Web-Based Computer-Mediated Communication - Semantic Scholar

Web-Based Computer-Mediated Communication: An Experimental Investigation Comparing Three Communication Modes for Determining Software Requirements Rosalie Ocker

Jerry Fjermestad

School of Business Administration Pennsylvania State University at Harrisburg 777 W. Harrisburg Pike Middletown, PA 17057-4898 [email protected]

School of Management New Jersey Center for Multimedia Research New Jersey Institute of Technology Newark, New Jersey 07102 [email protected]

Abstract

Within the process of developing software, requirements determination is a most critical activity as work accomplished in this upstream development stage impacts the remainder of the software project [4, 7, 10,36]. Briefly, requirements determination is essentially a collaborative process of communication [21]
involving the interaction of software analysts with peers, superiors, subordinates and users [4]
for the purpose of acquiring, sharing, and integrating knowledge [37]
in order to develop a mutually shared understanding regarding the software to be developed. Although numerous researchers have stressed the importance of effective communication in the determination of requirements [e.g. 2, 3, 6, 21, 22, 25, 37], communication problems remain a persistent problem. Miscommunication and misinformation resulting from poorly understood software requirements are carried forward throughout the project, resulting in costly errors and delays in project completion. Valusek and Fryback [35] classify communication problems associated with requirements determination into three categories: within obstacles, among obstacles, and between obstacles. Within obstacles evolve around the cognitive limitations of humans as information processors and problem solvers, and thus deal with things such as limited memory and recall. Requirements determination techniques that help overcome within obstacles focus on enhancing recall. Between obstacles are concerned with problems that arise between the various groups of workers that must communicate with one another in the process of determining requirements. These obstacles result from a lack of a common language between the diverse groups that must interact in order to formulate software requirements. Finally, among obstacles are the result of the necessity to balance the various and often conflicting needs of multiple users. Among obstacles involve tradeoffs and the weighing of multiple needs after determining individual user requirements. Various researchers [e.g. 5, 8, 27, 29] have suggested that, in order to make substantial progress towards

This experiment is the third in a series of empirical studies designed to explore the effectiveness of various modes of communication for groups working on determining software requirements. This study focuses on the effectiveness of matching communication media to stages of group work. Three modes of communication are compared: (1) face-to-face (2) Web-based asynchronous computer conferencing and (3) combined – a sequence of FtF, Web-based asynchronous computer conferencing, and FtF communication. Teams of graduate students determined the requirements for a computerized post office as a course assignment, over a two-week time period. Concerning creativity, asynchronous groups outperformed combined groups who, in turn, outperformed face-to-face groups. Combined groups produced higher quality solutions than face-to-face groups. There were no significant differences for quality between the combined and asynchronous groups.

1.0 Introduction Developing effective software is still an elusive undertaking. Although many advances have been achieved in the field of software engineering over the last twenty years, software that fails to satisfy user requirements remains a serious shortcoming [e.g. 2, 3, 5]. Given that many firms operate in highly competitive, rapidly changing environments, another dimension adding to the complexity of software development is that software teams are expected to produce creative and innovative software solutions to meet these new competitive challenges. To further complicate matters, this need for effective, innovative software occurs under the time pressure of increasingly shorter development cycles [5, 14, 20].

Communication within Requirements Determination:

1060-3425/98 $10.00 (c) 1998 IEEE

increased effectiveness of software solutions, we must find ways to improve communication while providing an environment that promotes, rather than stifles, creativity. Cusumano and Selby [8] state: “In particular, coordinating the work of a large team building many interdependent components that are continually changing requires a constant and high level of communication and coordination. It is difficult to ensure that such communication and coordination take place while still allowing designers, engineers, and marketing people the freedom to be creative. Achieving this balance is perhaps the central dilemma that managers of product development face - in PC software as well as in many other industries.” [p. 61] Indeed, requirements determination provides perhaps the best opportunity to instill creativity into the development process [5], as it is here where developers work together with users to ‘figure out what to build’ [21]. Kraut and Streeter comment that, to address the need for better communication in software development, computerized tools that facilitate distributed meetings, such as computer conferencing [18], are “likely to be useful, by opening up the meetings, making them more efficient, and providing an archive”[24, p. 80].

