This project was supported by a grant from Electronic Data Systems. The authors would like to thank Kipp Verner and Greg Collier for valuable insight and assistance on this project.

This report summarizes the activities and findings of the 1997 Global Engineering Methodology Study (Gemsteam). Gemsteam is a collaborative research program of the Michigan State University Department of Telecommunication and the Case Center for Computer Aided Engineering and Manufacturing, sponsored by EDS. The goal of the research program is to investigate methods for improving the efficiency and quality of engineering design teams in geographically distributed contexts. The collaboration between these departments was launched in 1996 with a pilot study performed during the summer of that year. Using four EDS interns working at MSU, the effects of telecommunications technologies on distributed work groups were investigated. The experience and findings of the summer pilot led to an extended collaboration with groups from both Singapore and Germany.

In the spring of 1997, the Gemsteam research team conducted an experiment on globally distributed teams, involving engineering students from Michigan State University, the National University of Singapore, and the University of Kaiserslautern in Germany. Each team included both electrical and mechanical engineering students. During a four month time period, six teams of engineers completed an industry sponsored design project using either standard or emerging Internet-based communication technologies. Employing pre- and post-test questionnaires, a daily communication log, a review of the quality of the interns・ engineering designs, and in-depth interviews, data was collected to assess the impact of the various communication tools. Outcomes of interest were objective measures of the quality of the engineering solutions developed, the communications behavior exhibited by each group, perceptions about the usefulness of the communication tools, and team members・ satisfaction with the group experience. The usefulness of the media were judged on dimensions such as ease of use, understandability of information being communicated, and confidence in the information shared via a particular medium. The roles various media played in supporting team communication were established based on the frequency and timing of their use.

One central finding of the study is that, despite the constraints imposed by extreme geographic distance, cultural and disciplinary differences, the distributed teams were able to successfully coordinate their activities using the available communication media. The teams in the study produced several high quality designs during the four month period of collaboration. However, not all media were evaluated and used equally by the teams. Given time-zone differences, asynchronous media were clearly the preferred mode of interacting across locations. Email was the most frequently used medium, with the telephone, fax, and file transfers used as supplements. Internet-based video conferencing was used infrequently as the students found it difficult to use and inconvenient to schedule. Face to face interaction did occur among the MSU-based team members, but was not possible between team members in different locations.

Detailed findings are reported below in the following four sections. The first section discusses the reasons for performing research on globally distributed design teams and offers research questions in relation to overall patterns of communication. Next, the project design of the study is reviewed, followed by results and discussion of key findings. The implications section presents the offers findings for firms using globally distributed design teams. Recommendations for future research conclude this report.

The need for geographically distributed, multi-function, and multi-cultural teams arises from many of the pressures faced by modern industry. Globalization of markets and competitive pressures have forced industry to reduce time-to-market cycles, while at the same time increased the difficulties of coordinating activities across widespread operations. The engineering design process is one area where product development time can be reduced. Concurrent engineering, where different functions in engineering design and manufacturing are worked on simultaneously, rather than sequentially, is one approach firms use to reduce product development time. Yet, people with different functional specialties are often found in different locations, making coordination dependent upon electronic communications. Moreover, with the increasing internationalization of firms, effective coordination among distributed teams, even those from within the same specialization, becomes an even greater necessity. Such cross-locational teams can be separated by large distances with extreme time zone differences. Moreover, the different styles of communication, work, and project management across cultures hinders group efforts to coordinate activities and work effectively as a team. Although many new communication technologies have reduced the overall costs for information sharing, it is not clear whether they can yet serve as a viable substitute for co-location of engineering teams.

The Gemsteam project contributes to the growing base of knowledge about the extent to which emerging computer-based communication media can support distributed teams. Specific questions addressed by this research included:

1. What types of communication media are preferred by globally distributed design teams and to what degree do these preferences change over time?

2. To what extent are synchronous media such as the telephone and video conferencing used? To what extent do groups rely on asynchronous media such as electronic mail?

