An investigation into the feasibilty of providing intelligent support for computer mediation decision making groups
This thesis investigated the claim that the adaptation to the keyboard interface of a computer-mediated (CM) decision making group leads to differences in the style of communication when compared to that of a face-to-face (FTF) group. More importantly it examined the possibility that changes in satisfaction with the process and the decision outcome are determined not by the mode of communication, but rather the style of communication the decision makers employed in response to the keyboard interface. The decision processes of CM and FTF groups were examined using a simulated panel of enquiry presented via computer databases and containing inconsistent and incomplete shared information that could only be resolved through collaboration between the group members. An analysis of the communication styles employed in real-time CM and FTF groups (Experiment 1) revealed a tendency of CM discussions to exhibit a preference for a normative style of communication exchanging a proportionally high number of value statements and indications of preference, and for. FTF groups to rely proportionally more heavily upon factual and inferential statements. A paradigm for enabling intervention into the decision making process through the monitoring and coding of all group communication was developed (Chapter 2) which permitted the real-time analysis of the differences in communication style and aimed to reduce the differences in communication style. Using this paradigm and the norms for communication of the two forms of group (CM and FTF) established in Experiment 1, a series of studies examining the communication process were undertaken. Experiment 2 explored the possibility of intervening into the communication process using e-mail based support messages that conveyed the discrepancies between a CM groups communication style and the style a group might be expected to employ where it communicating FTF. Two configurations of support messages that each attempted to shape the communication style of CM decision panels to resemble those of FTF panels were considered. It was found that alerting users to their communication style and instructing them to increase or decrease certain styles of communication enabled them to more closely resemble the communication process and satisfaction levels of FTF groups. Experiment 3 considered the possibility that the presence of a monitoring system, rather than the content of the support messages provided, was the key issue in securing changes in the communication style of CM groups. Having established that it was indeed the content of the support messages that enabled CM groups to operate as if communicating FTF, attention turned to effects of the support. By easing the interpretation of the feedback through two configurations of visual feedback, Experiment 4 attempted to increase decision makers adherence to the content of the support messages. This study suggested that visual feedback alone was not sufficient to elicit the desired changes in communication style and that the text-based communication was required. Moreover, Experiment 4 considered the impact of support messages themselves, considering whether the support acted as continual assistance to the users or whether it merely trained the users to communicate in the desired way Conclusions from this study were slightly inconclusive, however, given that changes in communication styles had been achieved a further analysis of the content of the messages was undertaken. This final analysis (Chapter 7) revealed effects of confirmation bias within the communication and intervention steps that can on occasionally overcame such biases. The possibilities for the development of real-time intervention into these processes are considered and the findings interpreted in the light of existing theories of CM communication and recent developments in computer-based communication.