GSS/CMC Research Traditionally, face-to-face meetings have been a central component of collaborative work in organizations. However, new computer and telecommunications technologies are permitting groups to conduct work in a variety of ways and thus to extend the concept of collaborative work beyond traditional face-to-face meetings. In the corporate arena, it is becoming increasingly commonplace for teams to "mix-and-match" interaction media over an extended period of time to accomplish their work (e.g. [9]). Given the realities of today’s business environment, there is a pressing need to conduct research that reflects the new methods that organizations are employing to accomplish complex work [e.g., 23, 26]. To date, out of the approximately 140 experiments [13] conducted in GSS/CMC area, only three studies [11, 12, 15] investigated the usefulness of combining computermediated communication with other communication media (i.e. face-to-face and telephone) over a period of time. The research described in this paper was undertaken to explore the impact of various modes of group communication to support requirements determination, while at the same time addressing the gap in the GSS and CMC research. This paper presents results of a third experiment comparing the effectiveness of groups using various forms of communication media to conduct complex problem solving activities over a period of time. The experiment focuses on the effectiveness of matching communication media to stages of group work. Three

modes of communication are compared: (1) face-to-face (2) Web-based asynchronous computer conferencing and (3) combined – a sequence of FtF, Web-based asynchronous computer conferencing, and FtF communication The outcome variables of interest are the quality as well as the creativity of the groups’ software requirements definitions.

2.0 Previous Research Phases of Group Development The study of group development is concerned with the growth process groups experience over time. Group development is exhibited as groups progress through stages, whereby different dominant behavior is demonstrated by the group in each stage. There are three fundamental types of group developmental models - linear-progressive, life-cycle, and pendular (see [17] for a comparison of models). With the linear-progressive model, the group is seen as progressing sequentially through a series of stages over time. Generally speaking, members join together, gain initial familiarity with one another, engage in and resolve a struggle with authority, establish norms of work and conduct and move forward in performance. Tuckman’s [33] formulation of forming, storming, norming and performing follows these stages. The life-cycle model can also be categorized as a linear-progressive model, but it adds an additional terminal stage as the final stage of group development. Tuckman and Jensen’s [34] later fashioning of forming, storming, norming, performing and adjourning addresses the need for an ending stage of group development. Overall, theorists of the linear-progression and life-cycle models are essentially in agreement that the stages of group development are: orientation, dissatisfaction, resolution, production, and termination. The pendular or recurring model suggests that groups do not progress in a linear fashion, but rather experience a pendular movement back and forth which ultimately takes on a cyclical nature because of the persistence of issues rather than their resolution. This model of group development is more prevalent in open-ended groups of a long-term nature. Tuckman’s Original Model [33] Tuckman’s model as originally devised (without the terminal stage) distinguishes between interpersonal stages of group development and task behaviors exhibited in the group. This latter distinction concerning task activity is most prevalent to our study of groups, as we are concerned with task-oriented groups working on requirements determination.

1060-3425/98 $10.00 (c) 1998 IEEE

Tuckman’s stages of task-activity development in groups are as follows: 1. forming: This stage is characterized by groups orienting themselves to the task at hand. It is in this stage where groups lay out many of the ground rules for accomplishing the task. Group members familiarize themselves with the task and the way in which the group will go about accomplishing the task. Group members must decide on the type of information needed and how to obtain it in order to accomplish the task. 2. storming: In this stage, groups exhibit an emotional response to the task as a form of resistance to the demands of the task on the individual. This stage is not typically evidenced for groups working on impersonal, intellectual tasks; it is prevalent in so-called therapy groups where the task requires self-understanding and self-change. 3. norming: The open exchange of opinions is the dominant task behavior of this stage. Information is exchanged and acted upon so that alternative interpretations of the information can be formulated. 4. performing: This final stage of group development is where solutions to the task are devised. The type of solutions relevant to our research will be impersonal and intellectual in nature

Phases of Group Decision Making Hirokawa and Johnston [19] have drawn upon various academic disciplines to provide a theoretical framework for the study of group decision making. Their model is organized according to stages which they identify by time period, and, for the most-part, parallels Tuckman’s original model [33] of task-oriented group development. Time 1: introduction of problem stimulus: When group members first encounter the task, each member functions individually using his or her own cognitive schema. There is no group deliberation at this point in time. This timeperiod has no direct correspondence to Tuckman’s model. Time 2: initial interaction: During initial interaction, group members communicate regarding their perceptions, evaluations, and conclusions concerning the task, sharing their assumptions and beliefs and receiving feedback from group members. Communication skills of group members are important during these activities. It is at this point that the group begins to develop a set of agreed-upon procedures and interpretations that guides its attempts to complete the task This time-period corresponds to Tuckman’s forming stage. Time 3: discussion of positions and preferences: During this time period, group members continue the debate begun in the previous time period, presenting their opinions regarding the appropriateness of various alternatives. Persuasion is an important attribute of group members

associated with this stage. This time-period parallels Tuckman’s norming stage. Time 4: formation of group decision: During the final time period of group work, the group discusses individual preferences and rationales which leads to the formation of a consensus decision. Tuckman’s performing stage relates to this time-period.

