Nicole Clark

The differential use and effect of knowledgebased system explanations in novice and expert judgment decisions

Explanation facilities are considered essential in facilitating user interaction with knowledge-based systems (KBS). Research on explanation provision and the impact on KBS users has shown that the domain expertise affects the type of explanations selected by the user and the basis for seeking such explanations. The prior literature has been limited, however, by the use of simulated KBS that generally provide only feedback explanations (i.e., ex post to the recommendation of the KBS being presented to the user). The purpose of this study is to examine the way users with varying levels of expertise use alternative types of KBS explanations and the impact of that use on decision making. A total of 64 partner/ manager-level and 82 senior/staff-level insolvency professionals participated in an experiment involving the use of a fully functioning KBS to complete a complex judgment task. In addition to feedback explanations, the KBS also provided feedforward explanations (i.e., general explanations during user input about the relationships between information cues in the KBS) and included definition type explanations (i.e., declarative-level knowledge). The results show that users were more likely to adhere to recommendations of the KBS when an explanation facility was available. Choice patterns in using explanations indicated that novices used feedforward explanations more than experts did, while experts were more likely than novices to use feedback explanations. Novices also used more declarative knowledge and initial problem solving type explanations, while experts used more procedural knowledge explanations. Finally, use of feedback explanations led to greater adherence to the KBS recommendation by experts-a condition that was even more prevalent as the use of feedback explanations increased. The results have several implications for the design and use of KBS in a professional decision-making environment.

Explanation Provision and Use in an Intelligent Decision Aid

Intelligent decision aids have been widely adopted by organizations in an effort to capture, retain and disseminate the knowledge of individuals within the organization. To date, these efforts have met with mixed results. Two primary limitations have been the usability of systems (users either cannot or do not wish to use systems) and the inability to transfer knowledge from such systems to less expert decision makers. Explanation capabilities have been perceived as a potential means for improving the effectiveness of such systems in terms of both usability and knowledge transfer. In this study, four advancements are put forth. First, a working prototype system for highly complex decision making is used as the foundation for implementing a full-scale, embedded explanation facility. Second, the development of the explanation facility is carefully grounded in the extant prescriptive literature on explanation types and delivery processes—including feedforward and feedback. Third, a development strategy is formulated and demonstrated as a generalized process of explanation development and integration. Fourth, high-level professional decision makers use the system and provide evaluative feedback on the usability and value of the explanation facility. Overall, the results of the development efforts and subsequent evaluation are very positive and provide a solid foundation for future research and practice efforts related to explanation provision within the context of intelligent systems. Copyright

A generalized model of decision‐making processes for companies in financial distress

Decisions by insolvency experts are vital to the future of a company in financial difficulties and impact the financial returns to creditors and other stakeholders. This paper describes a generalized model of the decision-making processes used by insolvency experts when dealing with companies in financial distress. The model includes identification of high-level factors that are important in making insolvency decisions and how these factors are used in the decision-making processes of insolvency experts. In particular, financial and non-financial factors are required for judgments about relevant stakeholders and the future of a company. Validation experiments show a high degree of conformance between the model and a variety of actual cases.

Unveiling the black box of human-computer interaction: The need for replay process tracing in decision aid research

The extant accounting behavioral research related to decision aid impacts on users decision-making has been limited over time due to the focus on decision outcomes to the neglect of the underlying decision processes and strategies. A primary barrier in moving the decision aid literature forward into the exploration of the black boxof human-computer interaction is the paucity of methods for accurately capturing the interaction process. This paper presents a conceptualization of a technique we refer to as replay process tracing that allows the researcher to recreate the precise session that a user has participated in with a decision aid. We subsequently operationalize this conceptualization through a working prototype model with advanced replay process tracing functionality.