3. What are the patterns of cultural, functional, and locational communication in international engineering design teams?

4. Which media do groups feel elicit the greatest ease, confidence, and understandability in terms of group communication?

5. Which media do groups feel contribute the most to group progress?

6. How do globally distributed engineering design teams perform compared to physically proximate engineering design teams and do multimedia applications enhance performance?

Background

The research questions presented above were investigated using the following project design. The study involved 6 engineering teams, four teams had four members and two teams had three members. On each team, two of the members were located at Michigan State while one member was a National University of Singapore/EDS intern and the other was located at the University of Kaiserslautern in Germany (on Teams 1-4). In addition to being cross-cultural the teams were also cross-functional. The subjects from Michigan State were mechanical engineering students while those from Germany/Singapore were electrical engineering students.

Three engineering projects, provided by EDS, constituted the work for the teams during the course of the study. The first project required an update to the design of an automotive instrument panel. The second project required the design of a power adapter for a portable computer that drew power from a source other than the computer keyboard port. The third project required the design of a desktop computer data adapter which recognizes camera and printer data to enable automatic switching.

The students at Michigan State were given project descriptions and four students signed up for each project based upon their interests. On each project, two MSU students were then randomly assigned to a control or experimental condition, and each worked with one intern from Singapore and one from Germany (on Teams 1-4). The experimental condition provided access to augmented communication technologies while the teams assigned to the control condition were provided access only to those communication technologies that are widely available. The standard technologies included phone, voicemail, fax, email and file transfer (FTP). The augmented technologies included the standard technologies with the addition of Internet-based video conferencing and electronic white board tools.

The study occurred during the 4 month period from January to April during the Spring Semester at Michigan State. The longitudinal nature of the study provided a realistic setting for the generalization to industry engineering design teams whose tasks usually require several months to complete. It also enabled the research team to examine how communication media usage evolved over time.

Therefore, the overall research design was quasi-experimental with the 6 teams working on three projects over 4 months. On each project, one team was assigned to the control condition and the other to the experimental condition. This design allowed for comparisons among the different projects to control for effects that different projects would have on design quality, team satisfaction, etc. Additionally, the design facilitated comparisons between groups with different levels of access to communication technologies.

Forming the teams according to the research design proved to be difficult. The German and Singaporean counterparts were not able to provide 6 part time participants each. Therefore a compromise was reached. The compromise resulted in three subjects in Singapore, with each subject being assigned to two teams each. In Germany where the participants were limited to part-time work, only four participants could be found. The result was that two of the six teams did not have a "German" counterpart. Another difficulty resulted from the nationalities of the team members. Although all of the Singaporean members were indeed Singaporean, two of the "German" counterparts were actually Spanish students on a foreign exchange program in Germany. The difference in nationality presented a problem in terms of making cross-cultural comparisons as more than three cultures were actually represented in the project. Also, for the Spanish students on exchange in Germany, English was a third language, as opposed to merely a second language, which was the case for the other "German" students. This did not necessarily need to present a problem, in one case, however, the student・s English proficiency was not as good as the other students.

The communication technologies available to the teams were divided into groups of standard technologies and augmented technologies. The standard technologies were those considered to be most commonly available in the engineering field. The augmented technologies included those in the standard group, with the addition of Internet-based video conferencing tools and electronic white boards. The choice of an Internet-based system was made to minimize the cost of both hardware and software, as well as communication costs. Using the Internet for these sessions eliminated the need for ISDN lines and their usage sensitive call charges. The inexpensive nature of these technologies helps make the research results applicable to firms with a wide range of communication technology budgets.

A list of the available technologies for each group are shown in the following table.