Work content and communication mode Kiesler and Sproull [23] speculate on the type of group communication and mode of work which is most appropriate, according to the stage or time-period in which the group is working. Compared to computer-mediated communication, these researchers theorize that face-to-face meetings are a more effective means for defining issues, securing commitment, and decomposing the task -- all activities occurring during the initial (i.e., forming) stage of group work. However, during the discussion (i.e., norming) stage of group work, groups need to share information quickly and frequently. Kiesler and Sproull speculate that groups can benefit from mechanisms that allow for the quick and frequent transmission of information in a form that will permit other members to interpret and use it. Indeed, participants in both the Eveland and Bikson [11] study and Galegher & Kraut [16] study found computer-mediated communication to be effective for sharing information. Finally, groups in the late stages (i.e., performing) are working to develop a consensus decision and complete their task. Groups in this stage may find face-to-face communication more effective. Table 1 shows the relationship between Tuckman’s taskoriented group development stages, Hirokawa and Johnston’s time period model of group decision making, and work content and communication mode.

Minority influence and group creativity Prior studies (see next section) were designed around a theoretical framework based upon minority influence theory [28] and research on group creativity and innovation [29, 30, 31]. This framework consists of a model relating characteristics that have been found to foster group creativity to characteristics that are prevalent in computer-mediated groups. The model proposes that, all other things being equal, groups that use a distributed asynchronous form of computer-mediated communication will produce solutions exhibiting higher levels of. creativity than groups that do not use this form of communication.

3.0 Research Framework and Hypotheses Prior studies on requirements determination Initial study

1060-3425/98 $10.00 (c) 1998 IEEE

In a previously published study, we reported on research comparing the effectiveness of groups using different modes of communication (face-to-face versus distributed asynchronous via computer conferencing) and different problem solving approaches (structured vs. unstructured) [30]. All groups worked for two weeks developing the requirements and high-level design for an automated post office. It was found that the quality of solutions produced by the asynchronous computer conferencing (asynchronous-CC) groups was judged to be marginally higher than that of the face-to-face (FtF) groups while the creativity of solutions produced by the asynchronous-CC groups was judged to be significantly higher than FtF groups. Problem solving approach did not significantly impact creativity or quality. Second study Building on our initial study, we conducted a second study which expanded the exploration of the effectiveness of various modes of communication for groups working on requirements determination [31]. The primary research question investigated in the second study was: Are groups that use a combination of both face-to-face and computermediated communication more effective than groups that use a single mode of communication? Four communication conditions were explored: (1) Face-to-Face (FtF), (2) distributed asynchronous computer conferencing (asynchronous-CC), (3) co-located synchronous computer conferencing (synchronous-CC) and (4) combined communication - a sequence of FtF, asynchronous-CC and FtF (combined). Both the asynchronous-CC and synchronous-CC groups communicated using only a computer conferencing system, however the asynchronous-CC groups were distributed in space and time, while the synchronous-CC groups worked simultaneously within the same room. The FtF groups met twice and group members had no communication with one another between meetings. Results showed that combined groups were rated significantly higher than groups in the other communication modes in terms of creativity and quality. From a methodological standpoint, there were several limitations of our second study. The FtF and asynchronous-CC groups were the same groups used in the initial study while the synchronous-CC and combined groups were not used in any prior study. Since the second study included groups from two experimental conditions from the initial study, it was impossible to randomly assign groups across experimental conditions. Also, the second study spanned a time-period of two years. Table 2 presents a summary of the research findings of the first two experimental studies. Current study The current study is designed to avoid the methodological limitations of the second study and explore two research questions:

1.Do combined groups produce higher quality solutions than groups using a single mode of communication? 2.Are asynchronous groups more creative than FtF groups? Three communication conditions are explored in this experiment: (1) face-to-face (FtF), (2) Web-based distributed asynchronous computer conferencing (Webasynch), (3) combined communication - a sequence of FtF, Web-asynch, and FtF communication (combined). Both FtF and combined groups met twice. While the FtF group members had no communication with one another between meetings, the combined and Web-asynch groups communicated using the Web-based computerconferencing system.

