Approaches to user modeling

The three existing approaches to user modeling-GOMS (goals, operators, methods, and selection rules), production simulation, and mental models-are examined. Each approach is described, relevant experimental research is reviewed, and each approach is summarized in terms of its advantages, limitations, and applicability to the system design process. It is suggested that these approaches are inadequate to account fully for the interplay between human information processing, user characteristics, computer systems, and the demands of the various tasks. An alternative approach to user modeling that utilizes structural covariance analysis is presented. A theoretical causal model of the human-computer interaction, which incorporates the user, system, and task characteristics, is described and discussed in terms of applying structural analysis to the theorized pattern of causation among the user, system, and task. How this approach can provide useful information for guiding the design process is discussed