Measurement

Five types of measurement instruments were used in this study. First, a pretest was administered at the beginning of the project to insure the relative equivalency of the groups. The second instrument was a daily diary. The diary instrument took the form of a web-based questionnaire. The students were instructed to fill out the form for each communication event. This effort provided a record of the choice of media and hence the frequency of its use. The form also captured students・ assessments of the ease of use, understandability, and progress made using a particular communication medium. Diaries have been criticized as data collection tools because of the difficulty of insuring that subjects record their opinions each day. Having the diary form on the world wide web allowed new methods to be developed to overcome this difficulty. To encourage the students・ discipline in filling out the form a script was written that verified that each student submitted at least one form each weekday. (It was possible for students to submit one form indicating that no communication had occurred on a particular day.) If a student did not submit a form, the script automatically sent them a ．reminder email.・

In addition to gathering data from the research subjects via questionnaire, each student performed an in-depth interview approximately half-way through the study period. The interviews allowed the students to reply specifically to questions about team cooperation, media usage, cross-functional and cross-cultural communication issues. In the interview, students were asked follow-up questions on data entered into the diary form, thereby further emphasizing the importance of their contributions submitted via that measurement tool.

At the conclusion of the study, the students answered questions on a post-test questionnaire. The questions in this survey measured the subjects attitudes toward distributed group work, the media used, and the design quality of their solutions (i.e., a series of questions regarding completeness, accuracy, appropriateness and creativity of the design).

In the following paragraphs, the findings of the study are discussed with reference to the three questions posed by the overall research project. The results are drawn from the wide range of measurement devices employed in this study and represent both quantitative data as well as qualitative interview data.

The first goal of the study was to explore the capabilities of Internet-based video conferencing in supporting globally distributed design teams. To this end, frequency of use of these Internet-based communication technologies was examined. Unfortunately, the results indicate the students found Internet-based video conferencing difficult to use and that they were generally dissatisfied with its performance; it was considered burdensome by some of the students. Of the total number of communication events reported by all three of the augmented technology teams, only 7% of those events were via the Internet-based video conferencing modes. This can be compared to email which represents 61 % of the communications events.

Through interviews and the diary forms, the students indicated that team video conferencing sessions were less productive than interactions using other media. The reduced productivity was due to the additional time necessary to establish the connection and subsequently maintain it. Frustration also came from the unreliability of the medium. A high quality connection could change at any time, resulting in a switch to another medium. The switching of media disrupted the flow of the meeting, in addition to being a waste of time. Overall, the students reported technical problems on 78% of their video conferencing calls. In addition to evaluating the video conferencing the students also had the opportunity to use electronic white board tools included with the video conferencing software. Some students felt that in conjunction with a video conference, the white board tool was quite useful. However, if the video conferencing connection was poor the white board tool along with the conference was abandoned.

The students・ experience with Internet-based video conferencing made them an important resource for assessing its present usability as well as the medium・s future potential. During interviews, in response to questions about the usefulness of a hypothetical video conferencing system that was reliable and easy to use, the students expressed mixed attitudes. Some of the students felt being able to see one another was an important factor in their ability to function as a team while others believed it was unnecessary to see their project partners. The relationship between seeing project partners and design quality could not be estimated by the quantitative data as productive use of the medium did not occur. This lack of augmented technology use led to the abandonment of the differentiation between augmented and standard technology teams. The remainder of this analysis treats all groups as equal in terms of available technologies.

The second question of concern to this research asked "what patterns of communication are established by globally distributed design teams?" Here "patterns" refers to several characteristics of the groups・ communication, including overall frequency and timing of communication as well as types of media used during certain time periods. Also examined were what patterns exist for communicating with different types of team members (i.e. same-function or cross-function team members).

Frequency and Timing

The following graph shows time series data of all communication events broken down by media. A pattern of increased communication approaching weeks 5 and 10 and sharply decreased communication during the final two weeks of the project is apparent. During weeks 5 and 10 project reports were due, with the final reports and presentations occurring during weeks 13 and 14. This general pattern reflects the need to resolve ongoing issues at key points in time in the design cycle. Resolving issues led to "peaks" in communication (i.e., an increase in reported communication events), particularly immediately prior to a due date. During the final project phase in which the design had already been agreed upon and implemented, the remaining task was writing the report. This task required a lower level of consultation and thus communication frequency decreased.