Hypotheses on Quality and Creativity Drawing on research summarized in Table 1, the combined groups should experience a better match between task/decision making activity and mode of communication. The combined and FtF groups should be better able to plan and organize their work and thus experience an improved initial interaction (forming stage) compared to the Web-asynch groups. During the norming stage, where group members must gather, share and discuss information, the combined groups should outperform the FtF groups, as they will have the ability for continued communication through means of asynchronous communication. Although the Web-asynch groups will also have the ability for continued communication during this middle discussion phase, we speculate that these groups will not be as effective due to their less-effective initial stage. In the performing stage, although FtF communication is recommended, the effectiveness of the FtF groups should be lacking due to their poorer performance in the preceding stage. Typical of computermediated groups, the Web-asynch groups should experience difficulty in reaching a consensus decision in the performing stage [13]. Thus, the combined groups should be better able to integrate their work and reach a group consensus than the other groups, as they reap the benefits of matching the appropriate communication mode to the stage of group work. Finally, as found in our second study, combined groups produced higher quality solutions than single mode groups. Thus, we hypothesize that: 1.1 Combined groups will produce higher quality solutions than Web-asynch groups. 1.2 Combined groups will produce higher quality solutions than FtF groups. In our first study, it was found that distributed asynchronous computer conferencing groups were significantly more creative than FtF groups. Although these results were not replicated in the second study,

1060-3425/98 $10.00 (c) 1998 IEEE

research on minority influence theory [e.g. 28] and group creativity/innovation [e.g. 38] (see 29 and 30 for a detailed explanation) indicates that creativity is enhanced when groups are not constrained by time. Similarly, communication among group members is believed to be important for encouraging creativity. Therefore, due to the increased ability of the asynchronous-CC groups to stay connected with group members and the relative lack of time pressures compared to both the synchronous-CC and FtF groups, and based on the results from our initial study, we again expect that asynchronous groups will produce more creative results than FtF groups. Based on the appropriateness of fit of communication mode with the stages of group work (see Table 1) and results from our second experiment concerning combined groups, we also expect that the combined groups will be more creative than the FtF groups. As for a comparison of combined groups to Webasynch groups, although both conditions incorporate asynchronous communication, we anticipate that the combined groups will be more creative; again, we base this on the appropriateness of fit between communication mode and stages of group work. Additionally, results from our second study showed that combined groups were more creative than asynchronous groups. Thus, we hypothesize that: 2.1 Web-asynch groups will produce more creative solutions than FtF groups. 2.2 Combined groups will produce more creative solutions that FtF groups. 2.3 Combined groups will produce more creative solutions than Web-asynch groups.

4.0 Method There are a total of 27 groups included in this study (see Table 3). All groups met face-to-face for training. When working on the actual experimental task, groups in the combined and FtF conditions met for two face-to-face sessions, occurring exactly two weeks apart. Groups in the asynchronous condition conducted all work using the conferencing system over a two-week time period. Task: The Computerized Post Office (CPO) is the task used in this experiment. Groups were required to reach consensus on the initial requirements of the CPO and to submit these requirements in a formal report at the end of the experiment; each group produced a single report. The report was to cover the functionality of the CPO along with implementation considerations and was also to contain a description of the user interface design. This is a slight modification of the APO task used in our prior studies (groups are asked to devise services to be offered immediately and within the next five years).

Subjects: Subjects consisted of graduate students in the CIS and IS majors at the New Jersey Institute of Technology (NJIT). For their participation, all subjects received course credit. The majority of subjects had course work and/or job experience directly relevant to systems design. Group size ranged from 4 to 7 persons for all groups. Subjects were scheduled to meet based on availability for scheduled sessions. Technology and Facilitation: All of the computermediated groups (asynchronous, and combined) communicated using the Web-EIES version 3.0 computer conferencing system developed at NJIT. The interface permits the subjects to use Netscape or other web browser such as Internet Explorerto access the conferencing system. Each computer-mediated group communicated in its own conference set up on Web-EIES. The conferences were minimally facilitated. The conference facilitator's role was that of a technical assistant, helping groups with equipment problems and answering questions of a technical nature. Training: All groups met face-to-face for training and used the same practice problem, called Entertainment for Dutch Visitors [32]. Groups using the computerconferencing system were trained on the communication features of Web-EIES and worked on the conferencing system using the practice problem. FtF groups also worked on the practice problem, without using technology. Groups in the combined condition received the same training as did the Web-asynchronous groups. Procedures: After training, groups in the FtF and combined conditions remained for one additional hour to begin work on the CPO task. This meeting constituted the first face-to-face session for these groups. At the end of the two-week time period, these groups reconvened for a second face-to-face meeting, which lasted up to 2-1/2 hours. Each of these groups was provided with a computer with word processing software for their second meeting. Although instructed not to communicate with other group members during the interval between the first and second meeting, group members in the FtF condition were permitted to work independently on the CPO during this time period and bring their work to share with the group at the second FtF session. After training, the groups in the Web-asynch condition were excused and told they would now be communicating asynchronously using the computer conferencing system for the next two week period. They were instructed to return in two weeks to complete the questionnaires and be debriefed. Groups in the combined condition communicated asynchronously using the computer conferencing system during the two week interim between the two face-to-face sessions. For the second face-to-face meeting, groups in the combined condition were provided with a computer with word processing software and a terminal to connect to

