member), Miller, Donald (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-703318

Unique identifier UC11343879

Identifier DP22378.pdf (filename),usctheses-c17-703318 (legacy record id)

Legacy Identifier DP22378.pdf

Dmrecord 703318

Document Type Dissertation

Rights Steinfield, Charles William

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