Figure 2: Communication Frequency

Another finding displayed by the above chart is that the preferred mode of communication was email, and this preference changed little during the course of the project. Students・ assessments of email usage will be discussed in a subsequent section. It will be noted here, however, that the overall preference for email was not surprising, although the constancy of it was. The above chart demonstrates that when near a deadline the teams・ reliance on email continued. This is unexpected because under time pressure people usually require immediate feedback. This logic would predict an increase in the use of synchronous media, in this case either face-to-face, phone, or video conferencing modalities. However, with the time zone differences that separate the team members, it was difficult for them to communicate in a synchronous mode. As a result, despite the need for immediate response, the teams continued to rely on email.

Cross-Locational, Cross-Cultural, and Cross-Functional Communication

The teams in this study had several barriers to overcome in their communication. Take, for example, a phone conversation between a German engineer and an engineer located at MSU. This communication would have occurred over a large geographic distance, crossing several time zones; it would have had a cross cultural component as the two engineers were from distinctly different cultures. Finally, this communication would have been cross-functional as the MSU student was a mechanical engineer and the German student was an electrical engineer. Within the constraints of our research design, cross-locational communication is always cross-cultural but may or may not be cross-functional. Communication between Singapore and Germany is an example of communication which is cross-locational and cross-cultural but not cross-functional. The frequency of communication events between the locations are shown in the following table.

The distribution of communication events among the locations which is shown as percentages in the bottom row of Figure 3 are shown pictorially in Figure 4.

Figures 3, 4, and 5 (below) illustrate that, despite the time differences and cross-cultural nature of cross-locational communication, students did communicate with their overseas counterparts. For Teams 1, 3, and 4, communication among the American students accounted for only roughly 30% of the total communication, indicating that a significant amount of "other" communication was taking place. By examining the percentage of US-US communication by project (i.e., Team 1 vs. Team 2; Team 3 vs. Team 4, Team 5 vs. Team 6), it shows that projects generated similar levels of communication (although Team 2 can be considered an outlier). Moreover, the intra-functional (i.e., ME-ME and EE-EE) and inter-functional (i.e., ME-EE) combined communication reiterates the point that similar amounts of communication occurred within and between functional area specialties (a point that will be elaborated on later in this section).

Team	Germ-Sing	US-Sing	US-Germ	US-US	Total	Intra-Funct.	Inter-Funct.
1	22.56	18.46	23.59	35.38	100	57.94	42.05
2	14.12	12.16	16.47	57.25	100	71.37	29.63
3	34.28	14.49	20.49	30.74	100	65.02	34.98
4	33.42	18.21	16.03	32.34	100	65.76	34.34
5	0.00	42.70	0.00	57.30	100	57.30	42.70
6	0.00	36.87	0.00	63.13	100	63.13	36.87

The students in the study were well prepared for this international communication. Before joining the project, many of the students had cross-cultural experience in terms of travel and employment. Over half of the students indicated they had traveled to a country where the native language was not their own, and just under half also had foreign work experience. These attributes were identified during the pretest questionnaire and may be an artifact of the self-selected sample of students.

Due to the difficulty of quantitatively separating out the cross-functional, cross-locational and cross-cultural effects, the students were asked to express their opinions about these factors in the in-depth interviews. When asked about the impact of the cross-cultural factors for the outcome of the project, the American, German and Singaporean students alike indicated they felt the cross-cultural factors had no important effect on the project. Small problems with language, either written or spoken, were overcome easily. Students indicated no additional frustration due to the small amount of extra effort this created. Additionally, several of the students felt the ability to interact with people from different cultures made the project more interesting.