1060-3425/98 $10.00 (c) 1998 IEEE

Web-EIES. The groups were instructed that they had 21/2 hours to complete the task. All groups had a leader who volunteered for the role at the end of the training session, sometimes with encouragement from the facilitator. Each leader was responsible for ensuring that his/her group submitted a formal report at the end of the two-week experimental time period. For each FtF group, the facilitator completed an observation form at the end of the second session, rating the group on such aspects as general group atmosphere and type of coordination used. It was also noted whether the group completed their work early, or had to rush to finish the assignment. Debriefing: All groups were debriefed in a face-toface session. All participants were questioned regarding their adherence to the rules for communication outside the two sessions. Dependent Variables: A pair of expert judges measured the dependent variables of quality of solution and creativity of solution. These judges had academic and/or professional experience in systems design. The judges were the same judges as in Ocker, et al. [31] and were instructed to use the same procedures. All groups' formal reports were printed using the same word processing package and aspects of each group's mode of communication were masked. Judges rated each group on various aspects of the design (e.g., functionality and interface), written presentation, and overall quality of the analysis [32]. The level of creativity contained in each group's design was also measured by the panel of judges, using a single category called “creativity of solution” [1].

5.0 Results The GLM procedure in SAS for Windows, release 6.12, was run to test the effects of mode of communication on the dependent variables. All statistical means are calculated using the least-square means calculation. In presenting results, significance levels of 0.05 or better will be considered "statistically significant." Levels between 0.10 and 0.05 indicate findings that suggest a relationship may exist and will be considered "marginally significant." Table 4 presents the descriptive statistics for the dependent variables. ANOVA results are contained in Table 5 for all dependent variables. The expert judges had acceptable levels of agreement when rating the quality of solution (Cronbach’s alpha = 0.784) and for creativity of solution (Cronbach’s alpha = 0.714). Quality Measure: This expert rated category was judged on a scale from one (poor) to seven (excellent). Combined groups were rated higher than asynchronous groups, as predicted, but there were no significant differences (4.82 versus 4.43, p> 0.269). Therefore,

hypothesis 1.1 was not supported. Combined groups were rated significantly higher in quality than face-to-face groups (4.82 versus 4.13, p > 0.0498), providing support for hypothesis 1.2. Creativity Measure: The expert judges rated the creativity of each group’s solution on a scale from one (poor) to seven (excellent). Hypothesis 2.1 was supported; asynchronous groups were rated significantly higher in terms of creativity than those of the face-to-face groups (5.88 vs. 4.63; p> 0.004). Hypothesis 2.2 was marginally supported as combined groups were judged more creative than FtF groups (5.18 vs.4.63, p>0.065). Hypothesis 2.3 was not supported; combined groups were judged less creative than the asynchronous groups (5.18 vs. 5.88; p>0.024).

6.0 Discussion and Conclusions This study continued our exploration of the effects of different modes of communication on groups working on requirements determination. In particular, this research focused on the usefulness of matching face-to-face and computer-mediated modes of communication according to the stages of group work. Table 6 contains a summary of the empirical findings for the current study. As was hypothesized, asynchronous groups were rated significantly higher than FtF groups in terms of creativity of solution. These findings replicate those of our first experiment. Likewise, as expected, combined groups outperformed FtF groups, both in terms of quality and creativity. These results replicate those obtained in our second study. In contradiction to results from our second study, our hypotheses predicting that combined groups would outperform asynchronous groups, both in terms of creativity and quality, were not supported. Although asynchronous groups were judged higher than combined groups on quality, the difference was not significant. However, as for creativity, asynchronous groups were not only rated higher than combined groups, but the difference was significant. This result was the opposite of what we predicted. Thus, our findings on creativity show that asynchronous groups scored higher than combined groups who scored higher than FtF groups: asynchronous > combined > FtF Our findings provide partial support for the appropriateness of communication mode/stages of group work model (Table 1) in that combined groups outscored FtF groups on both the quality and creativity measure. Thus, combining FtF and asynchronous modes of communication according to the stages of group work was

