Task Structures

Although the input-process-output scheme provided a framework for the analysis of GDSS effects, it did not suggest the situations under which the use of GDSS can have a positive impact on group decision making. Of all the factors which have been suggested as moderating the effects of GDSS use in the input-process-output framework, the variable ‘group task’ is emerging as an important variable. Poole, Seibold, and McPhee (1985) point out that: ‘the general variable, group task, is emerging as an especially important variable, often accounting for as much as 50% of the variance in group performance’ (p.88). DeSanctis and Gallupe (1987) note the importance of tasks and task differences in GDSS research. Kraemer and Pinsonneault (1989) provide a conceptual framework for analysing the impacts of GDSS on group process and outcome. Task differences are depicted as an important class of contextual variables for GDSS use. Several authors studying group decision making (Hackman and Morris, 1975; Hiltz, Johnson, and Turoff, 1986; McGrath, 1984; Poole and Hirokawa, 1986) also point out the importance of the task in group decision research. Poole and Hirokawa (1986) declare that: ‘In the pantheon of factors determining decision behaviour and outcomes, task stands in the first position, both in terms of evidence supporting its effect strength and in terms of a theoretical linkage to decision processes’ (p.26). Task differences should be expected to affect significantly the communication process as well as the task performance (McGrath, 1984).

In spite of the importance of the task dimension and the elapsed time since its importance was identified, little effort has been devoted to a systematic analysis of the differences between tasks and how these differences affect group performance. GDSS research has generally been based on empirical evidence gathered in relation to a single task. Nor has the growing volume of research on GDSS focused on a specific class of task. Studies have been lumped together in efforts to understand process and outcome variable effects of small group decisions. Little systematic consideration has been given to the differences in tasks from one study to another (Pinsonneault and Kraemer, 1990). Although reviewers of GDSS literature (for example, Dennis and Gallupe, 1993; Benbasat and Lim, 1993) nave pointed out the importance of the moderating effect of task differences, most empirical research on GDSS has still failed

to take into account the ways in which effective work in groups is contingent upon features of the group task (McGrath and Hollingshead, 1993).

Group tasks can refer to the characteristics of the group’s substantive work (McGrath, 1984). Group tasks can be classified according to the degree of complexity (Kraemer and Pinsonneault, 1989), for example, simple versus complex tasks; the type of the task (McGrath, 1984), for example, decision task versus idea generation task; and the structure associated with the particular task (Steiner, 1972), for example, additive, disjunctive versus conjunctive tasks. GDSS researchers have studied some aspects of task differences and their effects on GDSS use. Gallupe (1985) compared the effects of GDSS support between a simple and complex task and suggested that GDSS was more suitable for a complex task. Hollingshead, McGrath and O’Conner (1993) studied the effects of GDSS among different task types and found that there were no differences in performance between computer groups and face-to-face groups for idea generation and decision making tasks, but face-to-face groups performed better on negotiation and intellectual tasks than did their computed-mediated counterparts.

These research studies focused primarily on the complexity and the type of task which groups perform. The structure of the task has received no attention. According to McGrath (1984), types of tasks can be categorised according to what the group must accomplish during the course of its meeting. Major group goals include idea­ generating tasks, creativity tasks, intellectual tasks, decision-making tasks and cognitive conflict tasks. However, within the social psychology literature, Steiner (1966, 1972) and others (Littlepage, 1991; Michaelsen, Watson and Black, 1989; Zaccaro and Lowe, 1988; Zaccaro and McCoy, 1988) have suggested that the performance and success of a decision group depend basically on the structure of the group's task. Task structure can be defined as the overall configuration of the problem space (Newell, 1980) that underlies the task. In every task performed by a group, there exists a set of collective or shared purposes which get transformed into a set of strategies for accomplishing these tasks (McGrath, et al., 1993). Task structure provides a procedural orientation or the ‘rules of the game’ for how members in the group make decisions. Task structure affects the group's need for problem analysis

and procedural orientation and planning. That is, with different task structures, the group faces different procedural requirements for accomplishing its objective(s).

Some researchers also argue that task structure is a more fundamental concept than the type of task (Steiner, 1972; Gouran and Hirokawa, 1983; Hirokawa, 1980, 1990). Even within the same task type, there may be different task structures. Within a decision task, each group can make their group decision differently. Steiner (1972) identified three types of task structures typically imposed upon decision groups: additive, disjunctive, and conjunctive.

In an additive task, each group member contributes a part to the group decision and the success of the task is determined by the aggregation of individual effort. Group performance is determined by the aggregation of individual effort. Each group member has similar responsibilities and information. According to Zaccaro and Lowe (1988), each group member must maximise his or her own individual performance in order to maximise the overall group effort in an additive task.

In a disjunctive task, a group selects one optimal solution from an array of solutions proposed by individual group members (Littlepage, 1991; Steiner, 1972). The success of the decision group depends on whether a member who has the ability to solve the problem exists and whether group members recognise and accept the superior contribution of an individual’s solution to the exclusion of all others. The success of a disjunctive task is therefore heavily influenced by the performance of the members who make the greatest contribution.

In a conjunctive task, however, the successful decision can only be achieved when all the group members maximise their efforts because all group members have unique information. Conjunctive tasks differ from additive tasks because each group member has different information. Conjunctive tasks also differ from disjunctive tasks because no one group member has enough information to suggest the correct answer or optimal solution. A group achieves a successful outcome only when all of the information held by individual group members is accurately communicated to other

group members. The whole group will fail even if only one member fails to contribute to the decision task. The success of the decision task is heavily influenced by the performance of the member who makes the least contribution.

According to Steiner (1972), even within a decision task, different task structures can exist and these task structures within a decision task will alter how a group approaches the task and eventually affect the behaviour of the group members and the performance of the decision groups. Although research on GDSS is starting to provide some empirical evidence for how decision performance across different group tasks are affected by GDSS support, the primary focus has been on the type of task (idea generating task versus decision task). Although the main objective of GDSS technology is to provide support for the group decision process, we still lack empirical studies of how GDSS technology may affect group decision performance under different task structures. This thesis is an attempt to bridge this gap in current GDSS literature.