When asked specifically if intercultural factors had led to misunderstandings, several students qualitatively replied that misunderstandings had occurred but they were not due to intercultural factors. The most difficult communication, they felt, was between functional areas. The ability of the mechanical engineers to understand the electrical engineers・ technical jargon was stated to be a bigger barrier for successful communication than even national culture or differences in English language ability (English was the language used by all teams, however for some subjects English was their second or third language and proficiencies in written and spoken English did vary). Assuming that people are less willing to communicate if communication is difficult, a communication barrier between electrical and mechanical engineers would lead one to expect the majority of the communication to occur within functions rather than across functions. Examining the distribution of communication events among teams 1-4, a pattern which supports this expectation can be seen. Figure 6 shows the subtotal of communication events occurring between the American students themselves (mechanical engineer to mechanical engineer) and between the Germans and Singaporeans (electrical engineer to electrical engineer) was higher than the subtotal of communication events between the Americans with both the Germans and Singaporeans (mechanical engineer to electrical engineer).

Teams	Germ-Sing	US-US	Sub-Total	US-Sing	US-Germ	Sub-Total
1	44	69	113	36	46	82
2	36	146	182	31	42	73
3	97	87	184	41	58	99
4	123	119	242	67	59	126
Total	300	421	721	175	205	380

While examining cross-functional communication another pattern emerged. The number of total communication events was higher for certain projects than others, as noted in Figure 7; that teams working on projects which relied more heavily on the electrical engineers communicated more. Figure 7 also shows that the percentage of cross-function and within-function communication was similar in these groups (approximately 2/3 within function and 1/3 across function). This trend suggests that there was a 2 to 1 ratio of within to across group communication, despite the variation in frequency ranges from 59 to 213 total events. Clearly, this pattern needs to be replicated with further empirical observations, but may shed light on the functional differences noted previously in qualitative interviews.

The next section probes further into these findings by investigating the reasons behind the use of the available media via assessments of ease, understandability, confidence, and progress.

It is also important to know the reasons why subjects chose one medium over another. For example, email was shown to be the most frequently used communication medium but little has been said about what makes this medium so popular. In this section the research question on how these technologies supported the teams will be answered.

On the diary form students were asked to judge the communication medium along three dimensions: ease of use, understandability of information communicated and confidence in the information communicated. Such criteria are frequently used to study the information channel effects of media usage. Ease of use, understandability, and confidence are categories that reflect media richness, a theoretical perspective which focuses on the equivocality (or relative ambiguity) of messages and media. The results of this study are displayed in Figure 8, which shows that face-to-face communication was rated highest on all three of these dimensions. Email was rated second in the user assessment, with telephone coming in third. The frequencies of all other media were so small that the assessments of their use were lumped together in the "other" category. Correspondingly, user scores on these three dimensions were the lowest of all media.

It is not surprising face-to-face communication rated the highest on these three measurements. It has been shown in previous research that the capabilities of this form of communication far exceed the others. In the present study, there are additional factors which would boost the relative score of face-to-face even higher. Due to the nature of the research design, face-to-face communication as a means for interacting with their teammates was available only to the American students. The students were from the same culture and were also from the same engineering discipline. Therefore, it would be expected that communication between collocated persons of the same discipline and the same culture would score high on dimensions of ease, understandability, and confidence.

Email ranked second in terms of all of the evaluative dimensions. By providing a written record of communication, high scores on confidence in and understandability of information were expected. It is surprising however that email scored higher than the telephone in terms of ease of use, understandability, and confidence, especially in light of past media richness studies which purport that the telephone is a "more rich" medium than email. Thus, it is important to examine why email was ranked so high in the user assessments.

The frequent use of email in the project is a reflection of the students・ communication patterns in their daily lives outside the project. Students, who have continually changing schedules, appreciate the convenience of this medium. Prior to the project, two-thirds of the students reported they used email on a daily basis. For the American students, access to email is available at a wide variety of locations and for some this also included access from home. The international students, on the other hand, indicated their access to email was restricted solely to their work environment. In interviews, students explained that email also provided an advantage over other communication media for those students whose first language was not English. These reasons explain why people who have used the telephone many more years than they have used email felt more comfortable communicating via email.