1060-3425/98 $10.00 (c) 1998 IEEE

more effective than using the single communication mode of FtF, for both creativity and quality. Table 6 Summary of Empirical Results Dependent Hypothesis Prediction Variable 1.1 Combined > Asynchronous Quality

Creativity

Result Unsupported

1.2

Combined > FtF

Supported

2.1

Asynchronous > FtF

Supported

2.2

Combined > FtF

2.3

Combined > Asynchronous

Marginally Supported Unsupported

Based on this model, however, we also predicted that combined groups would outperform asynchronous groups. This did not occur. In terms of creativity, asynchronous communication was found to be significantly better than either condition that incorporated FtF communication (i.e., FtF, combined). These findings, provide support for the theoretical model from our first study, which incorporates minority influence theory and creativity antecedents with aspects of GSS to predict that FtF communication will inhibit creativity and that distributed asynchronous communication will enhance creativity. In trying to explain these results on creativity, it should also be remembered that the task used in the current study was a slightly modified version of the task used in the prior two studies. It was felt that the APO task, as originally used, was not open-ended enough in terms of allowing for creative responses. Thus, in the current study, the APO task was modified to include not only a description of an automated post office for the immediate future, but also for five years in the future. Additionally, we designed the current study to avoid the methodological shortcomings of the second study (i.e., non-random assignment of subjects to conditions over two years). This, too, could account for the different results between the second and third studies.

FtF groups, and that the communication mode had nothing to do with the results obtained. Another alternative explanation concerns the timing of the group interaction; asynchronous groups were able to co-locate their work closely in time (i.e. near the end of the second week), while FtF groups were forced to split their effort over two weeks. Perhaps being able to accomplish the task without a two week “interruption” might account for the noted differences between treatments. In considering these explanations, it should be remembered that this experiment was designed to explore the ways in which work is accomplished in organizational settings, while still maintaining a controlled laboratory setting. The FtF treatment simulates the traditional mode of organizational work whereby work is accomplished during meetings by the group as a whole, with individual group members working on tasks between meetings. Hence, in the FtF treatment, individual team members were permitted to work on the task during the two week interval between the first and second group meetings; individuals could bring their work to the second group meeting. Additionally, the amount of time given to FtF groups seems adequate, as all of these groups completed the task in the allotted meeting time, with many of the groups finishing without using the full amount of scheduled time. However, in order to fully address the alternative explanations, a fourth communication mode which allows unlimited FtF communication should be added to the experimental design. As it stands, our findings show that mode of communication can have a significant impact on the effectiveness of group work, both in terms of quality and creativity. Although the face-to-face mode of work is still most prevalent in organizations, firms are increasingly using asynchronous technologies to conduct business. By intensifying their use of these technologies, and combining them with the more traditional face-to-face work mode, companies can experience increased levels of innovation and effectiveness.

Future Research Limitations Although it is tempting to attribute the experimental findings solely to differences in the communication media, several alternative explanations exist. This study did not control or evaluate the level of effort put into the task by groups who used asynchronous communication; however, the amount of interaction was controlled for the FtF groups. Both the asynchronous and combined groups potentially have the ability for more group communication then the FtF groups, due to the ability to “log on” to the conferencing system at any time, and for any length of time. Thus, one alternative explanation for the outcomes found is that groups using asynchronous communication were simply able to work on the problem more than the

The current experiment compared the effectiveness of three modes of communication. However, many more modes and combinations exist (see Table 7). In order to more fully explore the appropriateness of communication mode matched to the stages of group work, an experiment should be conducted which compares the performance of groups working in the following conditions: Table 7 Future Research Condition

Initial Stage (Forming)

1060-3425/98 $10.00 (c) 1998 IEEE

Middle Stage (Norming)

Final Stage (Performing)

1

FtF

FtF

FtF

2

Asynchronous

Asynchronous

Asynchronous

3

FtF

Asynchronous

FtF

4

FtF

Asynchronous

Asynchronous

5

Asynchronous

FtF

Asynchronous

6

Asynchronous

Asynchronous

FtF

7

media choice

media choice

media choice

Additionally, this paper develops the relationship between stages of group development and communication mode (see Table 1). However, only hypotheses which focus on the final group solution were tested. Explicit hypotheses matching the different stages of group work to the communication media should be developed and tested.

