Modern societies rely heavily on groups to make important economic and political decisions. However, a review of research on group processes shows that progress has been slow in the delineation of the conditions that promote or impede efficient, accurate group judgments. One reason for the slow progress is that research methods and data analysis in this area are varied, difficult to compare, and often substandard. In this review, the authors summarize alternate methods of analysis and provide detailed illustrations of the best methods for assessing and analyzing group judgment accuracy Increased accuracy is a common justification for using groups, rather than individuals, to make judgments. However, the empirical literature shows that groups excel as judges only under limited conditions. Hill's (1982) review found that groups tend to perform around the level of the second best member in most tasks, including group judgment. Hastie (1986) identified several task differences that moderate the relative accuracy of group and individual judges, but he also concluded that there were few, if any, task conditions under which groups consistently

Agreement is an important social outcome often poorly handled by computer-mediated groups, presumably because the computer cannot transmit the necessary rich information. A recently proposed cognitive model suggests richness is not the key to social agreement and that group agreement can be generated by the exchange of anonymous, lean text information across a computer network. This experiment investigates this theory. Self-chosen groups of 5 completed three answer rounds on limited choice problems while exchanging a few characters of position information. These asynchronous, anonymous computer-mediated groups generated agreement without any rich information exchange. The key software design criteria for enacting agreement is proposed to be not richness but dynamic many-to-many linkage. The resulting “electronic voting ” may be as different from traditional voting as e-mail is from traditional mail. It may also imply a new generation of groupware that recognizes social influence. To be effective, it is not enough for groups to simply produce decision task solutions. They also need agreement about, and confidence in, those solutions to implement them (Boje & Murnighan,

Kerr, MacCoun, & Kramer, 1996). Teamwork is not guaranteed to provide either of these advantages. With respect to (1) combining complementary inputs, increasing the size of an organization tends to reduce its overall efficiency unless there is also an increase in departmentalization and standardization of tasks (Blau, 1970). The latter features, however, reduce flexibility of response in a changing or novel environment (Donaldson, 1995). A related problem is goal displacement, in which specialized units lose sight of the larger organizational purpose, and pursue their own goals as if they were fixed ends rather than means, which should be reevaluated when conditions change (Scott, 1998). With respect to (2) better decisions, groups may be affected by socialization biases, such as groupthink, which induce conformity rather than diversity of thought (Janus, 1972; March, 1996.). For this reason, group decisions tend to be better when individuals think about the problem independently befo

Abstract not found