A simulator to test four models of advising systems

CARMART is a simulation of a new-car sales environment, used to teach about to research computer-based advising. CARMART illustrates four increasingly powerful advising situations:Data query and report facilities (MIS) Consultant advice on past history (CS) Logical models (ES—Expert system) Mathematical models (DSS) This scale of specificity is based on the degree of assist from the computer and the amount of pre-existing knowledge necessary to construct valid models. MIS is based upon data models and relies heavily upon either analyst or user knowledge to assist decision-making. CS requires similarity models and rely upon ad-hoc analyses of past history to provide advice. Logical modeling is based upon implicational or associational relationships between situation descriptions and outcomes or payoffs. Finally DSS and its relatives depend upon causal models which can compute payoffs for specific situations and therefore require high-quality theories (but less user knowledge). CARMART provides a vehicle for teaching students about these four modes of advising given a specific 19-factor description requiring a decision from the student concerning sales technique. In addition, CARMART is helping to test hypotheses about the applicability of each mode of advising to situations which differ in the ease of access to systematic knowledge.