Acknowledgments Partial funding for this research was provided by the National Science Foundation (9015236), the New Jersey Institute of Technology under SBR Grant number 421090, and The Pennsylvania State University at Harrisburg. Special thanks are extended to Yu-Fang Lu, George Olsen, Les Roberts, and Kevin Walsh.

References 1. Amabile, T. M. 1983. The social psychology of creativity: A componential conceptualization. Journal of Personality and Social Psychology, 45,2 (1983), 357-376. 2. Bostrom, R.P. 1984. Developmnet of computer-based information systems: A communication perspective, Computer Personnel, 9, 4. 3. Bostrom, R.P. 1989. Successful application of communication techniques to improve the systems development process. Information and Management, May 1989, 279-295. 4. Byrd, T. A., Cossick, K.L., and Zmud, R.W. 1992. A synthesis of research on requirements analysis and knowledge acquisition techniques. MIS Quarterly, March 1992, 117-138. 5. Couger, J. D. 1996. Creativity & Innovation in Information Systems Organizations, Danvers, MA: Boyd & Fraser. 6. Cronan, T.P. and Means, T. L. 1984. System development: An empirical study of user communication. Data Base, 15, 3, 2529. 7. Curtis, B., Krasner, H., & Iscoe, N. 1988. A field study of the software design process for large systems. Communications of the ACM, 31, 1268-1287. 8. Cusumano, M.A. and Selby, R.W. 1997. How Microsoft builds software. Communications of the ACM, 40, 6, 53-62. 9. Cutosky, M.R., Tenebaum, J.M. and Glicksman, J. 1996. Madefast: Collaborative engineering over the Internet. Communications of the ACM, 39, 9, 78-87. 10.Dunn, R. Software defect removal. New York: MacGraw-Hill, 1984.

11.Eveland, J.D. & Bikson, T.K. 1989. Work group structures and computer support: A field experiment. ACM Transactions on Office Information Systems, 6, 4, 354-379. 12.Finholt, T., Sproull, L., & Kiesler, S. 1990. Communication and performance in ad hoc groups. In Galegher, J., Kraut, R., and Egido C. (eds.). Intellectual teamwork: Social and technological foundation of cooperative work. Hillsdale, NJ: Lawrence Erlbaum. 13.Fjermestad, J, and Hiltz, S. R., 1997. Experimental studies of GDSS: An Assessment of variables studied and methodology. The Thirtieth Hawaii International Conference on System Sciences, vol. 2, 45-65. 14.Ford, C.M. and Gioia, D.A. 1995. Creative action in organizations: Ivory Tower Visions and Real World Voices, Thousand Oaks: Sage. 15.Galegher, J. 1990. Intellectual teamwork and information technology: the role of information systems in collaborative intellectual work. In J. Fulk and C. Steinfield (Eds.), Organizations and Communication Technologies. London: Sage Publications. 16.Galegher, J. and Kraut, R. 1990. Computer-mediated communication for intellectual teamwork: A field experiment in group writing. Proceedings of CSCW 90, 65-78. 17.Gibbard, G., Hartman, J., and Mann, R. Group process and development, in G. Gibbard, J, Harmant, and R. Mann (eds.) Analysis in Groups. San Francisco: Jossey-Bass, 1974. 18.Hiltz, S. and Turoff. 1993. M The Network Nation, Cambridge: MIT Press. 19.Hirokawa, R.K. and Johnston, D.D. 1989. Toward a general theory of group decision making: Development of an integrated model. Small Group Behavior, 20, 4, 500-523. 20.Holtzblatt, K. & Beyer, H.R. 1993. Making customercentered design work for teams, Communications of the ACM, vol. 36(10), pp. 93-113. 21.Holtzblatt, K. and Beyer, H.R. 1995. Requirements gathering: the human factor. Communications of the ACM, vol. 38(5), 3132. 22.Kaiser, K.M. and King, W.R. 1982. The manager-analyst interface in systems development. MIS Quarterly, 6, 1, 49-59. 23.Kiesler, S. and Sproull, L. 1992. Group Decision Making and Communication Technology, Organizational Behavior and Human Decision Processes, 52, 96-123. 24.Kraut, R. E. and Streeter, L. A. 1995. Coordination in software development. Communications of the ACM, vol. 38(3), 69-81. 25.Martin, M.P. and Fuerst, W.L. 1984. Communicaitons framework for system design. Journal of Systems Management, March 1984, 18-25. 26.McGrath, J.E. and Hollingshead, A.B. 1994. Group Interacting with Technology, Sage, Thousand Oaks, CA. 27.Nagasundaram, M. and Bostrom, R.P. 1995. The structuring of creative processes using GSS: A framework for research. Journal of Management Information Systems, 11, 3, 87-114. 28.Nemeth, C.J. Differential contributions of majority and minority influence. Psychological Review, 94 (1986), 23-32. 29.Ocker, R. 1995. Requirements definition using a distributed asynchronous group support system: Experimental results on quality, creativity and satisfaction. Unpublished Doctoral Dissertation, Rutgers University. 30.Ocker, R., Hiltz, S.R., Turoff, M., Fjermestad, J., 1995-96. The effects of distributed group support and process