For each communication event, students were also asked to rate the progress they made toward the project and working together as a team. Their scores are shown by medium in Figure 9 above. Here again, face-to-face communication received the highest rating. The telephone, however, received the second highest rating. Therefore, although students felt more comfortable, easier, and confident with email, they accomplished more over the phone. Face-to-face and telephone communication are synchronous which leads to the conclusion use of synchronous media yields higher results. However, it is likely that use of synchronous media requires greater coordination effort such as time to set up meetings and phone calls, in addition to having to rearrange schedules and make commitments to be present at such times. This extra effort explains the relatively low frequency of use of this medium as compared to email.

The face-to-face communication medium, which proved to be the most productive medium, was not available for cross-locational communication events. These communication events were conducted primarily by email. Did this lead the students to rate the productivity of email higher for cross-locational communication, when face-to-face was not available, and did the time zone differences which makes cross locational phone calls difficult to arrange also affect the rating of progress? Progress ratings for cross locational email and phone communication are shown in Figure 10. The progress rating for email is nearly the same as that shown for all communication in Figure 9. Whereas the progress rating for phone is higher for cross-locational communication than that for within-location communication. To answer the above questions, the progress rating for email did not change when face-to-face was not available. Also, although not necessarily related to time zone differences, the ratings for progress on phone calls across locations was quite higher. This may be because the extra effort to make the calls forced participants to maximize the usefulness of the communication event. Figure 10 also shows that compared with email, the most frequently used medium, communication across locations via the telephone was more productive. Based on these comparisons of progress ratings for email and phone in cross-locational communication, the telephone plays an important role, although seldom used by these teams. Concluding this section, the results suggest that team progress toward a goal relies on some form of synchronous communication, even though users may find it greater ease, understandability, and confidence in asynchronous communication. The final results section ties these results to overall performance criteria for these global engineering teams.

The final research question explores the quality of design and teamwork as factors in the overall performance of these teams and the results are listed in Figure 11.

Figure 11 shows the results of expert evaluations of the teams・ final project designs. The evaluation was performed via a questionnaire where the teams・ performance was scored on a scale ranging from 1-5. The survey consisted of items to capture evaluations of project design quality as well as an evaluation of how well the members worked together as a team. In terms of quality of design, Teams 2, 6, and 4 ranked highest, while based on working together as a team the best teams were 3, 4, with 1 & 6 tied for third. The highest overall ranking therefore went to Team 6 which had neither the highest quality nor the highest teamwork score, but performed well in both. Figure 11 also demonstrates Team 5 as an outlier in terms of their low score on performance and it should be noted that this team also had a significantly lower frequency of communication (a total of 89 communication events--see Figure 3) compared to other groups. This may indicate that a minimum amount of required communication exists to promote team work and to produce a quality design. Although this consistently low relationship exists between the various scores for Team 5, it does not exist for the other teams.

In relation to other design teams, it is important to note that these globally distributed teams can be quite successful. The student design teams in Gemsteam were ranked number one and two in a competition consisting of Gemsteam globally distributed teams and teams whose members were all students at MSU (and met in typical face-to-face classroom setting). This evidence suggests that distributed teams can be adequately supported using communication technologies, as long as the right mix of technologies is available.

Summarizing this report, the previous results and analyses are framed in terms of the six research questions proposed at the outset of this study. First, the preferred medium of choice by these groups was email, as noted by the usage statistics for these teams. The reliance on email even during peak periods highlighted the constancy of email use especially when project deadlines neared. Second, synchronous communication media were not utilized as much as asynchronous media. This was explained by noting the time-zone differences among team members and the relative difficulty of implementing computer videoconferencing. Third, the importance of locational, cultural, and functional differences were noted and, from both qualitative and quantitative results, point to the within and across functional differences as being the most important factor in terms of amount of communication. Whether one was a mechanical or electrical engineer seemed to make the most difference in communication patterns of these groups. Fourth, it was found that while face-to-face communication had the greatest ease of use, understandability of information communicated, and confidence in the information communicated (as expected), email was surprisingly the second choice over the telephone (refuting past conceptions of media richness). Fifth, in terms of group progress, the telephone was found to be more useful than email. This suggests that, despite the troubles of coordination and implementation, synchronous media still played an important role in these groups. Finally, these globally teams performed rather well compared to face-to-face teams, scoring 1st and 3rd in a design team competition at MSU.

