Communications of the ACM

We are more likely to lose when using email to make initial contacts, develop relational understanding, achieve consensus, or influence correspondents while working under deadline or other external pressures.

This point is underscored by the results of an exploratory study I conducted of email use by members of software development project teams [11]. Members of such teams frequently use email to augment FTF meetings, though it seems unlikely that they view email as having equivalent value for each particular meeting. This notion raises the questions: When is email effective? and When is it ineffective? To answer, I studied teams that had just completed a 14-week software development project as part of an undergraduate software engineering course at a Midwestern U.S. university. The project deadlines and technical challenges put them under substantial external pressure and required a high level of interactivity. We organized 13 four-member teams across course sections, so members were encouraged to communicate via email rather than seek mutually acceptable times for FTF meetings, though they were free to choose how they communicated. In open-ended responses, individual participants identified a total of 32 separate activities for which they felt FTF or email communication with other team members had been particularly effective or ineffective. The compiled activity list was returned to all participants as a survey instrument (n = 52) with instructions to rate the effectiveness of email and FTF communication in supporting team projects for each activity. Subsequently, I statistically grouped the activity ratings into six categories (see Figure 1).

Participants rated FTF communication higher than email for all activity categories except speed of communication. Striking differences emerged in how participants viewed certain categories for each medium. Predicting some of these differences would have been difficult using existing communication metaphors; for example, execution tasks received one of the highest ratings for FTF communication yet was by far the lowest-rated activity in email. As is often the case, results from the exploratory study raised more questions than they answered, prompting follow-up studies in the areas of task, socialization, and interpersonal influence.

• Generation. Generating ideas, brainstorming, planning, scheduling;
• Choice. Choosing among sets of solutions and by consensus;
• Negotiations. Resolving conflicts of viewpoint and motivation; and
• Execution. Performing to an objective standard and competitive activities

The revised instrument was expanded from the exploratory study to include the negotiation task type and generalized to be applicable to a wider range of contexts than the original software development. As in the exploratory study, participants were surveyed following a 14-week treatment period, but this time I measured three distinct groups. The first two were surveyed about email use in software development projects in which undergraduate teams developed 3GL applications in the first group (n = 57) and database applications in the second group (n = 36). As in the exploratory study, the projects were designed to put substantial time and intellectual pressure on team members and require a high level of interactivity. Members of the third group were not involved in project teamwork. I surveyed these participants about their overall email usage during the previous 14 weeks, when they received and sent email for both an undergraduate technology course in which they were enrolled and their own personal purposes. This treatment was designed to limit pressure on participants and require only a relatively low level of interaction with their peers (n = 62).

The research purpose of the three-group design was to test for the differences in ratings of media effectiveness by task type between low- and high-interaction groups (discriminant validity) and similarities between the two high-interaction groups (convergent validity) as a means of establishing the instrument's construct validity. However, the results also provide additional insights concerning the role the communication context plays in assessing a medium's effectiveness (see Figure 2).

A major difference emerged between generation tasks—where ratings are high and strikingly consistent—and other task types. Generation tasks, such as brainstorming, have been used in the majority of empirical group technology studies completed to date; these studies tend to drive our conceptualization and understanding of the fit between tasks and technology. The results raise troubling questions about the generalizability of the generation-task literature to other task settings. Second, both of the high-interaction groups rated email effectiveness lower for choice, negotiation, and execution tasks than the low-interaction group rated it. This finding suggests that people exposed primarily to low-interaction email use may overestimate email's utility in high-interaction situations. Third, the high-interaction groups disliked email for choice tasks in particular, possibly because many of the problems in their self-managed teams required consensus [2]. It is unlikely that even experienced low-interaction email users would have expected and prepared for this specific contingency of high-interaction use.

From a task perspective, it appears we are more likely to win when using email for generation tasks (such as proposing and sharing ideas, planning, and arranging events) and for low-interaction contexts (such as dyadic communication characterized by flexible deadlines). We are more likely to lose when using email in high-interaction contexts emphasizing choice, negotiation, and execution tasks. In practice, losses associated with high-interaction contexts can be particularly insidious, as a project may go fairly well in its early stages when generation tasks predominate, only to falter during the latter stages when communication needs increase for choice, negotiation, and execution.