1060-3425/98 $10.00 (c) 1998 IEEE

structuring on software requirements development teams, Journal of Management Information Systems, 12, 3, 127-153. 31.Ocker, R.J., Fjermestad, J., Hiltz, S.R., Turoff, M. 1997. An exploratory comparison of four modes of communication for determining requirements: Results on creativity, quality and satisfaction. The Thirtieth Hawaii International Conference on System Sciences, vol. 2,568-577 32.Olson, J.S., Olson, G.M., Storrosten, M. and Carter, M. 1993. Groupwork close up: A comparison of the group design process with and without a simple group editor. ACM Transactions on Office Information Systems, 11, 4 (1993), 321-348. 33.Tuckman, B. 1965. Developmental sequence in small groups. Psychological Bulletin, 633, 6, 384-399.

34.Tuckman, B. and Jensen, Mary Ann. 1977. Stages of smallgroup development revisited. Group & Organizational Studies, 2, 4, 419-427. 35.Valusek, J.R. and Fryback, D.G. Information requirements determination: Obstacles within, among, and between participants, in Information Analysis: Selected Readings, R. Galliers (ed.), Addison Wesley, Reading, MA, 1987, 139-151. 36.Vessey, I. and Conger, S. 1993. Learning to specify information requirements, Journal of Management Information Systems, 10, 2, 177-201 37.Walz, D.B., Elam, J.J., and Curtis, B. 1993. Inside a software design team, knowledge acquisition, sharing and integration. Communications of the ACM, 36, 10, 63-77. 38.West, M.A. and Farr, J.L. Innovation and Creativity at Work, Chichester: Wiley and Sons, 1990.

Table 1. Relationship between stages of group development, decision making time-period, and communication mode Tuckman's stage of taskoriented group development

Hirokawa & Johnston's timeperiods of group decision making

Task/decision making activity

Communication mode

not applicable

introduction

individual members reflect on task

not applicable

forming

initial interaction

planning work

face-to-face communication

norming

discussion

individuals gather information; share information and opinions with group

asynchronous computermediated communication

performing

formation of decision

integrate work of group face-to-face communication members; develop consensus decision

Table 2 Comparison of Results from Prior Experiments

Ocker et al., 1996 Ocker et al. 1997

Creativity Asynch. > FtF (significant)

Quality Asynch > FtF (marginally significant)

Combined > Asynch., Synch., FtF (significant)

Combined, Asynch. > Synch., FtF (significant)

1060-3425/98 $10.00 (c) 1998 IEEE

Table 3 Experimental Design Experimental Condition

Number of Groups

Training

Meeting 1

14 Day Interval

Meeting 2

FtF

8

Yes

Yes

No Interaction

Yes

Asynchronous

8

Yes

No

No

Combined

11

Yes

Yes

Web-Based Asynchronous Communication Web-Based Asynchronous Communication

Yes

Table 4 Least Square Group Means for Dependent Variables

Creativity of Solution Quality of Solution

Asynchronous

Face-to-Face (FtF)

Combined (FtF & Asynchronous)

5.881,2

4.631,3

5.182,3

4.43

4.131

4.821

Means with the superscript 1 or 2 are significantly different from each other. Means with the superscript 3 are marginally significantly different from each other.

Table 5 ANOVA Results for Dependent Variables

Creativity

Quality

Source

DF

MS

F

Pr>F

model

2

6.28

8.07

0.0018

error

27

0.77

model

2

2.26

2.15

0.14

error

27

1.06

1060-3425/98 $10.00 (c) 1998 IEEE