In conclusion, the Gemsteam project provided an in-depth examination of the ability of communication technologies to support globally distributed design teams. The project highlighted the importance of first-hand experience in globally distributed teams solving real-world problems. It also detailed the dynamic communication patterns in various computer-mediated environments. Results of this research suggest support for an appropriate mix of communication technologies including both synchronous and asynchronous media.

The findings of this research have implications for the implementation of globally distributed engineering design teams. These implications can be divided into three categories. The first category is concerned with supporting geographically distributed design teams using communication technologies. The second and third categories are concerned with the implementation and management of globally distributed teams.

The right mix of technologies includes both synchronous and asynchronous communication technologies. As was shown in this study, face-to-face communication, a synchronous medium, was rated highest on (1) ease of use, (2) understandability of information communicated, and (3) confidence in the information presented. It was also rated highest in terms of progress made while using this medium. However, face-to-face communication is not possible for distributed teams. It was believed at the outset of this study that Internet-based video conferencing might serve as second best substitute for face-to-face communication. The results show that the technical problems associated with Internet-based video conferencing prevented it from being an effective communication tool (although ongoing technology advances may change this in the near future).

The remaining communication technologies include email, phone, fax and file transfer. Email was the most frequently used communication mode and scored high on ease of use, understandability of information and confidence in information communicated. Email is an asynchronous medium and helps offset the negative effects of time zone differences. Phone, although scoring somewhat lower on dimensions of ease, understandability, and confidence in information, was rated higher than email in terms of making progress. This ranking was even higher for cross-locational communication. Also, although infrequently used, file transfer and fax rated high in terms of progress. Thus, the differences in strengths of these communication technologies demonstrate the need to provide distributed teams with an appropriate mix of technologies. That mix should include both asynchronous media and synchronous communication modalities which enhance the teams ability to communicate using words, data, written documents and drawings.

This mix of communication technologies enables teams to overcome the time zone differences they face. Time zone differences force distributed teams to rely heavily on asynchronous media. The reliance on email in this project can partially be explained by the convenience of the medium and the written record it provides. However, the teams・ reliance on the medium even at times immediately before deadlines indicates the centrality of this medium to their communication patterns. Peak times manifest themselves as crisis events or deadlines in organizational work processes and offer ways to tie this research to organizational design teams.

The relative importance of asynchronous communication for these teams also has implications for the design of new communication tools for distributed organizational work teams and stresses the importance of a reliable asynchronous medium for global cooperative work. Moreover, understanding the different notions of asynchronous communication across cultural, functional, and locational settings becomes extremely important in global teams. Perhaps instead of focusing on the development of enhanced synchronous communication tools such as video conferencing, more attention should be paid to creating more sophisticated asynchronous tools. Email that allows teams members to draw sketches on the screen or incorporate scanning capabilities into the email software might help. Also, systems that allow for quick and easy posting of documents on the web could replace email as an asynchronous communication medium. At the very least, the results of this study should affect expectations for the supportive capabilities of new synchronous communication modalities. The need to work around time zone differences will impact the use of even the most high quality video conferencing system.

The second category of implications of this research further builds upon the experiences of working with distributed teams. This category discusses what other forms of support in terms of (1) meetings, (2) supervision, and (3) expert access are required by globally distributed teams

One of the findings of this research project is the need to develop a substitute form of a formal team meeting. The time zone differences inherent to these geographically distributed teams made it practically impossible to hold a team meeting where all members were present. Michigan State students had weekly meetings with their supervisors where issues were discussed and resolved. The students in the remote locations (Germany and Singapore) could not participate in these discussions. The result was a somewhat sequential process of information sharing where outcomes of meetings were passed on to the other locations. Practically, this result can be applied to global organizations with facilities around the world, and offers potential cost-savings by handing off information to other team members around the world (one can envision an "around the clock shift"). Such a process can be applied to global concurrent engineering teams, which develop a product and share information simultaneously, by focusing on asynchronous media to support global group communication.