• Social initiation. Making initial contacts and developing relationships;
• Social influence. Achieving influence goals and implementing influence strategies; and
• Social intimacy.Maintaining existing relationships by sharing humor and emotional information.

Several interesting findings emerged when I compared how the low-interaction and high-interaction groups rated these socialization factors (see Figure 3). First, email support for social intimacy was rated significantly higher than either social initiation or social influence, corroborating earlier studies that found email to be relatively effective for maintaining ongoing relationships (such as those in business settings) [2]. Second, ratings of email support for social initiation were much higher in the low-interaction group than in the high-interaction groups. High-interaction participants found it frustrating to use email for developing relationships. This finding may be an important qualifier in research exploring development of close relationships via electronic media [4], as it suggests that achieving similar outcomes is more difficult in high-interaction settings. Third, participants overall found email relatively ineffective for achieving goals and implementing strategies of social influence.

When socializing, it appears we are more likely to win when using email to maintain ongoing relationships, perhaps through such simple expedients as passing along Internet humor and initiating relationships in low-pressure contexts (such as seeking out and contacting others who clearly share our personal interests). We are more likely to lose when using email to make initial contacts, develop relational understanding, achieve consensus, or influence correspondents while working under deadline or other external pressures.

• More work. Initially, at least, email requires more work than other interpersonal communication media to accomplish similar results [9, 11]. It's more difficult to get to know people via email [10] and much more difficult for group members to choose or negotiate via email [9];
• Apprehension. Some people are surprisingly apprehensive about using email for persuasion and negotiation, though it's not clear how to identify these people in advance [12]; and
• No guidelines. Intuition and experience using other forms of communication don't appear to offer good guidelines for using email; people often fail to use email effectively [8, 12] and may even dumb down their choice of influence strategies [7].

None of these points implies that email can't or shouldn't be used to support e-commerce. They simply caution that specialized planning and training are needed to win with email in e-commerce settings. Indeed, e-commerce technology has bypassed the pitfalls in some of the situations mentioned here; for example, the structural design of some online auctions largely obviates the need for either the buyer or the seller to choose among influence strategies.

Project Teams in Distance Education.Distance education has evolved from the instructor-centered lecture format, where dyadic student-instructor communication depends on a dedicated communication medium, to encompass a variety of other formats. There is no technical reason why distance education courses cannot include team projects among geographically distributed students. Indeed, sophisticated asynchronous learning networks provide significant feature enhancements over email and promise increased support for student group work [11]. However, administrators may be tempted to use email to support distributed teams due to its low cost and ubiquity. The study findings suggest using email to support student projects can to lead to problems—especially in projects involving high external pressures and interaction requirements:

• Generating ideas. Email is easier to use for generating ideas than for other tasks, putting at risk teams that procrastinate based on a false sense that using email for choice, negotiation, and execution tasks are equally trouble-free [9, 11];
• Disassociating from the process. Teams frequently include members whose apprehension about email results in disassociation from the process, potentially initiating a downward spiral in the way they are viewed and treated by other team members [8, 12];
• Initiating projects. In the same way it's difficult to use email for developing relationships and influencing others in high-interaction settings, it may be more difficult for distributed teams to use email to get projects off the ground, develop internal leadership, and stay motivated during a project [7, 10]; and
• Inability to mitigate. Distributed student teams typically lack access to the mitigation techniques used in business settings (such as securing funding for ad hoc FTF meetings, controlling participants' schedules to manage synchronous activities, and developing relationships that last) [11].

Simply substituting email for FTF meetings has the potential for changing student project teams from winners to losers. Yet this possibility shouldn't automatically prevent email and other computer-based systems from playing a role in distance education. Just learning to use email in team projects has pedagogical value. Email is widely used by business work teams, and experience in educational settings should be expected to promote career preparation [10]. Moreover, some of these issues may be addressed by designing systems that add team support and by designing curricula that structure team activities to a greater degree than might be necessary or even desirable in traditional FTF courses [11]. For distance education, the key danger is the assumption that email can replace other forms of communication without regard to the substantial effects such substitution can have on student motivation and productivity.