The supervision of globally distributed teams is also a subject for further research. The supervisory model in place on the Gemsteam project was one in which a supervisor was present at each location. The result was varying degrees of supervision were available to different team members. Students felt that differences in supervision led to different priorities and expectations among members of the team. Supervision is a problem even for collocated cross-functional teams. In this case the distance merely adds a new twist to an old problem. The possible configurations for supervision are rather obvious: (1) one centrally located supervisor or (2) several supervisors distributed at various work sites. However, with globally distributed teams, the time zone differences make this supervision increasingly difficult. A centrally located supervisor will have subordinates across many time zones. In the case of several supervisors distributed at sites, the time zone problem is minimized for communicating with their own people and communication between supervisors becomes a greater challenge. Experience from this study suggests focusing on asynchronous computer-mediated communication environments such as electronic bulletin boards, scheduling packages, and group Web pages, which allow team members to form a common group interface. Using such asynchronous tools can serve as a symbolically shared workspace and help compensate for the time zone differences between supervisor / subordinate or supervisor / supervisor communication. In this way, one supervisor can offer one-on-one or group "office hours."

The discussion of establishing supervision for the distributed teams relies on the expertise of supervisors to help students in project design. Because of the cross-functional nature of these teams, expertise may need to come from a wide-range of engineering disciplines. This model allows students to assume a more "self-directed work team" approach by focusing less on supervision and more on the use of experts on specific problems in their design project. Another possible solution would be to integrate experts (ideally from organizations who produce the engineering problem) on the student design teams. Students could query a list of functional specialists to help solve specific design issues using software programs which can help identify and connect experts. This option allows student teams gain experience with practicing engineers and allows organizations to use student teams to help solve real organizational design problems.

The final category of discussion explores the configuration of globally distributed design teams. One finding of the Gemsteam study suggests that at least two team members be located at each site. The engineers in Germany and Singapore, where they were the only member of their team, felt that the negative effects of distance (such as the lack of the possibility for immediate face-to-face feedback during the entire work day) would be reduced by having at least two team members at each site. The engineers also felt that having a second team member at their location would provide another person to interpret the meaning of issues discussed during team meetings. The need for an another interpretive voice stems from both language differences and cross-functional differences. It must be realized, however, that such imbalances often occur in real-world organizational settings; that given operational constraints, teams must foster creativity and support productivity despite imbalances in membership.

There are three main recommendations to this study to be addressed by future research: (1) the use of Internet-based videoconferencing; (2) an emphasis on control of variables; and (3) the use of cultural and team-building training to improve team performance.

The first recommendation for future research involves Internet-based video conferencing. The impact of Internet-based video conferencing and white boards on distributed teams was not assessed due to the performance of the medium. The possibility that this medium can support distributed teams in ways which are not addressed by email or phone still exists. It is suggested, therefore, that these research questions be addressed at a future time when the quality of video compression and bandwidth available via the Internet improves the overall performance of this medium. If cost constraints are lifted, the research question might also be investigated through the use of higher quality ISDN-based video conferencing.

A second recommendation for future research is to place a greater emphasis on control of variables which can confound the performance of teams and make comparisons difficult. Future studies should be planned in such a way as to minimize the differences between student groups. Also, equal levels of supervision should occur at all locations to control for variability in student performance based on mere effort alone. A careful examination of research projects should be made to ensure balance in the responsibilities of different engineering disciplines which was found to affect overall communication frequency.

Finally, the present research design was passive in its design concerning students・ use of the communication tools. Students chose the media for their own purposes. Future studies might actively use these communication technologies to increase team performance. Two directed communication activities that may improve team performance are cross-cultural communication training and team building. Both forms of training have been shown to improve team performance in face-to-face settings, but their impact in a computer mediated environment with global teams has not yet been addressed.